to: Richard Adams, Martin Wiseman, Chris Warner, Peter Milne, Simon Dawes, Trevor Sexton, Paul Austin, John Harrison, Roland Meins, Andrew Dean, Bob Taylor
from: A.W.Sloman
subject: Progress to 3rd January 1991
I got back from Japan and Australia intact and on schedule; this week has been largely devoted to writing up my visit report and filling in my expenses claim. Trevor Sexton and Peter Milne have distracted me from this with long discussions of what Closed Loop is actually doing, as opposed to what it ought to be doing. I've been telling them loads of things about Closed Loop that I ought to have told them earlier (if they turn out to be true) and they have been destroying my hypotheses about the "monkey nuts" problem - I've yet to produce a convincing explanation of why m by n has to be less than 256.
I should be able to get back to the Trigger board next week, which ought to make Paul Austin a lot happier, since he has been stuck with debugging the Processing outputs since I left.
He has done it well enough to get most of the functions of the Waveform Processor - Digital Mark 2 up and running; he is now trying to get the Recirculate Latch to work, which is where he left WPD Mark l.
Martin Wiseman has got both the Waveform Processor -Analogue Mark 2s working; with improvements. The Analog Devices AD-96l7 seems to be inferior to the Comlinear
CLC-400 in most of the places we want to use it, and we are probably going to end up using both.
We have lost two CLC-22l Filter Grid drivers to scintillator flash-over - at £l20 quid each - and it is to be hoped Trevor can prevent this happening again by finishing off the grounding.
Andy claims to have had a productive week at home – the Christmas decorations are down, and he has reduced his sleep debt.
Chris Warner has had a fun time with the column — he has now established that our tetrode gun works best if you run it as a triode (another Lintech feature turns out to be a bug) and he has worked out how to model the fringing fields from the blanking plates - the effective width (and transit time) of the present plates is about double the physical width.
(Note added 2013-03-04. It later turned out that Chris Warner hadn't been running the gun as a tetrode in the way he thought he was and it did work better as a tetrode. Score one for Lintech.)
to: Richard Adams, Martin Wiseman, Chris Warner, Peter Milne. Simon Dawes, Trevor Sexton, Paul Austin, John Harrison, Roland Meins, Andrew Dean, Bob Tavlor
from: A.W.Sloman
subject: Progress to 11th January 1991
We still haven't come up with a convincing explanation of the "monkey nuts" problem. Peter Milne and Andrew Dean have been working on the Delay Lists, and they have now worked out that the Delay List needs to start with two No-Ops of 1.5usec and 140nsecrespectively, and that the leading and trailing dummies have to be longer than the unblank period.
With these changes the sampling system works reliably up to the "monkey nuts" limit for simple Delay Lists; with multiple sub-lists there is a "monkeys times sub-lists" limit above which the Output Store only contains zeros, and ,another limit on "nuts" above which the Output Store freezes.
It looks as if something is going wrong on the Waveform Processor - Digital Mark l when it is left to Accumulate for too long - Andrew Dean suggests that it might be a logic output which has been tri-stated in error, and drifts off - but whatever it is will be a sod to find. If we are lucky the WPD Mark 2 won't have the same error...
Paul has got the Recirculate function working on the WPD Mark 2; it isn't working perfectly because the Latch and Encode outputs from the Mark 2 Trigger Board are not quite right.
I’m now working my way through the Processing Pulse generator on the Mark 2 Trigger board, working out what it does, and what it ought to do. Paul has established that the design has to be changed, and I'm spending a certain amount of time checking out the original specification produced by TDS.
Martin Wiseman has spent a couple of days working through the Blanking Board specifications parts A and B with the circuit diagram, and he is now working on the power supply for the Acceptance Test Box.
He has replaced the CLC-221's that blew up, and ordered six more - two to get the second board working, and four more as spares.
Stephen Fisher finished off the linking of the grounds on PPl on Monday - we haven't lost any more CLC-22l’s since then, but I'm not sure that any have been hooked up to the column since then.
Chris Warner is continuing to measure gun brightness, but he hasn't replicated last week's measurements. Of course the conclusion drawn last week's measurements was wrong because
of a misunderstanding about the power supply for the tetrode grid - Chris was getting highest brightness with the maximum positive bias on the tetrode grid, not zero as it had
appeared.
This week, of course, highest brightness usually appeared with zero volts on the tetrode grid, so he has replicated the result reported last week.
Roland Meins has got his Sparc station; on Monday we had to borrow a copy of version 4.03 of the operating system on tape from some friendly academics so that Roland could start installing his software. He has now got to the point where he has got the CAD software installed after a fashion, but it doesn't yet work.
Bob Taylor and I recently (and independently) came across a professional lead forming and trimming machine suitable for bending and cropping the leads on our ECL flat packs; it is expensive - about 7,000 pounds - but Richard is trying to get it bought on the capital budget.
to: Richard Adams, Martin Wiseman, Chris Warner, Peter Milne. Simon Dawes, Trevor Sexton, Paul Austin, John Harrison. Roland Meins, Andrew Dean, Bob Taylor
from: A.W.Sloman
subject: Progress to l8th January l99l
I am still working my way through the Processing Pulse generator on the Mark 2 Trigger board; it now seems clear that I gave Bob Ward a duff specification for the Processing Pulse outputs required from the Mark 2 Trigger Board when producing Multiple Encodes per Sample. I think TDS share some of the blame for this, but it is embarrassing.
Paul Austin has told me what he thinks the Mark 2 Waveform Processor actually needs, and I have checked his specification rather carefully, and rewritten the relevant bits of the Trigger Specification Part A.
I've roughly worked out the changes I'll have to make to the Trigger Board to meet this specification ~ we should end up with much the same number of packages, but with two fewer bistables and two more delay lines - and I am redrawing the affected parts circuit diagram on Metheus.
Once I've got a detailed design I'll check the tolerances again, but it looks as if I can guarantee 25MHz sampling with worst case components.
Peter Milne and Andrew Dean have had another exciting week on the Delay Board; honours are about even - Andrew has found and fixed two bugs in the hardware, while Peter has found and fixed a couple of bugs in the software. One of the hardware bugs caused the Delay Board to miss "Stop" pulses from the Waveform Processor, and one of the software bugs had much the same effect, but it seems there is at least one more bug still left in that area.
The "monkey nuts“ barrier seems to have moved up to 2048, but we still don't have clue about what is going on.
On Wednesday afternoon Bob Taylor and I had a visit from the guy who sells the lead forming and trimming machine; he brought along the "universal" tool that bends and crops all sizes of package, but only one side at a time — the ECL flat-pack we bent and cropped would almost have sat squarely on a board, but one corner was up a bit.
Apart from that the tool looked very nice. We want to buy the specialised tool that bends and crops all four sides of an ECL flat-pack in one operation, and guarantees coplanarity to +/-2 thou. Not only is it faster than the universal tool, but it is also cheaper.
Paul Austin hasn't been doing much on the Waveform Processor - Digital Mark 2 this week; he got Burst Mode going last Friday, and on Monday he was able to confirm that the “pixel dropouts" have vanished on frames 3 to 64; the first frame is still blank, probably because the relevant address offset isn't right, and the second frame seems to show five or ten random drop-outs, but the rest are clean.
With this behind him he went on to tackle the latest version of the Timebase Interface (the one with the interrupts enabled); it is pretty much working, although he still has to cut and link the board to convert the Interrupt Status register into an Interrupt Address register - the mod has been worked out, but not yet implemented.
His Sue has fallen a little behind her project plan - their child was due yesterday. Management are not yet talking about cancelling the project.
Martin Wiseman has spent most of this week sorting out the manufacture of the Acceptance Test board; we should get the bare boards in a couple of weeks, and Avantel expect be able to get at least one of them back to us, loaded, within a week of us giving them the bare boards.
Chris Warner has been having fun with scan coils; Nick Campbell has worked out an even better scheme for making prototype scan coils (for the S.360 FE) than the one Chris poached from him last October, and Chris has now got Tony Slater drawing up jigs in the latest style for Specials to make. With a fair bit of luck, Barry could be in a position to wind and assemble a set of the new scan coils by the end of next week.
Chris has been mainly occupied writing up the last fortnight’s work on gun brightness. He is resisting my suggestion that he submit it for publication in the Journal of Irreproducible Results, but he is concious that gun brightness is rather sensitive to gun alignment, and both he and Trevor have been finding the Gun Tilt and Align controls rather coarse recently; Trevor is sure that they used to be okay (although they only drive 8-bit DACs) and suspects that the new transformation matrix (which converts Tilt and Align inputs into upper and lower coil drives) might be incrementing one of the DACs by several bits at a time.
Chris's wife, Julie, seems to have caught a mild dose of the Leica~Cambridge disease — their new child is now eighteen days over-due…
to: Richard Adams, Martin Wiseman, Chris Warner, Peter Milne, Simon Dawes, Trevor Sexton, Paul Austin, John Harrison, Roland Meins, Andrew Dean, Bob Taylor, Tony Slater
from: A.W.Sloman
subject: Progress to 25th January 1991
I've worked out what I want to do to the Processing Pulse Generator for the Mark 3 Trigger Board, and I had got about half-way through up-dating the circuit diagrams last Friday afternoon.
This was pushed aside by the "monkey nuts" problem; Peter Milne picked up the fatal flaw in my system design on Friday evening, as an avoidable defect in the way the Subtract pass works, and while walking to work on Monday, I realised that the same flaw in my logic would mess up Update l passes for negative Feedback shifts.
The problem is that when you down-shift a negative 2s-complement number, you have to back-fill the high bits with "l"s, and we back-fill with "0"s for both positive and negative numbers.
We can partially fix Update l by a cut-and-link mod to the Mark 2 Waveform Processor - Digital, which will allow Update l passes to give the right results for Feedback shifts of -l, -2. -3 and -4 (at which point we run out of convenient unused OR/NOR gates).
There is a pretty straightforward design mod which we could implement on the Mark 3 version of the WPD, which would solve the problem for all negative Feedback shifts (from -l
to -15) but it means squeezing two extra l00l58 shifter packages onto an already crowded board.
What I propose to do is to restrict the Feedback shift range to +8 to -7, rather than +8 to -15. This reduces the number of l00l58’s in the Feedback shifter from 6 to 4; I only need add one 100158 to inject the high order sign bits, so we end up saving a package.
By reducing the negative Feedback shift range from -l5 to -7 we reduce the period over which the Sampling Crate can do autonomous noise reduction by a factor of 256; if we use an m.n greater than 256 the Syntel has to take over the job of down-shifting the accumulated deviations and up-dating the Expected Waveform Store.
The Syntel will do the arithmetic more slowly than the Waveform Processor, but to 32-bits rather than 16-bits: since it won't have to update the store more than once every few seconds the slowness is immaterial, and the extra precision is handy. Peter Milne says the program only involves a few lines of code.
Andrew Dean and Peter Milne have now installed the Mark 2 Trigger Board and Waveform Processor - Digital in their Sampling Crate; the system didn't work first time round, because of a dry joint in one of Paul Austin's mods to the Trigger board, nor second time round either, when my soldering proved even worse than Paul's, but with more careful soldering the system now seems to be working rather well; Andrew's TaxiChip transmitter circuit on the Mark 2 Trigger board is a lot more reliable than AMD's original circuit on the Mark l.
This has allowed Andrew to progress to a higher class of bug-hunting on the Delay board - he has fixed a flaw in the Coarse Delay Counter circuit, where Yohannes drove a CEP input as a count disable, where he should have driven all the S0 and S1 inputs in parallel. Andrew is now wrestling with the Load inputs on the two 10G061 synchronous counters that do the Fine Delay, amongst other things.
The next level of bug-hunting is going to be to look at the Time-to-Digital and Digital-to-Time converters on the Delay board, and check them out for jitter, stability and linearity.
Paul Austin has been somewhat preoccupied this week — his daughter Florence was born at 2.00pm on Sunday (only three days behind schedule) and he has been on leave since Wednesday lunchtime, and won't be back until next Wednesday morning.
Martin Wiseman has spent most of this week on the Acceptance Test board; his circuit diagram now shows the pin numbers and decoupling capacitor numbers assigned by PCA when they laid out the board. He also delivered a couple of Blanking Driver kits to Avantel for assembly, and has spent an hour or so discussing blanking drive levels with me and Chris Warner.
The up-shot of our discussion was that we can't get the plates close enough together to work with a GaAs driver, but a 200 micron spacing should let us get away with a +/-3.5V swing which we should be able to get out of a Step Recovery Diode, and a moderately vicious drive circuit (another pair of HP HXTR-5l04 microwave transistor at forty quid each, for starters).
Chris Warner has been distracted by the appearance of Nicholas Christopher James, shortly before midnight on Saturday, only 19 days late, a performance the EBT might envy. Of course, one would imagine that when Dave Hall estimated the time scales for his wife's pregnancies he looked up the data for human beings, decided that nine months was too long, and rewrote the time scale on the basis that he had married a gerbil - meanwhile, we would seem to be in the process of finding out whether the EBT is a white elephant or a blue whale.
Tony Slater has just about finished drawing the first jig for the new style scan coils — the original refinements have been further refined. The drawings for the other three jigs can be generated on the CAD system by changing the critical dimensions, and Tony expects to finish them in a couple of
days.
to: Richard Adams, Martin Wiseman, Chris Warner, Peter Milne, Simon Dawes, Trevor Sexton, Paul Austin,John Harrison, Roland Meins, Andrew Dean, Bob Taylor,Tony Slater, Ian Murray
from: A.W.Sloman
subject: Progress to lst February 1991
I've now documented a pair of solutions to the "monkey nuts" problem — both the cut and link mod to the Waveform Processor - Digital Mark 2, and the design change for the WPD3. The circuit changes for the WPD3 have been entered onto Metheus, and Paul Austin has a copy to check, and both changes have been written up, with a discussion of the software implications, in a document that I've mailed to Paul Austin and Peter Milne.
This has left me free to get back to the Processing Pulse Generator for the Mark 3 Trigger board; I've finished transferring my pencil sketches onto the circuit diagram, and I've already noticed a couple of minor defects - any one of them would stop the circuit working, but they are minor in the sense that it looks as if they will be easy to set right.
Andrew Dean has spent an exciting week bug-hunting on the Delay board. Last week's apparent victory over the Coarse Delay Counter proved illusory - with the ECL PALs we've got, the Coarse Delay counters can't be made to work properly at 100MHz. Besides this, with the design we have got we can't guarantee that the Fine Delay counters will work reliably with worst case components at 800MHz, so Andrew is planning to cut the Gordian knot by clocking the Delay boards at 400MHz — we can pull this off the Trigger board fairly easily.
This would theoretically degrade our minimum delay resolution from 5psec to l0psec, but this is several orders of magnitude better than the resolution we can actually see, or are likely to be able to see for some time. In principle we could redesign the Fine Delay area so that it would run at 800MHz, but this would involve ripping up large areas of the current design and finding room for more GaAs latches, neither of which is at all attractive.
Meanwhile, Peter Milne has been using the Sampling Crate quite happily; the Delay board is producing the wrong delays consistently and reliably, and he and Andy have managed to
get the system working with single sample sub-lists (as required for Trigger frequencies above about 12MHz).
Paul Austin is now working up Burst Mode operation with the Mark 2 boards in the Sampling Crate. So far he has got the pixel clock hand-shaking, but not yet the line or frame clocks.
Against the remote chance he ever gets a spare moment, I've asked him to look at the possiblity of replacing our 7 state machine 22Vl0 PALs with a single Altera device, and the 4 Taxichip link PALs with another - with luck, this should leave us enough spare terms to put some test states into the state machine, and enough spare space on the board to track in a bus for diagnostic data.
Martin Wiseman has spent the week multiplexing between the Acceptance Test box power supplies, the driver for the Pythagoras chip, and a step recovery diode drive for the flash-over blanking.
It may be that we can avoid the step recovery diode by using a relatively new Gigabit Logic part - the 16G06l dual pin driver ~ which can provide enough voltage swing for a balanced drive to plates spaced 0.2mm apart, although the speed may be marginal.
Martin's analysis of the step recovery diode drive seems to indicate that we can get reasonably close control of the storage time if we take reasonable precautions, like putting a crystal oven around the diode and stabilising the forward current through it. This is better than I had expected - and the step recovery diode should be quite fast.
Chris Warner has spent most of this week on the tetrode gun. but he has done some work on the blanking plates, and expects to do more next week, before the meeting on Wednesday.
Tony Slater has finished the drawings for the new scan coil jigs, and production are now making the jigs - we might get some new scan coils in a week or so. Tony has now moved onto the new blanking head, and has acquired an expression of more or less permanent incredulity.
Ian Murray started with us today. Peter Milne tells me that Ian has about as much experience with X-Windows as it is possible for anyone to acquire (unless there really is a time machine - and if there were someone, somewhere, would have met their timescale, and the world would have beaten a six lane highway to their door).
to: Richard Adams, Martin Wiseman, Chris Warner, Peter Milne, Simon Dawes, Trevor Sexton, Paul Austin, John Harrison, Roland Meins, Andrew Dean, Bob Taylor, Tony Slater, Ian Murray
from: A.W.Sloman
subject: Progress to 8th February 1991
I've spent this week on the Processing Pulse Generator - once I'd got the circuit diagram into reasonable shape, I used Bob Ward's simulation scripts to test it under Hilo on Metheus, and the simulation is now running quite nicely.
I had to spend quite a time working on the model and the driving programs to get it to run at all, but once it was running I only had to make one change to the basic circuit to get it right. Changing the actual circuit diagram so that it matches the working model isn't going to take too long, but there is a certain amount of tolerancing left to do. which may take up a day or two next week, not to mention up-dating the Spec Part B.
I think I'll have to put off up-dating the Spec Part B until I work out a cut-and-link mod for the Processing Pulse Generator on the Mark 2 Trigger Boards - we need it before we can finish testing Recirculate, and start making proper use of sampling pulses wider than l5nsec.
Andrew Dean has been bashing away on Metheus l, up-dating the Delay Board circuit diagrams, and working with Peter Milne on the Sampling Crate; they have finally tracked down the dreaded “WP busy" fault — it seems that Reset passes generate a Sampling Loop Completed interrupt (they shouldn't, but there isn't enough PAL space left to stop them) and the software was recognising that as the real Sampling Loop Completed interrupt, sometimes before the real interrupt had occurred. Once identified it was easy to fix in software - one more bug swatted.
Paul Austin got Burst mode going on Wednesday; he took 63 images of a magnification test grid off the column into the RAM Store using his own scripts. Initially he had a problem with the End of Frame detector on the Waveform Interface card, but he has modified it, and it now seems to work reliably. Today Peter Milne is going to try it with his software.
Martin Wiseman has had another diversified week; he has put in some more design time on a step-recovery diode drive for the "flashover" unblanking; he has started building the power supplies for the Acceptance Test box - Avantel should have finished the boards but Martin is not proposing to ski down to Melbourne today to pick them up - and he is waiting for Richard to buy the bits from which he can build the driver for the Pythagoras chip. Graham told Richard to try to get the Pythagoras bits for free, but the attempt is expected to be abandoned early next week.
Chris Warner has completed his analysis of his l mm blanking plates at a 0.2mm spacing; the transit time is about 80psec with a 700V beam (worst case) and the potential difference between the plates required to halve the intensity of a 2kV beam (worst case) is 2.3V.
Unfortunately, we don't know how much potential difference is required to reduce the beam intensity by a factor of one thousand — if the beam profile were Gaussian, it would be
7.8V (too much for direct drive from GaAs), but in fact the beam profile probably has a much sharper edge, so 5V might do it; Chris is going to do some measurements on in an effort to find out.
We had a meeting on the new blanking scheme on Wednesday afternoon, when Andrew Dean pointed out that if we left the fast plates open-circuit, and the l6G06l GaAs driver drove
the cable from +2.5V to —2.5V, the plates would see a +/-5V swing for a time equal to up to the twice the propagation delay through the cable (about 8nsec at present).
At the time we were worried about the double amplitude pulse reflecting back and forth along the cable, but since the 16G061outputs can be tristated, we could switch out the 16G061 driver and switch in a second l6G06l to present the cable with 50R to 0V to terminate the pulse by the time the pulse had been reflected back to the driving end.
None of this looks incompatible with 100psec blanking, which is no guarantee that we will make it; it isn't the problems that you anticipate that foul up projects, and Sod's Law suggests that some unexpected problem will make this discussion sound like an optimistic irrelevance when we actually try to get the system working.
Amongst other matter discussed at the blanking meeting were the problems of assembling and locating our blanking plates 0.2mm apart, and the problems of cleaning off the contamination after sustained use - Trevor’s flame-thrower has a lot to recommend it, but it would rather constrains our choice of materials and assembly techniques. Tony Slater (spelt it right this week) sat through all the brain-storming with remarkable forbearance.
Barry Barker has started winding the new scan coils on Tony Slater’s new jigs, and the new coils look a lot neater than their predecessors. These coils are going to get assembled onto the liner tube with rubber bands and glue in the first instance, but if they work properly, Tony and Chris Warner are going to work up a more formal assembly scheme.
Bob Taylor was going to send the new, thicker stage plate up to Leeds today, for Teflon coating. There as been a suggestion that he should hire a dog-team and sled it up, but he prefers to delegate the problem to the courier service.
I've had a couple of interesting discussions with Peter Milne and Richard Adams about Logic State Mapping. Peter has come up with a neat idea for using a variant of Burst Mode to let us pick up multiple waveforms more or less simultaneously into the RAM Store; it means modifying the Waveform Interface to spread a Waveform vector across a succession of worm-holes, and modifying the Scan Generator to hold a table of x,y vectors which it can spit out at line scan rates, as if it were doing a line scan, but by the time we have gotten around to replacing the LIP with an image store that can handle multiple cursors, we should have redesigned the Waveform Interface and replaced the Scan Generator.
Ian Murray seems to be digging into the User Inter-Face with exemplary speed - I saw one of the screens come up in lab just now, and Simon has just used the mouse to to control stage X/Y for the first time.
Roland Meins brought up the CAD Interface yesterday – large chunks of it are working, and he hopes to have enough of it going for a demonstration by the end of next week. This means leaving out quite a lot of features (like rotational alignment) that will take a lot of straight-forward programming effort, and others (like lock-stepping between sub-circuits) that are going to take an uncertain amount of digging around in the Cadence software.
(Notes added 20/01/2013: Trevor Sexton's “flame-thrower” was his butane gas fired cigarette lighter, which he'd adjusted to give an intense needle flame admirably suited to burning carbonaceous deposits off platinum apertures.
The residual gases in the nominally evacuated electron microscope column are long chain hydrocarbons, and they form a single atom thick layer on the entire inside surface of the column,
If this layer is bombarded by energetic electrons, the hydrocarbon chain loses hydrogen and crosslinks to form an involatile insulating layer that stays put and eventually gets thick enough to carry a fairly high negative charge – comparable with the electron beam voltage, which tends to push the beam around and messes up the electron optics.
Electron microscope columns have to be opened up from time to time to allow these insulating layers to be burned off the components where they build up. Trevor's job – as a service engineer – involved doing this from time to time.
Cadence sold a widely used integrated circuit design software package. The electron beam tester was expected to be used to compare what software simulations had predicted would happen on an integrated circuit with what happened in reality.
This – in fact – had greatly reduced the market for electron beam testers, because it made lot more sense to use computer simulation to predict what the circuit would actually do before you'd spent $100,000 on the masks sets that let you make the circuit. Electron beam testers were relegated to role of testing the precision of the simulations, rather than debugging a prototype mask set before you made a pre-production mask set. It's not clear whether Graham Plows had missed the marketing implications of the arrival of integrated circuit simulation software, or had merely ignored them. In due course Peter Crawley didn't.)
to: Richard Adams, Martin Wiseman. Chris Warner, Peter Milne, Simon Dawes, Bob Taylor, Paul Austin, John Harrison, Roland Meins, Andrew Dean, Tony Slater, Ian Murray
from: A.W.Sloman
subject: Progress to 15th February 1991
I'm still working on the Processing Pulse Generator; when I toleranced my approach to the Mark 3 Trigger Board, I found that in Single Encode per Sample, it limited the maximum sampling rate for the system to 22MHz, rather than the 25MHz we have been claiming.
The problem is the sloppy tolerances on the Newport delay lines - we should be able to make 25MHz fairly easily by replacing most of the delay lines with clocked delays derived from the 800MHz clock via the l0G06l synchronous counter we already use for Multiple Encodes per Sample.
Richard has persuaded me to defer detailing this approach until after I've worked out a cut and link mod for the Processing Pulse Generator on the Mark 2 Trigger Board. So far I've reconstructed Bob Ward's Hilo model of the relevant parts of the Mark 2 Trigger Board, and got it working, which took about a day. I'll now use it to test the cut and link mod which should be faster, over-all, than doing the mods directly on the second Mark 2 Trigger Board, and less potentially destructive.
Andy has modified his Delay Boards for operation with a 400MHz clock, and I've modified the first Mark 2 Trigger Board to provide a 400MHz output to the Delay Boards.
The combination is working but Andy promptly found the next bug, which sometimes randomised the Delay List entries while they were being loaded. Andy has devised a mod which has cured this (but deprives the board of the capacity to accept block mode transfers), and so far testing the mod has revealed a bug in the Timebase Interface, which Paul Austin now thinks he has fixed, and a bug in the software generating the Sampling Lists.
Testing Burst Mode has got to the point where we now know that when we pick up 8-bit images, we pick up the low order 8-bits of the 16-bit output from the Waveform Processor. Paul Austin has devised a mod to fix this, but it hasn't worked yet, and he is waiting to get back to the system to find out why. Meanwhile he has been working on the Timebase Interface, mostly up-dating the circuit diagrams to bring them into line with the latest PCB.
Martin Wiseman has spent most of this week testing the Acceptance Test board; he did put in a little time on Beam Blanking - we now have a current spec for the 16G061 GaAs driver chip, which makes less extravagant claims about its speed than the preliminary spec, so now we also have price and delivery on a 50psec step recovery diode as well.
Chris Warner has been doing some measurements of beam current as a function of voltage across the blanking plates, but he hasn't told me how they are coming out.
Tony Slater has assembled a set of the new scan coils onto a liner tube, but Simon has no immediate plans to fit it to the column - the existing coils are working adequately.
There has been more discussion about Logic State Mapping – I now understand Peter's point about the advantages of a upping the number of samples the Sampling Crate can handle from l000 to 4000, when you actually want 30,000 samples to compete with Schlumberger. The idea of using Burst Mode to fill the RAM Store with a huge logic state map turns out to have been fairly dumb - if you want to store that much information you do have to put it into the RAM Store, but it is 15 times quicker to acquire the information in Waveform Mode, and put it into the Frame Store one or two megabytes at a time.
Roland has had a huge new package of software from Cadence, and we had to go to our friendly academics again to borrow the SparcStation operating system 4.03 tapes, so that Roland could reconfigure his workstation to accommodate the new software.
The last Trevor Sexton joke is that his dismissal is actually an exercise in cloak and dagger disinformation — he will now apply for a job servicing the Schlumberger machines, and discover all their dark secrets, so that when he comes back he will be able to say "but they all do that sir, and the Schlumberger machine does something much worse”.
(Note added 28th January 2013. Trevor Sexton was the hyper-thyroid field service engineer who had started off with Cambridge Instruments, moved to Lintech and moved back to Cambridge Instruments when Lintech went bust.
He was an entertaining character, but impatient, and his impatience with doing his travel and accommodation claim forms finally got him fired from Cambridge Instruments. Richard Adams reproached himself afterwards that he hadn't been as nit-pickingly careful in vetting Trevor's expense claims as he should have been. The accountants who did go through them all – nit-pickingly carefully – found stuff they didn't like. Nothing that suggested that Trevor had been getting more money out of the system than he should have been, but there was apparently was some evidence that he'd improvised a bit from time to time.
He contacted me out of the blue at the beginning of January 2004, from an e-mail address at VideoJet.com, which was presumably one of the ink-jet package labelling companies which got going around Cambridge in the late 1980s spun off out of one of the Cambridge contract research firms.
We exchanged a couple of e-mails at the time, but when I sent him a change of address e-mail in 2011, I didn't get a response.
In general, the engineers talked to the service engineers when they were around. In the field they spent a lot of time with the people who were using our machines – as opposed to the people who had bought them – and could be informative about which features worked, and which didn't. Trevor had been very informative about the hardware used to monitor the position of the electron microscope column with respect to the integrated circuit being looked at on the original Lintech electron beam tester – it was easy to knock out of alignment, and tedious to realign, and we'd consequently gone to some trouble to make sure that the Cambridge Instruments electron beam tester used a more robust system, as shows up in my weekly report for the 15th June 1990.)
to: Richard Adams. Martin Wiseman, Chris Warner, Peter Milne, Simon Dawes, Bob Taylor, Paul Austin, John Harrison, Roland Meins, Andrew Dean, Tony Slater, Ian Murray
from: A.W.Sloman
subject: Progress to 22nd February l99l
I've completed and tested the cut and link mods to the second Mark 2 Trigger Board; it should now provide Encode, Latch, Write, Memory and Squelch pulses in the right numbers and with the right timing to let us use multiple encodes per sample, working into the Recirculate Latch on the Waveform Processor.
Paul Austin should be able to do the corresponding mods on the WPD quite soon now, after which he will be able to test the combination.
The cutting and linking was relatively quick - it meant roughly a dozen cuts and a dozen links. At the same time I did Andrew's mod on the Trigger Disable inputs, which was another three cuts and three links, and no great trouble, and the 400MHz mod, which was an absolute pest - getting the Filtotex cable soldered to all the right places was quite awkward, and very time-consuming.
Testing was a bit maddening as well - I charged off up a couple of blind alleys before I got the board working – but once I had got Sampling Pulses coming out of the board, I also had the rest of the zoo. The basic cut and link gave me Latch Edges that lagged the relevant Encode Edges by 3nsec, where I actually wanted a 2nsec lead - this was entirely due to propagation delays across the board, rather than in the logic, and a couple more cuts and links enabled me to reverse the situation - the Latch Edges now lead the Encode Edges by 4nsec, which is more than would be ideal, but within tolerance.
I also found two missing tracks on the circuit board, which meant that the number of Encode Pulses varied in rather odd steps, but the gaps have now been linked across, and the Encode Pulse count varies as one would expect.
I've still got to finish documenting the cut and link mods - there is a circuit diagram on Metheus representing the circuit after most of the cuts and links, which I will update on Monday. Then it is back to getting the Mark 3 design to run at 25MHz.
Andrew Dean seems to be happy with the Delay Board at the moment – he has updated the circuit diagrams for the Mark 3 Delay Board on the Metheus system. He and Peter And Paul have spent some of the week arguing about leading dummies, where there is still a gap between theory and practice, although the practice works reliably. They have also been trying to get Burst Mode working.
By and large, Burst Mode is working - which is to say for everything except images. Paul finally got the byte-swap mod working on Wednesday - the mod was correct in conception, and almost correct in implementation. He lost Tuesday to wrestling with the Timebase Interface and the Scan Generator - the Timebase Interface is now fixed, but he still has to modify the Scan Generator to provide a Scan Reset input.
Peter has now got Burst mode running under interrupt control from his software, rather than .Paul’s scripts, which is useful progress, but we still have a bug that causes the system to drop out of Burst Mode after a hundred or so lines, when we are picking up images in Line Mode – if necessary we could revert to picking up images in Frame Mode, but it leaves the Plinth Control card un-monitored for the duration of the burst, which can be rather long.
Sorting out the number of leading dummies has cured a couple of other minor problems, such as the time-offset in between the sampling window and the actual period sampled. It is worrying that Paul Austin's theory and Peter Milne's practice don't entirely agree, but Peter's practice seems to be working at the moment so the question can hang fire for a while.
Martin Wiseman is still testing the Acceptance Test Board - he has fixed the trivial flaws in implementation, and is now wrestling with a fundamental flaw in the specification – a requirement for a precise and linear ramp with a l usec rise-time. What would be a trivial exercise for a slower rise-time is proving quite difficult to achieve at l usec.
Chris Warner has been writing up his beam current measurements - everywhere where he can measure leakage current when the beam is blanked, we only need 3V to blank
the beam, but he can't come to any conclusion at spot size l, and extrapolation suggests that there we will still need 7V.
Barry Barker has started assembling the fat stage, although the top plate hadn't got back from Leeds when I last saw him.
(Note added 28th January 2013. The vacuum chamber top plate – which carried the electron microscope column around above the integrated circuit we were looking at, which was mounted in a fixed and non-moving socket – was electroplated at Leeds with a proprietary mixture of teflon and nickel.
There was a groove all the way around the top of the fixed box, carefully sized to accommodate a long rubber O-ring, which preserved the vacuum inside the box while the top-plate was sliding across it, but the atmosphere pressure acting on roughly a square metre of top plate is roughly ten tonnes of air, so both the plate and the top of the box were “teflon-plated” so that we could use relatively small electric motors drive one across the other.
Cambridge was a high-tech centre, with lots of small specialist suppliers for all sorts of useful services, but for the “teflon-plating” we had to ship the parts up to Leeds.
Electroplating nickel is easy enough, but teflon doesn't ionise, so getting it into the electroplate layer is a bit more complicated,. Very finely divided teflon - “colloidal teflon” - does pick up an electric charge when suspended in water, and you can move it around by electrophoresis. Getting the conditions right to allow you to “electroplate” a surface with a useful proportion teflon particles embedded in a nickel matrix is what specialised suppliers get paid for.)
to: Richard Adams, Martin Wiseman, Chris Warner, Peter Milne, Simon Dawes, Bob Taylor, Paul Austin, John Harrison, Roland Meins, Andrew Dean, Tony Slater. Ian Murray
from: A.W.Sloman
subject: Progress to 1st March l99l
I think I've now got a counter/shift register solution for the Processing Pulse Generator on the Mark 3 Trigger Board - the core of it is a 100136 ECL counter/shift register clocked at a 200MHz which is not much lower than the guaranteed maximum clock rate of 250MHz, so the design has some interesting features - and I am now amending the Trigger Board drawings (854696-3 sheet 4) on Metheus to reflect the new design.
I'll still have to do detailed tolerancing on the detailed design when I complete it, but the Hilo model seems to show reasonable margins.
Andy distracted me earlier in the week, to work on the second Mark 2 Trigger Board, mainly to check that the Convert Edge generator was working correctly, which it was - but it worked rather better when I took a 75R resistor off one of the taps on the delay line.
While I was fiddling with the board I recalculated the frequency response and damping around the phase-locked-loop controlling the 800MHz oscillator, and decided to increase the damping resistors (R219 and R220) from 240R to 2k7.
This appears to have reduced the jitter on the 800MHz clock from l.6nsec to 60psec, which is rather dramatic, although not too important - the l.6nsec jitter seems to have been generated by a roughly l ppm frequency modulation (+/-l kHz in 800MHz) with a period of around 400Hz, which would have been hard to detect with the sampling system.
When it was first tested in the system, late this afternoon, it seemed to be working fine - even Recirculate seemed to work after a fashion.
I was able to get at the first Mark 2 Trigger Board earlier this afternoon to check out its Convert Edge Generator – I had to cut an extra track to get the mod working correctly. I also had yet another look at the Unblank Pulse generator and found that the shortest ranges had failed again ~ this means another 10G002 at £33.75 and another think about the design.
Andrew Dean remains happy with the Delay board - Peter Milne has found no fault with it this week either - and has spent most of the week modifying the second Mark 2 Delay board to match the first, with a view to using both in combination, for Delay by Events, and for auto-calibration of the Time-to-Digital and the Digital-to-Time converters.
Andrew and Peter and Paul Austin have resolved their differences about leading dummies - there are now no leading dummies in Open Loop, and only one in Closed Loop. I want to disrupt the consensus by generating Encode pulses during Processing Passes, which will change some of the Accumulator presets by one, but not before the Mark 3 sampling crate.
Paul has spent most of the rest of the week getting the last of the bugs out of Burst Mode. Peter now finishes off programming the WPD for a burst by writing data to the dummy address 602E, rather than the real address 6020, and the Length of Burst register is now being set.
Paul has been moderately successful in modifying the Scan Generator to allow it to be reset - the mod seems to work, but it takes around two hundred l.5 usec wide Reset Pulses to reset
the Frame Integrator (the capacitor has a worst case discharge time constant of 50 usec, and suffers from charge soak - roll on the digital frame scan).
Burst mode is working rather better than it was last week - everything seems to have worked in isolation, and we expect to collect and display 62 images sometime on Monday.
Martin Wiseman is still working on the Acceptance Test Board. The latest and it is to be hoped, the last serious mod involves getting the 100136 counters reset inside l0nsec - half the 50MHz clock period - which is tight. Martin is more than usually quietly confident, so it ought to work.
Chris Warner has finished writing up his report on vacuum lubricants, and has done some more work on the ion beam column, which now needs an electron flood gun, its own Z-axis motor and a double-skinned gas-handling system.
Chris has also processed Simon Dawes measurements of final lens current versus working distance, beam voltage, and extraction voltage, and concluded that we can get working distance to within a millimetre at working distances of a couple of millimetres, where we need this sort of accuracy.
We also get enough precision to think about compensating magnification and rotation against the small changes in final lens current required to keep the surface of a tilted specimen in focus.
Simon is going to have to repeat the measurements for the new, "thin" final lens on PPl, once PPl has been Trevored into alignment (for future reference, a "Trevor" is an undocumented procedure known to have worked some other time on some other machine).
Barry Barker has got the fat stage assembled and pumped down. The top plate doesn't seem to flex at all, but it sinks as a whole by 400 microns, and the side-plates splay out by 120 microns on each side. Bob Taylor and Chris Warner are contemplating the situation - the vertical movement is not of itself a problem, but the side-ways movement is part of our column position error budget.
(Note added 30-07-2017 : The difficulty of getting a several 100136 synchronous counters to function as single long synchronous counter at 200MHz wasn't a new discovery. The 1988 Delay Board specification had a paragraph on the subject which Johannes Gebrehiwet hadn't taken as seriously as he should have.)
to: Richard Adams, Martin Wiseman, Chris Warner, Peter Milne, Simon Dawes, Bob Taylor, Paul Austin, Roland Meins, Andrew Dean, Tony Slater, Ian Murray
from: A.W.Sloman
subject: Progress to 8th March 1991
I haven't got any further on up-dating the Trigger Board drawings (854696-3 sheet 4) to reflect the shift-register-based design for the Processing Pulse Generator — I should get back to this later today.
I've been distracted by Trigger Board Mark 2, version l, which has now been cut and linked to much the same state as version 2, except that it now produces the right number of Latch edges in recirculate (I noticed that I had freed up a flip-flop during the cuts and links ...). The 800MHz clock isn't as good ~ there is still 500psec of jitter over periods of some milliseconds but this appears to reflect the occasional injection of a spot of phase noise into the Phase Locked Loop, which is damped out within about a millisecond. Grabbing the board to stop it vibrating under the influence of the fan seemed to reduce the incidence of these bursts of phase error.
The cutting and linking took about a day in total - it took as long again to find out which bits I'd cut and linked wrong, and the 75R resistor which had gone open circuit the last time it had been ripped off the board.
In the process I endeavoured to fix the Unblank Pulse Generator - IC53 had failed again - but the replacement I'd put in failed immediately, despite the changes I'd made to protect the input that fails. So I changed the circuit so that the short pulses come out with the right width even though IC53/9 has gone leaky and we will see how long that lasts.
I'm going to phone Gigabit in California on Monday evening, then I think I'll redesign the circuit around the Plessey SP93802BHG dual comparator - marginally slower than the GBL10G002 (0.8nsec typical, versus 0.5nsec), but adequate to the job. It means an extra component to fit on the Trigger Board, but we can lose ICl07 and its associated components, so there should be a nett saving in board area.
Andrew Dean's Delay Board Mark 2, version l continues its exemplary reliability. Version 2 is giving him some trouble at the moment, now that he is testing his 400MHz mods on that board, but we should soon be in a position where Paul can start testing the second version of the Waveform Processor Digital Mark 2 in the second Sampling Crate.
He and Peter have spent most of this week on P2 getting Burst Mode going (with Paul), getting Recirculate going (or multiple encodes per sample if you want the descriptive name), getting dire S-curves and putting in the "photon gun“ - a green LED on the stage, firing photons at the scintillator to produce a test signal more predictably than the real electron gun.
The consensus on leading dummies still seems to be holding up - and with the right number of leading dummies we now get 64 images in Burst Mode. Since Graham hasn't yet finished
working out his plausible explanation of why we can only display 62, we are not planning on telling him for a while.
Martin Wiseman is still working on the Acceptance Test Board. The latest mod involves tuning the current through the step recovery diode to get a decent step - this involves changing only a resistor and a transistor, but the new transistor has a different pin-out from the old - so not a “serious” mod, but not quite down to “frivolous” either.
Chris Warner is now writing up the working distance work on the fat lens, and trying to generate a program for producing an estimate of working distance. He also spent two days at CMS using their ion beam to cut holes in passivation, without benefit of an electron flood gun to neutralise charging. It is slow.
Repeated measurements on the fat stage now suggest that the top plate flexes by 300 micron under vacuum, and the side-plates only splay by l2 microns. The force required to ,move the fat top-plate around on the evacuated chamber is much less than was required with the thin top-plate - Barry can move the fat plate by hand - so there are good prospects of an appreciable improvement in stage-positioning reproducibility.
to: Richard Adams, Martin Wiseman, Chris Warner, Peter Milne, Simon Dawes, Bob Taylor, Paul Austin, Roland Meins, Andrew Dean, Tony Slater, Ian Murray
from: A.W.Sloman
subject: Progress to 15th March 1991
I've now up-dated the Trigger board drawings ( 854696-3 sheet 4) to include the shift-register based Processing Pulse Generator. The drawings are not perfect, and I haven't done the tolerancing. Metheus 2 was out of action until this morning so I ended up doing the work on Metheus l on a slightly out-of-date version of the circuit stored on Metheus 3, but the files on Metheus 2 are now up-to-date.
I haven't touched a Trigger board all week - version 2.2 in P2 needs modding up to version 2.1, but it doesn't need fixing as urgently as we need a quasi-working Trigger Board. Version 2.1 is still working fine for Paul Austin.
My phone-calls and a fax to Gigabit in California have not produced a solution to the failing GBLl0G002, so I've ordered a bunch of MAX9687CSE dual comparators — only a
third as fast as the GBL10G002 or the Plessey SP93802, but useable. When I read the detailed data sheet for the Plessey chip, it turned out to want 0V, -5V and -l0V power supplies and offered an input voltage range from -2.4V to -7.3V, which wasn't much help, and the delivery was lousy.
I spent half of yesterday and most of today on the two Mark 2 200psec Blanking Boards, taking out connectors and putting in replacements - the parts list for the board is pretty much right, but it was kitted from the Mark l parts list.
I also tidied up the two Mark 5 Blanking Drivers. Both the Blanking drivers were loaded with SMA connectors without the standoff - Avantel said that the stand-offs "didn’t fit", whereas they are snug-fitting on a solder~free SMA connector. I've had the connectors off one of them, fitted the stand-offs and replaced the connectors - but that revealed that the stand-offs had been made from nylon rather than PTFE (they melted), so I'm not doing the other board until Bob Taylor gets me some stand-offs made in PTFE.
Andrew Dean spent the first part of the week finding out for himself that the LIP is a museum of duff circuit design – he and Paul still haven't solved the noise problem with the Slow Scan Card, though they think that putting a ground plane into the Slow Scan Card would help.
After that Andrew got involved in getting blanking going, and helping Peter Milne with the T-curve, and helping Simon Dawes with the aftermath of the move - including the repeated failure of the Suppressor Drive on the Waveform Processor - Analogue. Today he has started on a cable diagram for the EBT, showing what block is connected to what. This will eventually end up on Metheus.
Paul Austin was working with Andrew on the LIP at the start of the week, but has moved on to testing the WPD Mark 2.2 . He has found and fixed the usual problems - a duff chip, an open circuit track and a couple of solder bridges, but he got stuck on working out why the low-order byte of the adder was coming out wrong. Just now he has discovered the lead that hadn't been soldered down to the pad.
Martin Wiseman has got the step-recovery diode going on the Acceptance Test Board - it now gives a 3V step with a transition time of 130psec, 10% to 90%. His Pythagoras chip driver is now working, now that we have a socket into which we can plug the chip - the chip was too thick for the first socket, and the current socket was too deep for the chip until Barry Barker milled the top off it, but the third socket (with a different pin~out) will be just perfect when ,the new PCB for it comes in....
Martin also had to replace two CLC-22l Suppressor Grid driver amplifiers, which blew up in quick succession on the WPA - we have got five CLC-22ls left, and Martin and Andrew have worked out a protection network to go onto the column end of the drive cable, while Martin has glued heat-sinks to the CLC-22ls so we may have improved the mean time between failures.
Chris Warner has been talking to Tony Slater about the new Blanking Head - how we assemble it and connect it up, and they seem to be coming to a conclusion. Chris also spent some time at the University Library, reading up on the chemical compounds used in ion beam deposition - they are all horribly toxic.
While he was there he read up a bit on electron beam flood guns, required to neutralise the charge deposited by ion-beam erosion and deposition.
Yesterday he did something more practical — covered two of the new scintillators with dirty aluminium, so we can use them in the column (albeit briefly).
Peter Milne has had a busy week getting things going. We have seen S-curves (there must be something wrong with the TTLD) and T-curves (they looked okay) and Peter has got Stan
Davidson's Simple Autogain working for Open Loop (but not for Closed Loop).
(Note added 3-08-2017: the T-curve was plot of detected secondary electrons as function of the delay between a narrow Unblanking pulse - 10nsec or less – and the subsequent Sampling Pulse from the Trigger Board that drove the analog-to-digital converter on the Waveform Processor – Digital some 120nsec later, when the primary electron pulse had had time to get down the column to the specimen, and the secondary electrons they'd knocked out had had time to get back up the column through the final lens and the filter grid to the scintillator, to be converted to a flash of light the would be detected some thirty nanoseconds later at the output of photomultiplier, amplified, buffered and sent on to the analog -to-digital converter.
The delay from the Unblank Pulse to the Sampling Pusle was digitally programmable on the Trigger board by a string of AD9500 programmable delay generators.)
The histogram program also seems to be working, and the results make sense, although they suggest that we are detecting many fewer secondary electrons than the primary electrons hitting the specimen. Peter has deferred using it in an Autogain program until after the demonstration -I spent a day or so earlier in the week working out how to use it to get some of the signal-to-noise advantage of single-electron counting while retaining the capacity to count up to 15 electrons per sample (the approach may be patentable, but the patent would not earn significant license fees).
(Note added 7th February 2013. Stan Davidson was the software equivalent of Neil Richardson at Lintech in the mid-1980s, who wrote the menu-based software that controlled Lintech's pioneering electron-beam tester, and the routines that automatically controlled the photomultiplier voltage (and thus it's gain). Like Neil Richardson, he had physics Ph.D., and – again like Neil – he'd been over-loaded to the point where he'd had to keep everything simpler than it should have been.
I can't remember what the initials LIP stood for - probably linear image processor - but I think the LIP was the analogue scan generator and video processor for the original Lintech electron-beam tester, as designed by Tim Frost under Ralph Knowles' direction.
The “histogram” program collected 256 successive digitisations from 256 individual electron “flashes” onto the specimen. The idea was that if the photomultiplier gain was set too high, the most frequently appearing output would be closer to the maximum than the middle, but in practice it was easier to rely on the fact that any above-maximum response was registered as a maximum output, and cut the gain if that slot had more hits than any other. This dealt with the problem that arose when we were reduced to detecting single electrons from time to time, where any program that tried to get the average signal level right in that regime ended up setting the photomultiplier gain much too high.)
to: Richard Adams. Martin Wiseman, Chris Warner, Peter Milne, Simon Dawes, Bob Taylor, Paul Austin, Roland Meins, Andrew Dean. Tony Slater, Ian Murray
from: A.W.Sloman
subject: Progress to 22nd March l99l
I haven't done very much on the Trigger Boards this week - Trigger Board 2.2 has now been modified to the same state as 2.1, so it too produces the right number of Latch Edges in Recirculate, and gives the right pulse widths in the range 2 nsec to 40 nsec even though IC53 has gone leaky.
I called up Gigabit in California again this week, and while the applications engineer claims that my problem with IC53 has been processed by engineering, the promised response has yet to appear.
I've spent most of the week on the Blanking Boards; both are now working. Blanking Board 2.1 has had its delays trimmed rather brutally and gives Unblank Pulse Widths of 500 psec, 800 psec, l.2 nsec and l.6 nsec. Board 2.2 now gives 500psec, 950psec, l.6 nsec and 2.l nsec but I intend to modify it to bring it closer to Board 2.1.
The Blanking Driver plugged into Board 2.2 has the PTFE stand-offs under the SMA connectors - they don't melt. Bob Taylor had got them out of Sheet Metal by Wednesday.
Andrew Dean has spent most of the week in the gold-fish bowl, helping Peter and Simon with P2 - some of this has just been keeping the head amplifier output, the Unblank pulse and the Encode pulse visible on the scope, so that they can see what “pulse width” and “Encode delay” are actually being generated.
In the process he found that the Encode pulse was being inverted on the WPA, so the ADC was encoding on the trailing edge of the Encode pulse; this has been corrected by cut~and~link and will be corrected in the layout on the next issue of the WPA. I can't think how I missed it on the original circuit diagram.
Andrew has also made some progress with the cable diagram - all the blocks are in place, and he has started putting in the cables.
Paul Austin has now got WPD 2.2 working - he had to find a couple more dry joints and solder bridges before he could get the Averager going or persuade the board to give the right results in Recirculate, and there was an open circuit which stopped Burst Mode from working for a day or so, but the board is now ready for testing with the system software on PPl, when we can find a spare Sun to drive the system.
Once WPD 2.2 has been proven on PPl, it is planned to try the "nuts" mod, which should allows us to use "nuts" values up to 5 (7 if Paul uses all the OR/NOR gates we aren't actually using at the moment).
He is now working on the Slow Scan card in the LIP, reorganising the 0V tracking and adding a grounded screen in an effort to reduce the common mode noise across the card.
Martin Wiseman has spent most of the week on various aspects of the Pythagoras chip and its sockets; some of the time was spent in the gold-fish bowl, watching Simon use the chip and
the drive circuit, but meanwhile the new printed circuit board for the new socket had had extra holes drilled in the wrong places...
Martin has also put in some more time on the Acceptance Test board, writing it up, designing a box to put it in, and even looking at the time lag to the step recovery diode's step, which turns out to be rather temperature dependent.
Chris Warner has been doing a bit of sitting with Simon, a bit of looking at the electron flood gun for the ion beam column, a bit of chasing around looking for carbon deposition systems. and a certain amount of work with Tony Slater on the new Blanking Head. He says it constitutes more activity than progress …
Peter Milne has been busy trying to keep his auto~gain and his auto-encode-delay algorithms going despite the appearance of "blanking induced drift" on the column – it came on sufficiently slowly and erratically to suggest that the problem was a bug in the software, or perhaps a bug in the physics, leading the "worry committee" to explore a couple of blind alleys.
Everybody on this list has been invited to the party on the lst April, but not everybody has told me whether they are coming - we need to know early next week, so we can order the beer. We will have to take our chances on whether the EBT section of the party is a celebration or a wake, which will influence the volume of beer likely to be consumed.
to: Richard Adams, Martin Wiseman, Chris Warner, Peter Milne, Simon Dawes, Bob Taylor, Paul Austin, Roland Meins, Andrew Dean, Tony Slater, Ian Murray
from: A.W.Sloman
subject: Progress to 28th March l99l
I haven't done very much to the Trigger boards this week, though I have now worked out what I am going to do to the Short Pulse Generator, and what I want to do to the Long Pulse Generator, Richard and Andrew willing.
I finally managed to get to talk to somebody useful at GigaBit on Monday evening; he hasn't got a clue why the 10G002’s are going leaky either, and asked me to send him some of the parts I'd blown up — I've put five of them in the post ~ but this leaves us with no real alternative to redesigning the short pulse generator; I'm planning on racing the two halves of a Bt-622 dual delay generator against one another to generate pulses down to 2 nsec wide.
This choice of device puts the upper limit on "short" pulses at about 28nsec. and I'll either have to move the long pulse generator down a bit, so that the shortest "long" pulse is shorter than the longest "short" pulse, which wouldn't take more than five minutes, or use my new, all-digital long pulse generator.
I've done a bit more work on the Blanking Board - some of it on Saturday afternoon to show solidarity with the workers - and both boards are now producing much the same output pulse
widths – 490 psec, 760/780 psec, l.26 nsec, and 1.76/l.74 nsec.
I've also found a small systematic dependence of pulse width on average inter-pulse interval, which seems to arise in temperature changes in the driver transistors on the Blanking Driver daughter board. Peter Milne may have to adjust the off-set on the shortest Unblank Pulse by up to 1V to compensate for this - if he doesn't the beam may not unblank at all at the highest repetition rates – but otherwise this effect should not be a problem.
Doing a Mark 6 Blanking Driver Board with BFG31 driver transistors ought to reduce the effect — they have much lower thermal resistance to ambient than the HP parts we are using at the moment.
Andrew and Paul and Martin and Simon have spent most of the week in the gold-fish bowl with Peter, attempting, with some success, to preserve all our jobs.
Chris Warner has been stuck (with me) with task of sitting around looking the worried - didn't we do well. He also put in some time with Tony Slater on the new blanking head, negotiating a bit of extra width for the board, so that we can get the tracks past the SMA connectors.
Steve Fisher has got an image out of PPl - the column has to come apart again to get the C2 coil in straight, and to get the 100MΩ resistor on the scintillator properly insulated, but it is still a mile-stone (to be more precise, somewhere between an inch-pebble and a foot-rock, but it's all progress).
Bob Taylor called up Richard from Sweden today, to find out if it was worth coming back, and was reassured.
to: Richard Adams, Martin Wiseman, Chris Warner, Peter Milne, Simon Dawes, Bob Taylor, Paul Austin, Roland Meins, Andrew Dean, Tony Slater, Ian Murray
from: A.W.Sloman
subject: Progress to 5th April 1991
I've spent most of this week on Metheus, working on the Trigger board circuit diagram. Sheet 3 (854696/a0cV3-2.sch) which now includes a Short Unblank Pulse Generator based on the
Brooktree Bt-622 dual delay generator, properly integrated with the current AD-9500 based Long Unblank Pulse Generator.
This version of sheet 3 will be saved, and I'll proceed to amend it further, to replace the current Long Unblank Pulse Generator with the digital system.
I spent about an hour yesterday writing a short memo on the Waveform Processor - Digital Mark 3; we ought to replace the eleven 22V10 PALs with a smaller number of the big Altera PALs, both to reduce the number of packages on the board, and to provide more program space to let us extend the repertoire of the WPD. We should be able to provide a bunch of specific memory-to-memory transfer operations which would make testing the WPD a lot easier, and it is possible that we might be able to find enough board space to give us at least read access to the Accumulation Store.
Martin Wiseman has been checking out his second Acceptance Test board; he has got it working pretty much like the first one.
Paul Austin has been working on up-dating the Timebase Interface drawings on Metheus. He hopes to start working on Burst Mode again today.
Andrew Dean, Chris Warner and Bob Taylor have been on holiday all this week. Andrew comes back next Monday; Chris Warner and Bob Taylor are back on Tuesday.
Tony Slater is waiting for Graham to approve the new Blanking Head and for Simon to fit and test the new Scan Coil assembly (which should go onto PP1 early next week).
Steve and Simon have got the column on PP1 aligned at last - they are now going to do some more component swapping to try and find which bit had been messing them about.
to: Richard Adams, Martin Viseman, Chris Warner, Peter Milne, Simon Dawes, Bob Taylor, Paul Austin, Roland Meins, Andrew Dean, Tony Slater, Ian Murray
from: A.W.Sloman
subject: Progress to 12th April 1991
I've spent most on this week working on the digital Long Unblank Pulse generator for the Mark 3 Trigger Board; I've got a circuit design which gives the desired output when simulated on the Metheus (854696-3/bill.sch) and I've spent the last couple of days learning about "metastability" and doing worst case tolerancing - there is another day or two of work to do before I've toleranced enough of the signal paths to be entirely confident of the design. Only then will I start redrawing the relevant circuit diagrams.
Meanwhile, back at the Short Unblank Pulse Generator on the Mark 2 Trigger Board, Gigabit Logic have sent me a fax saying that there was nothing wrong with any of the five 10G002’s that I'd sent them. This destroys my hypothesis that the 0.5V drop I measured across R341 (100R) was leakage current from the input to the 10G002 – IC53/9. The next hypothesis is that the internals of the 10G002 were oscillating at around a GHz or so (see Gigaherz application brief AB-2 for a discussion of a similar effect in the 10G010/11) so that what I was measuring was the DC component on a volt or so of high frequency AC coming out of the input. Certain integrated circuit comparators have always been prone to this sort of behaviour - starting with the first, the uA710. The oscillations tend to persist even after you have got the obvious bugs out of the layout and the decoupling, and the solution is usually to design out the comparator.
Andrew Dean has had a quiet week; he has written up Simon's start-up and recovery sequences for P2 which will also serve for the other machines (with occasional changes of name), and has spent some time discussing "digital gain" with Peter Milne and gotten started on getting the system documentation right, by listing every board in the machine and assigning each one to an engineer. With 43 boards and five hardware engineers we could all be quite busy, even after we palm off the bought-in boards onto Peter.
Paul Austin has modified the Timebase Interface to allow Peter to read the Plinth Control Card during Burst Mode; he has tested it and the mod seems to work.
He also cleaned up the TaxiChip Receiver inputs on one of the Waveform Interface cards; one of the screens had been intermittently shorting to ground or to +5V.
Burst Mode now seems to be working again - Peter had 64 decent (albeit identical) images on Wednesday. We still aren't sure what actually stopped it working before the
demonstration.
The second of the Mark 2 Waveform Processor - Digital boards seems to work in the lab crate, but not in P2 - Paul is going to have to debug it on P2 until he can localise the fault closely enough to allow him to duplicate it in the lab crate.
When he has got that sorted he can finally get on with trying the mod to fix the arithmetic with negative shifts, after which comes the Slow Scan board in the LIP.
Martin Wiseman recalls spending most of the week humping junk around the lab, but in his spare moments he put the Acceptance Test Board into the Acceptance Test box, and did some thinking about the Scanned Blanking Driver board.
Chris Warner has spent some time this week admiring Bob Taylor's drawing of the low voltage electron "flood" gun for the FIB system, but more time on the Blanking Head.
Graham rejected the printed-circuit board blanking head, on the grounds that a chunk of printed circuit board material at the blanking head would have a high enough vapour pressure
to impair the gun vacuum. Chris is now looking at getting Wesley-Coe to put four 50R impedance coaxial glass-to-metal seals into the nose of our Blanking Head, right up against the relevant blanking plates; Bob Taylor, who used to work for Wesley-Coe, seems to think that this is straightforward, and it looks as if it shouldn't be too difficult to couple the glass-to-metal seals to RG402 or 405 semi-rigid cable without going through the bulk of an SMA connector, or introducing intolerable impedance discontinuities.
Tony Slater is reputedly unhappy about having a couple of weeks of drawing effort shredded, but when directly questioned he is philosophical about it.
(Note added 2013-02-07. I suppose we could have tried for an alumina substrate for the in-column blanking board, but it would have been expensive.)
to: Richard Adams, Martin Wiseman, Chris Warner, Peter Milne, Simon Dawes, Bob Taylor, Paul Austin, Roland Meins, Andrew Dean, Tony Slater, Ian Murray
from: A.W.Sloman
subject: Progress to 19th April 1991
I've spent this week working on the Mark 3 Trigger board; I've toleranced the digital Long Unblank Pulse generator and the design looks reasonably robust. It would look less robust (but still practical) if I had taken the 10G021 propagation delay data at face value - for this one part, Gigabit do not specify minimum propagation delays in their data book, but when I talked to them about it, they quoted minimum propagation delays of 390 and 410psec for the conditions I was interested in, nearly doubling the period over which I could rely on the output signal.
I'm now trying to start tolerancing the revised Processing Pulse generator; as I started identifying the critical timing paths it became clear that the circuit divides naturally into half a dozen tolerably self-contained blocks, and I'm going to redraw the circuit diagram as discrete blocks on two sheets - the present circuit diagram just fits on one sheet, but is pretty much totally incomprehensible.
Andrew Dean has drawn a block diagram of the Delay board this week. He has tested the draft procedure for running up P2 and now Simon and Paul are repeating the exercise.
Tom McNamara has started on the layout for the Mark 3 Delay Board and Andrew will have to check out the Time-to-Digital and Digital-to-Time Converters on the Mark 2 Delay board
before he can sign off the new layout.
Andrew has nearly finished assigning engineers to boards for the system documentation - the crates are pretty much dealt with but there are a few isolated boards still up for adoption.
Paul Austin has had a successful week on the Waveform Processor - Digital; he got the WPD Mark 2 working on P2 by identifying the problem as a wobbly Add/Subtract line and replacing the PAL whose wobbly output caused the problem.
Paul then went on to implement the carry propagation mod - initially with a minor bug in it - and we can now use negative Feedback shifts from -1 to -6. Peter has set up the software to exploit this, but Closed Loop still isn't working.
Peter Milne has spent most of this week working up the multiple waveform display - it now works, but all the waveforms come from the same spot. Next week, while Peter is off on holiday, Ian will arrange things so that different waveforms can come from different spots.
Martin has got the Acceptance Test board working in the Acceptance Test box; the original power supply cables were too thin, and Martin has now rejigged the separate power supply with remote sensing, and thrown in an extra, thick, 0V return cable for the low voltage negative supplies.
He has spent the rest of the week on beam blanking, in part looking at circuit design for the Scanned Blanking Driver - which now looks like a second daughter board on the "200psec Blanking Board" (855034 is a less emotive title) - and for the rest, cutting up and reassembling coaxial cables and measuring the impedance discontinuities for Chris Warner.
Chris Warner has spent the week multiplexing in pretty much his usual fashion - Monday he was off sick, which isn't usual - putting in time on the ion beam, the blanking head, and the alignment of PP1.
On the ion beam the plan to use the sliding seal plate as a gate valve for the ion beam column and electron flood gun has had to be slightly modified to include a scheme for compressing the 0-ring seal before it has to seal against air.
On beam blanking Chris has refined his new set of ideas to the point where he has started badgering Tony Slater to start drawing them.
Chris has also put in some time looking at the alignment of PP1 with Steven Fisher - when I saw it, the alignment was awful, and Chris is worried that the tetrode gun is warped in some way.
Bob Taylor is wading through the job of working out how to get P1 and PP1 up to the state that P2 will be in when we get the fat stage fitted - it is going to take some time.
Tony Slater is reorganising the electrical connections into the column at the TTLD - we should end up with the photomultiplier mounted a bit further out from the column to give Simon and Steven enough clearance to get at all the bolts in that area.
to: Richard Adams, Martin Wiseman, Chris Warner, Peter Milne, Simon Dawes, Bob Taylor, Paul Austin, Roland Meins, Andrew Dean, Tony Slater, Ian Murray
from: A.W.Sloman
subject: Progress to 26th April 1991
I've spent this week redrawing the circuit diagrams for the Mark 3 Trigger Board - the Processing Pulse Generator is now spread across two sheets, and I think the circuit is now reasonably intelligible.
I'm still trying to squeeze the Unblank Pulse Generator onto sheet 3, but I'm increasingly tempted to split off the Processing Loop Generator and the Convert Edge generator.
I still haven't toleranced the Processing Pulse Generator - now that I am over my cold and can concentrate again this goes back into the priority list.
Paul Austin, Martin Wiseman and Andrew Dean spent a lot of this week getting Closed Loop running. It now works (on P2 with the simulation box) from scripts, or from Pete's software if you prevent the Syntel from writing to the Expected Waveform Store. This passes the parcel to Peter Milne for the moment. An attempt to try Closed Loop on a specimen foundered when they couldn't get an image, but I imagine that we will try again.
P2 shuts down on Monday for most of the week to allow Barry to put on the "fat" stage.
Paul Austin has also done some work on the Timebase Interface; his recent mod to allow Peter to continue to monitor the Plinth Control card during Burst Mode was almost perfect, but occasionally let the system read back duff values, because the VME bus signals were changing more than once between ticks of the 5.0MHz clock on the Timebase Interface. Paul has now elaborated his mod to include a bit of asynchronous latching, which should perfect it.
Last week's problem with the Add/Subtract line on the WPD recurred early in the week; it turns out that the TTL version of this signal was terminated to -2V through a 75R resistor, which did rather unsavoury things to the TTL circuits on the track until it was removed.
Martin Wiseman squeezed in some work on the blanking electronics - he has measured the propagation delay through the current blanking board as about 7.5nsec (including the
daughter board but not the output cable).
Simon Dawes and Andrew Dean spent most of Monday and Tuesday sorting out the cable diagram - they were at it for ages.
At the same time Simon was sorting out the paperwork for the Sampling and Display unit power supplies that were to be shipped to Motorola; it took a couple of days as he chivvied the paperwork through the system until he could get a shipping number. ‘
(That was for one of the old Lintech EBT 1 machine Lintech had sold to Motorola a few years earlier).
Since then Simon has been trying to test Ian's new user interface. So far he has found the fault on the modified Timebase Interface, and is now stuck because wpcon won't work. When I last looked he was modifying the Cursor board in order to improve the interface to the Syntel; it wasn't picking up 16-bit data from the 8-bit bus as reliably as it should have.
Chris Warner has been multiplexing, as usual. He has spent most of his time on PP1, helping to sort out the alignment. They have found a few problems - one of the more irritating faults was that the limit sensors on the final aperture plate were not centred, so that if the plate was well off centre in X the Y limit sensor never saw the plate, rather messing up the software centering routine, not to mention the adjacent mechanics.
And the PP1 LIP had got its brightness and contrast settings way off, which made it almost impossible to see the beam when source imaging: once these problems had been solved it
proved possible to get the column moderately well aligned (from spot sizes 2 to 10 at 1.5kV), and they took a large area photo of a magnification calibration grid to document the scan distortion with the old scan coils.
The old scan coils are now being replaced by the latest scan coils; Tony Slater is waiting to see if they are okay so that he can start getting the. new scan coils into the system, which involves a lot of new drawings - for the hardened steel winding jigs, the assembly drawing, and the potting fixture that production will need - as well as the tidying up on the existing drawings.
Tony Slater and Chris Warner went to Wesley-Coe on Wednesday to talk about glass-to-metal seals for the vacuum feed-throughs on the blanking head; they came away confident that they can get 3mm OD seals, which gives Tony enough room to to put in an O-ring seal between the blanking head and the bellows flange.
Tony is actually working on the new office layout at the moment; the builders are about to start putting in services, and Richard is going to have to okay the current layout on Monday morning (jet-lag permitting) so that the builders have it before they start putting in sockets.
Bob Taylor has finished planning the up-dates on P1, PP1 and PP2; once the "fat" stage is on P2 and working he will be able to start ordering the long lead time parts. He and Mick East are off to Brands Hatch on Thursday to get a racing start on ordering the next fat stage from Lucchesi-Carins (who did the Line-Tester stage and got it right), but there are a lot of other components (about 20K-worth per machine). Once Barry has got P2 back together Bob plans to get him to do a hunt for the missing bits of P1, P2 and PP1 - there is a few thousand quids worth of hardware which seems to have been made, but can't be found.
to: Richard Adams, Martin Viseman, Chris Warner, Peter Milne, Simon Dawes, Bob Taylor, Paul Austin, Roland Meins, Andrew Dean, Tony Slater, Ian Murray
from: A.W.Sloman
Subject: Progress to 3rd May 1991
I've done some more work on the Mark 3 Trigger Board this week. On the circuit diagrams, the Unblank Pulse Generator has been split off from the Processing Loop Generator and the Convert Edge Generator, but the digital circuit to generate the long Unblank pulses is still only half drawn.
On the tolerancing side, I've made some progress through the new Encode Edge generator, but there are still a number of critical signal paths to be worked through.
On Tuesday Martin Wiseman and I went to the Analog Devices seminar at Heathrow - I don't know what Martin got out of it, but I was inspired to try to sell Analog Devices and Comlinear the floating video output circuit from the Waveform Processor Analog; I had thought that the circuit was "obvious to those skilled in the art" but since it doesn't seem to be, it may well be patentable.
Andrew Dean has lost this week to administration and appraisals; he did try to get the second Sampling Crate going on PP1, but kept on losing the trigger; this happens with both trigger inputs on both trigger boards, so he is now hunting for a problem in the software (an auto-trigger routine that has passed its sell-by date or the like).
Paul Austin spent all week crawling through the LIP, and he was finally rewarded yesterday by finding a spot of dodgy timing on the VPC; he has bodged the clock to prove that this is the source of the noise on the image, and he is now tolerancing propagation delays along the clock and data paths to see if he can find a mod that is guaranteed to eliminate the problem.
Martin Wiseman spent Monday on the Cursor Card with Simon Dawes; they have solved the multiple cursor problem (ask for two, get a third cursor for free) by cutting and linking in seven more ICs - mostly latches - to properly pipe-line the cursor co-ordinates from the processor to the frame sync. By going over to 10-bit wide latches on the next issue of the board they can reduce the number of extra latches to two.
On Tuesday Martin was at the Analog Devices seminar, and since then he has been working on the circuits for the scanned blanking; he has finally got a purchase order signed for the 16G061 GaAs parts that will actually drive the plates, and Janine should have ordered them by lunch-time. Naturally, I've just discovered that Triquint makes a similar part...
Chris Warner seems to have spent most of this week with Steve Fisher on PP1. They have fitted the new scan coils, which work, and seem to give less distortion at low magnification. The column was initially very astigmatic, but after they dismantled the column, replaced all the apertures, and degaussed everything that would fit inside the degausser, the astigmatism went away. It isn't clear where the astigmatism was coming from - Steve suspects the screws used to pull down the scan coil plate, which were initially quite strongly magnetised.
Steve has gone through a magnification calibration procedure with the new coils, but there isn't enough gain adjustment in the system to cope with the difference between old coils and the new - we will have to change the gain of the scan amplifier board (which should only mean changing one resistor) before we can do a proper calibration.
The Scan Generator board we are using on PP1 is the first of the Leica-Cambridge manufactured boards; the layout is not the same as on the Lintech manufactured boards - it was reconstructed from the circuit diagram by a sub-contractor - and the new layout around the Line and Frame ramp sources is distinctly less compact. We have already changed the frame ramp capacitor for one with less dielectric absorption, and it looks as if there might be a similar problem with the line ramp capacitor; we should be able to tighten up the layout at the same time as we change the capacitor.
Simon Dawes has spent most of this week sorting out the mods to the plinth that he and Trevor made to get rid of vibration and stray magnetic fields; just the change from 115V AC fans to 24V DC fans changes lots of cabling. He also arranged to ship an oil mist filter to Southampton, and almost finished building up a TTLD for the Bell-Northern Canada retrofit - he will be off to Canada at the end of the month to install it.
Ian Murray has given up working on the user interface until Simon has a machine to test the interface on; Ian is now working on the Multiple Waveforms software, getting it to display properly on the Sun.
Roland Meins is working on his schematic capture package from the Cad/Cam group in California, trying to find out if it is a reasonable basis for the schematic-to-layout facility that Graham wants to see in the CAD side of the user interface.
Tony Slater claims to have spent most of this week being appraised. He put in about one day solidly on the blanking head, and perhaps another day in dribs and drabs. He has spent much the same sort of time doing mods for Bob on the Graftec. Next week he and Bob move into the new office area, which should wipe them off for most of the week.
Barry Barker has got the fat stage and the column onto P2 - they still have to fit the thin C3 and the new adaptor plate (Chris Warner has put them somewhere safe, otherwise they would be being fitted today). The next stage in the reassembly is for Chris to measure the force required to move the stage in X and Y with the chamber at air, after which they will try and pump the system down and measure again under vacuum.
Barry has got himself a decent sized surface plate - one of the two that EBMF mechanical assembly haven't got room for any more (though they have got room for the little surface plates that they got from Philips). Barry traded the other one to the tool room in exchange for the semi-permanent loan of a height gauge to go on his surface plate.
Bob Taylor is now working on the backlog of mechanical mods; he took a break yesterday to visit Lucchesi-Carins at Brands Hatch with Mick East, and came back happy that they will be
able to do justice to the fat stage. Apparently Lucchesi-Carins is in an unprepossessing building under a railway viaduct - so they will probably be in business next year (it seems that purpose-built factories are up there with new company logos and Queen's Awards to Industry as signs of incipient bankruptcy). Inside, the building is full of people who know all about precision machine and finishing - Bob was most impressed with the precision milling machine with the diamond cutter that left a perfect mirror surface on copper and perspex. John Sturrock said the place had twenty Eddie Bainbridges (a legendary machinist here, whom we made redundant a couple of years ago).
(Note added 2013/02/08. John Sturrock was the brilliant mechanical engineer responsible for the mechanical design of the Cambridge Instruments Electron Beam Microfabricator stage.
He was hearing-impaired, with a hole in his frequency sensitivity from a few hundred Hertz to a few thousand Hertz, but could lip-read impeccably in English, French and German. Mike Penberth and I had to make sure to trim back our moustaches to expose the top of the upper lip when we expected to spend any time communicating with him. The rest of our - full - beards weren't a problem, but the exact shape of the upper lip had to be visible.)
to: Richard Adams, Martin Wiseman, Chris Warner, Peter Milne, Simon Dawes, Bob Taylor, Paul Austin, Roland Meins, Andrew Dean, Tony Slater, Ian Murray
from: A.W.Sloman
subject: Progress to 10th May 1991
This is an apology for not writing a progress report; I haven't made any progress this week, and it wouldn't have made much sense to spend time yesterday working out how much progress we'd all made in the first half of the week – P2 seemed to be nearly together again when I last saw it, Paul Austin had hopes of having fixed the LIP, and Bob Taylor and Tony Slater had moved office into a rather cramped half-way house, but this scarcely justifies a progress report.
I spent Tuesday and Wednesday worrying about the risks involved in poking tungsten hexacarbonyl into the EBT chamber, as we will when we get around to using the Focussed Ion Beam for making tungsten links on specimens. I think the tungsten hexacarbonyl will dissolve in the oil in the vacuum pumps, which seemed to surprise Graham and Chris, but even if this happens it shouldn't create any unmanageable problems.
Thursday was the Linear Technology seminar at Heathrow - part of the loot now sitting in its bag on my desk are three floppy disks which could be used on our new PC — one is a filter design program, one is a demo version of PC-Spice with the Spice models for Linear Technology's op amps, and one is a program for working out noise figures for circuits built around Linear Technology's op amps. Feel free to browse.
The Trigger board got nowhere in all of this - I'll have to immerse myself in it on Tuesday morning, assuming I get back from the annual meeting of Scotch (single malt) Perspectives in Psychology on Monday evening, as planned.
to: Richard Adams, Martin Wiseman, Chris Warner, Peter Milne, Simon Dawes, Bob Taylor, Paul Austin, Roland Meins, Andrew Dean, Tony Slater, Ian Murray, Maha Mahalingham, Madhusudan Patel
from: A.W.Sloman
subject: Progress to 24th May 1991
I've made quite a lot of progress with the Mark 3 Trigger Board this week; the circuit diagrams now should include the right collection of components, although there is still some tinkering to do to get them connected and numbered absolutely correctly.
I've made some more progress with the tolerancing; the Encode Edge generator has also acquired another 100131 D-type (we get three in a pack) to prevent it clearing itself too early, but I've still got to finish tolerancing that block and I've still got the 200MHz shift register block to go.
For monitoring tungsten hexacarbonyl I've had a short form catalogue from Horiba, who seem to have a range of equipment which might do - either Chris or I will have to do a literature search on the IR and UV absorbtion spectrum of tungsten hexacarbonyl, probably via Chemical Abstracts, before we can come to any useful conclusions.
Andrew Dean hasn't made much progress this week - he is supposed to be healing up nicely (one can only suppose that the EBT was trying to get its own back on the hardware engineers that keep poking around in its guts). Steve Fisher administered as much first aid as he could remember in the thirty-odd seconds before Ozzy Marks arrived and took over.
When Andy gets back he has a new problem to sort out - Gigabit have decided to discontinue the 10G024 part that we use on the Delay Board, and while it looks as though we should be able to replace it with the 10G023, Andy will have to look into it.
Bearing in mind the distractions, Peter Milne has had a tolerable week on PP1. We even had Closed Loop working for half an hour yesterday, until we told Richard and the machine got the sulks.
The filaments are still failing, if less often - Peter has found and fixed two bugs in the software that might have been blowing the filaments, Steve Fisher has replaced the AD7581 8-bit, 8-channel ADC at G9 which monitors the filament current, amongst other things, when it hasn't blown up (it blows up quite often), and it turns out that as the Alpha-Repeater EHT set warms up it delivers more filament current even though the software is holding the control signal constant.
We may even be able to do something about the Alpha-Repeater; the one on P2 has quit, and Chris Warner is in the process of replacing it with a unit from Start Electronics, which may not have the same problem.
In amongst this Peter has got Ian Murray's new user interface up and running, and he is pretty happy with it - amongst other things Ian has given him a much more flexible system for setting up "facilitation" which may even make it user-configurable. I used the interface for a couple of minutes last night, and it didn't crash, so I'm happy with it too. The new interface code is now in the main software library — there is lots of tweaking left to do, but Ian is happy to see it in use at last.
The multiple waveform software is coming on - we can now put up to four black and white fixed cursors on the display, as well as the original colour cursor, although it is still some way off being finished.
Martin Wiseman has spent most of the week putting the new blanking board into Isis; it should all be in early next week, and with luck, on the Valid system by the end of the week. He has also made a start on writing up the final mods for the Waveform Processor - Analogue, and spent some time working on PP1; the Plinth Control card and its duff ADC is now his responsibility, and he got called in yesterday to help find out why the Sampling Crate was behaving oddly.
In fact we tracked down the oddness of the behaviour via the Trigger board, which was triggering at 240kHz on a 74kHz output from the pulse generator, until you disconnected the pulse generator from the specimen, when the Trigger board detected 74kHz as it should. We cured the problem by connecting the pulse generator ground to the chamber ground; once that was done you could connect the pulse generator up to the specimen without confusing the Trigger board, and we could get Open Loop, and - for about half an hour – Closed Loop waveforms.
Martin was expecting to do some measurements on the scan coils on PP1 while Chris and Steve put the Start Electronics EHT set on PP1 - since it has now been put on P2, the opportunity didn't arise.
Paul Austin has had a quiet week, mostly up-dating the Waveform Processor - Digital to issue 3; correcting the circuit diagrams on Metheus, and writing up the list of changes for GS Designs. This should be completed by the end of next week. He has also modded WPD 2/2 match WPD 2/1 - both boards now offer the options to
1. Disable writing to the Expected Waveform Store in Update 2.
2. Disable the Expected Waveform Store output in Update 1, so what you read from the Output Store after Update 2 is just the accumulated signal.
3. Monitor the carry output from the adder for evidence of overflow.
Paul has had some distractions - he has been reviewing image processors as bought-in replacements for the LIP, and the phone has been running hot with salesmen arranging visits to present their particular candidate. The DataCube Maxvideo 20 is apparently quite interesting - you can get a great deal of power into a singe VME slot, if the VME crate can take the weight of the board...
Chris Warner has been working on P2 (and reports that it is cooler in there this week). So far he has found that at least some of Simon's problems were caused by the failure of the Alpha-Repeater EHT set, and he has wired in the Start Electronics EHT set. It seems to work, though we haven't had an image yet. It is a bit higher than the Alpha-Repeater set, and only after we have tested it for a bit will we cut down the nylon pillars in the EHT rack, and install it properly.
Chris has also been thinking about a more comprehensive column alignment procedure - our present approach is based on the Cambridge Instruments approach, which relies on careful mechanical alignment to get C3 right before you start, and tends to leave our C3 misaligned. We get decent alignment between spot sizes at a single beam voltage, but changing the beam voltage messes things up.
Roland Meins is off to Cadence on Taiwan tomorrow – the tickets and his passport plus visa arrived back from the travel agent this afternoon, and he has a hotel booked.
When Rapid Recall showed up this week to try to sell us more ASIC’s they told us that ES2 can now provide schematics in the EDIF-2 format, and Roland is supposed to be getting these files for two of our ASICs — on of them (B6436 – Helen Watters’ Serial Interface) for which he already has GS2 files.
Tony Slater has been dividing his time between C3 mods and the new lab layout - every time I saw his screen during working hours it was displaying the new lab layout. He and Bob have been suffering a bit in their half-way house - there are only two mains plugs in the area, and all the computers and terminals are plugged into one of them via a long series of distribution blocks - so far Ron Hartley has crashed the lot once by pulling out the wrong plug, and John Sturrock crashed everybody down-stream from him by plugging in an electric heater and blowing the fuse in his distribution block.
Bob Taylor has spent most this week on the mods to get PP1 and PP2 up to P2; the parts are now 90% loaded, but we won't be ordering them until we have got images on P2, and can check stage positioning accuracy with the fat stage in place.
Bob Taylor and Steve Fisher went over the motor control rack this week to make sure that there was enough spare room in it for another motor control card to control a separate column Z motor for the ion beam column, if we ever get around to fitting it.
Steven spent most of the week on PP1; apart from swapping blown filaments and duff integrated circuits, he also had to replace the turbo-molecular pump, which had grown noisy, and change the oil in the rotary pump (it was reputedly two years old) and renew the oil mist filters on the outlet of the rotary pump.
On Monday afternoon he got the opportunity to test the half of the first-aiders course he completed before he left Marconi; apparently it was the right half because Andrew survived his care for the thirty seconds or so it took Ozzy Marks to get there from the workshop.
Simon Dawes made it to Canada, got his tools and spares through customs and found that his TTLD parts had been delivered on-site, if not to the EBT bay itself. Since then he does not seem to have needed to phone, and we expect him back next Tuesday, though he wasn't planning to spend much of Tuesday at work.
Graham Plows was seen at Semicon West earlier this week, and should be in Singapore today, and in Korea on Monday. I wonder who he is selling what now, to paraphrase an old song…
to: Richard Adams, Martin Wiseman, Chris Warner, Peter Milne, Simon Dawes, Bob Taylor, Paul Austin, Roland Meins, Andrew Dean, Tony Slater,Ian Murray, Maha Mahalingham, Madhusudan Patel, Steven Fisher
from: A.W.Sloman
subject: Progress to 31st May 1991
I've made a bit more progress with the Trigger board, but the tolerancing still isn't finished, and the 200MHz Shift Register block also seems to need an extra 100131 D-type to get the timing margins right; on Wednesday I was distracted by Richard's request that I look at the integrated circuit modem chip that Chartered Semiconductor have sent us to test, and yesterday I got exasperated with the exploding filament problem and went through the relevant bits of the Alpha-Repeater EHT set and the Plinth Control Card circuit diagrams.
Examining the Chartered Semiconductor modem chips in the EBT is not going to be as easy as it might be - the chip doesn't seem to have a reset input, nor any output that we could safely use to trigger the sampling system. I drafted a letter to Chartered Semiconductor asking for more help, which we faxed to Graham, and he apparently faxed them a bowdlerised version.
I didn't come up with anything wildly exciting on the exploding filament problem; it seems the Plinth Control card uses a feedback loop to control the voltage applied across the filament, such that the filament current is held stable, and it is by no means obvious that this feedback loop is stable, but there is no clear evidence of oscillation on PP1, with the Alpha-Repeater controller.
Of course P2 has been fitted with the Start Electronics EHT set, whose filament drive reacts to a negative control input by latching up in a state where it delivers 6A into the filament for as long as the filament lasts - we can avoid this state by breaking the feedback loop on the Plinth Control card, and without the feedback the filament drive worked fine, after Peter had tweaked the ramp-up software a little.
Andrew Dean has spent this week like last week, concentrating on healing his wounds - if the doctor okays him today he should be back next week.
Peter Milne has spent most of the week on PP1; Closed Loop appears to have been working, but signal amplitudes were very low and the signal was noisy - Peter is now testing the theory that this was the signal actually appearing on the surface of the specimen, and the much bigger signal coming out of the HP pulse generator was getting attenuated somewhere en route to the specimen surface. He is switching over to the strip-line specimen, which comes with a proper connector, so he can be sure that the signal is getting to the body of the specimen, and he has had the surface plasma etched.
Meanwhile the new user interface seems to be working fine and Peter has wrapped up the multiple cursor display software; next week he and Ian Murray plan on testing this with the multiple waveform display software.
The Sun Sparc 2 has arrived, although it isn't working yet - it needs an extra cable and a copy of the operating system. It has been renamed, along with the other Suns, after the various islands of Greece. It is Hyades; the others are Rhodes, Kos, Corfu and Crete. Spetse is going back to Sun in part-exchange for Hydra.
Martin Wiseman has spent part of this week - when he could get at our PC - putting the last of the Blanking Board circuit diagram into Isis. He will be giving copies of the circuit diagram to Mahdu next Monday. He tried to get it loaded onto the Valid system today, but it turns out that Isis doesn't put power supply pins into its net-list, where Valid does; watch this space for further developments.
When he couldn't get at the PC (Richard is aware that we would like another) Martin got on with updating the Waveform Processor - Analogue. He hasn't been able to get at PP1 to measure the scan coils’ inductances and resonant frequencies, and there is no immediate prospect of an opportunity.
Paul Austin has spent most of this week on his mods to the Waveform Processor - Digital. The first two work, and Peter proposes to exploit them (sometime) to acquire a complete S-curve in a single call to the Sampling Crate.
The third modification - to allow us to monitor for arithmetic overflow during data accumulation - is still giving trouble. Paul is now diverting the hardware averager to count overflows, rather than to average the incoming signal, but it isn't clear that it is counting the right number of overflows.
Paul has also put in a little time on the LIP and on finding a replacement for the LIP - DataCube have found us another board to think about.
Chris Warner has installed some more software on the PC - Z-Match II, the cheap microwave circuit analysis program, and its companion program, Analyser II, are both on the hard disk now.
Chris spent most of the week on various aspects of the ion beam; partly talking to Bob Taylor about getting a one off adaptor plate and some temporary fitting to get the column onto P1, partly shifting the P1 console into the lab (sceptics had claimed it wouldn't fit) so that he and Steve could think about the rewiring it for the extra electrical bits, and partly preparing his talk to the group - he is lecturing us on Focussed Ion Beams on Wednesday the 5th June at 3.00pm in the board room.
Roland Meins has reported back from Taiwan by fax with half a page of tiny but surprisingly legible handwriting; he is working happily with Cadence, and they have found that our design data for the Plessey chip has bugs in it.
Rob Coombs of Rapid Recall has rung back to say that he is working on getting us EDIF-2 files on some of the Leica-Cambridge ASICs - he was rather disappointed to find that Roland was inaccessible on Tiawan.
Bob Taylor has spent this week on the adaptor parts to fit the first Focussed Ion Beam column onto P1; he would have liked to have gone out and ordered the bits for the next fat stage, but until we get imaging again on P2, we can't test the stage positioning nor order the next stage.
Steven Fisher started off the week by replacing the turbo-molecular pump on PP1 again. This time the windings had burnt out, and the replacement was installed with all the cooling options Steve could lay hands on, including a duct added to the cooling fan to get the air-flow all the way to the pump.
Since then he has been reworking the motor-control rack on P1 to make room for the FIB column Z-motor control, which is all done bar the additional wiring and the front panel. He also spent some time with Chris Warner looking at the other wiring in the console.
He is on holiday next week, on the south coast, somewhere near Bognor Regius, and probably close enough to Crawley to pick up a new turbomolecular pump from Edwards if we lose another of the Balzers pumps...
Barry Barker has spent a happy week sorting out the mechanical side of P1 and working on the coil winding machines. He is looking forward to getting his hands onto PP2.
Simon seems to have spent most of this week sorting out his expenses from Canada - since Trevor we are being ultra-careful and it took Simon a while to find his last few receipts - but he is now getting P2 back on the air, not greatly helped by having to cope with the new EHT set.
Graham is in Korea, trying to retrieve our relationship with Samsung; in Singapore he sold Chartered Semiconductor a demonstration on the modem chips mentioned earlier - I hope
he doesn't come back from Korea with another can of worms.
to: Richard Adams, Martin Wiseman, Chris Warner, Peter Milne, Simon Dawes, Bob Taylor, Paul Austin, Roland Meins, Andrew Dean, Tony Slater, Ian Murray, Maha Mahalingham, Madhusudan Patel, Steven Fisher
from: A.W.Sloman
subject: Progress to 7th June 1991
In an effort to get Bob Taylor to read something about progress on the electronics and software, the weekly report will now have a head-lines section and a sex symbol on page three (if the report stretches to 3 pages).
○ I've got the Trigger board redesigned but not documented.
○ Andrew Dean is back and checking the Delay Board layout.
○ Paul Austin has an overflow counter for the WPD.
○ Martin Wiseman found a broken wire on P2, but is mostly Blanking.
○ Peter Milne has fixed the flashing on the multiple cursors.
○ Ian Murray has a new Motif.
○ Roland Meins is still hunting bugs on Taiwan.
○ Simon Dawes has got images on P2.
○ Chris Warner told us all about the FIB.
○ Bob Taylor is FIBing, and attending meetings about mods.
○ Tony Slater is working on mods, and not Beam Blanking.
○ Barry Barker has reassembled P1.
I think I've finished the redesign of the Trigger Board; sheets 1 and 2 of the circuit diagram will have to be updated to reflect the changes in the High Impedance Trigger/External Clock, I'm going to have to rewrite the parts list and I've got to write detailed notes for GS Designs on the constraints on the layout, but the dodgy part is finished.
Andrew Dean came back to work on Wednesday; his main preoccupation over the next couple of weeks is going to be checking out PCA’s layout for the Mark 3 Delay board, but he is also going to make an effort to get everybody informed about the reporting procedures for Hazards and Dangerous Occurrences.
Paul Austin was off on Monday, fixing his house and his car, and on Thursday, learning about TI’s latest DSP chip which would replace most of the Waveform Processor - Digital. In the meantime he has got the WPD counting over-flows (strictly, every fourth overflow) into what used to be the hardware averager. It isn't perfect yet but Peter Milne pretty happy with what it gives him.
Martin Wiseman spent Monday crawling around P2 with Simon, trying to get the EHT working - he eventually traced the problem to a broken wire in our cabling. He has spent the rest of the week on the new Blanking board, getting his circuit diagram into a form that is satisfactory once loaded on the Valid system.
Martin has been promised time on PP1 next Thursday to measure the properties of the new scan coils; he is not sure how much confidence to place in the promise.
Peter Milne has spent most of the week getting the timing right in the program that generates the multiple cursor images; it now switches between frames in the DP RAM card during frame blanking like it was intended to do, and the display no longer flashes. It doesn't yet work with wpcon, but I imagine it will eventually.
We still haven't got Closed Loop on PP1, but it is working on P2, albeit with some residual bugs.
Ian Murray has a copy of the latest version of Motif, which he is going to install on all the Suns. He has also devised a new standby screen for our Suns, which has a couple of Leica logos creeping around the display.
The new User Interface seems to be stable, and he is waiting to test the Sun end of the multiple waveform software.
The Sun Sparc 2 is still waiting for Sun to deliver the tape of the 4.1.1 operating system - it should have arrived this week, but still should be here before Roland gets back on Monday week.
Roland is still bug-hunting on Taiwan in the Cadence software (note corrected spelling of Cadence, courtesy of Chris Warner). He won't be able to get high-lighting from schematic node to layout track on the Plessey circuit until he can get the bugs cleaned out of the Plessey data.
Simon Dawes has got P2 imaging again - to get the Start EHT set to give the right filament current he had to break the feedback loop from Filament Monitor to Filament Drive on the Plinth Control Card, and to keep it giving the right filament current he had to take the cover off the EHT set, even though it didn't feel as if it was getting hot.
To get extraction voltage Simon had to get Barry to cut down the back of the extractor electrode, which was shorting to the old drift tube. New, shorter drift tubes have been drawn and made, but they weren't fitted when C3 was changed.
At the moment the minimum working distance is 16mm longer than it used to be - Tibor was going to draw a new Z-block to suit the fat stage, but he got made redundant. A new, shorter Z-block has now been drawn and made, and will be fitted on Monday.
Chris Warner had a short session on P2 with Simon, looking at the positioning performance - it looks appreciably better than it was with the thin stage. If the column is centred on a point, then driven away in an arbitrary direction and brought back to the same coordinates, the image is restored to within ten microns (it was 20 microns with the thin stage). An anti-backlash manoeuvre, ensuring that the stage always approaches the target from the same direction, should reduce this error quite a lot. More impressive, where the old stage used to wind up for about 5 micron before the image actually moved, the new stage visibly responds to single (l micron) increments from the stepping motor – it doesn't usually move exactly one micron, since the sub-stages still have to wind up a bit, but it is a lot easier to manoeuvre.
This is definitely a sex symbol, though it wouldn't do for The Sun.
♀
When Chris can get the time on P2 he will measure stage positioning systematically and document it in detail, in order to be able to tell Graham what he can offer our potential customers.
Chris also got Simon to align the column at 1.2kV, which he did quickly, and without any difficulties.
Away from P2, Chris spent quite a while preparing his talk for the group on our proposed FIB system and the existing competition.
Bob Taylor divided his week between drawing up adaptor parts to get the first FIB column onto P1, drawing up a different set of adaptor parts to allow for column-Z movement on the next system, and attending meetings about mods. This didn't leave any time for actually cleaning up any mods.
Tony Slater spent his week cleaning up mods; these have an even higher priority than finishing off the design for the new Blanking Head.
Barry Barker has pretty much got P1 back together, and hopes to have it pumping down by Monday. The plinth assembly isn't exactly complete - Trevor and Simon had been pinching bits off it for months, and it wasn't possible for all the missing bits to be replaced immediately.
to: Richard Adams, Martin Wiseman, Chris Warner, Peter Milne, Simon Dawes, Bob Taylor, Paul Austin, Roland Meins, Andrew Dean, Tony Slater, Ian Murray, Maha Mahalingham, Madhusudan Patel, Steven Fisher
from: A.W.Sloman
subject: Progress to 14th June 1991
o I'm still documenting the Mark 3 Trigger board.
o Andrew Dean is still checking the Delay board layout.
o Paul Austin has found the Burst Mode bug.
o Martin Wiseman has measured his scan coils.
o Peter Milne has got Closed Loop and Burst Mode working.
o Ian Murray is permanent.
o Roland Meins is back from Taiwan.
o Simon Dawes has written loadsa parts lists.
o Chris Warner has found some bugs in the operating software.
o Bob Taylor has completed one mod, despite loadsa meetings.
o Tony Slater has worked a full week (on the new C3).
o Barry Barker has got P1 pumping.
o Steven Fisher has got an issue 4 Plinth Control Card loaded.
I'm still working on the Trigger board circuit diagram; sheet 2 is okay (I think) and I've done most of the changes to sheet 1. The process has involved writing a draft company standard on inductors, but in the improbable absence of similar garden paths I might be able to finish it off next week.
Andrew Dean has spent most of the week checking his issue 3 Delay board layout. He was also involved in the processes of getting Closed Loop and Burst Mode working.
Paul Austin spent most of the week marking up the Waveform Processor - Digital circuit diagrams and preparing notes for GS Designs. We took the package down to Knebworth yesterday afternoon, and Bob Moore should be onto it full-time early next week. Paul is on holiday today and next week, but he rang in this morning with an after thought on the overflow counter, which Bob Moore will put in after Paul gets back, and before GS Designs start the 100% check on the revised layout.
Paul took time off from documenting the WPD to find and solve the problem with Burst Mode; he managed to trace the problem back to the counter that keeps track of the number of words transferred to and from the Waveform Interface on any one pixel - it wasn't clearing when it should. The system software was loading it as if it was part of the scratch-pad memory (which inverts its data between write and read) so the counter wasn't counting down to zero and disabling the relevant bus. The bug was intermittent because if the last byte of Taxichip data was all zeros – as it often was - the bus would be just as disabled as if the counter had counted down to zero.
Martin Wiseman has passed his new Blanking board circuit diagrams on to PCA. After being frustrated by the bugs in Isis while drawing the Blanking board, he is now being frustrated by the slowness of Metheus while he up-dates the Waveform Processor – Analogue.
He wasn't frustrated in his ambitions to measure the inductances of the new scan coils on PP1 - he got his time on PP1 yesterday, as promised, and measured the inductances, which are roughly halved. This is enough to make the scan amplifiers unstable at high magnifications with the existing compensation; Martin is now starting to think about revising the compensation.
Peter Milne spent the week getting the software to produce Closed Loop and Burst Mode reliably - getting them running reliably isn't the end of the job, but it is a serious milestone; something like the end of the beginning.
Peter's next job is to work out how to set up the system so that Burst Mode images have decent brightness and contrast - my original system design assumed that these could be tuned up in the LIP, but the LIP is currently programmed to display Burst Mode images just as the Sampling Crate made them, and we haven't the time to reprogram the LIP. There are going to be a lot more of these sorts of problems.
Right now, Peter is installing Sun0S 4.1.1 on the Sparc 2 - it finally arrived this morning, and should be ready for Roland when he gets in on Monday.
Ian Murray has become a full-time member of Leica-Cambridge; Personnel will put him through their introduction to Leica-Cambridge on Monday, as they do with all new starters.
Ian has installed the latest version of Motif on all the Suns - the 2/360s, the Sparc 1+ and the Sparc 2, so we now have identical versions of the program on all the machines, for the first time. There were some teething troubles - Chris Warner was definitely grinding his teeth before they were all sorted out - but the nasty medicine was probably good for us.
Roland returned from Taiwan this morning, with his collection of stuffed and mounted bugs. He has found and fixed the bug that was causing his crashes; he now knows his way around SYMBAD and should be able modify it more easily than before. And he has seen the Cadence CAD package running under Sun0S 4.1.1 - according to Cadence it didn't run under 4.1 because 4.1 was full of bugs.
Simon has spent about half the week in the usual way - getting images on PP1 and P2 for the rest of us to lose. Most of the rest of the week went on preparing parts lists for the EBT service centre spares kits, and up-dating the cable list.
He has persuaded Start Electronics to come and fix their EHT set - they will be here at 2.30pm on Wednesday.
Page 3
Chris Warner has done a bit more work on the P1 conversion this week, but most of his time has been spent trying to do measurements on stage positioning, and on the relationship between magnification, working distance, C3 current, EHT and extraction voltage for a focused image. Mostly he lost the image before he could measure anything, and on Wednesday he was here until 7.30pm waiting to get the ion pump going so that Thursday's successful attempt to get Burst Mode going could start at 8.00am.
He expects to actually get his measurements sometime next week, now that he has exposed the latest generation of bugs to the attentions of the fixers.
The beam blanking hardware design is in trouble – it includes a metal part which Ozzy isn't sure that he can make. Mike Dean has recommended getting Wesley-Coe to quote, but if they can't hack it, it's back to the drawing screen.
Bob Taylor has spent most of the week going to meetings about mods. He managed to find a free hour to actually complete one mod, but that was the only one this week. He also got the last of the bodge ion beam parts loaded, which pretty much exhausts the list of things that could be loaded; all the other parts that ought to be loaded are held back until the relevant mods are completed.
Tony Slater has worked a full week for the first time for ages; he spent the week putting the mods onto the drawings for the new C3.
Barry Barker has got P1 pumping, with a gash anode section to block the hole. To get a C3 they are going to have salvage the badly scored C3, by turning it down below drawing tolerances. It will be okay for in-house use.
Barry is now working on the fibre-optics harness for PP2; we are going to need it fairly soon.
Steven Fisher has finished loading the first Issue 4 Plinth Control Card. He had to make up a new cable 82 to match the connector actually fitted to the new card, and he is now rewiring the plinth power supply to put the right voltages on the right pins for the Issue 4 board.
When this is all done the new board will be fitted to PP1 and Chris Warner will get the old Issue 2 board for P1.
(Note added 2013-02-09. P1 and P2 were research prototype machines. PP1 was the first production prototype machine.
Marketing would eventually have sold all of them if we'd put the machine into production, if experience with the electron beam microfabricators was anything to go by, where engineering and production always wanted to have a working “reference” system on-site, but some customer would always want to get their hands on a working machine earlier than production could produce a new one.)
to: Richard Adams, Martin Wiseman, Chris Warner, Peter Milne, Simon Dawes, Bob Taylor, Paul Austin, Roland Meins, Andrew Dean, Tony Slater,Ian Murray, Maha Mahalingham, Madhusudan Patel, Steven Fisher
from: A.W.Sloman
subject: Progress to 28th June 1991
o I've got through to sheet 4 on the Trigger board - 3 more to go.
o Andrew Dean is designing a new column simulation box.
o Martin Wiseman is waiting to resume curing scan distortion.
o Peter Milne has got multiple waveforms going with substrate subtraction.
o Ian Murray has framed his windows.
o Roland Meins has fixed the lock-step bug.
o Simon Dawes has been demonstrating against unreliability.
o Chris Warner is immersing himself in the flood gun.
o Bob Taylor and Tony Slater are on holiday.
o Barry Barker has got the wafer stage assembled.
o Steven Fisher has exercised all the motors on the assembled wafer stage.
Despite losing most of Monday and Tuesday to the presentation to the U.S. Sales guys, I've now marked up sheets 3 and 4 of the issue 3 Trigger board circuit diagram, and written the accompanying layout notes - in the process I've found and corrected a couple of minor errors (mainly missing termination resistors) which I am going to correct on Metheus as soon as John Gibbens gets it running again.
The Trigger board ought to be down at GS Designs by the end of next week.
Andrew Dean has gone through the Delay board design for delay by events and look-ahead mode; delay by events ought to work, and Peter is writing the software that should let Andrew test it. Look-ahead mode is going to need a firmware change, and Andrew is going to work through that again before irretrieveably blowing one of our ECL PALs.
Andrew also put in his time on the presentation; today he is working on the Mark 2 column simulation box.
Martin Wiseman has finished up-dating the Plinth Control card circuit diagrams, on Metheus, so that Steve has got a set of circuit diagrams he can mark up. Before Martin got kicked off PP1 he had got about half-way to curing the scan distortion - the TV rate images are now okay, but the slow scan images are still bent. Martin has also put in his time on the presentation, helping with the Burst mode images, and he also found time to get the second Waveform Processor - Analog back into the same state as the first one; this meant undoing the unsuccessful mods he was trying out around Christmas.
Paul Austin has had a fairly successful week on the overflow detection mod for the Waveform Processor - Digital; at one point he had it working perfectly for Single Encode per Pulse operation, and pretty well for Multiple Encodes per Pulse, but his attempt to solve the residual problem introduced another fault, which he is still trying to sort out - he doesn't feel very successful at the moment.
Peter Milne has demonstrated the multiple waveforms package with substrate subtraction to the U.S. Sales guys, and very impressive it is too. The Burst Mode autogain routine seems to be giving pictures which aren't quite as contrasty as an operator would aim for, but with quite enough contrast, nicely centred at mid-grey, to let you see what is going on.
Ian Murray is now in Venice, on holiday. His achievement for the week was to merge his Motif screens from the User Interface, with the Leica border from Graham's desk-top publishing package, so that he could print out the screens as overheads, without any subsequent cutting and pasting. I can't wait to see his holiday snaps...
Roland Meins has been working on the colour set-up for the CAD package for most of this week; he had to find a subtle bug in the lock-step function after Monday's demonstration, but that is fixed now. He has promised to chase Rob Coombs of Rapid Recall for the promised EDIF2 schematics for one of our ASICs.
Simon Dawes has had a busy week demonstrating. The machine has invented a new fault - when the stage is moving to the right the rubber boot gets trapped, jams the motor, and jacks the sliding seal plate off the top of the chamber, for an instant air-admit. Disconcerting.
Simon is looking forward to the new, lower Z-block. At the moment he is operating with a 6mm working distance, which is way too high. It means that we need a high extraction
voltage to keep the local field effects even moderately suppressed, which has meant flash-overs …
Chris Warner checked out his scheme for sorting out working distance against a range of new measurements, and came up with working distance within +/-1mm of the right value for working distances up to 7.5mm, getting proportionately worse at longer working distances. It looks as if there is a systematic component in the residuals, so he may be able to do a little better.
Chris has also spent time on the presentation, and he is now having another go at the theory of the flood gun for the FIB.
Barry Barker assembled the wafer stage early in the week, and Steven Fisher was able to demonstrated it with all four motors (X, Y, Z and rotate) operating under computer control
- a first. We still have to repeat this with the wafer stage under vacuum.
The column sections for PP2 are now at the heat treatment stage, and Barry hopes to get them next week. The sliding seal plate for P1 is now being plated, and that too should be available next week.
to: Richard Adams, Martin Wiseman, Chris Warner, Peter Milne, Simon Dawes, Bob Taylor, Paul Austin, Roland Meins, Andrew Dean, Tony Slater, Ian Murray, Maha Mahalingham, Madhusudan Patel, Steven Fisher
from: A.W.Sloman
subject: Progress to 5th July 1991
o I fixed Trigger Board 2/2, but issue 3 still drags on.
o Andrew Dean is checking the corrections to the Delay board layout.
o Martin Wiseman is on holiday until Monday.
o Peter Milne is trying to test the corrected Syntel software.
o Ian Murray has done some more desk-top publishing.
o Roland Meins has installed a new release of the DRE software.
o Simon Dawes is off to Motorola next Friday.
o Chris Warner is sub-editing the paper for Graham's next conference.
o Bob Taylor is on holiday.
o Tony Slater has been beam blanking.
o Barry Barker has got the modified seal plate into P1.
o Steven Fisher has got the ion pump onto PP1.
I'm still writing layout notes for sheets 5 and 6 of the issue 3 Trigger board; it might have got down to GS Designs this week if the buffer on the 800MHz oscillator on Trigger
Board 2/2 hadn't died - it took me all of Wednesday afternoon to decide that it was the 10G012 buffer that had died, and all Thursday morning to replace it and to remake the five wire links and three coaxial cable links that make the circuit work.
Today I received an up-to-date data sheet (revision G, but still preliminary) for the Bt-622 delay circuit, as used on the issue 3 Trigger Board, and I am going to have to make minor changes to sheet 3 to accommodate the sloppier tolerances.
Andrew Dean has finished his design for the Mark 2 Column Simulation Box, and is now waiting for someone to build it - probably Gerrard.
Andrew is now checking over Tom MacNamara’s corrections to the issue 3 Delay board layout; he has corrected the Metheus drawings to fix the faults that Tom and Jack found when they were checking the layout against the circuit diagram.
Andrew was also involved in the Test Strategy meetings; he is working on a card to plug into the front of the WPD to give the test engineers easy access to the Encode pulses.
Martin Wiseman should be back from Yorkshire on Monday.
Paul Austin thinks he has finished the overflow detection mod for the Waveform Processor - Digital; the revised drawings have been posted down to GS Designs, who have pretty much finished changing the layout. We should have pen plots from GS Designs for checking sometime next week.
Peter Milne has spent most of this week trying to test his revised software for the Syntel; so far he hasn't been able to get it to work. Since the Trigger Board 2/2 didn't fail suddenly, he had an interesting Tuesday trying to separate the problems with his software from the problems produced by a progressively more erratic 400MHz clock for the Delay Board.
Ian Murray has made it back from Venice; he has spent part of this week helping to facilitate Graham's trip to Como in September, by producing progressively prettier screen dumps for Chris Warner to put into the paper Graham is presenting at the voltage contrast conference at Como.
Ian and Peter have also put in some time making the stage mimic move more smoothly.
Roland Meins has got a new release of the DRE software (whatever that is) up and running on his Sun. He has put in some more time on the Plessey data, without coming to any conclusion - Richard is chasing Plessey to see if we can get more data from them.
Roland has chased the EDIF schematics on our ASIC with Ron Coombs, but Ron Coombs has passed him on to another engineer.
Simon Dawes has had another busy week with demonstrations; he gave three on Wednesday, for Ian Few of Northern Telecom (Canada), for E-beam final test, and for E-beam production.
Yesterday he was mostly acquiring Burst Mode images with Chris for Graham's paper. Next Friday he is off to Motorola, to fix Steve Morris's EBT1 and Dave Perrin’s - he hopes to have both going within a week.
Chris Warner has spent some time with Tony Slater while they sort out the detail of the Blanking Head design; Wesley-Coe have found a sub-contractor who can make their special pins,
so that is moving forward again.
Chris has spent most of the week putting together stuff for the paper Graham is going to present at Como - it was due on the 14th June, but Chris and Graham had both contrived to fail to notice this and were scheduling their effort for the end of August.
Tony Slater has had the usual distractions from Beam Blanking - the parts for the new C3 are starting to come back, and the problem that got blamed on the rubber boots last week, is now blamed on the mis-positioning of the cut-out in the chamber lid.
Barry Barker has put the modified sliding seal plate onto P1, ready for the FIB column; PP2 is still waiting for its column sections. Barry has put in some time this week designing a chamber "barrow" - basically the bottom half of the EBT plinth - with vacuum pumps and the stage actuators, where they can pump down the column and chamber, and perhaps set up the Magnascales. When the chamber and column are set up, the spare set of rails on the barrow will eventually be lined up with the rails on the new plinth, and the chamber and column slid across.
Steven Fisher has put the ion pump onto PP1, and has put the Ion Pump Monitor into the controller rack, but he hasn't get the HT connected in yet - there is still some debugging to do. The new Plinth Control Card is working on the bench; John Ford has been looking at it this week to familiarise himself with the design.
Despite his name, John Ford is interested in Lotus’s – he has been restoring an Elan for a few years now.
to: Richard Adams, Martin Wiseman, Chris Warner, Peter Milne, Simon Dawes, Bob Taylor, Paul Austin, Roland Meins, Andrew Dean, Tony Slater, Ian Murray, Maha Mahalingham, Madhusudan Patel, Steven Fisher
from: A.W.Sloman
subject: Progress to 12th July 1991
o I took the Mark 3 Trigger board down to GS Designs this morning.
o Andrew Dean has signed off the Mark 3 Delay board; it goes off for photo-plotting early next week.
o Paul Austin has solved the interlace problem on Burst Mode.
o Martin Wiseman is working out how to add safety to the Plinth Control Card.
o Peter Milne has almost finished testing the new release of the Syntel software.
o Ian Murray has been standardising the file formats.
o Roland Meins has produced a resizeable CAD window.
o Simons Dawes is sunning himself at Austin, Texas.
o Chris Warner has been PEEKing at the Blanking Head.
o Bob Taylor has been sorting out drop-offs.
o Tony Slater has been sorting out drop-offs, and the Blanking Head.
o Barry Barker got several new Z-blocks onto P2.
o Steve Fisher has been hunting vac leaks on P2.
I took the marked-up circuit diagrams for the issue 3 Trigger board down to GS Designs this morning, with 21 pages of layout notes; Gary Askew thinks he should be able to
complete the revised layout by the end of August.
The issue 2 Trigger board developed another fault yesterday, but it was just a dry joint, and took less than an hour to find and cure.
Andrew Dean has signed off the issue 3 Delay board layout, and it will be sent for photo-plotting as soon as the package has been checked by Claire Brant, probably Monday or Tuesday next week.
Andrew spent most of this week on PP1 with Paul Austin finding faults in the Sampling Crate - as well as the fault in my Trigger board, they found a fault in the Delay board; the prototype buffer daughter board had given up, so Andrew replaced it with one of the PCB versions.
They also found a fault which could be cured by changing the software; there wasn't enough delay between the last sample in the Sampling List and the entry that generated the SLC, so that the Waveform Processor was going into the processing loop before the data from the last sample had been written into memory.
Paul Austin is pretty confident that the overflow detection mod is working; he has had some trouble testing it this week (see above) but he is hopeful that when the SLC occurs in the right place his residual errors will go away.
Bob Moore should have finished the issue 3 layout by now, and will be producing check plots on Monday, so we hope that the overflow mod is okay.
Meanwhile Paul has found and fixed the interlace problem - he had found that inverting the Interlace signal on the Naffbus Interface didn't help, but when he found that there was a second, different signal of the same name on the board, and inverted that, he cured the problem.
Martin Wiseman now believes that Scan Amplifiers built to the Lintech circuit diagrams should be adequately compensated to deal with the new scan coils, but he still has to check the real Scan Amplifiers against the circuit diagrams, and he can't do that until Paul gets off PP1. In the meantime he has been working on the problem of adding safety interlocks onto the Plinth Control Card, not to mention the problem of hooking up the existing safety interlocks so that they do something.
Peter Milne has almost finished testing his revised software for the Syntel; it was going pretty well when P2 went down with a scintillator flashover, and stayed down as various Z-blocks got fitted and rejected, then pumped down very slowly, until Steve found the vacuum leak.
In the meantime Peter has been putting in some time on generating labelled cursors - I've seen them, but they have this tendency to bleed across the image, which is probably another aspects of the LIP-synch problem.
Peter has also been giving some thought to software schedules and delay by events (that is events as defined by external clock edges into the trigger board, rather than visits by sales engineers and the Leica board).
Ian Murray has spent the week standardising formats; his ambition is to get the User Interface to digest all of the file formats it is likely to see - I talked to him a bit about "delay by events" which seems to have been an irresponsibility.
Roland Meins has now got the CAD package to give resizeable windows, like all the other major windows on the User Interface ("pop-ups" are not resizeable).
His pursuit of the EDIF2 schematic data for our in-house ASICs has run into something that sounds a bit like the run-around; Ron Coombs at Rapid Recall passed him onto another guy, who is on holiday - Graham is interested, and since his was one of the crucial signatures on the original ASIC orders it may be that he can invoke more vigorous cooperation.
I haven't heard anything about Simon Dawes’ progress with Motorola's machines at Austin, which suggests that they haven't broken anything new; he shouldn't be having any trouble with his secondary aim, to increase his chances of getting skin cancer.
Chris Warner spent most of the week on P2, making a new scintillator and joining in the fuss about the Z-blocks. He found time to talk to Tony Slater about the blanking head; alumina has been replaced by PEEK as the dielectric of choice - PEEK has about a third of the dielectric constant, which means lower stray capacitances; we aren't too sure about its high frequency dispersion and dissipation, but there probably isn't going to be enough of it to matter.
Bob Taylor came back from Portugal to face an interesting collection of drop-offs; the welding shop have got the sliding seal plates and the plates should be going back to the sub-contractors for re-machining early next week.
to: Richard Adams, Martin Wiseman, Chris Warner, Peter Milne, Simon Dawes, Bob Taylor, Paul Austin, Roland Meins, Andrew Dean, Tony Slater, Ian Murray, Maha Mahalingham, Madhusudan Patel, Steven Fisher
from: A.W.Sloman
subject: Progress to 26th July 1991
o I'm still working on a parts list for the Mark 3 Trigger board.
o Andrew Dean has done some more work on safety.
o Paul Austin is checking the WPD-3 layout; six errors so far.
o Martin Wiseman is getting on top of the issue 4 Plinth Control Card.
o Peter Milne has got the new Syntel processor board up, if not running.
o Ian Murray has set up a threshold program for the logic analyser job.
o Roland Meins has finished his improvements to the CAD user interface.
o Simon Dawes fixed the other EBT1 at Austin as well.
o Chris Warner finally located the vacuum leak on P2.
o Bob Taylor has been modelling the EBT2000 on Aries.
o Tony Slater has detailed most of the blanking end of the Blanking Head.
o Barry Barker is on holiday.
o Steven Fisher has got the second issue 4 Plinth Control Card under way.
I've continued working on the issue 3 Trigger board; I'm now reasonably happy with my model for the transistor in the 800MHz local oscillator, and propose to tune out the inductive reactance at its emitter with an eighth wavelength (3.0cm at 800MHz) 35R transmission line. The revised circuit diagram includes two new resistors which should allow me to correct fairly substantial differences between the model and reality (up to a factor of two at least), but the circuit is pretty insensitive to small differences and ought not to need trimming once we have got the model right.
In the process I became aware of the limitations of the formulas for strip-line impedance and propagation delay in the GigaBit data book; both my microwave books have better formulas, which give about twice the width and 9% more propagation delay per unit length, for a 35R microstrip.The errors are much smaller at 50R.
I've spent rather more time working on the parts list – I've now just finished the resistors, having found six resistors (out of 453) that I had to renumber - and I'm going through the IC’s so that I can add the extra decoupling capacitors before I do the capacitors. It is extremely tedious.
Gary Askew has nearly finished creating the net-list for the issue 3 board; he modified the existing issue 2 netlist for sheets 1 and 2 of the drawing, but is generating new net-lists for sheets 3 through 6 - sheet 7 isn't much changed and he expects to get that by modification.
Andrew Dean has had my stuff accepted by Stores; he also returned several more thousand pounds worth of DRAM chips from Paul Austin's stock which haven't been accepted yet.
Apart from this yuppie high finance, Andrew has been concentrating on safety and interlocks - Tony Slater is already moaning about the number of mods Andrew has generated, and Andrew expects to generate some more, now that he is getting some confidence in his understanding of the IBM Safety Manual.
As usual, Andrew has also been involved in sorting out this week's crop of bugs; the "ticking vacuum" problem on P2, and the "invisible violations" on the TaxiChip link to the Plinth Control Card (PCC). In the process he added a switch to one of his Delay boards so that he could reliably generate violations on the Delay board.
He also got the opportunity to whip the EHT unit out of P2 to trace the 0V and ground connections, which proved educational — the scintillator EHT return was taken to the wrong 0V.
Paul Austin spent the first three days of this week modifying the first of the issue 4 Timebase Interface cards to include the Burst Mode mod, and a new mod to allow us to monitor violations detected at the plinth and signalled back to the Timebase Interface as Taxichip commands. He also fixed a couple of errors in the artwork. The board is now working; the other issue 4 board will be modified in a couple of weeks.
Paul has spent the rest of the week checking the WPD-3 netlist (from GS Designs, on a 5.25 inch floppy) against his circuit diagrams. Now that he has got the GS Designs net-list on the Vax he can use the "Find" command in the editor to pick any arbitrary pin on the circuit diagram, and find out what else the net-list thinks is connected to that pin (node). This is a lot faster than doing it by hand. He has found a couple of errors.
He is going to check the tracking down at GS Designs; when the layout is displayed on their computer system you can get it to high-light all the tracking (a node) attached to a specified pin on a specified device, which should be a lot easier than tracing the tracking through layers of pen-p1ots.
He expects to make a separate visit with Maha to check the ATE pads; again it should be easier on the computer at GS Designs. Maybe we should put CAD Star on Chris Warner's 386?
Martin Wiseman is still modifying the issue 4 Plinth Control Card (PCC) on PP1. The door interlock and the 68040 access mods are completed and working but Martin has now got
involved in the "invisible violations" problem; there is a PAL on the board that should react to Taxichip violations detected on the PCC by sending back a Taxichip command – it doesn't.
Moreover, the same PAL handles the transmission of status information back to the Timebase Interface, and doesn't allow enough time for the data bus to settle (25nsec versus 57nsec worst case settling time), which may explain some of the erratic feedback (gun vacuum, filament current) from the PCC.
Martin is now engaged in redesigning this PAL; he has implemented some quick and dirty fixes on the PPC, which now works intermittently, but he needs to redesign the PAL to get a solution that we could use in production.
Peter Milne has been working with Martin Wiseman on getting the issue 4 PCC working; he has to reverse a lot of addresses to match the PCC spec (which the issue 2 PCC didn't) but he is rather relieved to find that it was the PAL on the PCC which was producing the "gun vac Ready" errors.
On the long term stuff, Peter has got the new Syntel processor card (VM022) up and running well enough to find that the G-64 and and VME I/0 blocks have moved to different addresses. He has got the new cross compiler running on a Sun and it is very much faster than the old compiler (except when Roland crashes the Sun) and he is trying to get the new Microware OS-9 package running so that he can use the network.
Microware are not being very helpful - they offer telephone support for 90 days after purchase, and Peter couldn't actually start installing the new operating system for six months, so they are being difficult about starting the 90 days from now.
Ian Murray has been beavering away on bug fixing as usual; he has also put a threshold process into "Results" against the eventual logic analyser, tidied up the Burst Mode and the wafer stage control windows, and talked to EBMF about the rationale behind three-point chip alignment (tilted chips).
Ian has tried importing long simulated waveforms from Roland's CAD package, but it showed up a need for dynamic memory allocation in the file handling system, which we haven't got yet.
Ian is proposing to put a permanent Leica logo in the top right hand corner of the screen to give us some "commercial identity". It sounds like a good idea.
Roland has finished polishing the CAD UIF for the moment - he is now waiting on feedback from Simon. Nothing much is happening about schematics; the CAD/CAM group schematics package hasn't appeared, and Rapid Recall haven't rung back about the EDIF-2 schematics for our ASIC.
Simon Dawes is now back from Austin, having fixed both EBT1s and tried 15 varieties of American beer, fourteen of which he liked. As soon as he got back he was sucked into the "ticking" problem with the P2 gun vacuum, and has now replaced the original firing unit (which he had never liked) with a new one.
The new one hasn't "ticked" yet, but it isn't properly out-gassed and may yet tick at lower pressures (though it seems unlikely).
This has temporarily relieved Chris Warner from nurse-maiding P2, which took up a lot of his time this week, what with finding a chart-recorder and hooking it up to the ion pump to record the "ticks" in the first place - roughly three seconds of tick every five minutes, quite regularly.
He also spent time on the 386, using Super Project to lay out the FIB project timetable; he locked up Super Project once, which is standard for a new user.
And he spent quite a lot of time with Tony Slater arguing out the detail of the Blanking Head.
Tony Slater has spent almost all of this week detailing the blanking end of the blanking head; assuming that we can find a high vacuum Araldite to lock everything in place.
Mike Willis (who has become our plastics expert, by virtue of a course on the subject) was persuaded to call CIBA-Geigy for a data sheet on the subject on Tuesday; it still hasn't
arrived, so we are pressing on in ignorance.
Bob Taylor has spent most of the week putting the EBT-2000 design onto Aries; this information can then be used to produce 35mm slides of the structure from any angle, with any rendering or shading you care to specify, which should help the publicity material no end.
Barry Barker is on holiday.
Steven Fisher got a source image on PP1 before he passed the first issue 4 PCC over to Peter and Martin. Since then he has wired up another mains distribution unit so that Mike Sheldon (from EBMF test) can start testing the second issue 4 PCC. John Ford has gone back to EBMF test; something about the EBT seems to have worried him.
Steve also put in some time working out how we got issue 1 DC Motor Drive Cards, when Richard ordered issue 2; the answer was that Production pay no attention to issue numbers, and the issue 2 boards hadn't been released when production kitted the boards. Be warned.
to: Richard Adams, Martin Wiseman, Chris Warner, Peter Milne, Simon Dawes, Bob Taylor, Paul Austin, Roland Meins, Andrew Dean, Tony Slater, Ian Murray, Maha Mahalingham, Madhusudan Patel, Steven Fisher
from: A.W.Sloman
subject: Progress to 2nd August 1991
o I found and fixed a bug in the Mark 3 Trigger board design, but the parts list still isn't finished.
o Andrew Dean has done some more work on safety.
o Paul Austin has produced the issue 5 Timebase Interface.
o Martin Wiseman is still immersed in the Plinth Control Card.
o Peter Milne has archived images from the Syntel.
o Ian Murray has put a Leica-EBT2000 logo on the Sun screen.
o Roland Meins has started work on the net-list window.
o Simon Dawes has got P2 imaging again.
o Chris Warner has got the Plinth Control Card wired up on P1.
o Bob Taylor is finishing off the detail drawings for P1.
o Tony Slater has moved onto the Blanking Head bellows section.
o Barry Barker is assembling two new TTLDs.
o Steven Fisher has got PP1 imaging again after the move.
I've continued working on the issue 3 Trigger board; the bulk of the capacitors are now on the parts list, including the decoupling capacitors, where I felt obliged to negotiate a little re-numbering with Gary Askew. While I was waiting to talk to him I put in a little time on the Trigger board Spec B and found a bug in the design, which would have stopped us using External Clock inputs from 62MHz to 1GHz.
The design would almost certainly have worked when Bob Ward thought it up, but since then Peter Milne and Andrew Dean have taken to generating Trigger Disable only at the end of a sub-list, rather than taking the signal high at the start, which means that the divided External Clock would now be disabled throughout the sub-list.
The modification to solve the problem was minor, wiring in two unused GaAs NOR gates on sheet 1, and swapping two PAL outputs on sheet 7, but it does mean that Peter will have to change the software driver a little - if he wants to count divided External Clock he will have to set LOAD by writing to 851C, and if he wants to sample with Delay by Events on a divided external clock he will have to clear LOAD by writing to 850C.
Andrew Dean has continued to work on the safety of the EBT; he now has a list of things that have to be changed, though few of them have been formulated as mods yet. At present he is looking at the EHT cables and connectors.
The Mark 3 Delay boards are now being made at Printech; they should be here on the 12th August, which gives us a week to kit them. Andrew has an up-to-date parts list, and is going off for a fortnight’s holiday today. That's timing.
Paul Austin has finished checking the Mark 3 Waveform Processor - Digital. He and Maha went down to GS Designs on Wednesday to check out the critical tracking and the test pad locations respectively. Paul found some split lines and misplaced terminating resistors, which Bob Moore fixed on the spot. Maha was content with what he saw. Paul also got Bob Moore to do a last minute mod to get "violation monitoring" and "ISG" to work - the PAL that generated the TaxiChip strobes didn't know about either condition, and the relevant signals had to be wired up to its inputs.
Paul has also up-dated the Timebase Interface circuit diagrams to issue 5 and is giving them to PCA this afternoon; they can work on the layout while he spends next week on holiday.
Martin Wiseman is still modifying the Plinth Control Card (PCC); he has got his new PAL worked out, programmed and fitted, so the status information is now coming back reliably, and violations detected at the PCC are now signalled back to the Timebase Interface (though nobody has detected one yet).He has also fixed the bug that made it possible to air-admit onto a hot filament.
This week's progress has been on squeezing some of Andrew's safety mods into one 22V10 PAL, including the logic to allow us to move the chamber open/close control to a pair of push buttons on either side of the chamber, both of which have to be pressed to move the chamber. Next week he expects to hand-wire a daughter board to get this onto the issue 4 PCC with the column safety switch and the door interlock.
Peter Milne has spent most of this week setting up the facility to archive images from the Dual Port RAM (DPRAM) card on the Syntel; he has written the software to disguise the DPRAM card as a disk, with the images as files, and can use some of the standard file-transfer utilities to move images across the Ethernet to the Sun (amongst other possible destinations). It takes 5 seconds to pull an image out of the Syntel, and 20 seconds to put it back. At the moment it takes 2 minutes of typing to do either, but with the new version of 0S-9 it will be possible to automate the process.
This week Microware have been persuaded to be much more helpful about the new version of 0S-9 and not only have come through with the 90-day telephone support, but also are promising a free up-grade, so Peter is planning to dive into the up-grading software next week.
Ian Murray has been working with Peter on the image transfers, and has managed to transfer them on to Graham's Microsoft Word desk-top publishing package so that they can be printed out in black and white on Graham's laser printer; Graham wants to incorporate colour images in his publicity material, and Ian is not yet sure that we can make this work with a printing sub-contractor.
Ian has also been working on the User Interface, and has a version with the Leica logo in the top right hand corner, with "EBT2000" right next to it in slightly less ornate lettering. This version also offers much simpler control windows, with most of the control buttons buried on the next window down.
Roland Meins is now working on the net-list window for his CAD interface, so that a node on the net-list can be high-lighted with the corresponding track on the layout (or on the schematic if we ever get one to display).
The CAD/CAM software group still haven't come up with the schematics package - it isn't even in England yet – and nothing has happened at Rapid Recall about the EDIF-2 schematics for our ASIC.
Simon Dawes got P2 imaging earlier this week, in time for the photo session, and hasn't been seen since. He and Chris Warner had some fun in the process; the scintillator was flashing over due to a crack in the Perspex in the TTLD, the TTLD was covered with vacuum grease, and there was a broken wire in one of the stigmator coils which Simon unwound and rewound by hand; the reassembled stigs aren't entirely satisfactory, but Chris is as suspicious of the wiring as of the rewound coil.
Chris Warner has spent most his week getting the PCC from PP1 hooked up to P1; everything except the TaxiChip links are now connected - the TaxiChip links will need new cables
made.
He has contacted CMS about the FIB column, without getting a satisfactory response - he and Graham are paying CMS a visit next week.
My optimism last week about the source of the vacuum "ticking" on P2 was misplaced; the gun was still ticking on Monday. Allan Walley has come up with another hypothesis - flashovers in the 7kV connector to the ion pump - but we haven't tested that yet.
Bob Taylor hasn't done any mods this week either; now he is working on detailing the FIB column housing for P1 - we are going to use the fast, expensive PMT we use on the EBT column for the FIB column detector, because this makes for the easiest design.
Last week's work on Aries was less effective than hoped - the 35mm slide maker can't use the Aries video and Richard is chasing other photo-record units.
Bob has found a way to save £85 on the rodless pneumatic cylinders that move the stage, by mounting a solenoid off the cylinders; it also makes the system more flexible (against Graham's next inspiration).
The Test Strategy meetings have accepted and improved Barry Barker's "barrow" for setting up and testing the column, chamber and Sony scales off the plinth, then transferring the whole assembly at the last moment.
Tony Slater is still detailing the Blanking Head - he has got to the bellows section now. Production don't want to touch the components for the blanking end of the assembly - Rees-Wilson said they were too small and too precise, so Mike East is putting them out for sub-contract.
The vacuum behaviour of the adhesives we might use to put the stuff together is still a mystery; CIBA-Geigy sent us their standard package, which says a lot about over-lapping your joints, and nothing about vapour pressure, while Loctite talked about "virtually zero" outgassing, which is unhelpful.
We are going to buy some Araldite 2001 and Loctite 406 and see what they do to our gun vacuum - Chris proposes to put each sample on a 1cm disk, assemble it into the gun and pump
down.
Barry Barker has been assembling two new TTLDs all this week. While he has been waiting for the glue to dry he has put the ion-beam column housing onto P1, and cleaned up some of our gear for the photo-call.
He has a new scan coil in his cupboard, waiting for a Scan Coil PCB so that he can finish wiring it up - Chris Warner says that the PCB has been ordered.
Steven Fisher has been mainly occupied in moving PP1 - fortunately they got it moved before the rain came through the roof. Since they were moving it anyway, they rewired the console, but despite this Steve got an image out of it last night, but with the P1 gun. And the software said that it took 2.9A to get the filament emitting ...
(Note added 2013-0212. The TTLD is the “through the lens detector” which collects the low voltage secondary electrons emitted from the integrated circuit that you are looking at, which is – obviously enough – sitting below the third and final magnetic lens in the electron microscope column – and uses a series of grids to persuade the electrons to travel up through the bore of the final lens into the space above it, where they are exposed to the – crucial - “filter grid” which is set at a voltage about three volts more negative the voltage at the point on the integrated circuit being imaged (at the instant you are imaging it) which lets about half low voltage secondary electrons get through to see the roughly 10kV positive voltage we put on the outside of the scintillator, leading them to make a suicidal dive into the scintillator, where they generated a small flash of light, which we picked up with a small, fast and expensive photomultiplier tube.
Mike Penberth reckoned that we didn't get more than one photon into the photomultiplier tube for every electron hitting the scintillator, as in every other Everhart-Thornley detector.
http://en.wikipedia.org/wiki/Everhart-Thornley_detector
The cute part of this system was the business of taking the electrons up through the final lens. The first electron beam testers, like the machine I worked on in 1982, were susceptible to local electric fields above the integrated circuit being tested, and an inconvenient positive track next to the point you were looking at could hoover up a lot of secondary electrons and lead you to get the wrong idea about the voltage on the point you were looking at, when you were looking at it.
Simon Garth – like Graham Plows and Everhart and Thornely – did a Ph.D, in Professor Oately's group – and his thesis - Garth, S. C. J. (1985). Electron beam testing of operating integrated circuits. Ph.D. Dissertation, University of Cambridge – describes how adding a magnetic field to the system could minimise local field effects.
http://www-g.eng.cam.ac.uk/oatley/nixon.html
The Lintech through-the-lens detector was rather more practical than the system that Simon Garth had set up at Texas Instruments in Bedford – who supported his Ph.D project.
I got to see it in 1984 when Texas Instruments wanted Cambridge Instruments to commercialise it. At that time their system used a “Mulvey” unipolar lens, and I committed the solecism of asking them how they cooled the lens – Mulvey lenses were notorious for running hot – and Simon Garth told us that they didn't bother. They just ran it for twenty minutes and left it to cool off for a few hours between runs, which meant that Texas Instruments didn't have a design that they could commercialise.
By the time that Texas Instruments did have a practical system. Graham Plows and Ralph Knowles had put together a much more practical scheme at Lintech
The first version of the Lintech TTLD that we put together at Cambridge Instruments was another disaster, and I had the glory to pointing out why then too. The filter grid that sat above the final lens (in the space between it and the intermediate lens) was supposed to be hemispherical, and had started off as a bit of wire gauze shaped over a mandrel.
This wasn't felt to be posh enough to impress Cambridge Instrument's customers, and someone had redesigned it as a carefully machined thimble, with circular holes drilled through it to let the electrons through.
Sadly, whoever had chosen the hole sizes hadn't read Eberhard Menzel's 1981 Ph.D. thesis, which spelled out – in German – how small the holes had to be in relation to the spacing from the thimble to the scintillator. In 1983, the senior Cambridge Instruments sales representative in West Germany had given me a copy of Menzel's thesis – which I'd read and kept (and I've still got it). There was lot more to it than just the information about filter grid sizes – Menzel had done a very thorough job.
When I got to see the new filter grid, I knew exactly why it wasn't working – and I'd had enough of an idea of what might be going wrong to insist on seeing it. We went back to the wire gauze, which worked fine.)
to: Richard Adams, Martin Wiseman, Chris Warner, Peter Milne, Simon Dawes, Bob Taylor, Paul Austin, Roland Meins, Andrew Dean, Tony Slater, Ian Murray, Maha Mahalingham, Madhusudan Patel, Steven Fisher
from: A.W.Sloman
subject: Progress to 9th August 1991
○ I've finished the Mark 3 Trigger board parts list.
○ Andrew Dean came back from holiday to design out an obsolete GaAs chip on the Mark 3 Delay board.
○ Paul Austin is on undisturbed holiday.
○ Martin Wiseman has wired up his daughter board for the Plinth Control Card.
○ Peter Milne is getting archived images back into the Syntel faster.
○ Ian Murray has got the UIF set up for image archiving.
○ Roland Meins has got the netlist window tested and working.
○ Simon Dawes has fixed the stig problem on P2 and got an image out of PP1.
○ Chris Warner has got P1 into the lab, and most of the console/plinth links made.
○ Bob Taylor has finished off the detailed drawings for P1.
○ Tony Slater is on holiday.
○ Barry Barker has shipped two TTLD fibre-optics assemblies to lapping and honing.
○ Steven Fisher is on holiday.
○ Mahdu Patel is correcting the parts list for the Mark 3 Delay board.
I finished the parts list for the issue 3 Trigger board on Tuesday morning, and passed it on to Richard, who had given it to Rodney Brown by lunch-time, by which time I'd found three omissions - so far I haven't found any more omissions.
I then started in on up-dating the Spec Part B for the Trigger board, and found a drawing error on the circuit diagram, but I'm now working on the circuit to drive the Chartered Semiconductor demonstration chip, and have found yet another problem - the chip contains a 17-stage shift-register scrambler, which implies that everything downstream of.the scrambler will take 54 or 108 seconds to cycle back to the identical state, which is going to slow us down a bit.
Fortunately the scrambler isn't always active, and when it is active we should be able to feed it a stream of bits that will persuade it to generate a pattern that repeats itself every 3.3msec - I think I've worked out a way of doing it, but I'll try to consult Mr. Lo in Singapore before I start wiring up this reinvented wheel.
I'm off on holiday in France next week - back Sunday next.
Andrew Dean got one of those ominous phone calls on Wednesday - there he was sitting quietly at home, redoing his kitchen, when I called him up to tell him that Gigabit had made the 10G045 obsolete a year ago, and Janine couldn't buy any for the Mark 3 Trigger board. It is an 8-bit digital comparator, but since we only compared 4-bit words with it, Andrew was able to replace it with the 10G002, which is cheaper and faster.
Andrew came in on Thursday to check the timings and up-date the circuit diagrams - PCA now have the corrected circuit diagrams, but they won't be able to get back to the job for
4 weeks, and we will have to wait 6 weeks for the corrected photoplots.
Despite last week's progress report, the original photoplots hadn't gone off to Printech, so we didn't lose any money on junking the work in progress.
I'm a bit surprised that we hadn't found out that the 10G045 had gone obsolete - Pronto were punctilious about informing us about this year's crop of deletions, and Andrew had already dealt with the withdrawal of the 10G024, but this year's list of deleted components didn't include the 10G045, presumably because it was deleted last year.
Paul Austin is going to have to pacify Maha when he comes back from holiday; the decoupling capacitors on the Waveform Processor - Digital don't appear on the net-list, which
makes life difficult for ATE.
Martin Wiseman has wired up his daughter board, and tested some of the functions; the chamber open/close is now under dual push-button control, but the door inter-lock hasn't been tested yet, and the HT shut-off hasn't even been wired to the Plinth Control Card.
Peter Milne is still working on his image transfer code; it works and he has speeded up image transfer back into the Syntel by providing bigger buffers, but he hasn't yet got his file locking working to arbitrate between access to the D-P RAM card for archiving or for display.
Getting NSF onto the Syntel so that he can automate the image transfer is being more troublesome than expected; the new Ethernet software is incompatible with the old Ethernet cards. It would only take a PROM change to bring the old cards up to the new standard, and Peter is trying to negotiate this; alternatively Microware might supply an earlier version of the software, but they are not enthusiastic.
He has also reorganised the software so that the system can be booted up from a terminal alone, without using two windows on the Sun, and installed the Ansoft "Maxwell's Field Simulator" on the Sparc 2 for Graham. Graham wants to use it to redesign and simulate the TTLD - Julian Davey is prejudiced against the Ansoft software already, and Graham hasn't even demonstrated it to him yet.
Ian Murray has set up a User Interface Dialogue for saving and recalling images; it won't do anything until Peter has finished the Syntel software, but the surface layer is complete. Ian has also put in an icon for the CAD interface, and organised a scheme for archiving tolerable variants of the User Interface, so that if you don't like the current version of the User Interface, you can always go back to the last one you liked enough to archive.
Roland Meins has completed his net-list window and tested it on the Plessey data. On the schematics window, the CAD/CAM software group have now got the schematics package in
England, and Roland expects to get it next week, with a Data Express converter on the same tape. It isn't clear what he is going to test it on - our Valid system is supposed to include an EDIF converter, so we could perhaps use the circuit diagram for one of the S.300 boards as an example.
Rapid Recall have passed him on to ES2 for the EDIF schematics for one of our ASICs, but it isn't clear that ES2 are willing to produce the schematic.
Simon Dawes finally sorted out the problem with the stigs on P2 - the rewound stig coil was shorting to its aluminium former, which was in electrical contact with the threaded mounting stud, which had been screwed so far into the Tufnol mounting block that it had come out into the centre and made contact with the drift tube, which ought to have been at +35V, but was in fact at +5mV, and injecting about half a milliamp into the stig coils.
It was quite fun to find, but easy to fix - Simon replaced the offending stig coil with one he begged from Barry. And Bob Taylor has written out a mod to change the stig coil formers from aluminium to delrin which will stop it happening again, as well as giving us a chance of seeing TV rate modulation of the stig field - with aluminium formers you'd expect the high frequency modulation of the stig current to mostly generate eddy current in the former, rather than changes in the field in the centre of the column.
With the stigs fixed, Simon got images and waveforms to Peter's satisfaction, and the system is now waiting for Carter and Carter to come back and photograph the screens.
Simon is now on PP1 - he has got an adequate image with a tungsten source, and is now trying to get the gun vacuum good enough to let him fire up LaB6. When I last looked he was still cooking the gun with heating tape.
Chris Warner and Graham Plows went over to CMS this week, and came back with a promise that the ion beam column would be delivered by the end of next week. In the meantime, Chris has got P1 installed in the new lab, and linked up plinth and console - even the TaxiChip cables are now in place.
Barry is servicing the P1 rotary pump, and the chamber should be being pumped down real soon now. When Chris has got an ion pump vacuum in the column he plans to put in a test sample of hot-cured Araldite 2001, as purchased at Mackay's last Saturday morning. Presumably Chris will drive over to Milton to buy the Loctite 406 while the column is pumping down. It is to be hoped that one or other adhesive has a low enough vapour pressure to be used in the new Blanking Head.
Bob Taylor has finished all the detailing for the FIB column housing, and production are getting on with making the parts, despite the flood of work from Esprit – production weren't warned about the Esprit work, though they had expected it, and they are still dealing with the work they were warned about.
Tony Slater is on holiday.
Barry Barker's fibre-optic assemblies for the two new TTLDs are now in Lapping and Honing, having the ends of the fibre-optic bundle polished. Barry is occupying himself servicing the rotary pump for P1. He still has to wind a new stigmator coil to replace the one Simon used to fix P2, but that isn't going to hold up the next scan coil assembly - he is still waiting for the PCB!
Steven Fisher is on holiday.
Mahdu Patel is going through the parts list for issue 3 Delay board, and he has found that it includes a number of resistors no longer fitted to the issue 3 board and some mis-assigned decoupling capacitors, while missing out the ECL PALs (which don't seem to have been issued yet). I wonder what Andrew will be dropping on me when I get back from France?
(Note added 2013-02-17. Dr. Julian Davey was Cambridge Instruments resident mathematical physicist. He looked like an un-made bed, and was somewhat odd, but he was very good at his job. If he didn't like the Ansoft software, he was probably right.
Graham Plows was - at the time - in the process of setting up a new company “Technology Sources” in Newmarket which was to sell Intuisoft's programs, amongst others. He took Mike Penberth and eventually Nick Cambell with him. I'm sure he was taking incidental advantage of Julian Davey's considerable expertise in this area for the benefit of his incipient company.)
to: Richard Adams, Martin Wiseman, Chris Warner, Peter Milne, Simon Dawes, Bob Taylor, Paul Austin, Roland Meins, Andrew Dean, Tony Slater, Ian Murray, Maha Mahalingham, Madhusudan Patel, Steven Fisher
from: A.W.Sloman
subject: Progress to 23rd August 1991
○ Boris Yeltsin has improved our chances of selling an EBT-2000 to Russia.
○ I'm working on the Spec B for the issue 3 Trigger Board.
○ Andrew Dean is working on the Spec B for the issue 3 Delay board.
○ Paul Austin has been working on Burst Mode problems.
○ Martin Wiseman has passed the issue 5 PCC over to PCA.
○ Roland Meins has got the schematic package up and sort of running.
○ Simon Dawes has almost got an image out of P1.
○ Chris Warner has been chasing pictures for the brochure and his paper.
○ Bob Talyor has done three mods, and written three more.
○ Tony Slater has been sorting out cables.
○ Barry Barker is on holiday.
○ Steven Fisher is working on PP2.
I got back from France last Sunday evening well fed and well rested, and perfectly fit to tackle anything Leica-Cambridge could throw at me, but I'm back to normal now. I've spent about half this week on up-dating the part B specification for the Trigger board, from issue 2 to issue 3; there is a lot left to do.
Yesterday I broke off to draw the diagram for a circuit to drive the Chartered Semiconductor demonstration circuit (an SC11046 modem chip). So far I've created symbols for the chip itself and a set of connectors which allow me to represent the chip plugged into our universal test socket, cabled back to a DIN416l2 64-way a,c connector. This leaves all the interesting bits still to do, like the circuit to defeat the 17-stage scrambler in the modem chip.
Andrew Dean came back from reconstructing his kitchen on Tuesday, and has been working on up-dating the part B specification of the Delay board from issue 2 to issue 3; he has also been involved in the usual meetings - on the safety mods, on Martin’s involvement in the FIB work on P1, on monitoring the temperature inside the Sampling Crate, and on getting decent contrast on Burst Mode images.
Paul Austin has been working on the LIP and on Burst Mode problems; on the LIP he has fixed the problem with the interlace, put in a proper "front porch" for the line synch, cured the pixel drop-outs and stopped the Cursor board from messing up the line and frame sych. The white line down the left-hand side of the monitor hasn't been fixed yet – Paul knows where it comes from, but hasn't yet worked out how to get rid of it.
The work on Burst Mode is at an interesting state - at the beginning of the week frame-at-a-time Burst Mode was working fine, but line-at-a-time Burst wasn't working at all. Now neither of them works - Paul found that Pete's software was resetting the Scan Generator on every line, but when Peter killed that reset command, Burst Mode died with it. At the moment it looks as if the Naffbus has also died; Peter and Paul seem to have decided that some of the sockets in the RAM-Store backplane have worn out.
Paul and Martin between them have tidied up the interaction between the Scan Generator and the Waveform Interface in Burst Mode; Martin reprogrammed the EPROM on the Scan Generator so that a single reset discharges the frame and line capacitors, while Paul reprogrammed the state machine PAL on the Waveform Interface to wait for the end of the frame blank before starting to acquire an image.
Paul has received the photo-plots for the issue 3 Waveform Processor - Digital artwork, and up-dated the circuit diagram for the Ion Pump Monitor to Jack Warner's exacting standards.
Martin Wiseman has achieved his ambition of getting shot of the issue 5 Plinth Control Card (PCC) before he goes off on holiday to the Ffestiniog Railway; they have a new general manager, so he may not feel quite the same need to practice peristroika on their board after he comes back from this year's holiday.
The issue 5 PCC even includes sockets for analog and digital temperature sensors for the Sampling Crate; Andrew and I have yet to work out which we want, and how and where we are going to fix whichever one it is to the Sampling Crate.
Martin's issue 4 Blanking Board is now on the Valid system, after a fashion; Martin's 8 sheets of Isis drawings have been collapsed onto one Valid drawing sheet, with all the connections turned to rubber bands; PCA have edited the net-list to put in the power pins, and to put five pins on each SHA PCB connector, and will create the PCB from this net-list. Presumably the Isis drawings will be photographed for the microfilm drawing register.
Peter Milne has spent most of the week on the Burst Mode problem; before Burst Mode had died completely he put in quite a lot of work on getting good, high contrast images in Burst Mode - and found that when the contrast was high enough to be acceptable, the whitest pixels were rolling over to black in the Waveform Processor. To avoid this he has to set up for slightly lower contrast - with a decent image processor he could put the contrast back into the images displayed, but we can't do this with the LIP, so he is proposing to do "histogram equalisation" in software, on the Syntel. It will be slow, but should illustrate one of the advantages of replacing the LIP with a decent image processor.
Peter has also put in some time on the new software for the Syntel; NSF still isn’t working, and it isn't clear why. Peter has written a long letter to Microware to wait until his opposite number comes back from holiday; since Peter is on holiday next week, the reply will probably come by letter too.
Ian Murray has spent most of this week creating pictures for the brochure and Chris Warner's paper. The latest version of the UIF, as used for the brochure, seems to be being superseded by a version with the menu bars across the tops of the windows, rather than down the side - this seems to be the latest fashion in the GUI business.
Ian has also found time to play around with the logic analyser code; he hasn't yet tried it on a real waveform, but this is getting close.
Roland Meins is off sick, but his beta-site schematic editor is up and performing as well as you would expect with beta-site code; well enough to provide pictures for the brochure.
Simon Dawes has also spent most of this week producing pictures for the brochure and Chris's paper, or more exactly, hanging around waiting to produce pictures.
He has also put in some time on P1, and found the cracked Perspex tube around the scintillator lead which was letting the scintillator flash over. There are other worries; it looks as if the Alpha-Repeater EHT set has lost an op-amp, and the plinth is still powered from the old EBT-1 Powerline unit that we've had to replace on the EBT-2000s because it isn't powerful enough, but P1 is pretty close to imaging again.
Last week Simon was mostly working on his cable bible, which is now up-to-date. He and Tony Slater are now working through the cable drawings, making sure that the connectors are right, and that the cables are long enough. There are roughly 300 cables on the EBT, so it is a long job.
Simon is off to Korea and Samsung next Thursday. This won't be a holiday.
Chris Warner has spent all this week chasing pictures for his paper, and the brochure; he now has an S.360 picture of the square hole the CMS column cut into the metalisation on an IC. He has also used the ion beam to cut a track, but the adjacent substrate charges up too much to allow him to get a picture of that with the S.360.
He has also spent time on P1 with Simon; he points out that P1 is still short of an ion pump, which means that he can't even use it to test the adhesives for the Blanking Head (although Nick Campbell has come up with some figures for Araldite 2001 and Loctite 406 - they both outgas at three times the rate of the equivalent area of Viton A).
Chris is on holiday next week.
Bob Taylor has done three mods this week, and written three new mods, bringing the number of outstanding mods up to 19. He has received most of the ion gun bits back from production.
The stage base plates passed their vacuum test, and are now back at Lucchesi—Carins for final machining; they should be back next week for plating, not too far behind the rest of the stage parts, so Barry should be able to start assembling the fat stages for PP1 and PP2 real soon now.
Janet has ordered two more sets of stage parts for the end of September; MRP2 seems to have been loaded with duff information - it expects to have the parts manufactured in 22 days (actual time will be about 12 weeks) and plated in 2 days (2 weeks).
Tony Slater is working on cables with Simon Dawes. He doesn't seem to be enjoying it.
Barry Barker is on holiday.
Steven Fisher fixed the column connector on PP1 on Monday; the sleeving hadn't be put on right, and one of the gun align coil drivers had been briefly shorted to ground, blowing up the driver.
For the rest of the week he has been getting on with PP2, mainly testing boards with Mike Sheldon. So far they have got an Ion Pump Monitor, a Column Control Card, and a Scan Amp - soon they are going to need a Syntel to drive them, and a RAM-Store crate to put it in.
(Note added 2013-02-22. Ian Murray was working on the User Interface software (UIF) and he was being influenced by latest fashions in graphical user interfaces (GUI).
It was all part of the process of making computers easier for people to use. Since I'd started my computing career by typing Fortran commands onto 80-column Hollerith cards, and proceeded to get a PDP-8 to do things for me by typing assembly language on a Teletype terminal, I found it all a trifle effete, but I was well aware that a GUI was easier to use than it's predecessors.)
to: Richard Adams, Martin Wiseman, Chris Warner, Peter Milne, Simon Dawes, Bob Taylor, Paul Austin, Roland Meins, Andrew Dean, Tony Slater, Ian Murray, Maha Mahalingham, Madhusudan Patel, Steven Fisher
from: A.W.Sloman
subject: Progress to 30th August 1991
○ I've been working on the Chartered Semiconductor demo circuit.
○ Andrew Dean has been reviewing the Waveform Interface board.
○ Paul Austin has been modifying the Waveform Interface board.
○ Martin Wiseman is on holiday.
○ Peter Milne is on holiday.
○ Roland Meins has updated the CAD user interface to the new standard.
○ Ian Murray has done some more pictures for the brochure.
○ Simon Dawes is off to Korea.
○ Chris Warner is on holiday.
○ Bob Taylor has done two mods.
○ Tony Slater has done some more work on the Blanking Head.
○ Barry Barker is assembling the column for PP2.
○ Steven Fisher is cabling PP2.
I've completed a circuit diagram for the Chartered Semiconductor demonstration circuit, faxed chunks of it to Mr. Lo in Singapore, and tried to phone him to talk about it. Since I now feel fairly sure that that circuit wouldn't work, my inability to contact him might be a good thing; I'm now at liberty to charge off up some other blind alley.
Andrew Dean has been engaged in a premature design review of the Waveform Interface board - Paul Austin hasn't yet finished his Spec B for the board, but Andrew needs to be clear on what is messing up Burst Mode. Andrew has also put in some time on his Delay Board Spec B.
Paul Austin spent the first half of the week working out how to modify the Waveform Interface card so that we could rapidly increment the line counters on the 8-Plane RAM cards - he has a scheme but it isn't tested yet. The modification should allow Pete to display a single Burst Mode plane while it is being accumulated (in line mode, if we ever get it to work).
Paul spent the rest of the week on P2, getting Burst Mode running from scripts, so that we can let Peter loose on it next week with some confidence that the hardware won't mess him about.
Paul on holiday next week, in a cottage by the sea in Norfolk.
Martin Wiseman is on holiday.
So is Peter Milne.
Ian Murray has spent part of this week creating more pictures for the brochure, or rather slightly different versions of earlier pictures; he hasn't been able to automate the process of getting the right picture, but he claims to have speeded up the procedure for getting a different picture.
Since the complaints about the brochure are now coming from the printers rather than from Marketing, we can hope that Ian won't have to create any more pictures for a while.
Ian spent the rest of this week reworking the UIF to the new style, and writing out a set of rules for constructing windows and menus and pop-up lists so that we can standardise the "look and feel" of the UIF.
He is off on holiday next week, to visit his family in the West Country.
Roland Meins has spent most of this week changing the CAD UIF to conform to Ian's new rules. He has been poking around looking for schematics to display on his schematics editor, but with little success so far.
Simon Dawes flew off for Korea yesterday evening. He didn't manage to get an image on P1 before he left, but he was doing okay on P2, and he seems to have sorted out the cables to the point where Tony Slater could finish the job.
Chris Warner is on holiday in northern France; like Graham, he won't be back until Tuesday.
Bob Taylor has done two more mods this week, bringing the total number of outstanding pre-production mods down to 17. There are of course 51 additional mods required for the
production batch, which I managed to ignore in last week's report.
Bob has also written a set of do-it-yourself instructions for Bell Northern, to allow them to reassemble their EBT-1 stage, which sheared the 0-ring in the sliding seal last night. To paraphrase Trevor Sexton "They've never done that before sir!".
Tony Slater is working on cables and the bellows assembly for the Beam Blanking Head. When I talked to him he was looking at a large bundle of cable drawings which Simon had superseded, nerving himself to drop them in the bin, while a rather smaller pile sat near to hand, waiting to be renumbered.
Barry Barker is building up the PP2 column in SEM Final Test; he has been hassled by Margaret from coil-winding who is making the next scan coil assembly. Since we haven't fully tested the last set of scan coils, Tony hasn't be able to up-date the drawings, which leaves Margaret a little short on assembly instructions. Martin Wiseman is supposed to have modified the Scan Amplifiers such that Chris could now put enough current through the scan coils to test for distortion at low magnification, so maybe we will get this problem sorted out fairly soon.
Steven Fisher has spent most of the week cabling PP2, with a little bit of time on up-dating PP1. He also joined Dave Cole in the hunt for the Auxiliary Control Panel for PP2 - this has gone missing, and since it includes two DESTAs (Ethernet receivers) at 200 pounds each, and 200 pounds worth of cable kit, the hunt has been pretty thorough.
to: Richard Adams, Martin Wiseman, Chris Warner, Peter Milne, Simon Dawes, Bob Taylor, Paul Austin, Roland Meins, Andrew Dean, Tony Slater, Ian Murray, Maha Mahalingham, Madhusudan Patel, Steven Fisher, Tony Edwards
from: A.W.Sloman
subject: Progress to 6th September 1991
○ I've been working on the Spec Part B for the issue 3 Trigger Board.
○ Andrew Dean has been working with Peter Milne to get Burst Mode right.
○ Paul Austin is on holiday.
○ Martin Wiseman has been fixing drawings.
○ Peter Milne has been working with Andrew Dean to get Burst Mode right.
○ Roland Meins has up-dated all of the CAD user interface to the new standard.
○ Ian Murray is on holiday.
○ Tony Edwards has joined the group.
○ Simon Dawes is servicing prospects in Korea.
○ Chris Warner has booked hotels at Como, at last.
○ Bob Taylor has done seven mods.
○ Tony Slater is ploughing through the cables.
○ Barry Barker is assembling the column for PP2.
○ Steven Fisher has got PP1 imaging again.
I haven't got any feedback from Chartered Semiconductor on the circuit to drive their chip, so I am not doing any more work on that until Graham gets back from Singapore (if at all).
I'm now back on up-dating the Spec Part B for the issue 3 Trigger board; I'm most of the way through describing the new circuits to control the Unblank Pulse width, and I've still got to describe the rather extensive changes in the circuits that generate the Processing pulses, that drive the Waveform Processor - Digital (WPD). At the moment I am finding about one minor error per day in the circuit diagrams for the issue 3 Trigger board; I correct them as I go, and fax the corrections down to GS Designs. Since Gary Askew at GS Designs has only just completed entering the net-list from the circuit diagrams, he hasn't been too worried by the corrections so far; I'm to go down there on Tuesday to okay the component placement (three quarters of the board has to be ripped up and rerouted to make room for the new circuits).
Andrew Dean has spent this week with Peter Milne getting Burst Mode working again. The problem was that the Scan Generator wasn't getting reset, which was a software bug.
When Peter fixed that, the Scan Generator didn't act as if it had been reset, because the reset outputs had to be latched into output registers before the rest of the world could see them, so Andrew and Martin Wiseman fixed that with an extra clock pulse. At this point, Burst Mode came back, running faster than ever before. When Andrew took the Waveform Interface off the Extender Board, and plugged it back into the rack Peter was able to produce a 56-frame sequence in 15 seconds - the images were a bit noisy, but perfectly acceptable. The noise reduced version, taken over 10 minutes, serendipitiously shows quite nice voltage contrast on a buried track, running under active tracks; that is through passivation.
Voltage contrast through passivation is the Holy Grail of the electron beam testing business; buyers seek it with persistent enthusiasm, and research workers report how they have achieved it at every conference, but nobody so far actually seems to rely on it.
Paul Austin is on holiday. When he gets back on Monday he will be pressed into doing the last of the mods to the Waveform Interface to let us try line-by-line Burst Mode.
Martin Wiseman seems to have spent most of this week fixing circuit diagrams. He spent Monday and Tuesday working on Metheus on the Cursor Board circuit diagram, putting in the latest set of mods to pacify Jack Warner and incidentally to provide an up-to-date circuit diagram for the latest issue of Cursor board. While he was working on the diagram Martin spent some time making it more intelligible, by grouping switches by function, rather than by package, and similar helpful adjustments.
He then moved on to the 386 and used Isis to up-date his circuit diagrams for the new Blanking Board, so that they loaded onto the Valid system as 8 sheets rather than one. But the Valid system still reroutes most of the wires, and leaves off all the rest - some with warning messages, and others without warning. Martin is not impressed.
Martin also helped Peter and Andrew sort out the Scan Generator board on P2, and helped Steven Fisher sort out the Scan Generator on PP1, where the two PROMs had mysteriously gotten swapped between their sockets.
Peter Milne has spent all week with Andrew Dean on Burst Mode, as mentioned above. By early next week he should be free to chase Microcall about the networking problems on the new version of the OS-9 operating system for the Syntel.
Ian Murray is on holiday.
Roland Meins has finished up-dating the CAD user interface to conform to the new rules. He is thinking about visiting ES2 in person to try and get some progress on the EDIF schematics for at least one of the S.300 ASICs.
Tony Edwards joined the group on Tuesday; at the moment he is using Simon's desk, but no doubt something better will get sorted out - maybe one of the computers will consent to double up.
He spent the last two hours printing off "* Under Development" labels to go on our new brochures - all the FIB options are starred in the brochure, but there wasn't a tag to tell you what the star meant.
Simon Dawes is busy in Korea - if Tony Edwards has got his desk, Simon seems to have been temporarily stuck with part of Tony's role, as the Korean agents trot him around as if he was marketing, rather than service. Since Schlumberger have ruined their prospects in Korea by being very slow in installing and servicing the one machine they sold there, Simon presumably sells the EBT just by being there.
Chris Warner has spent this week getting set for the conference at Como; the biggest problem turned out to be getting hotel rooms for the three days of the conference, but he finally got accommodated not too far from the conference.
Chris was able to find time to check out the latest scan coil assembly, so Barry can put that into the PP2 final lens assembly.
Bob Taylor has done seven mods this week, which ought to have reduced the number of remaining pre-production mods to ten.
Tony Slater has been up-dating cable drawings all this week - despite my earlier report, Simon hasn't given Tony enough information to let him sort out all the cable drawings, but Simon did provide enough data to keep Tony going for a while.
Barry Barker is still building up the PP2 column; Barry has not quite got all the components for the gun or the final lens, but it is getting close. Bob expects Barry to start on assembling the PP2 stage next Thursday; it was to have been earlier but Lucchesi-Carins scratched the lower stage plate, and Bob had to sent it back to have it reground. It won't be back until next week, and to get it plated by Thursday. Richard Adams has promised to drive it up to Leeds and back on Wednesday, to the nearest "plate it while you wait" shop.
Steven Fisher managed to get PP1 imaging again; PP1 still has the old-style final lens, and the extractor lead had broken in the old, familiar way. This wasn't the only problem - the gain and offset pots on the LIP had gotten tweaked to the point where a good image came out looking lousy, and of course there was the problem with the reversed PROMs on the Scan Generator; the Line PROM didn't work too well in the Frame socket, nor the Frame PROM in the Line socket.
Steven and Mike Sheldon have also done some more cabling on PP2; they are starting to worry about building up a new RAM Store crate.
to: Richard Adams, Martin Wiseman, Chris Warner, Peter Milne, Simon Dawes, Bob Taylor, Paul Austin, Roland Meins, Andrew Dean, Tony Slater, Ian Murray, Maha Mahalingham, Madhusudan Patel, Steven Fisher, Tony Edwards
from: A.W.Sloman
subject: Progress to 13th September 1991
○ I'm still working on the Mark 3 Trigger board.
○ Andrew Dean is back on the Delay board - two of his components have gone unavailable.
○ Paul Austin is fixing the Burst mode hardware bugs.
○ Martin Wiseman has been back on the Blanking board.
○ Peter Milne is fixing the Burst Mode software bugs.
○ Roland Meins has moved the chip mimic on his CAD interface.
○ Ian Murray has been working on the latest version of the UIF.
○ Tony Edwards has been to Como, and is writing a report.
○ Simon Dawes is back from Korea.
○ Chris Warner has been to Como, and is writing a different report.
○ Bob Taylor has completed three mods.
○ Tony Slater is working on the Blanking Head.
○ Barry Barker has assembled the C3 lens for PP2. It is going on PP1.
○ Steven Fisher is sorting out test jigs and crates.
Gigabit Logic made the 10G000A-4C quad 3-input NOR gate obsolete in 1989. We found this out on Wednesday; I should have checked our entire Gigabit list when we found out that the 10G045 had been made obsolete. Janine Heylan (in Purchasing) and I both did it yesterday, and there isn't anything else that we should have known about.
The 10G000A-4C is used on the Trigger board (three), the Delay board (five), and the Blanking board (two). It looks as if we can replace them all with 10G001-4Cs which are quad 2-input NOR gates; I can get by with the same number of 2-input NOR packages (I had two gates unused), Andrew thinks he can keep the Delay board working with five 10G001-4C more or less replacing the five 10G000A-4C, but Martin has to use three 10G001-4C to replace two 10G000A-4C, which is what you would expect from counting inputs. The change saves 38 pounds on my board, 64 pounds on Andrews board, and 2.74 on Martin's board.
I was most of the way through the Spec Part B for the issue 3 Trigger board when this came up; part of the section on the Short Unblank Pulse Generator is now going to have to be rewritten again.
I went down to GS Designs yesterday afternoon to look at Gary Askew’s component placement; we swapped a couple of adjacent components, but the placement looked basically sound. I took down copies of sheets 1 and 3 of the revised circuit diagram, so Gary should have up-dated the net-list to accommodate the 10G001s by now.
Andrew Dean started the week by finding out that Analog Devices were discontinuing the supply of the AD-9712 12-bit DAC that he uses on the Delay board; we are going to replace it with the AD-9768 8-bit DAC that we used on the issue 1 Delay board. Since we only drove the top 9-bits of the AD-9712, the theoretical loss of time resolution (10psec to 20psec) is tolerable, and I suspect that there will be enough noise dithering on the system to allow us to see 10psec increments anyway. In the process of checking out the circuits around the DAC Andrew found a bug in his timing; by the time he had sorted it out he had revealed a bug in my timings, so the exercise did have a positive side.
The 10G000A story started just as Andrew was getting on top of this lot.
Paul Austin has spent this week on P2 with Peter Milne, getting the low level operation of the machine set up so that we can View-Burst - that is line-by-line Burst Mode with one of the Burst Mode images coming up on the image monitor, ten or twenty lines at a time.
It was mostly detailed stuff, with Peter single-stepping the software while Paul monitored the hardware with a logic analyser; there have been minor changes in both software and hardware, and all that is left between us and View-Burst now seems to be some messy manipulations of the Naffbus-2 control lines. Paul thinks we might see it next week - Peter thinks that the Naffbus-2 control lines are a can of worms, and is less optimistic.
Martin Wiseman spent part of this week on continuing to sort out the Blanking board circuit diagrams; all the 0VA connections he had put in on the Isis system had turned to 0VD on Valid, and his diodes were all the wrong way round.
Martin also put in time on the Scan Generator on P2 for the View-Burst work, and on the Scan Generator on PP1 to minimise scan distortion. Neither job is finished yet.
And while Martin's immediate reaction to the 10G000A-4C story was to stay home on Thursday and get his gas boiler connected, he has now sorted out the necessary mods to the Blanking board. He has marked up the Valid drawing, which is essentially unintelligible, and he expects that PCA will sort out an intelligible drawing in the end. Optimistic.
Peter Milne has spent most of the week on View-Burst, with Paul Austin - quite successfully, as above. Peter has also blown up a Sun monitor, presumably by looking at it too intently when switching it on, but Sun replaced it under warranty. When he spilt coffee onto a Sun keyboard he thought about going the same route, but decided to clean up the keyboard first, which cured the problem.
Getting away from hardware for a minute, Peter has also set up a scheme for putting X-window pictures onto the image monitor, for the next video. They are strictly monochrome, and the image quality is naff, but when Paul gets his front porch mod into the LIP the quality may improve.
The incompatibility between the new version of OS-9 and the old Ethernet cards is still unresolved, but Peter has gotten some response out of Microware, and is visiting them on Monday week.
Ian Murray expects to release the next version of the user interface (UIF) next week; this week he has been working to make Results and Acquire look the same, and to provide a button for flipping waveforms from Acquire to Results in short order. There was also a request to let the time-cursors cross over, which he finally achieved – it wasn't as easy to do as he had hoped.
Roland Meins spent this week moving the small image of the whole chip in his UIF from one corner of the screen to a different window; he plans to ring up ES2 and talk to them about the CAD data on our ASICs, but he hasn't gotten around to it this week either.
Tony Edwards was at the Electron Optical Beam Testing of Integrated Circuits Conference at Como for the first half of this week; he is writing a report on the conference which will be circulated. Now that Simon is back he has moved to the desk previously occupied by Chris Warner's 386.
Simon Dawes got back from Korea today, after having fixed a lot of the problems on the Samsung R & D EBT-1, and put in some time training the new Korean EBT service engineer, Mr. Ham, who seems to be an improvement on his predecessor.
The Samsung EBT-1 shows quite a lot of vibration when used with long working distances with the wafer stage, on a suspended floor, right next to the air-conditioning plant, and he didn't have much luck in reducing the vibration. His copy of last week's video didn't arrive in time.
Chris Warner gave the paper at Como; he too is writing a report. He did take his copy of the passivation video to the talk, but ran out of time before he could show it. Graham seems to have compensated by claiming that the video that Chris did show was taken from a passivated specimen.
Chris has responded to seeing the Schlumberger machine's large field of view by renewing his enthusiasm to try out on the EBT the "fish-eye" scan mode someone found on the S.360.
The "fish-eye" mode gives a distorted image, but it could be useful, despite the distortion.
Bob Taylor has done three mods this week, and was expecting to finish a fourth today. And he has found some new, cheap, small DESTAs (for the Ethernet links) which simplify the safety mod for the Auxiliary panel, and save 200 pounds into the bargain.
Tony Slater has been working on the Blanking Head this week, being persuaded that it can be soldered together. Barry Barker also took a bit of his time, getting Tony to sort out the drop-offs on the new C3 casing that Tony drew – there was rumour of a mirror-imaged mounting block...
Barry Barker has assembled the new C3 for PP2. It is to be fitted to PP1 for testing, and PP1's final lens will be rebuilt to the new standard in time to go on PP2 ... Barryis now starting on assembling the stage for PP2.
Steven Fisher spent a little time on P2 on Monday – the filament remains off-centre, but not so far that we can't use it. The rest of the time was wiring; PP2 is almost completed and PP1 is updated. Steven is now getting onto putting test rigs together, and finding a RAM Store crate to drive PP2.
And Peter Milne wants to record a record - for the first time ever, P2 was still working when Simon came back from overseas.
to: Richard Adams, Martin Wiseman, Chris Warner,Peter Milne, Simon Dawes, Bob Taylor, Paul Austin, Roland Meins, Andrew Dean, Tony Slater, Ian Murray, Maha Mahalingham, Madhusudan Patel, Steven Fisher, Tony Edwards
from: A.W.Sloman
subject: Progress to 20th September 1991
○ I've got a first draft of the Spec Part B for the Mark 3Trigger board.
○ Andrew Dean is trying to keep ahead of Tom McNamara’s corrections on the Delay board layout.
○ Paul Austin has fixed most of the Burst Mode hardware bugs.
○ Martin Wiseman has been Blanking, Plinthing and Scanning.
○ Peter Milne hopes he has fixed all the View Burst software bugs, and is well into Logic State Maps.
○ Roland Meins has finished tidying up the context/overview window on the CAD interface, and is working on store/recall.
○ Ian Murray's new version of the User Interface has been run and tested - there are thirteen minor flaws to be fixed.
○ Tony Edwards has been digging out lists of Lintech customers, and learning to use the EBT-2000.
○ Simon Dawes has been down to Southampton again, and given Tony some more cable data.
○ Chris Warner has completed his. Como report, and started vacuum-testing the glues for the Blanking Head.
○ Bob Taylor has completed two mods, and generated three more.
○ Tony Slater is back on cables.
○ Barry Barker is assembling the stage for PP2.
○ Steven Fisher has wired up the new C3 for PP2, and it is on PP1.
I've finished the text for the Spec Part B for the issue 3 Trigger board; I've got some work to do on the Metheus simulations to get them into a state where the output is worth binding into the Spec Part B.
Gary Askew finally sent me his placement for the issue 3 Trigger board layout; a couple of the coax terminators are going to have to be moved, but I haven't found any other problems yet.
This report is late because I foolishly volunteered to e-mail a copy of our acceptance tests document to Dr. Tan in Singapore, then found that I had to type it into the VAX myself. By the time I had finished putting in a two page summary, Graham had lost enthusiasm, so the effort was pretty much wasted.
Andrew Dean is working on the Delay board; Tom McNamara is now working on the layout to replace the obsolete 10G045 with a 10G002. He knows about the problems with the AD-9712 and the 10G000A, but he hasn't yet had up-dated circuit diagrams to tell him how to fix them, though Andrew expects to give him the data on the AD-9712 replacement on Monday.
The situation with the 10G000A is confused; Andrew has been testing the vernier delay elements of the issue 2 Delay board and the 10G000A on that board has been acting as if its logic "1"
output level depends on mark-to-space ratio; if the gate is "high" for less than 0.1% of the time, the "high" level starts falling from the nominal -0.6V, to -1.2V (at 5 parts per million). Since the guaranteed minimum "high" voltage is -0.8V, this is worrying.
Andrew has called up Gigabit in California, but Dan Rodriguez couldn't produce any immediate response, though he thought that going over to the 10G001 wouldn't solve the problem.
The effect on the Delay board is to introduce some 400psec of jitter onto the Sampling Edge propagation delay. The performance of the issue 3 Trigger board Short Unblank Pulse Generator could also be seriously impaired by the same phenomenon. I'm thinking about looking at the Sony ECL gates.
Paul Austin spent most of this week on the last bug in the View-Burst mode; the 64-line bursts were intermittently missing a line, which turned out to be a fault in the PAL design on the Waveform Interface, which meant that a Reset on the Waveform Interface didn't always leave the interlace bit in the same state.
Paul then got the mod information together for the Waveform Interface Issue 3, and took it over to Syscom on Friday.
Martin Wiseman put in some more work on the Blanking board this week; the Valid netlist now seems to be correct, but the Valid circuit diagram is still unintelligible.
Martin has tried to put in some work on scan distortion on PP1, but since the final lens assembly (including scan coils) was taken off the machine on Monday and wasn't restored until Friday, he didn't get much done. In the meantime, he has done some work on up-dating and expanding the Spec Part B for the Plinth Control Card.
Martin hasn't yet been drawn into the debate about the mark-to-space sensitivity of the 10G000A; the effect should be visible on the Blanking Board, if we set up the conditions correctly.
Peter Milne has got the View-Burst software working, and has moved onto the Logic State Map acquisition software. His first pass through the software gave a Logic State Map of a kind, and clarified his thoughts wonderfully. He is now well into a second pass, which is rather more complicated, but should do pretty much what we want.
Peter is going down to Southampton on Monday, to talk to Microware about the problems with the new OS-9 and the old Ethernet cards.
Ian Murray has released the new version of the User Interface, and he is working on the 13 bugs that testing has so far revealed. None of them looks too difficult to fix. The new User Interface supports Burst Mode (but not View-Burst) with either Flicker or Difference to allow you to compare pairs of images out of the Burst. One bug is that it tries to allow you to use both Flicker and Difference at the same time, which isn't going to become a feature.
Roland Meins is continuing to work on the user interface to the CAD package; he has sorted out the overview/context window (the small image of the whole chip), and is now working on store and recall, enlarging the table of settings and data stored to give the user more context when the set-up is recalled.
The hunt for CAD data has broadened from ES2 to include the Institute of Microelectronics at Stuttgart. Roland has been on the phone to both of them this week, without hearing anything
that sounds promising.
Tony Edwards has spent part of this week trying to get together a list of all the people who bought voltage contrast gear from Lintech, and the rest of the week learning how to operate P2. On Monday he is going down to Southampton with Peter, but to visit Philips with the aim of getting a test chip with design information from them. We hope he has better luck than Roland has been seeing.
Simon's progress for this week involves a roof and one sill of the van he and Nick are converting into a passion-waggon. In between his efforts as a welder, he also contrived to get down to Southampton to fix what used to be Gerry's machine; the new user is still a little green, though they probably were not responsible for the short that appeared between the scan coils and the vector coils. Simon worked around that by disabling the vector coils; he goes back on Tuesday with a new scan coil assembly.
Simon has also finished going through the list of existing cables on the AS400, and checked that they are all right (or that the mod paperwork is in the pipeline) and he is now well on his way through the missing and changed cables, giving Tony the information he needs to get the drawings up-to-date.
Chris Warner has completed his report on the Como conference and is getting on with measuring the out-gassing rates from his alternative glues for the Blanking Head. He is also starting work on trying the "fish-eye" scanning mode on PP1.
Bob Taylor has completed two mods on the preproduction list, and added three more - although none of them are too complicated; one of them was to remove the hole in the desk-top that used to accommodate the digi-knob.
Tony Slater has been doing cables this week, including looking at the cable entry for the Blanking Head.
Barry Barker is assembling the stage for PP2 - when I talked to him he was at the point of lowering the top plate onto the chamber. He now has all the bits for the stage, so it should be straight-forward from here on in. He even has a second seal plate back from the platers, which he ought not to need for PP2.
Steven Fisher had a couple of days holiday, and spent most of the rest of the time wiring up the new C3 lens for PP1. There seemed to be a hole in the assembly data - Steven didn't know where the 100M resistor in series with the scintillator was to go, and there was a certain amount of ad hoc potting. Nobody seems to be responsible for putting the extra information on the assembly drawings...
to: Richard Adams, Martin Wiseman, Chris Warner, Peter Milne, Simon Dawes, Bob Taylor, Paul Austin, Roland Meins, Andrew Dean, Tony Slater, Ian Murray, Maha Mahalingham, Madhusudan Patel, Steven Fisher, Tony Edwards
from: A.W.Sloman
subject: Progress to 27th September 1991
○ I should be able to start checking the issue 3 Trigger board layout next Wednesday.
○ Andrew Dean is working at reducing the jitter on the Sampling Edge.
○ Paul Austin fixed all the significant Burst Mode hardware bugs.
○ Martin Wiseman is mostly writing part B specifications.
○ Peter Milne has got the View-Burst mode to an acceptable state.
○ Roland Meins is still working on store/recall for the CAD interface.
○ Ian Murray has just released a revised version of the new User Interface.
○ Tony Edwards has been working on the new brochure, but he is off to Deerfield next Tuesday, for three days.
○ Simon Dawes has been back to Southampton, but otherwise is stuck on cables.
○ Chris Warner has been looking at Scan Distortion.
○ Bob Taylor has completed several more pre-production mods.
○ Tony Slater is still on cables.
○ Barry Barker has assembled the stage for PP2.
○ Steven Fisher is working on the RAMStore from P1.
This week's excitement has been the consequences of Andrew Dean's discovery of the moving threshold on his 10G0000A-4C on the Delay board. Andrew got to talk to a design engineer
at Gigabit on Monday evening, who told him that "they all do that sir" (I never knew that Trevor Sexton ever worked for Gigabit), and suggested that Andrew tweaked the threshold
voltage fed into the part at Vbb (pin 4). This solved Andrew's problem.
The Short Unblank Pulse Generator on the Trigger board is very much more sensitive to threshold shifts than the Delay board, and I've redesigned the circuit to eliminate the 10G001 at IC42, replacing it with a 100K ECL part, the 100101, which is less than half as fast, but adequate to the job. I would have preferred to use the Sony SPECL part CXB-1100Q, which is almost as fast as the Gigabit part, but three times more expensive (at 92-70) and on at least 3 to 4 weeks delivery (everybody except Pronto quoted 12 to 16 weeks).
I've modified the relevant sheets of the circuit diagram on Metheus, and I took copies of the modified sheets down to Gary Askew yesterday afternoon; I've also completed an AS400
mod sheet for the board parts list, but I haven't yet modified the Spec Part B.
Gary Askew expects to send me marked-up circuit diagrams showing pin numbers and grid references on Monday, and hopes to have the board fully tracked by Wednesday. I hope to be
able to get the pin numbers onto the Metheus drawings over Monday and Tuesday, so I could take a set of up-to-date drawings down to Knebworth on Wednesday, but I'm not too optimistic.
Andrew Dean has been testing the time-to-digital converter on the Delay board; Peter Milne has given him a program that gives him a histogram of the digitised Trigger Offsets against the number of times that offset appears, which is a handy diagnostic tool.
Andrew was able to solve last week's problem – 400psec jitter in the input stage to the Delay board - by raising the voltage applied to the Vbb input from the -1.32V coming out of our ECL source to -1.15V. This week's problem is that one of the transistors in the time-to-digital converter is oscillating at 700MHz, which messes up the voltage that is being digitised, produces two large humps on the histogram, where it ought to be flat, and leaves us with about 400psec jitter on the output of the Delay board. We should be able to stop the oscillation.
Andrew spent Tuesday at a Tektronix seminar on EMC (electromagnetic compatibility) and came back with a fairly clear idea of how much it is going to cost us to prove that we conform to the relevant standards when they come into force. A bundle.
Paul Austin spent part of this week on View-Burst again, with Peter Milne. They got rid of the intermittent white line at the top of the picture by changing the software to write to the Waveform Interface twice at the start of the Burst; the line counters on the 8-Plane Memory Boards only get incremented on every second line (the other line goes onto the other Interlace) and the column counters weren't cleared until they had seen a line counter increment. Messy.
The intermittent black second line has been ignored for the moment; Peter thinks that he could bodge it out in software if necessary.
Syscom didn't want to do the mods to the Waveform Interface, and quoted us a silly price - they have given us the old schematic and layout information on disk and we are going to get the work done by a sub-contractor in St. Ives, who has the relevant software packages.
At home, Clair has finished modifying the Timebase Interface, so we should have new boards in a couple of weeks.
And Paul put in some time with Steven Fisher on the P1 RAMStore boards, which Steven is refurbishing as a test tool.
Martin Wiseman tried to put in some more work on high magnification scan distortion this week, but Chris and Simon's work on low magnification scan distortion rather cut across this.
Martin actually spent most of the week on the Spec Part Bs for the Plinth Control card and the Scan Generator; he also went to CMS with Chris Warner and Bob Taylor to inspect the ion beam column we aren't actually taking delivery of for three months.
Peter Milne has got the View Burst software working – and Ian Murray has got it into the latest User Interface. The Logic State Mapping software is working; the lines are coming up slightly out of order, but Peter has implemented most of the fix for that, and the line shuffling after data acquisition is relatively slow, but Peter knows how to get that down to four seconds, when he gets the time.
Roland's CAD software is taking up most of the 700Mb on the Sparc 2 disk, so that Peter only has room to archive six images, which doesn't help demonstrating Burst Mode.
Peter went down to Microware at Southampton on Monday, and they found the bug in their networking software; the Ethernet card makes a completely unnecessary "bus request"
whenever it is activated, and hangs up until it get a "bus grant". We could get a "bus grant" out of a bus controller, which is a tiny card which sits on the back of the VME backplane, if we could get a bus controller (170 pounds from Syntel, on 4 weeks delivery). Paul may make a bus controller out of PAL before then, and in the long term it is likely that we will be able to get away with wiring the "bus request" output from the Ethernet board onto the "bus grant" input.
Peter has also been reviewing the new User Interface, and is quite taken with it. The voltage and time cursors are very impressive - Tektronix should do as well.
Ian Murray has just given Peter the latest version of the User Interface, minus last week's bugs, plus a new set-up dialogue which is consistent with the rest of the interface, and Burst-View.
During the week Ian found and fixed an alignment bug between the Sun 3/260 and the Sparc 2 which explained most of the funny debug messages Peter had been seeing.
Roland Meins is continuing to work on the store and recall aspects of the user interface to the CAD package; he is getting to the point where he needs to talk about where he is going.
The pursuit of CAD data continues; the relevant people at ES2 were all out this week, and while the Institute of Microelectronics at Stuttgart has promised to quote him for a test circuit with design data, they haven't sent anything yet.
Tony Edwards spent most of this week rewriting the brochure to suit Peter Crawley; next week he is off to Deerfield for three days to discuss the US launch of the EBT-2000.
Simon Dawes’ passion waggon is almost all welded up – only the rear wheel arches still need welding - and the next job is to cover up the welding with loads of plastic padding.
Simon went down to Southampton on Tuesday with Steven Fisher, to finish the minor service on the EBT-1 – the problem with the vector coils turned out to be a problem with the scan generator, which they fixed. There was quite a lot to do, so they were not able to test the machine too thoroughly, but Gerry has rung back to tell them it is working fine.
Simon was involved in the scan distortion work on PP1 with Chris and Martin; he took out C3 four or five times. He has also done some more work on cables with Tony Slater, and is planning on up-grading the +/-24V supplies on the plinth to stop them tripping out quite so regularly.
Chris Warner claims to have spent most of the week entering a project plan into SuperProject, which ought to pacify Peter Mitchell. He also visited CMS with Bob Taylor and Martin Wiseman, and has got a sample of Loctite 406 into the gun of P1 which is pumping down as low as it did without the sample.
On Monday and Tuesday Chris checked the field size available on PP1 with the column fully retracted - we get 2mm with some margin. He also tried the fish-eye mode, and got quite
a lot larger field - closer to 10mm - but with very poor resolution. Adjusting the lens currents to give a smaller spot gave an impossibly noisy picture; we might do better with an auxiliary detector on the chamber.
He and Simon looked at the sources of the low-magnification image distortion by successively removing extraction, filter and suppressor electrodes - all these cut off the beam at low magnification at roughly the same deflection, and when the beam passes close to the electrodes, the associated electrostatic fields bend the beam to produce the distorted low magnification images with which we are familiar.
Bob Taylor admits to having completed several pre-production mods this week, but refuses to go into detail, for fear of having the information incorporated into a SuperProject plan and used in evidence against him.
Tony Slater is still working on cable modifications; he thinks that there are still plenty left to do, but he isn't quite sure what Simon still has up his sleeve.
Barry Barker has got new stage assembled; when I talked to him he was about to put it on a trolley and drag it off for leak testing. Production are going to make a lifting frame for picking up the assembled chamber - the existing eye-bolts are a bit low low and the rope could foul the ball-screws.
Steven Fisher got the new C3 lens onto PP1 and working on Monday. He went to Southampton with Simon Dawes on Tuesday, and since then he has spent some time with Tony Slater sorting out the cabling that used to go through the panel on the back of the photomultiplier box, but most of his time has been devoted to the RAM Store and LIP from P1, which is being refurbished as a production test tool. There were a couple of hardware problems - bent pins in the edge connectors, a superseded PAL and a duff RS-232 link – but now he is onto writing test programs to run on the Syntel processor.
to: Richard Adams, Martin Wiseman, Chris Warner, Peter Milne, Simon Dawes, Bob Taylor, Paul Austin, Roland Meins, Andrew Dean, Tony Slater, Ian Murray, Maha Mahalingham, Madhusudan Patel, Steven Fisher, Tony Edwards
from: A.W.Sloman
subject: Progress to 4th October 1991
○ I should be able to start checking the issue 3 Trigger board layout next Monday.
○ Andrew Dean has checked the modifications to the Delay board.
○ Paul Austin has been up-dating circuit diagrams on Metheus.
○ Martin Wiseman has finished the part B specification for the Scan Generator.
○ Peter Milne was involved in the demonstration; since then he has mostly been working on making Burst Mode even more reliable.
○ Roland Meins had to rebuild the Plessey design database after the demonstration.
○ Ian Murray has corrected a number of minor bugs in the revised User Interface, but it went pretty well at the demonstration.
○ Tony Edwards has had requests to get the EBT-2000 to Semicon West, May 1992.
○ Simon Dawes had a good demonstration.
○ Chris Warner has been immersed in SuperProject.
○ Bob Taylor hasn't completed any mods this week, but there are several in progress.
○ Tony Slater admits to having done some work on cables.
○ Barry Barker expects to have PP1 pumping again on Monday.
○ Steven Fisher is off sick.
I spent the first two days of this week working on the Trigger board specification; on Wednesday Gary Askew's marked up circuit diagrams arrived, and I spent Wednesday and Thursday on Metheus, adding pin numbers and preliminary circuit references to the Trigger circuit database. While I was at it, I threw in a table of power supply connections, and representations of the unused gates.
I put a set of the revised circuit diagrams in the post for Gary this morning. Gary now expects that the tracking will be fit for inspection on Monday, but I'm going to ring him before I drive down to Knebworth.
Andrew Dean has eliminated the 700MHz oscillation in the time-to-digital converter on the Delay board, by increasing the relevant base-stopping resistors; the circuit still rings rather viciously at 700MHz, but the ramp looks better; he still has to take another histogram to check the performance of the circuit as a whole.
Clair has completed the modification to the issue 3 Delay board; the revised layout is now the issue 4 board. Andrew has checked the modifications; Clair and Jack will now exhaustively check the new layout against Andrew's marked up circuit diagram and should be able to release it for photoplotting sometime next week.
Andrew is now putting the modifications onto the Metheus circuit diagrams.
Meanwhile Mark Diver is taking Andrew's Metheus cable diagram for the whole machine, and re-entering parts of it into Autosketch on Graham's PC.
Paul Austin has spent most of this week on Metheus, updating circuit diagrams so that we can get them archived before the Metheus system vanishes. The Waveform Processor - Digital up-dates involved adding the grid references now available on the issue 3 board, adding the decoupling capacitors, adding a table of the power supply pins on the ICs, and showing the power supply pins to the board. Paul also up-dated the Ion Pump Monitor and the Waveform Interface circuit diagrams.
In the process he found a drop-off in the issue 5 Timebase Interface layout - a 7432 has been pinned as if it were a 7402 and the board will have to be cut and linked to work. Clair and Paul seem to have agreed to share the responsibility.
Martin Wiseman spent most of this week writing Spec Part Bs; the Scan Generator spec is done (which means that I should be reviewing it) and the Plinth Control Card spec is coming on. Martin also managed to find a UK source for the insulating pad to go under the CLC-221 amplifiers on the Waveform Processor - Analog. He had to buy 150 (five year's supply) to meet their minimum order charge, but the 34.50 it cost would only have bought a couple if we had had to make them in-house. Martin went through 19 of the 26 suppliers listed in the TI Index before he found anyone who supplied the part, so he probably spent more than the 34.50 in engineering time on the search.
Peter Milne lost the first few days of this week to the preparations for the internal demonstration; since then he has been working on making Burst Mode more reliable. He has cured a few minor bugs, and improved the operation of the Autogain routine by making sure that we always take histograms over 256 samples (8 samples are not usually enough).
I'm trying to organise my ideas about the statistics of our signals to the point where I can write out a procedure for setting up the photomultiplier and the digital gain in the Waveform Processor; Peter needs a new Autogain routine like a hole in the head, but we can probably tweak the existing code to get close enough to an optimal procedure.
In between the demonstration and Burst Mode, Peter has found time to speed up the line-shuffling for the Logic State Map – he hasn't done a full map yet, but individual segments now get shuffled very quickly. The lines are still not being shuffled into quite the right order — Peter's first fix didn't work – but he hasn't run out of hypotheses to test.
And the user can now write arbitrary labels into the Dual Port RAM, which is to say, onto archivable images.
The bug in the new networking software is still unresolved; Peter is investigating various ways of getting his hands on a bus controller card, and may drive his VME crate down to Quin Systems just to try it with their bus controller card.
Ian Murray's latest version of the User Interface worked pretty well for the demonstration; there were a few minor bugs, most of which Ian has already sorted, and Simon has some specific ideas about defaults which Ian will incorporate as he gets clear on what Simon wants.
The Logic Analyser and Logic State Mapping are not in the demonstration; Logic State Mapping isn't quite right yet, but could go in fairly fast once Peter finishes it off. Image archiving is waiting on the new networking software, which in turn depends on the bus control card (see above).
Roland Meins was involved in the demonstration; it revealed that the Plessey data base had been corrupted and Roland has had to rebuild the data-base. On the CAD user interface, Roland has got about as far as he can on the store/recall without getting some more feedback on what we actually need.
Neither IMS-Stuttgart nor ES2 has come back to him about data on other integrated circuits; INS-Stuttgart may yet give us a quote, but ES2 seems to have decided that there is no money in giving us data on our ASICs, and are giving Roland the run-around.
Tony Edwards is back from his trip to Deerfield to discuss the US launch of the EBT-2000; they are talking about having a machine at Semicon West in May 1992. Shades of 1988!
Simon Dawes spent the first half of the week working up the demonstration on the revised User Interface; he seemed happy with the way the demonstration went went but he wants some minor changes in the User Interface, and more control over the colours on the screen. He has spent the rest of the week sorting out cables, some of the time with Tony Slater.
Chris Warner seems to have spent most of this week on the PC, loading stuff into a SuperProject plan. Richard has also been working on the plan, which has given Chris time to write a bit
about magnification calibration for the EBT, and to put an Araldite 2001 sample into P1 to check its vapour pressure/outgassing (it looks okay).
Bob Taylor won't even admit to having completed a mod this week, though he claims to have been working on several.
Tony Slater is equally unforthcoming - he admits to working on cables, but while I saw the Blanking Head drawing up on his screen earlier this week, he denies actually having worked on it.
Barry Barker has got PP1 back together in the clean area; he expects to start it pumping again on Monday, when he can go over to putting the new stage on PP2. We haven't got a C3 for PP2 yet, but the old C3 from PP1 is being refurbished at the moment.
Steven Fisher is off sick today, so I don't know what he has been up to.
to: Richard Adams, Martin Wiseman, Chris Warner, Peter Milne, Simon Dawes, Bob Taylor, Paul Austin, Roland Meins, Andrew Dean, Tony Slater, Ian Murray, Maha Mahalingham, Madhusudan Patel, Steven Fisher, Tony Edwards
from: A.W.Sloman
subject: Progress to 11th October 1991
○ I've finished checking the Trigger Board.
○ Andrew Dean's Delay board is about to go to photoplotting.
○ Paul Austin has finished up-dating circuit diagrams on Metheus.
○ Martin Wiseman is working on the part B specification for the Scan Amplifier board; he has found a bug in the implementation.
○ Peter Milne has used the EBT to find a bug in the Plessey chip driver.
○ Roland Meins has had qualified approval for his CAD UIF.
○ Ian Murray has got the new UIF code into the SCCS system.
○ Tony Edwards is reconstructing the brochure.
○ Simon Dawes had another good demonstration.
○ Chris Warner is getting loose from SuperProject.
○ Bob Taylor claims to be conforming to his SuperProject plan.
○ Tony Slater claims to be conforming to his SuperProject plan.
○ Barry Barker has got the ball-screws onto the next PP2 stage.
○ Steven Fisher's second image processor jig is almost working.
I spent all of Tuesday, and a long morning today, at GS Designs at Knebworth, checking through the issue 3 Trigger Board layout; there were quite a few minor changes, and a number of 75R terminating resistors have become 51R resistors, which means a change to the S.400 parts list (not yet done), but the layout isn't so cramped that the changes were difficult, and where I did adjust track lengths, it was to centre the typical propagation delay within the tolerance window, rather than to get within tolerance.
Now that I've got track lengths and routings, I've retoleranced a number of critical paths to include propagation delays (with allowance for capacitative loading of the tracks) and I'm feeling tolerably confident about the design. I can't guarantee a complete absence of drop-offs; I found another one last Wednesday, which would have inverted the short Unblank pulses, and there could well be others lurking in the schematic. On the Valid system you can simulate the same circuit diagram that gets turned into the printed circuit layout, which would have caught almost all of my drop-offs.
Maha has a net-list on floppy disk derived from today's layout; it lacks some of the decoupling capacitors, but is otherwise up-to-date to the latest circuit diagram.
Gary Askew will be checking the layout against the circuit diagram next week; he may - with luck - get the photoplots done before the end of the week.
I also did some more work with Peter Milne this week, on signal statistics - otherwise "the overflow problem". I ended up giving him a 10-page memo on the subject, which contained most of the information he wanted, and I was able to scribble the rest in the margin of page 6. I've thought of an additional wrinkle, which may add another page to the memo.
I haven't done anything about the over-temperature sensor on the Sampling Crate.
Andrew Dean tried his up-dated Delay Board on P2; it produced double Unblank Pulses, but we should be able to eliminate the second pulse by terminating the Sampling Edge inputs to the missing Delay Board. Andrew is building a dummy Delay board on an unloaded issue 2 PCB to handle this and other minor difficulties.
The issue 4 Delay board has now been checked by PCA and is ready for photoplotting; the parts list has been up-dated, and the new part numbers raised.
Andrew has also put in some time testing and fixing his column simulation box; it should work down to 50nsec Unblank Pulses and 80nsec inter-sample intervals. So far the biggest problem has been getting the right sockets onto the box to match the existing cables.
And Andrew got involved with the fixing of the Plessey chip driver on P2.
Paul Austin has finished his stint on Metheus on the circuit diagrams for the Waveform Processor - Digital and the Timebase Interface; they are now fit for archiving. Paul has now moved on to fixing the LIP; at the moment he is trying to restore the "front porch" on the synch pulse without messing up Martin Wiseman’s old mod to stretch the active line.
Martin Wiseman is still writing Spec Part Bs; he is now onto the Scan Amplifiers, where he had found that the Lintech-designed decade-switching resistors change the scan amplification by rather less than one decade; this might explain some of our problems in matching the CAD image to the SEM image.
Martin had to put in another session on the issue 3 Blanking Board - Valid had some more problems with the various power pins on the circuit - but Kevan Purton thinks that we have now got over the power pin problems, so Andy should be able to get on and lay the board out.
Martin also put in a little time fixing his driver for the Plessey chip - the wired-wrapped 0V wiring was a little light for the currents involved, and was dropping 80mV. Heavier wire improved matters appreciably.
Peter Milne achieved something of a break-through this week - he thought that he had got all the bugs out of Burst Mode, and the problems that he had left looked like there was something wrong with the Plessey chip or its driver. When hardware were finally persuaded to look, we found that the earth straps had been taken off the sockets (to make them look neater for the publicity photos) and not restored, so that the whole circuit and power supply was floating.
When Martin and Andrew had restored the earth straps, and beefed up the 0V wiring on the driver, the Burst mode images looked much nicer, and really impressed Dr. Brassel. The overflow problems also seemed much reduced, even though Peter has yet to up-grade the Auto-Gain routine.
The latest stage in networking saga is that Motorola have sent Peter a bus controller card (for free) and are now in the process of sending him the manual, so that he can use it. Pete is impressed.
Ian Murray has got the latest version of the User Interface (UIF) into the Source Code Control System on the Sun – this provides complete documentation of every change (until the code overflows the disk) and on the one hand makes the UIF source a good deal more secure, and on the other makes it harder to up-date. Ian must be feeling confident.
Ian has been doing some up-dating of the UIF with Simon (for user-definable colours) and with Chris Warner to provide hand-holding through the Device Under Test set-up procedure.
And Ian has fixed the Timebase bug, which overwrote the expanded timebase with the reference timebase when you had got the expanded waveform you wanted and tried to store it
(an irritating bug, that one).
Ian is now preparing a series of notes on the UIF for marketing; just a brief description of the buttons – where they are and what they do - as a sort of rudimentary manual.
Roland Meins seems to have got the zooming and locking very nearly right on the CAD UIF — Simon is pretty happy with it, and is now proposing to use the CAD’s figures for feature positions to calibrate the SEH magnification. We might get the user to highlight a third pair of points and calibrate orthogonality as well.
IMS-Stuttgart has come back with a quote for a small batch of integrated circuits (5 - 20) of about 5,000DM, with another 5,000DM for the documentation, which would take quite a lot of further work to convert to schematics that we could use. The figures are for non-exclusive supply – an exclusive design would be more expensive. Schlumberger are reputed to be interested; ICT/Advantest is not interested.
ES2 continues to give Roland the run-around - maybe we should try to get Peter Mitchell to talk to them.
Tony Edwards has got his L-plates on the EBT; between driving lessons he is rethinking the brochure.
Simon Dawes gave a rather successful demonstration for Herbert Brassel. Simon has put in some time testing the CAD UIF - magnification and zoom worked quite well, though there was a bug when the CAD magnification dropped below the 100x lower limit on the SEM.
Simon has also been talking to Tony Slater about cables, and to Ian Murray about the UIF, as well as trying to set up P2 for resolution (the fan on top of the Sampling Crate doesn't help).
There has also been a discussion with Russel Goodenough about carrying over holiday into next year - at present it seems that Simon will have to take 4.5 weeks of holiday in the next 9 weeks, which will rather crimp Chris Warner's SuperProjections.
Chris Warner has only had to spend two days on SuperProject this week, which has left him time to work on the UIF with Ian Murray, on the vacuum properties of our glues (okay) and on the vacuum feed-throughs for the Blanking Head (a 0.5mm hole isn't big enough to accommodate a 0.5mm coax inner from our conformable cable).
Two days on SuperProject were enough to give Chris the fixed idea that Simon's tasks wouldn't fit into the time available - he didn't say how much holiday time he was allowing for.
Bob Taylor and Tony Slater both refer me to their SuperProject task lists for their progress this week - unfortunately I only read science-fiction during my leisure hours, so I can't expand these references.
Barry Barker is assembling another stage for PP2 - since the original PP2 stage is now on PP1 - and he has built up the new stage to point of putting on the lead-screws, which is fairly well advanced.
Steven Fisher is putting together a second image processor jig; he has been getting live images, but there are no stored images so far. He has to modify the PP1 and PP2 mains distribution units fairly soon to incorporate the safety mods.
(Note added 2012/02/23. Russel Goodenough was in charge of personnel at Cambridge Instruments at the time. Personnel was never a particularly useful or helpful department in any of the UK companies I worked for.)
to: Richard Adams, Martin Wiseman, Chris Warner, Peter Milne, Simon Dawes, Bob Taylor, Paul Austin, Roland Meins, Andrew Dean, Tony Slater, Ian Murray, Maha Mahalingham, Madhusudan Patel, Steven Fisher, Tony Edwards
from: A.W.Sloman
subject: Progress to 18th October 1991
○ I've been working on the Trigger board Spec Part B.
○ Andrew Dean is testing his column simulation box.
○ Paul Austin is working on the LIP.
○ Martin Wiseman is adding +/-24V capacity to the plinth.
○ Peter Milne has improved the joystick.
○ Roland Heins is working out his sub-cell orientation.
○ Ian Murray is adding help to the set-up menu.
○ Tony Edwards is getting screen dumps from the Sun.
○ Simon Dawes is cleaning up the column on P2.
○ Chris Warner is designing the DUT set-up procedure.
○ Barry Barker has dismantled the PP2 column.
○ Steven Fisher has got duff stored images on the second image processor jig.
I've spent most of this week tidying up the issue 3 Trigger board documentation; I've given Richard a mod to change 35 of the 75R terminating resistors to 51R - the tracks were rerouted onto the buried signal layer - and make three other changes required by the layout. I've also reworked my Metheus simulation of the Long Unblank Pulse Generator to fit in with my spec part B description of the circuit, so I can bind the diagrams and the waveforms into the specification.
I also rewrote parts of my "signal statistics" memo for Peter Milne, to cover "bin 63" overflow detection with less than one electron per sample; Peter has coded the important stuff as the "Son of Bill" autogain, and it seems to fix the problems we had been having at low beam currents and narrow unblank pulses. There are some refinements that could go in once we are confident that the system does run into the problems they are designed to avoid; this won't be until we are working down below 0.1 electrons per sample, or lower, where the machine will be rather slow.
The issue 3 Trigger board photoplots won't be ready until early next week - we are planning to have contact prints off the masters sent directly to Printech from Knebworth, to save the day or two it would take to get them through our system.
I've made a start on the over-temperature sensor for the Sampling Crate; the Variomatic fans look to be useless, so it will probably be a snap-action thermostat or two on the top of the Sampling Crate.
Andrew Dean has made up his dummy Delay board, and put it into P2 - it stopped the double Unblank Pulses, as expected. He spent Tuesday with Peter on the difficulties in using the joystick to move the stage over short - single step - distances; one of the problems they found and sorted out was that the "gate" input to the stepper motor controller has a 100usec set-up time, which the software didn't provide, though it does now.
Andrew has also spent time on planning a test strategy for assembling Sampling Crates - one problem that he is trying to anticipate is that we don't have enough bench space in the Clean Area for two sampling crates with their power supplies and our test gear.
At the moment Andrew is testing his new column simulation box - the connectors he needed have arrived, and Simon has taken the column off P2, leaving a real opportunity for a column simulation box.
Paul Austin has spent most of this week on the LIP, trying to get the blanking, the synch pulses and the video all occurring at the right times; he has completed a mod which ought to do this, but he doesn't like the results he gets.
Paul also put in some time helping Steve Fisher with the second image processor test jig in the Clean Area - it is improving, but still imperfect - as well as checking the sub-contractor’s work on the Waveform Interface layout, where he found a couple of errors, one of them ours. The new layout should be finished by the end of next week.
Martin Wiseman took three days off and has spent most of the other two helping Simon to beef up the +/-24V power supplies on the plinth with two more 3A units; these should allow us to drop our minimum magnification to 50x. Martin also got involved with the Blanking board layout again; Valid managed to lose some of the gates from some of the multi-gate packages, which created alarm and consternation until it was fixed; the up-shot is that only now is the board layout getting under way.
Peter Milne added the "Son of Bill" autogain to the "Son of Stan"; the new software controls photomultiplier and digital gain for low beam currents and narrow unblank pulses, where
there is less than one detected secondary electron per ADC sample. It seems to do it better than "Son of Stan" did.
Peter spent Tuesday with Andrew Dean looking at the joystick control of stage movement; they found a problem with the "gate" input to the stepper motor drive card (see above) and Peter found a few other bugs; he thinks he has fixed all the bugs he found, but he hasn't been able to test all the fixes yet.
Peter has made some progress with the networking problem; with the Motorola bus controller card he can Telnet reliably to himself, but nowhere else - he is going down to Quin Systems next week to board swap on a working equivalent of our system until he finds out what is going wrong.
Ian Murray is agreeing a Device Under Test Set-Up procedure with Chris Warner so that the user can tell the system where the chip is and how high it is, amongst other things; they seem to have agreed Expert and Operator modes of presenting the choices, and short and long Help formats for the various buttons involved.
Page 3
Roland Heins has sorted out most of the residual bugs on the CAD UIF; the colours are now acceptable to Simon, but the cell orientations are still random. Roland expects to sort out the rest next week.
Roland is proposing to ring ES2 again this week about schematics, more to wind them up than in any real hope of getting the information.
Tony Edwards has got a printer to take screen dumps in colour off the Sun on P2, to give potential customers pictures of the windows in action; the new brochure is hanging fire until Peter Crawley gets back from his travels next week.
Simon Dawes spent today and yesterday cleaning the column on P2 (not before time); it is now reassembled and pumping. Earlier in the week he spent time with Roland on the CAD UIF (see above) and has written another memo on cross-probing from the net-list, where he high-lights the sub-cell random rotation bug.
Chris Warner has spent most of this week on the DUT set-up procedure for the UIF - some of it in discussion with Ian Murray, and some of it working out how to measure the collision distances and the silicon distances that we will need to let the user work safely and effectively; by forcing the user's hand in the set-up procedure Chris expects to reduce the data collection and curve fitting he will have to do to establish estimates of working distance and magnification.
Bob Taylor and Tony Slater are still hiding behind SuperProject, so I don't know what they are doing – Richard says that Bob has done 15 days worth of tasks in the last ten days, and Tony only 6, but Tony has done 3 days worth of work on the move, so his task estimates may be accurate to +/-10%.
Barry Barker has got most of the PP2 stage assembled and is now waiting for bits before he can blank off the column hole and getting it pumping - he expects to have it pumping early next week. Barry had to dismantle the PP2 column to provide Simon with the bits to replace the brass fittings that were still lingering in P2’s column; since the PP2 column is still waiting for a C3 lens and a new set of scan coils, this probably hasn't delayed the final assembly, but it has wasted quite a lot of Barry's assembly time. He doesn't mind - he got paid for it - but it makes the austerity program look a bit silly.
Steven Fisher is still working on the second image processor test jig; he has now got stored images, but some of the lines on the replayed are misplaced; Paul Austin spent an afternoon on it with him, but they are still mystified.
Steve is also working on getting PP1 back to work; at the moment they can't get the UIF on the Sun to drive the Syntel, but they have a fix to test.
to: Richard Adams, Martin Wiseman, Chris Warner, Peter Milne, Simon Dawes, Bob Taylor, Paul Austin, Roland Meins, Andrew Dean, Tony Slater, Ian Murray, Maha Mahalingham, Madhusudan Patel, Steven Fisher, Tony Edwards
from: A.W.Sloman
subject: Progress to 25th October 1991
○ My Trigger board is being photoplotted.
○ Andrew Dean has up-dated the Delay board circuit diagrams to issue 4.
○ Paul Austin is still working on the LIP, but on a different problem.
○ Martin Wiseman is working on the Plinth Control Card spec B.
○ Peter Milne is moving sub-directories over to the cross-compiler.
○ Roland Meins has worked out what he has to do to fix sub-cell orientation.
○ Ian Murray is putting a chip editor into the UIF.
○ Tony Edwards is rebuilding the Sales Demonstration package.
○ Simon Dawes has spent the week preparing for today's demonstration.
○ Chris Warner has worked out how to block the C3 light leak.
○ Barry Barker has dismantled the PP1 column to get rid of the brass bits.
○ Steven Fisher has had the PP1 stage moving about.
I‘ve spent most of this week working on the Software Interface document for the issue 3 Trigger board; it isn't half done yet, and already up to ten pages. In the process I found another drop-off in the issue 3 Trigger Board design; I faxed a last modification request to Knebworth yesterday to get Gary Askew to move three tracks by one pin each - it will give us one more Encode Pulse at any given Long Pulse width, and I think we need it.
Since’ Gary expected to set the board photo-plotting yesterday evening - that is the last of the cheap corrections. The issue 3 board ought to get to us on the 25th November; it seems unlikely that Purchasing will let us assemble a complete parts list by then - there are 30 shortages, and quite a few haven't been ordered yet, and of the others that do appear to have been ordered, some parts that should have been delivered last month are still outstanding, which gives cause for concern.
I've also put in about a day on the Sampling Crate cooling - I now know what I want to do, and I've got to write it up in the form of a mod.
Andrew Dean has spent some time on Metheus, putting together an archivable set of circuit diagrams for the issue 4 Delay board, including the IC grid references from the issue 4
layout.
He has also tweaked the column simulation box, so that it can handle slightly narrower Unblank Pulses, put in some time planning next month's performance tests on P2, and put in some more time working out where to put all the test equipment that we want moved into the Clean Area.
Andrew has been involved in working up P2 for today's demonstration; so far the demonstration team has found a dodgy 0V connection inside the Blanking Head, a fault on the Waveform Processor Digital, which prevented the Head Amplifier Offset from being set up correctly, and a fault on the Waveform Processor Analogue, which left the high voltage suppressor grid drive non-working.
A more mysterious fault appeared on the Encode Delay; this had to be extended past 120nsec for the first time on Wednesday, and it isn't obvious why.
Paul Austin has spent a large chunk of this week on the LIP; he can get the blanking and the synch pulses in the right places in two different ways, but he is now trying to find out why we only have 51.2usec of active video, rather than 52usec - we lose about 12 pixels out of our 768. Some of the work was done on the clean room LIP, some on the demo machine.
Paul also spent two days getting a RAM Store crate running in the clean room - it took a lot of board swapping but he finally got a Syntel which would talk to him.
He and Steven Fisher and Martin Wiseman had a joint session on a Plinth Control Card and a Timebase Interface; Martin traced the fault to the Timebase Interface card, where Paul
found and fixed a dry joint.
The issue 5 Timebase Interface layout is completed; the schematics have been updated, and the artwork should be going for manufacture soon.
Martin Wiseman is writing some more spec part B for the Plinth Control Card; he will be up‑dating the circuit diagrams on Metheus as soon as he gets his marked-up prints back from PCA.
Kevan Purton has not made as much progress as had been hoped on Martin's Blanking board; Valid was treating +5V, Vcc, -5.2V, Vee, 0V and GND as separate power supplies, so Kevan had to edit the netlist. Martin has had a netlist, which isn't much use to him because it seems that all the ICs have been renumbered; it also looks as if the gates have been shuffled around within packages - which wouldn't normally matter, but does when the layout is critical.
Peter Milne has finally started getting his software reorganised so that he can cross-compile on the Sun and exploit their elegant and powerful source-code control facilities; he has moved three out of ten sub-directories so far and is enjoying being able to recompile the lot in one minute (on Roland's Sparc 2 - on Peter's 3/260 it takes 10 minutes, which is still fast).
The software update from Quin didn't solve the Ethernet problem; Peter expects to get a fixed price quote from John Lamb at Quin to take our Syntel crate, with its Quin Ethernet card, and return it working.
Peter has been dragged into the preparation for today's demonstration for Ian Cruttwell; he found a bug in the routine for calculating the closed loop set-point, and closed loop now stays closed down to much narrower unblank pulse widths. There were also problems with the hardware - see Andrew Dean's week above.
There are a couple of known problems with "Son of Bill" as it is presently coded which will make it unreliable at very low nett beam currents; Peter thinks that it will take about a fortnight to extend "Son of Bill" to cope, and get rid of any bugs in the extensions.
Ian Murray is coding the detailed features of the Device Under Test Set-Up procedure - he showed me a package editor, with which the user can tell the EBT how many pins his package has, how wide and long it is, and where pin 1 is. The user has to tell the system how high the package is at the same time, which will make Chris Warner's job a lot easier until the users start keying in arbitrary numbers to save the effort of looking up or measuring the actual height.
Roland Meins has worked out how to extract sub-cell orientations from his design database; coding the solution is going to take about a month. Before he is committed to tackling that, Richard is getting him to spend another couple of days trying to find a techinque for cross-probing between the Plessey layout, net-list and schematic; there ought to be somewhere on the chip where it will work!
Roland is continuing to prod ES2 about schematics; they are still giving him the run-around.
Tony Edwards is rebuilding the Sales Demonstration pack, otherwise known as Dave Hall's Bumper Book of Prevarications.
Simon Dawes spent the beginning of the week dismantling and reassembling the column and TTLD; we had the Blanking Head apart at one point when the unsound, if heretofore reliable,
Lintech ground return went open circuit; the original metal-to-metal contact has now been reinforced with a bit of soldered wire.
Over the last couple of days Simon has been working up the assembled machine for today's demonstration; it is imaging really well, even if the Cursor card has gone inaccurate.
Chris Warner has been through the complete User Interface Specification; he hasn't said what he found. At a more basic level he negotiated a supply of black bathroom sealant to block the light leaks into the C3 housing.
On SuperProject he has worked out a "baseline" (that is, ostensibly feasible) project plan for the next couple of months; we have already fallen behind because PP1 wasn't available this week for Martin's magnification calibration work.
Almost all the parts for the new Blanking Head are now available, blanking plates excepted; Tony Slater still has to draw the input section with its four SMA plugs in a rather small area.
Bob Taylor assures me that he is still SuperProjecting to plan, whatever that may mean, while Tony Slater has decided that I am an unreliable reporter, and has put up a screen of no-comments.
Barry Barker has been dismantling the PP1 column this week to replace all the brass bits with aluminium; the PP1 column was no better than P2’s (see last week's report). He also stripped down the stage far enough to let him put the mu-metal shield on the turbo-molecular pump.
Steven has got the old issue software working on PP1 – he can't get‘ the latest issue software working without an extended memory card, which isn't available. With Martin and Paul he got the Timebase Interface working to drive the Plinth Control card - he had the stage motors moving the stage around. Steven has also done some testing on some Dual Port RAM cards.
to: Richard Adams, Martin Wiseman, Chris Warner, Peter Milne, Simon Dawes, Bob Taylor, Paul Austin, Roland Meins, Andrew Dean, Tony Slater, Ian Murray, Maha Mahalingham, Madhusudan Patel, Steven Fisher, Tony Edwards
from: A.W.Sloman
subject: Progress to 1st November 1991
○ My Trigger board photoplots are at Printech, waiting for an order.
○ Andrew Dean is on holiday.
○ Paul Austin is still tracking down the missing pixels.
○ Martin Wiseman is working on the Cursor calibration problem.
○ Peter Milne has moved some more sub-directories over to the cross-compiler.
○ Roland Meins may have found a source of demonstration chips at Manchester.
○ Ian Murray has found the "vanishing UIF" bug, and may have fixed it.
○ Tony Edwards has produced a new specification.
○ Simon Dawes is on holiday.
○ Chris Warner is wrestling with PP1.
○ Barry Barker has started reassembling the PP2 column.
○ Steven Fisher has made some more bench space in the Clean Room.
I've spent most of this week on the Trigger Board Software Interface document, and it isn't much closer to completion - I'm stuck on the Short Unblank Pulse generator calibration procedure, which is less of a mess than it used to be, but still rather incoherent.
I have finished my notes on the Sampling Crate temperature monitoring and raised a part number for the tab-mounted thermistor I think we ought to use.
Copies of the artwork for the issue 3 Trigger board have arrived here and at Printech; Printech won't start making a board until management have decided that the EBT project is continuing. We have yet to get the solder paste masters, the "assembly" drawing that identifies every component on the board (there isn't room on the silk screen to identify all of the resistors and capacitors) or the floppy disks containing the design files, netlist and Gerber files, but Gary Askew‘has promised them for early next week - since we don't actually need them for a few weeks yet, this should be good enough.
I've got to use the silk-screen plot to up-date the grid references on the issue 3 circuit diagrams, which shouldn't take too long - the schematics should then be fit for archiving.
Paul Austin is still looking for our missing pixels; some seem to be lost in the LIP and a similar number on the 8-Plane storage cards. Richard is thinking about living with 750-pixel video lines. Paul had a little trouble with the Clean Room LIP - its +5V power supply plugs into a dodgy connector on the LIP backplane, whose contacts had gone brown and oxidised, and the "+5V" rail had gone down to 4.2V. Paul soldered the leads directly to the board, and the rail is now up to 4.8V, just within tolerance; it now works better, but it is still swallowing a few pixels.
Martin Wiseman has spent most of this week on the Cursor Board - it doesn't give a linear relationship between cursor position in pixels, and the corresponding sampled, held and digitised scan voltage, and Martin is trying to work out where this non-linearity arises - he suspects the LF-398 sample-and-hold.
Kevan Purton is about 75% of the way through tracking the new Blanking Board; Valid hasn't messed up anything this week, and the package more or less confirms our choice of track widths - it thinks our 75R tracks have a characteristic impedance of 76R, and our 50R tracks, 53R. Close enough.
Peter Milne has got a bit further with moving the Syntel software over to cross-compilation on the Sun; the problems with the Ethernet links are still unresolved but John Lamb from Quin systems is prepared to look at the problem for 500 pounds per day; he would expect to solve it in less than a day.
Peter has been doing some more work on the autogain algorithm; he has worked out a way of extending Son of Stan, as Better Son of Stan, down to very low signal levels; Better Son of Stan still looks at bin 63 from the ADC output, but now ignores bins 0 to 5 (roughly), and accumulates extra samples if there are less than 100 (we may need to increase this) samples outside the ignored bins. This is tidier than switching to Son of Bill, and less dependent on dodgy statistical inferences.
Peter has also been looking at the stage motor problems with Simon Dawes; the stage still refuses to move the last 3 microns to the desired point about 30% of the time, and goes the wrong way on small moves about 20% of the time, but stage repeatability is within +/-3 micron at all times, and +/-1 micron 70% of the time. It looks as if there is a bug in there, but it may take ages to find, and is the performance bad enough to justify spending the time to find it?
Peter fixed the column Z problem; the stage X and Y processes were occasionally failing to terminate and locking out the Z motor drive.
Ian Murray has identified, and probably fixed the bug that cause the UIF to vanish from view on the Sun - a chunk of memory was being corrupted, and shouldn't be any more.
He suspects that the bug which causes the UIF screen to lock up is actually in the Sun Memory Manager, which makes it rather hard to fix; there is now a procedure for unlocking screen without reseting the Sun, which is an improvement.
Ian has spent most of the week on the Device Under Test Set-Up procedure (except that it now seems to be called Status Application). He has advanced to the point where the data which the users feeds in is now available to the operating software, so the hardware could now know when it is going to hit a user-specified sample.
Roland Meins has looked at the possibilities of getting getting cross-probing between the Plessey chip layout and schematic, and has decided that there aren't any that he could get to in less than a week's work.
Roland went to the Silicon Design show at Heathrow on Wednesday, and had a fun time bugging the ES2 stand about EDIF schematics; they found it harder to give him the run-around within ear-shot of prospective customers. More usefully, he found an ASIC contractor who uses the same schematic package as we do, and were pretty happy with it; they have spent 8-man months getting a moderately intelligent EDIF converter written for it, which is a useful number for Roland's SuperProject plans.
Roland and Chris Warner went to Chris's old department at Manchester on Thursday, and formed the impression that we might well be able to get chips and data from them rather economically; they don't have all the data we would need, but it sounds as if we might be able to get it.
Tony Edwards has completed the Sales Demonstration package, ready for Peter Crawley to take around the Far East next week; he has also prepared a revised specification for the EBT-2000, which only claims the level of performance we have already demonstrated - 500psec blanking, rather than 200psec, amongst other things. Sanity at last.
Simon Dawes has been on holiday since the move; he moved a lot ‘of heavy stuff on Tuesday, and my back did a lot of sympathetic twinging.
Chris Warner has passed a significant milestone; he has been here two years and is now qualified to received compensation for being made redundant - if he was working on some other project this might be of less significance.
He went to Manchester on Thursday with Roland, as mentioned above - Chris mentioned that it rained on Thursday, which Roland failed to report.
Chris has spent most of the week on PP1, and had a source image on it, before the Z-motor went nuts and mashed the column 0-ring.
Bob Taylor has spent most of this week talking to Arthur Nightingale about the work that Bob will be picking up when he moves over to EBMF next March to replace Arthur (about two man-years of work according to Arthur). But Bob will be back to SuperProjecting next week.
Barry Barker has finally collected all the replacement parts for the bits that were pirated off the PP2 column to replace the brass bits in the P2 and PP1 columns; he has now started reassembling the PP2 column. He still has to wind a new set of scan and stigmator coils for PP2.
Barry also put in some time this week locating the PP1 cladding and sorting out some minor drop-offs.
Steven Fisher has reorganised the EBT bay in the Clean Room to provide considerably more bench space; it looks as if there is as much as Andrew Dean thought that we would need.
He has also been working on PP1 with Chris Warner; they have been having trouble getting control of the stage motors through the UIF. And while PP1 has been pumping down, Steven has found time to do some more work on the Dual Port RAM Cards; they still aren't working.
to: Richard Adams, Martin Wiseman, Chris Warner, Peter Milne, Simon Dawes, Bob Taylor, Paul Austin, Roland Meins, Andrew Dean, Tony Slater, Ian Murray, Maha Mahalingham, Madhusudan Patel, Steven Fisher, Tony Edwards
from: A.W.Sloman
subject: Progress to 8th November 1991
○ Richard Adams has been made redundant.
○ Bill Sloman has been made redundant - but he has finished the Trigger board spec part B. .
○ Andrew Dean remains to archive the hardware, and keep P2 alive against an eventual buyer.
○ Paul Austin has been made redundant.
○ Martin Wiseman has been made redundant.
○ Peter Milne remains to archive the software and keep P2 alive against an eventual buyer.
○ Roland Meins has been made redundant.
○ Ian Murray has been made redundant.
○ Tony Edwards has been transferred to write manuals for the S.300.
○ Simon Dawes job seems to be unchanged; he is still keeping P2 alive against an eventual buyer.
○ Chris Warner has been made redundant.
○ Barry Barker is back on EBMF.
○ Steven Fisher has been made redundant.
Now I know why Graham Plows resigned …
(Note added 2013-02-24. Graham Plows had resigned about a month earlier – I don't know exactly when, but his last mention in the weekly reports is on the 20th September 1991, and in the next week's report Tony Edwards is re-writing the EBT-2000 brochure to suit Peter Crawley, who took on the responsibility for selling the EBT-2000 after Graham bailed out, and decided – after a couple of weeks of collecting data on potential customers – that we couldn't sell the 18 machines in the following 18 moths that were needed to get the cash flow right.
I don't know if Graham had been over-optimistic, or Peter Crawley was being unduly careful, but I had rather more faith in Peter Crawley's grasp of reality than I'd ever had in Graham's.
About a week before the redundancies had been announced. I'd had an interesting interview with Andrew Dean (who was – officially – my supervisor, though neither of us took that all that seriously) and Richard Adams (who was the project leader – and I'd been one of the many people who'd said that he was the obvious replacement for Dave Hall - and I did take him seriously) who wanted me to agreed to a change in job-title, from something like “project engineer” to “staff consultant” which was – with hindsight – a transparent attempt to get me off the list of people who were going to be made redundant when the project was canned.
I wouldn't have anything to do with it. Not because I was consciously intending to go down with the sinking ship, but because I'd met a number of “staff consultants” during my time in British industry, and spent three years as a “staff engineer” at ITT-Creed (1979-1982) and really didn't want to play that role.
I don't think that I was wrong. I'd already survived three rounds of redundancies at Cambridge Instruments. The first one had gotten rid of a lot of dead-wood, but the next two had lost us a number of people that we really should have hung onto.
I'd gotten shunted into working as a solo engineer on the Metals Research Crystal Puller after the Alvey shaped-beam electron-microfabricator project had been cancelled, which occasioned the second round of redundancies. This was a bit of step-down from supervising a team of hardware engineers for that project. The atmosphere in a factory after a round of redundancies isn't great.
It isn't exactly
Better to live on beggar's bread
With those we love
alive,
Than taste their blood in rich feasts spread,
And
guiltily survive!
But it was comforting to be able to walk away from Cambridge Instruments and not have to worry about how they were making themselves less viable, year by year.)
to: Richard Adams, Martin Wiseman, Chris Warner, Peter Milne, Simon Dawes, Bob Taylor, Paul Austin, Andrew Dean, Tony Slater, Ian Murray, Madhusudan Patel, Steven Fisher
from: A.W.Sloman
subject: Progress to 2nd December 1991
I've got a temporary job with Cambridge Mass Spectrometry for the next two months, assembling, testing and debugging an isolated and floating scanning crate for one of their scanning mass spectrometers. There is some prospect that this might lead to a permanent job with Kore Technology when this splits off from CMS when Kratos moves the business up to Manchester in the new year.
I also had a rather satisfactory interview with Xaar Ltd on the Cambridge Science Park last Wednesday, and I've got an interview with Elmjet on the 9th January (two days after Chris
Warner's).
Andrew Dean has picked up a list of jobs from AGD Recruitment, which I am enclosing with this report; he is still archiving, and awaiting a serious talk with Peter Mitchell about his long term prospects.
Paul Austin has an interview tomorrow with Rank Xerox in Welwyn Garden City, for a long term sub-contract job, and an interview with Camtec in Cambridge on Friday for a permanent job.
Martin Wiseman has had an interview with ED computers ( Paul Austin and I both got knocked back without interviews) but he didn't like them, and didn't think that they liked him either.
Peter Milne has been shifted to the EBML, and is now cobbling up a user interface for the Esprit machine based on the EBT interface; it actually runs on the EBT Sun, but is displayed
on the EBML VAX, which acts as an X-server. He says that it is very hairy.
Ian Murray had a sort of interview last week through an acquaintance, and expects a decision in a couple of weeks; he is now looking in London rather than Cambridge.
Simon Dawes is at home from now until Christmas, catching up on his accrued holiday. Leica‑USA seems to be interested in hiring him to work in Phoenix, and he expects to fly out there shortly, at their expense, so that he can spend a week out there seeing them and being seen by them.
Chris Warner’s name has gone forward from an agency to a west country firm, but he hasn’t got an interview out of them yet; he talks to Elmjet on the 7th January.
Bob Taylor has resigned from Leica-Cambridge and will start at Cambridge Product Innovation at Sawston on the 20th January, as Engineering Manager with a company car and a salary which is reputed to have defeated Peter Mitchell's attempts to bribe him to stay.
Tony Slater has an interview coming up in a couple of weeks, but in the meantime he is painting and decorating and building and labouring full time - weekends included - and says he is already booked up to Christmas.
Steven Fisher has had about five interviews, and one job offer that would have paid him so little that he couldn't afford to accept it. A couple of the interviews could still lead to job offers, but he is out there hunting for more.
Mahdu Patel has had a little sub-contract work from a friend of his, doing assembly language programming. He has had one interview so far (with Keyboard Controls), and has another
scheduled for later this week in Welwyn.
Richard Adams has an interview coming up in London early next week with SGS-Services and another with Diomed on the Science Park.
Late news extra - 5th December - I got called in for an interview at Diomed myself, and arrived this afternoon to meet Richard leaving. The job as described at the interview was rather different from the job Diomed advertised, and Tim Frost says that they had about 15 applicants, of whom they are interviewing about six. And Tim himself is threatening to apply.
From Leica-Cambridge there is a report that Nick Campbell has resigned, and one rumour via Mark (Dodgy) Saunders (ex-Lintech) suggests that he might be going in with Graham!
Peter Crawley seems to have tried to sell the EBT to Karl Zeiss of Jenna, who turned out not to have enough money for decent suits, let alone EBT’s. Their managing director was at Leica last week, but no deal was struck.
to: Richard Adams, Martin Wiseman, Chris Warner, Peter Milne, Simon Dawes, Bob Taylor, Paul Austin, Andrew Dean, Tony Slater, Ian Murray, Madhusudan Patel, Steven Fisher
from: A.W.Sloman
subject: Progress to 13th January 1992
Two of us have now got permanent jobs - Paul Austin and Steve Fisher. Mahdu Patel and Ian Murray have both got several months of contract work, likely to be extended to a year. Tony
Slater and I are still in temporary work, but Chris Warner has a bad back which is going to put a crimp in his job-hunting for the next few weeks.
I’m still working for Cambridge Mass Spectrometry at Bar Hill ~ I've started testing the crate I'm working on, but there are still some weeks of work to go. My interview with Elmjet was
cancelled (with Chris Warner’s) ~ the recession has frightened them out of hiring anybody. But I've got an interview with Domino (industrial ink jet printers) next Monday.
Andrew Dean is is now looking at EBMF stages for the Esprit project, which is probably more secure than archiving the EBT,
Paul Austin was offered a job just before Christmas, at Addenbrookes in Medical Equipment Design Maintenance, and he started on the 6th January. His first task is design a device to measure stomach volume ~ the patient swallows a balloon, and Paul’s machine then blows the balloon up until it fills the stomach, then tracks the stomach volume until the consultant loses interest.
Martin Wiseman had an interview with Cotag just before Christmas, which seemed to go fairly well, but he hadn't had a response from them last week.
Peter Milne has resigned from Leica-Cambridge and will leave on the 3rd February. Like Mike Penberth and Nick Campbell, he won't say where he is going, except that he isn't going to join
Graham.
Ian Murray's contract is for four months in the first instance, with some bank in London, and he will be commuting for the moment, but if the bank extends the contract they will probably move down.
Simon Dawes was still at home when I talked to him, still using up his accrued vacation. Leica—USA still seems to be interested in hiring him to work out of Phoenix, and he was hunting for his birth certificate when I called - Phoenix want to start applying for his work permit now, even though the deal with Leica~Cambridge hasn’t been entirely worked out yet, and could still fall through.
Chris Warner’s Christmas break with his parents in Switzerland was marred when he did his back in carrying his suitcase and his son into Heathrow on the way out – apparently it was his son's gymnastics that did the damage – and he saved a certain amount of money by not skiing around the Swiss Alps. His doctor thinks that three weeks of bed rest should fix the back.
Bob Taylor is away in Sweden - enjoying a break between Leica-Cambridge and Cambridge Product lnnovation. Apparently Leica-Cambridge let him go without the usual ceremonies, which is a bit odd.
Tony Slater is still painting and decorating almost flat out - although he did take a break over the weekend of the 4/5th January, his first weekend off since he was made redundant.
Steven Fisher has got a job as a Senior Test and Repair Engineer with Tadpole Computers (they didn’t even short~1ist me ... but that was for a Design Engineer job). Like Paul he was offered his job just before Christmas, and started on the 6th January.
Mahdu Patel’s sub-contract assembly language programming has apparently expanded into a 3-month contact, and is likely to run for at least a year. He has just spent three weeks in California with his daughter, visiting Disneyland and Seaworld, and other expensive scenery, so it sounds as if assembly language programming doesn't pay too badly.
Richard Adams had a promising interview with Brunel University before Christmas, but he hasn't heard from them yet, which isn't a good sign. He has an interview coming up on the
15th January with the Royal Borough of Greenwich, for the post of Principal Contracts Officer, which will be something new for him if he gets it. Neither Richard nor I got the Diomed job …
The latest rumour about Graham is that he is getting together with some of the Cambridge University E~Beam people to put together a metrology machine.
(Note added 2013-02-25. Brunel University is in North West London, a bit north of Heathrow.)
to: Richard Adams, Martin Wiseman, Chris Warner, Peter Milne, Simon Dawes, Bob Taylor, Paul Austin, Andrew Dean, Tony Slater, Ian Murray, Madhusudan Patel, Steven Fisher
from: A.W.Sloman
subject: Progress to 16th February 1992
Not a lot of progress to report ~ Martin Wiseman has a 3 month contact in Watford, and Simon Dawes is taking the job with Leica~U.S.A. Graham Plows has set himself up as a consultant on Electro-Magnetic Compatibility (or Interference, to those of us who have had to fix it) with a suite of computer programs for modelling electromagnetic fields; or so Peter Milne tells me.
I'm still working for Kratos Analytical at Cambridge Mass Spectrometry‘s old site at Bar Hill. The crate I'm working on is mostly tested and working at the module level, but it is still short one module and a couple of crucial connectors, and I’m filling in on other bits of the machine.
My interview at Domino went well, but not well enough – I got their rejection letter last weekend. I've got about six job applications outstanding, but no interviews.
Andrew Dean is still working on EBMF stages, mainly looking at the two Philips attempts at a 6" stage (in 1985 and 1986). He was surprised to learn that Gordon England had been made redundant from Leica-Cambridge - he had had the impression that Gordon had resigned. l wonder how Andrew got that idea?
Paul Austin is working away at Addenbrookes. His gadget hasn’t expanded any bowels yet (last month's report mentioned stomachs ~ apparently the project has slipped a little) but it is
coming along well enough.
Martin Wiseman’s contract in Watford started two weeks ago. He doesn’t like the commuting ~ about 80 minutes each way when the traffic is good ~ and he has certain doubts about the work, on an image-intensified TV camera for site security, where a funny fibre-optic taper couples an image-intensifier plate to a CCD~camera, but he is very happy to have the money coming in. The contract is to run for 3 months in the first instance, but it is likely to run longer - according to Martin his project plan has slipped two weeks in the first fortnight.
Peter Milne started at Link in High Wycombe on the 3rd February. working for Ralph Knowles, on the software for Link’s electron microscope controller. He is a little more positive about his commute — 65 miles in 65 minutes if he times it right - but I wouldn’t like it. Link is one of the companies to which Graham Plows has written, informing them of his availability as a consultant on Electro~Magnetic Compatibility, and Peter was interested to read his flyer.