from: A.W. Sloman
subject: Progress to 8th July 1988
I have measured the output of the 50lb weight head as a function of lab temperature with no load (output about 5mV) and with a 50lb load (output 7.976V~. Under both conditions the output varies by a millivolt or so over a working day, but there is no obvious correlation with temperature. The origin temperature compensation pot was tweaked a little, but the span compensation pot remains as it was set up for the 20lb load cell.
The revised Low Noise Head Amplifier artwork has gone out - We should have a pair of prototype boards by the end of week 16. I spent a couple of hours generating a revised rough parts list. I thought that I had already done it, but we couldn’t find any evidence that I had.
Rodger Hartley did 18.5 hours on the Low Noise Anomaly Shape Control board this week, and is now laying out the new parts of the board.
Phil Fick has now got the Low Noise Head Amplifier board specification, and has measured the board it is replacing to establish board size.
I put in roughly one day this week thinking about the Lintech machine and expect to do more next week.
I spent a couple of hours on Friday with David George, measuring and analysing the speed stability of the Motormatic controller on the crystal puller- It was better than we expected.
I have done a little phoning around for the S.360 magnetic ripple problem, mainly aimed at getting a steel chamber clad with a relatively thin layer of mu-metal, inside and out. This seems likely to be difficult and expensive.
to: DE.Nichols, Dave Hall
subject: Progress to 15th July 1988
I have spent a little time on the pullers this week, mainly talking to David George about stability of the pulling system on the 358 - which appears to be good - and of the representation of the pull position on the front panel and the computer screen -which can cause concern to all but the most sceptical observer.
Rodger Hartley has put in several days this week on the Low Noise Anomaly Shape Control Board, and has positioned all the new components on the board — this has lead to some duplication of component numbers, which he and I will have to resolve (it won’t take long) - and next week he should be able to get onto putting the tracks in between these components.
I also spent a little time on the ripple problem, talking to Nobel’s Explosives on explosive cladding. They confirmed the comment from the Welding Research Institute that it was only possible to clad flat plates, and went on to point out that it wasn’t economic to clad much less than a square metre of material because something like 100mm has to be trimmed off all the edges of any clad sheet.
The process costs about 500 quid per square metre per side. If we wanted to clad both sides of sufficient 12mm steel plate to bend up into three chambers it would cost us about 2,000 quid, plus about 1000 quid for the Mumetal cladding, and as much again once the three chambers had been bent up, welded (or nickel brazed) to have all three chambers heat treated.
These costs are all more or less in proportion. Mike Scott of the Welding Research Institute though that we would be better off nickel-brazing the Mumetal to flat sheet, and we would be able to do that on the material for a single chamber, rather than three.
Adding 1000 quid to the cost of the chamber is probably excessive for the S.360, but it might well be a useful added-cost option for customers with foul magnetic environments.
I spent the bulk of the week on Voltage Contrast, mainly concentrating on the specification of the Sample and Delay system for the EBT2; it has a lot on common with the equivalent part of Tektronix’s sampling scopes, notably the 11402 - and I am in the process of writing a memo to be sent to Tektronix as a sort of preliminary specification to their Custom Waveform Acquistion group. I am none too hopeful of being able to buy in the hardware, but I think it is worth a try.
to: D.E. Nichols, Dave Hall
from: A.W. Sloman
subject: Progress to 22nd July 1988
I have spent only a couple of hours on CPG work this week -mostly answering questions from PCA on the various Low Noise boards, and a little time spent talking with David George about stepper motor controllers.
The rest of the week has been devoted to thinking about the EBT2 development, mostly split between the Timebase and the Waveform Processing, although I put in a little time on the Image Processor today.
The aim is to be able to say roughly how many boards we will need to develop to provide multi-sampling at a useful rate, and thus to put some sort of time-scale onto the RE - this is coming on, but the number of boards is rather uncertain.
to: D.E.Nichols, Dave Hall
subject: Progress to 29th July 1988
Most of this week has been spent on Voltage Contrast — I’ve given Dave Hall estimated development times on the new boards that seem necessary for the EBT - Blanking Drive excepted.
I managed to miss the Component Standards meeting, which is irritating.
On the S.360 ripple problem, the chamber blank in 5CC steel has been flame-cut and shot-blasted. When this was compared with a chamber blank in 43C steel at the same stage, the 5CC steel chamber gave slightly less shielding than the 43C chamber. Both of them roughly halved the external magnetic field.
When the 43C chamber blanks are machined and can be fitted with front and back plates their performance improves - the external magnetic field is cut by a factor of six. It will be interesting to see how the 5CC chamber performs in this state.
I spent about an hour in total talking to John Foulkes about his magnetic pump for gallium, mostly about heat generation and dissipation in electromagnet windings - I think I have sold him on the idea of using neodynium/boron/iron permanent magnets from Philips.
Maurice Blanchard is stirring up the PCA department - Rodger Hartley has almost finished laying out the Low Noise Anomaly Shape Control board, so I am getting hassled to provide a parts list so that the board can be kitted for 1st August, when the prototype board is scheduled to come in. Rodger can’t provide the parts list - he is too busy laying out the board.
to: D.E.Nichols, Dave Hall
subject: Progress to 5th August 1988
The greater part of this week has been spent on the EBT~ I finally got around to giving Jim Rose a rough parts list for the Low Noise Anomaly Shape Control board, which is now largely kitted.
I got to see the layout for the board on Tuesday, after Rodger Hartley had gone off on holiday. I found a couple of minor faults, but they could be corrected by cut and link, so it seemed expedient to send the board out as it was and save the bulk of the corrections until we had gotten the board tested.
The prototype boards should be back around the middle of next week.
The revised Low Noise Head Amplifier boards are now loaded and effectively waiting to be tested - they are short two 10k1 resistors, but 10k0 resistors would do almost as well, for testing.
John Gibbens chased me for the Low Noise Sine Oscillator board dimensions yesterday, which an optimist might take to imply that board layout was imminent.
For the EBT I spent some time thinking about the implications of 50psec blanking - since my intuitions do not line up with those of Dr. Plows I would really quite like to see Julian Davey’s opinion.
For the last day or so I have been worrying about the shot noise in the EBT detector system, and I am now starting to rough out the Head Amplifier specification.
to: D.E.Nichols, Dave Hall
subject: Progress to 12th August 1988
Again, most of this week has been spent on the EBT. I have done some work on the Crystal Puller.- Derek asked me for a customer-level write-up on the low-noise modifications and he now has a draft document which definitely needs pruning in some spots, clarification in others, and expansion here and there. When he gets around to giving me his comments on it, I will put in some polishing time.
The Low Noise Anomaly Shape Control boards are now being loaded - the parts list and circuit diagram both have errors which are being corrected as the board is loaded. We are short of LTC-1052 amplifiers for the board. Alan Plews borrowed some of my stock some months ago, and while we had expected to see them in store by now, Jim Rose has had to order the replacements, which are not expected in until early next week. By then all the other components should be loaded.
I’ve got both the new Low Noise Head Amplifier boards, though neither is tested.
John Gibbens has started laying out the Low Noise Sine Oscillator board - he is doing it as much to keep his hand in as to help diminish the back-log, and won’t be full time - but he does expect to have placed the components before I go off to Australia, and to have tracked them by the time I get back.
On the EBT I am still writing short descriptions of the new boards and trying to get some idea of their cost in production. I’ve done the Head Amplifier and the two Timebase boards, and I am now well into the Waveform Processor.
I was distracted early in the week by the problem of the grids for the DVCS 1500 spectrometer - Agar Aids changed their manufacturing process some time ago, and the grids they produced could not be used to make effective spectrometers. The gold plating fell off the electroformed copper, and the exposed copper corroded to present an insulating surface to the beam.
At the time we complained to Agar and they made a number of test grids, gold-plated in variety of ways. The most successful were gold-plated over nickel-plating over copper - nickel is magnetic, but the magnetism associated with the nickel plating was too small to produce visible distortions on our SEM images.
There is now a mod in the system - 25326 - to change the affected drawings 612961/2/3/4 to specify this plating. The existing stocks of grids (near1y two hundred quids’ worth) have been withdrawn from stores to my cupboard for use in training and sales, and the new stock requested on the 4th August will be ordered against prints marked up to include the new plating specification.
Dave Hall and Trevor Sexton have found a lot of the information covering the new Lintech Image Store - I have spent some time scanning through it and I have a vague idea of what it is supposed to do and how it is supposed to do it.
to: D.E.Nichols, Dave Hall
subject: Progress to 19th August 1988
I have spent most of this week writing a specification for the Head Amplifier board for the new electron beam tester, tentatively labelled the DVCS 50M. I have just printed out a complete draft of the specification, which needs some tidying up.
I expect to draw a circuit diagram for the board on Monday, and give the circuit design group the complete package that evening.
I have completed my description and cost estimates for my other DVCS 50M boards. Dave Hall finds that the ESCs tie up with his estimates, which is surprising. I have talked him through the description of the Head Amplifier and the Waveform Processor, and we will go through the Timebase early next week.
I expect to earn a fair amount on overtime tomorrow, talking to Mark Saunders about the Image Store and Image Processor; if we can hook up the Waveform Processor to the NAF-Bus-2 we should be able to get almost all the benefits of burst mode imaging at 50Msamples/sec without the problem of distributing 50Mbyte/sec data around a frame store, which would mean that we won’t have to design a new Image Store. We will need a new Image Processor sometime, but that is a less urgent problem.
I have done some work for Derek this week - I’ve done some minor re-writes on my description of the Low Noise modifications to the weight cell to deal with the points he raised, and I’ve spent an hour or so talking to John Foulkes about his electromagnetic gallium pump. It is proving hard to convince him that an AC-excited pump is more practical than a DC system which requires 200A working into 100 micro-ohms.
The Low Noise Head Amplifier boards remain untested. The Low Noise Anomaly Shape Control boards are now fully loaded, and waiting to be tested - the printed circuit layout doesn’t leave room for the screening can, but we can probably test the board without it.
I’ll try to start testing the boards next Tuesday, in the hope of getting them both sorted out by next Friday - if they are not done then I won’t be able to touch them until the 26th September, when I get back from Australia.
to: D.E.Nichols, Dave Hall
subject: Progress to 26th August 1988
Too much of this week has been spent on the electron beam tester; the Head Amplifier specification didn’t get finished until Tuesday evening, and I spent a large lump of Wednesday preparing a block diagram for the EBT Timebase, after which the Circuit Design Group came back for layout detail on the Head Amplifier
This should have left me free to work on the Low Noise Weight Cell on Thursday, but I took a few moments in the morning to write a note on programming the EBT Timebase for software, which was about half done when I got grabbed by Dr. Plows and Mr. Cruttwell to explain the Timebase and the Waveform Processor which took the rest of the morning, and I lost about an hour in the afternoon finishing off the note for software.
In fact I got the Weight Cell powered up and running by 4.l5pm on Thursday. The new board needs 56R at R21 to stop output oscillations when it is driving 15 metres of cable, and both boards have been built with 20k top-adjust pots at X1R3, mainly because I got the parts list wrong. The circuit diagram calls for a part 098389 50k side-adjust pot, and this is what was put on the first PCB. There don’t seem to be any in stores. The board should work with 20k, but top-adjust makes it a pig to set up.
Friday morning I got the gain set up correctly - R1 had to be changed from a 147R 0.1% device to 113R 0.1% (part no. 097031 from stores) and the cell should now give a 7.40V swing for a l4kgm change in load (I measured a 5.286V swing for a l0kgm change in load).
I then went on to check the zero stability against temperature which was bad - about -10gm/C. I have set up VR3 to reduce this - probably to better than +/- l gram/C if my calculations were accurate. My first attempts to do this failed because I couldn’t get the right ratio between the voltages at TP1 and TP2 - which turned out to be a 6k2 resistor fitted at R16, where circuit diagram and parts list both call for 27k. Once I replaced this resistor with the right value, I was able to set up the desired ratio - the zero stability ought to be okay, but should be checked.
Friday also included some EBT involvement - an hour of interview in the morning, and two hours of meeting in the afternoon, and 1.75 hours with Mark Saunders talking about the Lintech Image Processor. My rather scrappy notes from that meeting are now stapled to today’s page in my day book. I now have a much better idea of how the Lintech Image Processor works, why it often doesn’t work, and why everybody wants to dump it - detailed comprehension of all these points is going to take more time.
The addition of 1.25 hours of overtime produced this and found the duff resistor on the Low Noise Head Amplifier board.
to: D.E.Nichols, Dave Hall
subject: Progress to 30th September 1988
Not a lot of progress this week - we landed back from Australia on the 24th and jet-lag can carry a lot of the blame.
The only significant lump of work I have put in has been checking out the artwork and circuit diagram for the Low Noise Sine Oscillator board (85457B) - this took about a day. I found nothing seriously wrong, but contrived to find some points to nit-pick about so that PCA would believe that I had checked the stuff.
From time to time during the week I have been helping Peter Milne with the new Vac Control card for the EBT. It looks as if Tim Frost made a significant number of minor errors around the Penning gauge input - he seems to have copied the S.240 circuit without having a clear idea of what it was doing. We will sort out some sort of short term mods next Monday, but the board layout will have to be up-dated.
I’ve made a start on the Timebase board specification, about a paragraph’s worth. There is a long way to go.
I propose to incorporate Brust’s pre-trigger sampling in to the Time-base system - I think it can be fitted into the “sample list” approach without much strain.
(Note added 31-07-2017 : The “pre-trigger sampling” was in fact based on the assumption that the circuit under test was being cycled through successive cycles of identical behavour so that it you kept on sampling through the cycle into the next cycle you could see what was happening just before - and just after - you'd detected the “trigger” signal that indicated the start of a new cycle. Since we were claiming to resolve 5psec time intervals, this implied that the interval between trigger signals should be stable to 5psec – unrealistic, but we never ran into a situation where it gave us any trouble. The idea meant that we had to have two Delay cards – one that could keep sampling while the trigger signal was coming in while the other would detect that trigger signal and get on with starting it's own sampling sequence. Expensive, but it turned out to have additional advantages.)
to: Dave Hall
subject: Progress to 7th October 1988
I have made some progress with the Timebase specification, but not a lot. The Vac Control board has taken about two days out of the week, and the User Interface another day.
The Penning gauge output from the Vac Control board now appears to be sensible; the software over-estimates the pressure by about a factor of ten, but the estimate goes up and down in the right direction, and in roughly the right proportion.
The “Penning struck” circuit now seems to work - for some obscure reason Tim ran the relevant comparator from the +5V rail, and when I switched it to the +15V I started to get the results I expected. We still have to sort out gains and tolerancing, but the mysteries seem to be behind us.
There is no guarantee that the rest of the Vac Control board is working properly - there are other mods of a messy and substantial sort on the board, and while Tim might have done better with these areas, it might be prudent to anticipate more difficulties.
My comments on the User Interface were largely confined to proposed modifications of the Waveform Acquire and Waveform Expansion menus - the Waveform Acquire menu would become an introduction to the Waveform Expansion menu, and the user would spend most of his time working with “expanded” waveforms — that is waveforms recorded over periods of arbitrary user-defined length and starting at an arbitrary, user-defined interval after the trigger.
To my mind, the initial proposal was shaped in terms of the limitations of the old S&D unit, and hampered the user in ways that no longer serve any useful purpose. My proposals are shaped by the limitations of the new Timebase, and hamper the user in ways which make our job easier.
to: Dave Hall
subject: Progress to 14th October 1988
Quite a lot of progress this week; I now have a much clearer idea of the mechanics of the Waveform Processing, and of the clock edges that will have to be supplied to keep it running, and I think I have a practicable scheme for the timing circuits on the Timebase board.
Neither of these developments has manifested itself in the Timebase specification, but they will.
The Vac Control card has taken up a little of my time and a lot of Peter Milne’s - I have modified the board again to remove the hysteresis from the “Penning Struck” output, and put it on the “Vac Ready” output, because losing “Vac Ready” after it has appeared is treated by the software as an excuse to turn everything off, with awkward effects. Uncertainty on “Penning Struck” is less troublesome.
Peter is modifying the board to enable him to write fourteen rather than seven logic level commands from his computer to the board - this requires a couple of latches and a monostable or two. I have been consulted, but accept no responsibility.
to: D.E.Nichols, Dave Hall
subject: Progress to 21st October 1988
I spent most of last Friday at the Analog Devices seminar. I found their session on High Speed Data Acquisition distinctly valuable; it brought home to me that even “flash” ADC’s are sensitive to the slew rates of the signals that they are digitising - there are variations in propagation delay through the comparators which carry out the conversion which degrade linearity for fast-slewing signals.
I may switch from the AD-9002 8-bit converter to the 6-bit AD-9006 to minimise this problem - both cost about the same. The AD-9006 could sample fast enough to allow us to accumulate across our minimum pulse (with GaAs logic) rather than aim to hit the peak of the pulse. We could use the same hardware to accumulate across wider pulses without changing the bandwidth of the analog part of the input channel, but the averaging logic would cost 600 quid.
I also found out about the AD-9712 12-bit 100MHz DAC, which has now been designed into the Filter Drive on the Waveform board, which now offers 20mV resolution over +/-l0V. This is 10-bit resolution rather than 12, but even the AD-9712 would be hard pressed to settle to 12-bits in 20nsec. By dropping the sampling interval to 30nsec we might make it.
The other interesting component is the fast ECL static random access memory for the Waveform Processor and Timebase boards. The most attractive device is the Hitachi HM100484-l0 4096x4 static RAM which would offer a l0nsec cycle time if Hitachi actually had it in production. The Advanced Micro Devices Am100474-15 is smaller (at 1024x4) and slower (at l5nsec) but remarkably cheap (at 7.37 quid each 100-up). Unfortunately, while Advanced Micro Devices are producing the device, they are not getting good yields of the l5risec part, and in practice it is no more available than the 100484.
For the prototype machine we will use the 100422-10 256x4 l0nsec device, which has the overwhelming advantage of being widely available (STC and Farnell both stock it). This part is fast enough to permit us sample at 50MHz, and to demonstrate this if the rest of the system can support the rate. By the time we get to production Advanced Micro devices may have solved their production problems, or we may chose to go for the 25nsec 100474 parts which are readily available now, and live with a 30MHz maximum sampling rate until fast parts become available.
The bulk of this week has been split between getting over a cold and completing a much more detailed preliminary specification for the Waveform Processor board - it turns out to be impossible to write a detailed Timebase specification without a much clearer model of the Waveform board (which I have now largely completed).
The downside of this highly satisfactory operation is that the Timebase specification is not a line further forward than it was last week.
to: D.E.Nichols, Dave Hall
subject: Progress to 28th October 1988
I have done about an hour’s work for the CPG group this week. The prototype PCBs for the Low Noise Sine Oscillator board came back early this week and one has now been loaded, and passed on to CPG Final Test for testing. This should happen in the next few days.
Most of this week has been devoted to writing the Timebase specification. The guts of the specification have been written, but none of the detail - that is, inter-block connection specifications, board inputs and outputs, connector allocation, test procedures and equipment, power supplies and environmental/mechanical.
In principle these can now be churned out. By the time Johannes and I have been through what I’ve already got we will probably have made enough changes to devalue a lot of this detail, and in any event I am going to devote the next couple of days to sorting out a specification for the Trigger board, which shouldn’t take too long - roughly half the existing Timebase specification has been copied into the Trigger specification with minimal changes.
Finishing off my preliminary description of the Waveform Processor board took up Monday and about half of Tuesday. It is not a specification in any real sense; it says what will be done, and how, but concentrates on the dodgy bits to the exclusion of fitting the nuts and bolts of the design together.
Dave Ward of TDS and Johannes both have copies. I haven’t yet had any significant feedback from either - Johannes has looked at it, but hasn’t formed much of an opinion yet.
to: Dave Hall
subject: Progress to 4th November 1988
This week the Timehase card got renamed as the Delay card, and the Trigger specification was brought up to the same level as the Delay specification.
I expect to get most of the detail into both next week, provided that I don’t have to expand or modify the Waveform Processor description for TDS.
I now have what looks like a layout for the Fast Head Amplifier; I’ll check it out in detail on Monday.
I’ve been having a few doubts about continuing to try for 50MHz sampling; there are a number of points where achieving 20nsec means putting in extra components.
In the long term I suspect that we will probably back off to something closer to 25MHz sampling, but it seems important to keep the prototype system as fast as possible so we can find out where the problems would be if we had to go faster than say 25MHz.
The biggest bottle-neck occurs on the Waveform Processor board where we might be able to halve the quantity of ECL RAM if we went quite a lot slower; the alternative is to use GaAs RAM, which is dramatically faster, and about three times the price -with GaAs we would want 8 devices at 450 quid, whereas with ECL we would need 16 chips costing 235 quid, or - in a slower system - eight slower, more capacious chips for about 80 quid.
to: Dave Hall
subject: Progress to 11th November 1988
This week Johannes Gebrehiwet put in 25.5 hours on the Delay board, and Bob Ward, 32 hours on the Trigger board (not including today).
Johannes has divided his time between back-ground reading - the spec and GaAs application notes - and setting up HiLo models of the GaAs and ECL synchronous counters (100136 and 10G061) and the GaAs D-type bistable (10G021), which both he and Bob will be using.
Bob has spent the week on the analog part of the high speed low impedance input to the Trigger board. This strikes me as a long time, and I have pointed out to him that there are a least six other blocks of comparable complexity to be designed if we are going to have a circuit design to release to PCA at the beginning of January. Bob is less familiar with long-tailed pair design than I would have expected, and for him this part of the board may be rather more complex than it appears to me.
TDS have done very little on the Waveform Processor - they haven’t received the formal order, and while they have not said that they are waiting on it, the timing of their reference to the absence of the order might have been intended to leave the implication.
I spent 2.25 hours with Mark Saunders last night, about an hour learning the exact state of the Image Store boards, and the rest of the time discussing the functions of the Waveform Interface board; the Waveform Interface board will apparently have to manage the refresh cycles for the Image Store dynamic RAMs, which I had not expected.
I had expected that it would have to drive the Scan Generator; Mark was not aware exactly how this was to have been managed, and was anxious about. the performance of the analogue Scan Generator circuits at low scan rates - we could use advice from Tim Frost on the subject.
It seems that it was accepted that the Image Store was not to be used in a “read/modify/write” mode so that an existing set of Burst Mode images could not be improved by adding in extra sets of scans.
My own progress over this week has been less than I had hoped; I have prepared a block diagram for the signal processing system, and ordered the Head Amplifier parts and some related components.
I’ve made no progress towards completing the Trigger and Timebase specifications. This may not represent a week’s slip, but it doesn’t help.
to: Dave Hall
subject: Progress to 18th November 1988
Johannes and Bob Ward seem to have spent most of this week on the Delay and Trigger boards respectively.
Johannes has blocked out large chunks of his board, and is getting down to detail design.
Bob is still working on the Trigger and External Clock inputs to the Trigger board - we now have drawings of both the Fast, and the High Impedance inputs. I have proposed what I conceive as an improvement to the Fast input - pushing up the threshold range from +/-2V to +/-4V which took me half a day to draw, and will presumably take Bob as long to incorporate into his drawing, if he accepts it.
Bob Anderson from TDS spent an hour or so with me on Wednesday evening, clarifying matters left obscure in my discussion of the Waveform Processor; he seemed to have an adequate understanding of what we were after by the time he left.
Phil Fick has found a printed circuit manufacturer - Printech Circuit Laboratories Ltd. (0245) 323 244/323 353 with experience in fabricating multilayer boards in PTFE (Teflon) based composites. If we settle for the low dielectric constant, low dispersion material in only the two outside layers of the board we can have as many layers of g1ass-epoxy in the middle as we can afford.
I have spent about 14 hours this week on the design of the Cursor board; we expect to be able to put the cursor on the TV signal on the output of the Image Processor so that the user can move it around smoothly; the column will not be being scanned at TV rate under these circumstances, but at Slow Scan 1 at best.
Even at Slow Scan 1, it may take up to 0.4sec from the time the user decides where the cursor ought to be, to the time that the Cursor board has captured the corresponding X— and Y-Scan voltages. We are giving the processor a couple of flags with the digitised voltages to indicate whether the voltages read are entirely up-to-date.
The nuts and bolts of this all seem fairly clear now; I have just written about four pages of description of a simple state machine to handle the process of digitising the voltages and setting the flags. Either Simon or I will have to spend a few more hours reducing this to a list of inputs and outputs; once we have done that it should be easy enough to fit most of the machine into a. PAL or a PROM.
I have made very little progress on completing the Trigger Specification and none on the Delay Specification, the Waveform Processor Specification or the Waveform Interface Specification.
The Trigger Specification now sports a complete list of Inter-Block signals for the first three blocks. The next three still have to be done. It isn’t a long job, but it is proving very difficult to get at it.
to: Dave Hall
subject: Progress to 25th November 1988
Progress since last week has not been satisfactory. On the bright side I have now tied off the Cursor Board, as far as I am concerned - Simon Dawes now has about ten pages of text about the state machine required to control the Cursor Board.
The Trigger Specification presents a darker picture. I have been through the text and tied off all the loose ends - or at least those of which I am conscious - and I am now working through the “Inter-Block Connection Specification”, nailing everything down. At a guess I have got through 75% of that, so a few more hours ought to see me through to the Input/Output Specification, which is followed by Test Procedure/Test Equipment, then Power Supplies and Environmental/Mechanical.
These will be ready before Bob Ward starts working on them; he is still on the Trigger Inputs. He should get through the External Clock inputs quite fast since the bulk of the circuitry is identical to the Trigger Inputs, but may be slowed up by the Local Oscillator. Bob may be able to poach the guts of the more demanding output blocks on the Trigger Board from Johannes’s Delay board, which may save drawing time. Neither Bob nor John Gibbens expect the Trigger Board to be available for PCA before mid-January.
I haven’t done anything on the Delay Specification; it seems to be sufficient to let Johannes go forward as it is.
Johannes has laid out the three major blocks of the Delay board, and spent most of this week sorting out the detailed logic of the first block - the Trigger Detector (including the circuit for measuring the off-set between trigger and local clock to 5psec. He has got down to the Boolean algebra defining the control signals, which is pretty far down into detail. Johannes still expects to produce a circuit diagram for his board by Christmas.
I haven’t done anything on the Waveform Processor specification, which is awkward, since TDS will be here on Tuesday afternoon. Much of the work on the Trigger Specification has been occasioned by my increasing clear ideas about the Waveform board so I will try to get some sort of skeleton specification together for Tuesday afternoon just to document the new detail.
The Waveform Interface is at the back of the queue; it would help if I could talk to Tim Frost about the details of driving the Scan Generator to generate a slow scan so that I know what the Waveform Interface has to generate during Burst Mode Imaging.
I have done enough work on the Scan Generator board to know that we can’t safely scan much slower than one frame every l0sec. This represents about 25usec per pixel, which represents the upper limit to the trigger-to-trigger interval; if the device under test can’t go through its test cycle in less than 25usec we can’t guarantee Burst Mode images.
This limit could be relaxed by a couple of orders of magnitude by reworking the Frame Integrator on the Scan Generator board, but granting the availability of a 16-bit DAC with l50nsec settling time, it would probably make more sense to design a digital scan generator board and completely eliminate this particular limit.
It seems that MicroPower Systems and Honeywell are both selling such a device, which promises even better price and delivery.
The Fast Head Amplifier board finally went out today for manufacture in ten working days - we should have it on the 12th December. The photomultiplier and socket for the board are due in this Monday; one of the zener diodes required is on shortage in stores, but should be available before the 12th.
to: Dave Hall
from: A.W. Sloman
subject: Progress to 2nd December 1988
Progress this week has been reasonable. The Trigger board specification is now complete through the inter-block connections section. TDS are now sufficiently into the Waveform Processor board design to want me to change the Trigger board specification to get it to provide the signal timings that they actually want, so there is more work to be done there, and not a lot of point in trying to finalise the specification.
Johannes seems to be making excellent progress on the Delay board. I am going to have to take some time next week reassuring myself that his sequencing and hand-shaking scheme will actually do what I want, but it sounds okay so far.
Bob Ward has completed the Trigger Input and External Clock Inputs on the Trigger board, and has almost completed the Local Oscillator block. He expects to start on the output blocks next Monday.
TDS are getting into the Waveform Processor board, but we haven’t got much to look at yet. Bob Anderson seems to have sorted out a fast but bulky scheme for the Accumulation Store, which will work at 50MHz.
I have got a Waveform Processor specification together but it is very half-baked at the moment, and needs a day or two more work before it would be worth passing it on to TDS.
I have talked to Peter Milne about the Signal Processor memory map, and about the software specification to Syscon for the Image Store - I wouldn’t say I entirely understand what is going on, but I expect to fairly soon.
We talked to Mike Rolfe about the manufacture of large surface mount boards, and he is to arrange a meeting with British Telecom, to whom we might sub-contract this work. Johannes is to prepare a slightly premature circuit diagram and parts list so that we can take a sample of the work to the meeting.
to: Dave Hall
subject: Progress to 9th December 1988
There has been some progress this week. The Waveform Specification is beginning to attain coherence. It needs an Inter-Block Connection section but until Bob Anderson at TDS puts down more detail about his memory/accumulator organisation I don’t know what sort of timing inputs he is going to want.
Johannes has spent this week detailing. He expects to have detailed but imperfect circuit diagrams for the Delay board on Monday, for Mike Rolfe to take to Newmarket Microsystems Ltd. Mike Rolfe did not come round this afternoon to pick up the circuit diagrams available, so they won’t get to Newmarket until Tuesday morning.
Mike Rolfe has yet to organise a meeting with BT’s SMD-board manufacturing service; this morning he said he was going to organise it today. I will ask him again on Monday. I have fixed a meeting with Roy Jacobson of BAC Dynamics of Bristol (0272 363975) for 11.00 am 20th December, to discuss SMD boards with them here.
I have to warn Jim Rose about the requirement for a technician to hand-build a couple of surface mount boards for the prototype; he will he back from vacation on Monday.
Bob Ward. is getting into the output buffers and delay generators on the Trigger Board. I have a block diagram, and a request to modify the Trigger board specification to use the signal name “Processing Loop” rather than the descriptive adjective “processing”.
The prototype Fast Head Amplifier printed circuit boards have arrived, and should be loaded soon. The XP-2982 photomultipliers from Philips have not arrived, nor their sockets (ordered at the same time), and without the sockets there will be little point in loading the board. We are also short a couple of 75V zener diodes; these ought not to be a problem, but I may have to talk Ken Lincoln into ordering a. few, since they are theoretically stores items.
Tim Frost is coming in on Monday at about 6.00pm to give me a short tutorial on driving the Lintech Scan Generator - he will need a letter defining his rate of pay before he starts talking, and I have promised to ring him on Monday if it seems that this won’t be available by 6.00pm.
to: Dave Hall, Bob Ward, Yohannes Gebrehiwet, Kevin Jackson
subject: Progress to 16th December 1988
The search for sub-contractors to assemble our surface mount boards goes on. Yohannes kept on up-grading his drawings until Mike Rolfe finally collected them, last thing on Tuesday afternoon for delivery to Newmarket Microsystems on Wednesday morning (to be fair, Mike was at Harlow on Monday).
Mike Rolfe has organised a meeting at l0.30am on Monday with a representative of British Telecom so we can talk about their SMD manufacturing set-up, and the guy from BAC Dynamics at Bristol will be here at 11.00a.m. on Tuesday to tell us about theirs. Mike has also found some Philips’ application notes on SMD layout which he has passed over to Phil Fick to give PCA some idea of what is required.
I’ve finally written my memo on the board material I expect us to be using for the Trigger and Delay boards and the housekeeping part of the Blanking Board. The memo is addressed to Phil Fick, but the rest of the team have copies.
I’ve got my specification for the Waveform Processor board into a moderately reasonable state, and was going to go over to TDS at Biggleswade to haggle about the nature and number of timing signals coming from the Trigger board to the Waveform Processor; unfortunately Bob Anderson is off sick with a sore throat. I have faxed over four sheets from the specification covering the “transversal filter” pulse stretcher for Dave Cox so that he can get on with the analog part of the circuit.
Yohannes is still tinkering with the detail design of the Delay board; he hopes to be able to give something to PCA by the end of next week. The detail to be finalised is mostly in the Trigger Detector logic and sequencing. We still haven’t finally made up our minds about the sorts of ECL PALs we will be using for the sequencing logic.
Bob Ward is still working his way through the Trigger board output blocks. I set back his design for the Sampling Pulse Generator block by a day or two by revealing that the signal “Processing Loop” had split into two. This feature is now written into his Trigger Specification, and a copy of version 6 has been deposited on his desk.
Kevin Jackson started work with us on Wednesday as a contractor. He is expert on microwave design, up to 10GHz and beyond - we hope he will be able to up-grade the Cambridge Instruments 0.5nsec Blanking Board to 200psec for Semicon West in May, and go on from there towards 50psec pulses for the production machine.
His first impression of the 0.5nsec Blanking Board seems to be that there is room for improvement, which is of course embarrassing for Tim Frost and me, but should cheer up Dave Hall no end.
I’ve had a couple of interesting discussion with Kevin about the problems of testing satellite receiver strips and other very high frequency circuits - it sounds as if it is a potential market for the EBT, since any physical probe introduces impedance discontinuities which disrupt the state you are trying to observe.
The EBT as presently conceived is a bit slow for microwave work; even with a 50psec pulse it couldn’t detect anything over 6GHz, and the beam duty cycle would rather low. If we used a microwave source to modulate the electron beam, the process of generating our low energy secondary electrons could be used as a mixer to generate an output signal modulated slowly enough for our secondary electron detector to respond to it. This is an obvious generalisation of the technique described by Brust and Fox in “Microcircuit Engineering 1985” pages 191-202, so we can’t patent it, but Mr. Jackson has the advantage over Dr.Brust of having revealed customers for the system.
If we ought to be selling the standard EBT to Tektronix as a sampling head, we might sell this version to Hewlett Packard as a detector. We will probably need to hire an HP Network Analyser or at least an HP Sweep Oscillator to determine and improve the input impedance of the Blanking Head; it might be interesting to see what else we could do with that equipment.
At a more mundane level, my photomultipliers and their sockets have arrived; no doubt the Fast Head Amplifier boards will be loaded soon.
On Monday I will have to get together with Yohannes and Bob and convert their preliminary parts lists into a very expensive order for Gigabit Logic GaAs logic parts.
Tim Frost came in last Monday evening to spend an hour tutoring me on the intricacies of the Lintech Scan Generator board. I think it was useful - there were some intricacies that I would have been a. long time finding.
to: Dave Hall, Bob Ward, Yohannes Gebrehiwet, Kevin Jackson
subject: Progress to 23rd December 1988
The usual sort of progress this week; not as much as we had hoped and some of it in unexpected directions.
We have now approached four groups for quotes for manufacturing our surface-mount boards; Newmarket Microsystems, British Telecom at Martlesham Heath, BAC Dynamics at Bristol, and Surtech Interconnection at Basingstoke.
Les Smart at British Telecom has come back to us over the phone with estimates of five to six thousand pounds to lay out individual boards, to which we have to add 420 quid for twelve photographic masters, and between 100 and 200 quid for surface mount tooling. The cost of loading individual boards will be 278 pounds on a three week turnaround, 499 for one week turnaround.
Since British Telecom claimed to have spare capacity both for laying out our boards and manufacturing them, this is distinctly attractive. If the other groups don’t give us their estimates very early in the New Year we may not have time to take them seriously.
We will have to visit British Telecom at some point before we commit ourselves, to make sure that the “spare capacity” is competent to do what we want; it might be unused because it is useless....
The board data on which all four quotes will be based is that supplied by Yohannes for his Delay board; he now expects to able to release a design for printed circuit layout on the 16th January; the problem is not so much finishing off the design as of doing enough simulation to be sure that the vital bits will work. We have fixed on the National Semiconductor PAL10016P8JC as the 100k ECL PAL for his board; there are four in the country, which should arrive here shortly, and six more are expected around the 8th February. Our PAL programmer will program this device.
Bob Ward has survived last week’s sabotage on my part and is now digging into the rest of the Sample Pulse Generator. I managed to simplify his system by killing off the Dummy Pulse - I know I’ve told Yohannes about that, because he assured me that it doesn’t change the Delay Board design worth a damn.
This week’s sabotage has been to introduce a new output from the Trigger board - the Write Pulse - derived from the Memory Edge, to actually drive the Write Enable lines on our ECL SRAMs. At the moment I am very tempted to ask that the delays from the Memory Edge to the leading and trailing edges of the Write Pulse should be separately programmable over the VME bus, and longer during the Smoothing Processing Loop than during normal accumulation or the Differencing Processing Loops. Bob shouldn’t get to find out about this until the New Year.
The Write Pulse is in the nature of a convalescence present to Bob Anderson of TDS who recovered from his flu sufficiently to spend yesterday morning over here talking about the Waveform Processor board. He had a number of questions about the Waveform Processor specification; most were answered by the latest version of the specification, which is nevertheless a long way short of completion.
The latest version of the Waveform specification includes a surprise requirement for a Suppressor Grid voltage drive, to track the Filter Grid Voltage with a 30V negative offset; fortunately it doesn’t have to track it very closely. I have yet to get the official reaction from TDS about this.
Kevin Jackson has completed a new layout for the output stage of the existing C.I. Beam Blanking board; PCA can’t do anything about it until Phil Fick gets back, but we should be able to turn it into a double-sided printed circuit board on microwave material (supplied by Kevin) in fairly short order.
We have also had useful advice from Kevin on surface mount and on high frequency connectors; Quadrant Connectors dropped in yesterday morning, and gave evidence of being able to provide the connectors we shall need in the rather small volumes we will need, on quite short delivery.
The Fast Head Amplifier parts remain kitted, rather than assembled - Shaun went off on holiday without telling Jim Rose that the boards and sockets had arrived. We should have the board assembled quite early in the New Year.