Chaos and Noise!
Malcolm Robertson and ‘chaotic source’ measurement.
In 1989 and 1990 Malcolm Robertson carried out a series of measurements on the mm-wave behaviour of quantum-well/multiple barrier (QW/MB) diodes supplied to us by Nottingham University and other sources. The above image shows him using a heterodyne receiver system to measure the output spectrum produced by one of these devices. The screenshots on the right illustrate how the resulting spectrum could either show a single frequency of what looks like wideband noise. The work involved him in designing the waveguide blocks into which the devices could be mounted and studying the way the resulting systems behaved.
For some time we had been evaluating various types of diode, including experimental Gunn devices from GEC Hirst Labs. In addition, we now were working in a joint-grant research project with Nottingham University. In particular, the wish was to develop improved mm-wave solid-state oscillators, mixers, and even ‘self-oscillating mixers’ – i.e. devices that could carry out both functions in one diode. The original arrangement for the joint project was that the postdoc taken on by Nottingham would spend a significant fraction of their time at St Andrews to become well-involved and familiar with mm-wave techniques and requirements for operating at and above about 100 GHz. But in practice this didn’t really happen, and the work at mm-wave frequencies was left almost entirely to Malcolm.
Initially, the plan was that the devices would oscillate to produce a controlled, single frequency. In effect, sinewave oscillators. But measurements quickly showed that some devices generated ‘noise’. i.e. when they oscillated, the result was spread across a wide range of frequencies, with a very complicated spectrum. This behaviour intrigued me because during the 1980s ‘Chaos’ had become a very trendy subject with the general public and various academics.
I also knew that many measurement systems required ‘noise sources’ rather than coherent single frequency sinewave sources. So, rather than this noisy behaviour being an unwanted nuisance, I realised that this behaviour might have practical applications which had not previously occurred to anyone! As a result we started to investigate the behaviour of the nonlinear diodes, and how it might be exploited. This idea may have come into my mind as a result of a I maxim I’d heard when I’d worked in industry: “If you have a weakness, make it a selling point.” i.e. if you develop something unexpected, look for a novel application for what you’ve made rather than throw it in the bin!
Thermal sources (‘black body’ radiators) were routinely used in the mm-wave region to calibrate the sensitivity or gain of detectors and measurement systems. The snag was that unless they were heated to a very high temperature, their output power spectral density (amount of power per Hz of bandwidth) was very low. So they were only useful for calibrating the performance of very sensitive and/or very wideband detectors. Heating such sources to high temperatures meant dealing with the practical problems of coping with dangerously hot materials, so was best avoided. The most common alternatives were based on arranging an ‘avalanche’ of electrons in a diode or tube filled with ionised gas. These gave higher powers, but their precise output level tended to be difficult to determine or keep steady, and lead to ‘chicken and egg’ problems. i.e. to calibrate a detector you needed a noise source that produced an accurately known power. But to determine the source’s power you needed a detector whose sensitivity was accurately known... However it occurred to me that the QW/MB devices might offer a neat way past this problem.
These devices offered the prospect of acting as a high level of ‘noise’ power when operating in a noise-like mode, but would switch to confining their output to one frequency when their operating conditions were altered. And these two modes of operation were linked, mathematically. So it should be possible to determine the details of their complicated (‘noisy’) output from their easier-to-measure simple sinewave output by understanding the physics of their operation. As a result, the behaviour of nonlinear systems of this kind became one of my main interests during this period. This also tied in nicely with Graham’s work on developing measurement systems to determine the sideband noise behaviour of coherent Gunn oscillators.
To Chris who has put up with my sore head whilst writing this book,
and to Derek Martin, who still knows a lot more than me about mm-waves.
In January 1990 my first book was published by the Institute of Physics. It was the result of countless hours of work at home. Sometimes I became so focussed on writing that I’d not noticed that the Sun had set and the room around me was dark as I typed! Thus confirming in Chris’s mind that academics don’t know what day it is! Once it was printed I sent a copy to Derek Martin because I knew it would never have been written without his taking me on at QMC, and being an excellent supervisor and teacher. I also sent a copy to my Mother, just so she could see it. The book covered all the basics of what I had learned about mm-wave optics, etc. And from then on I could just point new postgrads, etc, to a copy and say, “Read that, it’ll get you started”. Hence it let me dodge having to re-explain the same things to each new student who joined the St Andrews mm-wave group from then on.
On April 10th 1990 Ralph Yell from RSRE/NPL visited us to discuss funding work by Graham. RSRE was the Royal Signals and Radar Establishment in Malvern. (Later on it became DERA, then split into QinetiQ and DSTL). The NPL worked in conjunction with them on providing measurement systems to support their projects. Ralph and I were setting up a contract for Graham to work on oscillator sideband noise measurements for RSRE/NLP.
During April, George Radley also came to see me. We were paying for him to machine precision items, and this took him out of the main Physics Dept. workshop. As I’ve explained in previous pages, George and the head technician in the workshop did not get on well. Whilst we could bring in the money, our group could take George out from being under the control of the head of the workshop. That, and the chance to show just how good he was as a machinist, etc, suited George very well.
When George saw me he explained that the head of the mechanical workshop has been saying to him that our payments would end soon, and when they did George would be back under his control. George’s concern was quite clear - that his life would then be made unpleasant. As a result he said he would leave the department entirely rather than risk this happening. This was a real prospect because George was a good machinist and could expect to find a job elsewhere, avoiding the hassle of the Physics workshop. However it would be a real problem for us if we lost his skills, and I was concerned that he had, in effect, been threatened. So I had a meeting with Prof. Wilson Sibbett who was the head of the School of Physics at the time. He in turn then, erm, had words with the head of the workshop. We also set about ensuring that George would be likely to get promotions, and to fund him at a higher level for later contracts. In parallel with this we also arranged for financial support for another technician, Willie Smith. This was paid for by an NPL project, was useful for the work we did, and bolstered our case for ensuring our concerns were taken seriously.
A milestone was reached in May. On the 9th Graham’s viva took place, and he ‘won the race’ to be my first research student to gain their PhD! His external examiner was Nigel Cronin. We’d know each other and worked together for many years when Nigel and I been at QMC. Since then he had become a lecturer working at Bath University. The viva went well. That evening, having got together for the first time in ages, naturally Nigel and I went out for a drink that evening. I came home in a suitably ‘relaxed’ state and went to bed. The next morning when I was getting dressed I found that my credit card was missing from my wallet! I realised that I must have dropped it somewhere the previous night. So I phoned the card company and they inactivated the card, and set about sending me a replacement. On the 12th I found the missing credit card. It has fallen out of my pocket and ended up under the bed at home. I must have dropped it as I was getting undressed. By then, of course, it was no longer an active card, so I cut it up and put the pieces into the bin.
During 1990 I began to write an increasing amount of material for ‘New Scientist’ magazine. In June they published an article with the title, “Chaos on the circuit board”. This outlined the work I had been doing to investigate the behaviour of various electronic circuits that would act as chaotic oscillators, etc. More generally, a lot of the research other academics were doing into chaos at that time was based on using digital computers running programs. I also did use computer programs to some extent, but I also built various simple analogue electronic circuits so I could see how the ‘real thing’ actually behaved. I was mainly comparing the behaviour of QW/MB diodes at mm-wave frequencies with simple transistor-based electronics operating at audio frequencies. This was to try and determine what was happening without some of the snags which arose using digital computations to represent behaviour.
New Scientist 30 June 1990 - Chaos on the circuit board.
The above image displays excerpts from the New Scientist article. On the left, overlaying the front cover, you can see one of the illustrations that showed a simple oscillator circuit that would act in a chaotic, noise-like manner. On the page shown to its right you can see an illustration of the oscillation behaviour produced by a computer program that tried to mimic the transistor circuit.
A basic snag with using digital computations for such purposes is that they are what is called ‘finite state machines’. This limits their ability to mimic the real world. The simple transistor circuit shown above uses an arrangement which has two main sections. One is a two-transistor amplifier which has a non-linear behaviour. The other is feedback network consisting of a set of three inductors and three capacitors which link the output back to the amplifier’s input. The ‘state’ of this system at any instant is determined by six values – the current though each inductor, and the voltage across each capacitor. When we combine amplifier gain with feedback we can get a situation where the result oscillates. If the amplifier is linear this tends to produce a regular waveform like a sinewave or squarewave. But in this case the amplifier has a nonlinear spike in the relationship between its input and output. So with the right (or ‘wrong’, depending how you look at it!) choice of inductors and capacitors, the result is a chaotic oscillation behaviour which shows no obvious regularity and may never precisely repeat the same pattern. You can, of course, use a computer program running on a normal computer to mimic or model this arrangement, but this is where the problem may crop up...
For example, such a computer model requires six variable values to be stored as binary numbers to represent the inductor current and capacitor voltage levels at any given instant. Let’s assume these are stored on the computer as 32-bit numbers. That means the ‘state’ of the model of the amp + feedback is represented by a pattern of just 6 × 32 = 192 ones and zeros. Since each of these binary locations can only hold a one or a zero, that means their the number of possible combinations is 2 multiplied by itself (i.e. raised to the power) 192 times. OK, this is a very, very large number of possible values. But it is finite. And one of the properties of genuine chaotic behaviour is that it can be expected never to ever repeat exactly the same pattern or return exactly to any previous state. Indeed, the way the state of a chaotic system develops is infinitely sensitive to the precise state it was in at a previous time. Yet the digital numbers rigidly quantise the values they store and calculate, and can’t represent an infinite set of possible states.
Digital versus Analog.
The above illustration contrasts the digital and analog approaches. The digital program running on a computer uses a fixed set of locations in memory to store the current state of a system as a set of binary numbers. These values are then put though a calculation to estimate what the values should become a short time – e.g. a microsecond – later. The results replace the initial values and the process is then repeated over and over again. In effect the method ‘steps’ though time. Hence in addition to the voltage and current values being quantised, the process jumps forward via a series of quantised time-steps, which inevitable means a lack of precision.
The analog example on the right of the above diagram shows a photograph taken of the screen of an oscilloscope. The ‘scope inputs were connected to two of the capacitors in a small analog circuit that was oscillating in a chaotic manner similar to a device like a QW diode in its waveguide mount. Here you can see how the resulting voltages vary smoothly with time in a very complicated way. These voltages are only ‘quantised’ by the quantum-mechanical aspects of the behaviour of electrons, etc. And the oscillation process developed smoothly. So avoids the limitations of the digital computer model.
For many other purposes the limitations of the digital approach don’t matter. For example, it is easy to accurately model many linear systems like sinewave oscillators that way. But for non-linear systems capable of chaotic behaviour a computer program may fail to reliably represent the reality, and give no sign of what has been missed. In practice for the work we were interested in, an analog system with a few transistors, etc, was also much cheaper and easier to use! As a result we tended to use a mix of computer models, simple analog circuits, and real mm-wave devices to investigate this topic.
During the summer the SERC grant where were were working in conjunction with Nottingham University was drawing to an end. I was concerned that this hadn’t proceeded as I’d hoped. Novel devices were made, and did yield some promising results. But in practical terms this seemed to be a matter of being given some devices to try, but which weren’t really made in a way that could optimise or maximise their performance at mm-wave frequencies. In general, they were physically too large and had too high a capacitance, and this severely limited their ability at mm-wave frequencies. Despite initial agreements I had come to feel that Nottingham regarded the mm-wave work as a minor side-project, not part of the central aim of the grant. We had started discussing a follow-on grant proposal and I made my views clear in a letter I wrote on the 3rd of July to Martyn Chamberlain at Nottingham where I said:
“To be honest, I feel that the crux of this project is to develop devices which we can demonstrate work as good oscillators at frequencies above 100GHz. Without this, I don't know why we are doing it - or why SERC should support it (mind you, I never understand why SERC do anything they do!)”|
“We won't win any prizes for oscillators below 100GHz. ... Don't fall into the trap of thinking that 50GHz work is 'good modelling' and that waveguide systems are 'OK really'. Lots of good people have failed to get anywhere at high MMWave frequencies by thinking that way! Problems arise at 200GHz that you don't get a sniff of at microwave frequencies...”
Unfortunately, overall, the impression I’d formed was that Nottingham were happier to work at low frequencies and assume all their results could be scaled or modelled upwards. Whereas experience had shown me that mm-wave methods applied to more appropriate devices were vital in order to make real progress at higher frequencies. Modelling, etc, was very useful, but in the end, the need was for actual practical devices that could be shown to deliver useful performance! The project that was coming to a close was, for us, a disappointment in this respect.
During the first week in July Glenn White visited us and we discussed various ways to obtain improved local oscillators for the mm-wave region, and how to also improve the performance of the lock-loop systems which controlled them. One novel idea we were considering at Glenn’s suggestion was that cooled Indium Antimonide was thought to have a property of acting as an oscillator under the right conditions. So another of my research students, Darrell Smith, was investigating this.
Darrell was with the St Andrews mm-wave group from 1989 to 1992. His work was directed towards finding alternative forms of mm-wave oscillators. In the main this focussed on two areas. One was to see if it might be possible for us to make our own high-frequency vacuum-state devices like backward-wave oscillators (‘carcinotrons’). The other was to try and exploit a novel solid-state device property like the one Glenn had mentioned, or what were called ‘flux flow oscillators’ which some Japanese researchers were investigating at the time. Alas, Darrell’s work didn’t lead to any advances in these areas. Gunn and QW devices remained the main way forwards for mm-wave oscillators.
On the 6th of August Andy Harvey completed writing his PhD thesis and submitted it to the University for examination. Derek Martin had agreed to be his external examiner, so I contacted him and we arranged the date for Andy’s viva. A couple of days later John Duff visited to discuss the work being done by Mike Leeson, sponsored by EEV. The following week I travelled down to London for a few days. Dick Fifield was one of the editors of New Scientist and we’d arranged to meet and discuss possible future items I might write for them. At the end of August I had a meeting with someone from ‘Aquila’. This was an organisation that acted as a conduit between the primary standards work done at the NPL and UK companies who wanted to have their own calibration and measurement systems that were able to give them standardised results. They were interested in the work we were doing on ‘chaotic’ noises sources as a possible transfer standard and agreed to fund some of Malcolm Robertson’s work on this topic.
On the 11th of October I met Kate Charlesworth. She had been producing a series of full-page graphic stories for New Scientist which were informative and enjoyable. I was wondering if we might be able to co-work in some way to help explain various aspects of electronics and mm-wave developments to the New Scientist audience, or perhaps in an fully illustrated book. Sadly, nothing came of this as we were both too busy with other things. But I was very impressed with her work and think it made a great way to convey complex scientific ideas to people.
Andy’s PhD viva took place on the 16th of October. This followed the pattern I’d initially adopted for Graham’s viva. I arranged for the actual viva to be held in the morning so that lunchtime would serve as a good ‘break point’. I’d booked a lunch at a local hotel – The Parklands – which provided really superb meals. All being well, we could complete the viva in the morning, tell the student they’d succeeded in passing, and invite them to join the examiners for a celebration lunch. This has worked perfectly for Graham’s viva, and it did once again for Andy. In the back of my mind, ‘plan B’ was that if the viva threw up serious problems the examiners would take a break for lunch and argue over it about what to do, then resume in the afternoon. The ‘viva and then lunch’ approach became the standard one for all my following PhD students. Happily, plan B never seemed to be required!
In the evening after the viva Chris and I had dinner with Derek. We chatted about things in general. One particular topic we spoke about has stuck in my memory ever since. This was the issue of how the research councils might be reformed since neither of us was completely happy with some of their decisions or working methods. A particular problem was that they consisted of a set of ‘panels’ and ‘committees’, each having their own remit and area of interest. In practice, this then acted as a series of ring-fenced ‘support groups’ for specific established areas of research.
If you were well known for work in one of those areas, your proposals for a grant would come to their attention. The snag was that this structure tended to be closed and protective of the areas and the people already established in them. More than once, mm-wave proposals I’d submitted had been sent to an inappropriate committee because there wasn’t an appropriate one. Typically, I’d have a proposal sent to an astronomy committee because my work years earlier had been on receiver systems and telescopes for astronomy. Yet my actual proposals by 1990 usually weren’t for application in astronomy. And in many cases the proposal was for a type of work no-one else had been doing, so there simply wasn’t an appropriate committee whose members were familiar with the field.
The result was that the proposals I’d made had often been rejected – partly because the committee members had no interest in what I proposed, but also because they often didn’t have the background to understand what was I was proposing. I recall one particular example where I was told something I proposed was “impossible”. Yet in reality we already had a working system doing it in our lab and I was actually asking for some support to make an improved version!
During dinner, Derek made a suggestion which horrified me at the time. This was that, instead of the committees carrying out their mysteriously behind-closed-doors discussions the money should be shared out to university researchers who had a track record of innovation for them to decide to pass it on to. My initial reaction was that this would just entrench an existing, informal and unadmitted form of ‘old boys club’. But on reflection later on I realised that if the process was open, then it could be challenged and its results analysed in a way that the existing system evaded. Personally, though, I took a more radical view. That the research money be ‘spread thin’ and handed out to each active university researcher by giving each one about the same amount. But to then encourage them to join and work in co-operation, giving a top-up amount to those who did. Thus allowing active researchers to openly compete, combine, and assess each other without a few ‘ennobled’ individuals making all the decisions. I still think this a good idea, but I’ve not held my breath as I wait for it to occur...
Announcement of NPL/RSRE Contact for Graham and George’s work.
Also during October I received official confirmation of a contract being placed with the mm-wave group to build and supply RSRE with an instrument to accurately measure oscillator sideband noise. We’d been discussing this with Ralph Yell and others of RSRE/NPL to arrange what they required. The contract was to run for two years and employ Graham Smith and George Radley. I wrote a short press release announcing the event and it appeared in various newspapers. The above shows how it looked in the local “St Andrews Citizen” paper which effectively reprinted what I’d written.
On a personal level, 1990 had become a very difficult year for Chris. She was unwell on a number of occasions and her mother was diagnosed as having a cancer, and was expected not to survive past November because it had spread to her lymph nodes. As a result Chris was seeing her then-GP quite often due to depression, etc. I was also seeing a therapist about my feelings of anxiety, stress, etc. Chris’s father had avoided telling her at first how serious her mother’s situation was. But we got a phone call from her brother telling us what the situation was. At the time he and his family lived just two houses away from their parents.
Once she knew, Chris went down to Manchester to stay for a while with her parents. She discovered that her mother had needed a wheelchair to move about and was now essentially confined to bed. During the last few weeks of her mother’s life Chris spent more time in Manchester than in St Andrews. Her mother died during the first week of November 1990. She had been in bed on a morphine drip and we think she simply decided it was ‘time to go’, so stopped eating and faded away. Her parents had deliberately moved to be near Chris’s brother with the risk that they might eventually become frail or invalids in mind. So her father wouldn’t be totally alone, but it was still a dreadful time for all involved. At the funeral, Chris had a fit brought on by the distress.
Changes and experiments with her medication has also been adding to her symptoms rather than making things better. For a time she kept coughing up blood which turned out to be due to the medication she’d been put on to see if it helped. She repeatedly alternated between being seriously depressed, distressed, and confused and being hyperactive. These boughts tending to last a week to ten days each, alternating. The GP also on occasion recommended trying something which experience had shown caused a problem. An example of this was that it was suggested that Chris try taking Vitamin B6 supplements. She’d tried this in the past, and it had almost immediately caused her to have one or more ‘Grand Mal’ fits within a couple of days. These were much worse than her usual, more minor, seizures. At the time we sometimes had the feeling that her GP wasn’t really tacking the problems, but tended to fear/assume this was because little help was possible and we had to ‘make the best of it’!
During December I was engaged in discussions with Alan Costley who was working at JET (Joint European Torus) on making mm-wave and far-infrared measurements of the plasma. Andy Harvey and I were interested in supplying them with various quasi-optical components, etc. In particular, some ‘notch’ filters and an interferometer system to measure phase changes.
The notch filters were for a plasma diagnostics instrument that fired a high-power 140GHz beam into the plasma and then collected the scattered radiation for analysis. The idea being that the thermal velocities of the charged particles in the plasma would Doppler shift the backscattered radiation. By measuring the spectrum it would be possible to determine the temperature of the plasma along the illuminated beam path. The snag was that the resulting signal had at the center of its spectrum the high power ‘probe’ energy which would tend to blind a sensitive receiver. So the filter was to cut down this component, making the rest of the scattered spectrum easier to measure. As a result, the filter needed to be very narrow with a high rejection factor at the beam frequency, but pass nearby frequencies very efficiently. So it was tricky to design and build. The phase interferometer used a similar probe input beam, but was to measure the phase of the beam that emerged having passed though the plasma. Knowing the physical length of this path and the beam’s wavelength the result could be used to determine the refractive index of the plasma, and thus its density and electromagnetic properties.
Chris had essentially given up trying to find a paid job because the impression she got was that prospective employers wouldn’t hire her because of her epilepsy. Instead, by the start of 1991 she had settled into a routine of working 2-hour shifts at the St Andrews North East Fife Citizen’s Advice Bureau (NEFCAB) in St Mary’s Place. This was in rooms allocated in a council building there which had previously been a primary school. Over the years a steadily increasing number of people were coming in for advice and help with finances. Quite simply, more and more people were finding it hard to earn enough money to pay their bills. Chris found working as a volunteer advisor/helper for CAB clients worthwhile, but worrying. Even in a town like St Andrews which on the surface looked prosperous, it was clear that a rising number of people were struggling with poverty.
Unfortunately, it had also become clear that Fife Regional Council (FRC) were becoming less willing to pay for an organisation they didn’t control or clearly ‘own’ in publicity terms. During the Miners Strike in the 1980s the FRC had set up a ‘Rights Office Fife’ organisation to support the miners and their families. This, they could control, and take credit for. Regardless of the increase in the numbers of people seeking help from NEFCAB the Council took the approach which was summed up in the 1989-90 NEFCAB Annual Report by its Manager, Mary Kinninmonth: “Whilst the NEFCAB is in a ideal central location, the building is deteriorating daily. FRC advised us in April 1987 that buildings have generally outlived their lifetime...” In short, the message was, you’ll lose your offices soon and no alternative is on offer. NEFCAB did continue for a few years after 1991, but everyone expected it would close sometime soon. Which of course left Chris wondering what she might then be able to do, when that happened.
New Scientist 19th January 1991 - QM in the Home.
Having been successful at writing more than one feature article for New Scientist I was invited to write a leaflet they produced as part of a series they called ‘Inside Science’. Initially, a copy of each of these leaflets appeared in the center of an issue, as a ‘pull out’ item that could be kept. However the aim was that schools, etc, could then buy a set of these leaflets as teaching aids. As a result the above pull-out was included in a January 1991 issue. This was based on the material I’d written for my undergraduate courses in ‘Information and Measurement’ and basic electronics.
During February I was told that my application to the University for a ‘Sabbatical’ had been agreed. As a result I was nominally relieved of all non-research duties for two terms. These covered June-December 1991 and in 1992. In reality, of course, that still meant I had to give the lectures in other terms, and go on supervising my research students, etc. So was rather less generous than it might appear. However it would give me some extra time to work on new ideas, and to write things.
By the start of March I’d stepped down from the James Clerk Maxwell Telescope Users Committee. This was partly for personal reasons. Both Chris and I were very stressed at times, and tended to be unwell. I found it difficult to leave her at home travel because I worried about her. She kept experiencing periods of severe depression and anxiety, and we both were struggling to cope a lot of the time. In addition, though, I’d become increasingly unhappy with the way receiver development for JCMT was being done. I sent a letter to Ian Robson where I wrote:
“I'm afraid that I felt that the receiver situation had progressed beyond "the tail wagging the dog" to "the tail thinks it is the dog". To my biassed, semi-outsider's eye, the situation seemed to be firmly under the control of Cambridge & run pretty much for their benefit. The process whereby they held a tight grip on the main stream of receiver building was, I feel, very unhealthy and inefficient. This hasn't been helped by their tendency to ascribe receiver operating problems to everyone except themselves.
I did try to introduce the idea of 'modular' receiver work on a sort of 'continuous competitive development' basis. But that sank without trace. I also tried to push the idea that ROE (and others) should be allowed to compete on a rather more level playing field. No joy there, either. I finally decided that my ideas weren't going to ever coincide with anyone else's & I'm not really emotionally constituted to perform the role of Cassandra, so I decided to let everyone else get on with things without my carping.
My feelings were confirmed after I left by the way the ROE/SRON/St. Andrews proposal to work on advanced bits & pieces was treated. This was withdrawn without either our knowledge or approval following - so I believe - strong criticism from Cambridge. To me this whole affair was a fairly shameful demonstration of the real problem, which is 'political'.”
I really did regret breaking my remaining link with JCMT because I felt then – and now – that it was a wonderful project and I knew it would generate excellent results. I would have liked to have continued to help with instrumentation there. But I had come to feel that the development of receivers, etc, wasn’t as open, competitive, or cost-effective as it should have been to newcomers with alternative ideas or skills. However, having made my points, I was happy enough to concentrate on other areas where the St Andrews mm-wave group was making progress.
On 27th March I wrote to Martyn Chamberlain at Nottingham. In the letter I said that Malcolm Robertson had phoned and been told they now had some improved diodes which we hadn’t known about. I explained that we needed more, improved, diodes for Malcolm’s work to progress. But that although it had been agreed that Paul Steenson from Nottingham would spend more time here working with him, this still had not happened.
My Mother and Arthur came to St Andrews and stayed with us from the 9th to the 14th of April. On the 18th I wrote to the head of the Physics Department to let him know that RSRE had agreed to support a PhD student to help carry out work they were going to fund. The person we wished to get a studentship was Duncan Robertson – Malcolm’s younger brother. Duncan had done an undergraduate ‘summer project’ for us during the long vacation in 1989. During that time he’d designed, built, and tested electronic interface cards for the impedance measurement system that Andy was developing. He’d done excellent work and was familiar with what we did. So the aim was to get him a PhD studentship working with us.
By then however, SERC had changed the rules and CASE awards could now only applied for after a student had been identified. Hence the department would have to decide to allocate one of their studentships to him before a conversion to CASE status could be considered, formally bringing in its extra income. Fortunately, RSRE had already said they’d pay money to support his work even if he wasn’t given CASE status. The initial project work was to be on devices that could generate chaotic oscillations as a potential noise source for calibration purposes, etc.
Views of some sections of Iris and Ted’s Garden.
Chris and I then travelled down to Essex and stayed with her Aunt Iris and Uncle Ted. They lived in a home that really was out in the countryside. It was one of a small cluster of just two or three homes surrounded by open fields. Their garden was astonishing! I’d not visited there before, nor anywhere else like it. The house itself wasn’t particularly large. When I first looked out their back window I thought the garden looked lovely and quite big. However later on when we went out to explore I discovered that what I’d assumed were some trees at the end of the garden were in reality just a screen with a path though – leading to another large garden area! At the end of that, more trees and bushes, and beyond that yet another, and more beyond! It was about ten times longer and bigger than it had seemed from the house. In effect a series of gardens, each one with different features and plants, carefully screened from each other like a series of woodland glades. Iris, and Ted in particular, spent a lot of their time out in the garden, and the results were spectacular.
Ted was also a keen maker of ‘home brew’ wine. Normally I’m not much of a wine drinker and tend to expect home-made wine to taste more like vinegar or rocket fuel. But again I was really impressed when I tried some of the wine Ted made. In particular I’ll never forget the wonderful Parsnip and Elderflower sparkling wine he gave us. It really did taste like the best Champagne I’ve ever had a chance to try.
Another surprise was when Iris and Ted took us out in the car to Ongar, one of the local towns. Having driven around for a while we turned off the road between two houses and went down what seemed like small driveway. However we then emerged at the corner of a large open field, surrounded by the back gardens of houses. The field was covered with a large vineyard! Anyone who simply drove along the roads in front of the houses on every side would have had no idea it was there. It was run on a collective basis by a number of wine enthusiasts from the area. Having got there we did a small amount of work on the vines, but for me the main purpose of the trip was to be amazed to find this vineyard, tucked away out of sight in a small Essex town.
Whilst ‘down south’ we also visited various people there. Chris went back to QMC on the 22nd and met Peter Crew (still working there as a technician), Prof Guy Wilson, and Tony Marston who’d been one of the technicians in the old Far-InfraRed ‘Astro Group’. Chris and I also took the chance to go to a concert on the 23rd of April of music by Prokofiev at the Royal Festival Hall. We then returned home on the 24th.
Electronics World June 1991 - Chaos in electronics.
By the middle of 1991 I was enjoying writing technical articles for non-academic magazines. So I decided to try writing one for Electronics World. Given that my main research interest at the time was on Chaotic oscillators, etc, this was on Chaos in electronics. I submitted this, it was accepted, and appeared in the June 1991 issue. Because the magazine was aimed at a readership with a good knowledge of electronics I was able to go into far more practical detail than in New Scientist. Electronics World ceased being published many years ago, although copies are now freely available from an American website. Unfortunately, at the time I wrote this webpage their copy for June 1991 has some pages missing! However I have now produced a pdf version which can be downloaded from this link. It provides an explanation of how to design and build suitable analog electronic circuits, etc, to investigate this area.
Illustrations from the Electronics World article published in June 1991.
The above illustration displays some of the diagrams from the article. In this case it shows how a ‘varactor’ being driven by some d.c. bias and an input sinewave can generate a very complicated output that is nothing like a pure sinewave. The diagrams labelled as Fig 3a and 3b show how the output varies as you adjust one of the input voltages (the d.c. bias). These plots will be recognisable by someone who has studied chaotic behaviour as the show what mathematicians dubbed ‘bifurcations’. In this example the result isn’t actually chaotic output, but what I came to describe as ‘semi-chaotic’ behaviour. The output varies in a very complicated way, but does repeat after a given time. So the output remains periodic and repeatable if examined in sufficient detail for long enough, but seems chaotic if only observed for a short time. Each bifurcation doubles the ‘repeat length’ of the pattern. Having sent time examining this behaviour I realised it offered the potential of a source that behaved in a ‘noise like’ way for a long time, but was actually pseudo-random, and both the repeat period duration and the pattern could be switched in a controlled way. This realisation opened up a number of interesting applications...
During August Chris and I spent a few days in Pitlochry. Overall, we enjoyed the trip, but at the time I had become quite anxious about getting some work done, so we only stayed for four days rather than the planned week. We did, however, take the chance to see a play – J. M. Barrie’s “What every woman knows” – at the theatre there. We also bought two watercolour paintings from those on display at the theatre. These are of scenes from the landscapes in the area, painted by Margaret Roberts. When we got back home, Chris made some inquiries for her Father. He was considering the possibility of moving to St Andrews to be closer to us now that Chris’ Mother had died. However having thought about it, he decided later on to remain where he was.
New Scientist 28th September 1991 - More Democracy, please, at the SERC.
During the latter part of the 1980’s I had become increasingly unimpressed by the way the UK Research Councils operated. To me they had come to look very much like a group of ‘magic circles’ where people were appointed or were awarded grants largely on an ‘old boy network’ basis. (And, yes, they were indeed mostly ‘boys’ – i.e. male.) I’d found it vastly easier to get support for research from private companies or other sources that weren’t the SERC, and on more than one occasion I’d had an SERC application rejected for reasons that showed that neither the council nor its chosen referees had actually understood the proposal. Indeed, I kept having proposals sent to an inappropriate section for assessment. They then duly turned down the proposal because it wasn’t relevant to their interests. At a higher level, they seemed to be adopting policies that I thought counter-productive and I knew also upset their own research staff! A classic example of this was a proposal to merge the Royal Observatory Edinburgh with the Cambridge Observatory.
Having talked to various people about this, I stuck my head above the parapet and wrote an ‘opinion piece’ for New Scientist, making some of my views known. More than one person said I was ‘brave’ – i.e. code for ‘reckless and mad’ – to do this, but I’d decided that my main research support came from elsewhere anyway, so it couldn’t do me much harm. Indeed, it might even make them become more wary of refusing any future research grant proposals I submitted out of the risk of being perceived of doing so because I’d criticised them. On balance, though, I felt that those in charge didn’t actually care what I thought, one way or the other, so would either ignore it or be amused! Afterwards, I got more than one letter from well-established academics who thanked me for writing the item, sometimes admitting they’d have wanted to but decided that discretion was the better part of valour...
The proposal to merge the observatories was eventually shelved. I doubt this was because of my article. It felt good to write it, though!
On October 9th we received formal approval for Andy to work on the instruments and devices we had proposed for JET. This added to the funding he was able to bring into the group. I also had some discussions with Nigel Couch of GEC Hirst about Malcolm’s work on oscillators and GEC continued to supply experimental Gunn diodes for him. These devices proved very useful given that Nottingham weren’t supplying as many appropriate ones as we’d I’d hoped. Mike Leeson had continued to develop a quasi-optical dual polarisation CW swept FM radar system for his PhD project. At home we had finally decided it was time to have a new kitchen, replacing all the old worksurfaces, etc. So we started talking to a local firm about the design.
During November I wrote and sent off to the NPL a fresh proposal for another Metrology Prize. However for me perhaps the most lastingly memorable event of November 1991 was something that emerged from the local secondary school, Madras College. Their 5th and 6th year students had organised a small company to design, manufacture, and sell, a board game they had devised. This was called the ‘St Andrews Game’. News of this filled the top half of the front page of the ‘St Andrews Citizen’ newspaper of 29th November. it then went on sale in local shops.
Part of the board of the “St Andrews Game”.
Its a game based on having to race to see who can be the first to get around to a set of randomly chosen destinations in the town. Along the ways risking landing on a (yellow) square that leads to having to take a card that alters your fate in some way. It is quite good fun, and features various locations that were, and are, well known to townsfolk and students. Perhaps inevitably given who designed the game, destinations like the local ice cream shop, sweetshops, and pizza restaurants feature! I’m pleased to say we still have a copy of the game and play it. Good fun.
In stark contrast, December 1991 stands out in my memory because of one of those apparently small events that led to some very unwelcome consequences. On one of the last days before the holiday I was in my office tidying up some paperwork. I was sitting on an office chair and swivelled around to reach for some papers. One of the desk drawers was open and the bottom corner of the draw impacted my knee, the point driving in between the kneecap and leg joint at the side. It hurt at the time and I commented to myself that “I’ll feel that later!” as the initial pain quickly faded. Indeed I did...
My knee remained uncomfortable over the Christmas and New Year break. However during that time I didn’t need to do a lot of standing up or walking around. So I could rest my leg most of the time and avoid it becoming too painful or stiff. However when undergraduate teaching resumed for the first term of 1992 I had to stand up to lecture. This rapidly made my knee much worse. I summarised what then happened in a letter I wrote to a good friend, John Scott, later on in April...
“Did you know that there is a bag of watery stuff inside the knee? It’s a sort of shock absorber or something. (I'd heard of bags under the eyes, but not under the knees!) Anyway, little did I know but I'd dug a hole in said item. Just after Christmas I began standing up lecturing and so on. Sudden nasty pain in the knee.
Tendons inflamed and out of place. Doctors. Bandages. Saying 'Aaaah'. Then my ankle went, too. Then my whole leg swelled up like a party balloon. (A straight one, not a curly one.) Within a week I couldn't walk more than few paces at a time. Apart from being carted off to a hospital in Dundee for an ultrasound scan (they were worried I might have a blood clot - nope, no more of a clot than before) I spent weeks confined to home.
After two months I was fit enough to hobble back to work using a stick. This meant I was able to get there just in time to catch 'flu from someone who was getting over it. More 'Aaaah' (and 'Aaargh'!). Things seem fairly normal now (touch wooden head).”
It took about six months before I was able to walk without needing the stick at all. What I hadn’t told John about was the gory details of the dash to the hospital in Dundee. The swelling had become so big and painful that Chris asked a friend to take us to the DRI (Dundee Royal Infirmary) as a quasi-emergency. I couldn’t walk at this point, just hobble a few paces and then sit down. We arrived at about 10:30 am and found a sign at the hospital’s ‘dropping off point’ which said that cars could not stay there. So I had to wait on a seat nearby while the car was parked elsewhere. Chris went to find someone to ask for help and a wheelchair so I could actually get into the hospital.
She then had to struggle to bypass the ‘take a ticket and wait two hours’ system inside before she could get anyone to pay attention when she said I was outside, in the cold, and needed a wheelchair to be able to come into the hospital. The people there said then told her didn’t have any wheelchairs... The result was a long run-around with no-one initially taking any responsibility for helping as I sat outside in the cold with a painful leg. Chris made some fuss and went around seeking help. After a while someone did come out and wheel me inside, but without Chris knowing this had been done! All in all, it would have been a bit of a comedy routine if I’d not been in such pain at the time, and were weren’t worried that the problem with my leg might be very serious.
Chris eventually found out where I’d been taken and joined me. Over the next few hours I was given an ECG and an ultrasound scan of my leg. They also injected a marker liquid into my leg and did an X-ray. The concern they had was that I might have had something like a thrombosis. At about 6pm a doctor appeared but then vanished again before talking with me about the tests, etc. Apparently he’d been called to a meeting elsewhere. At about 7:30pm two other doctors came but couldn’t work out what might be wrong with me. By then it had been reported that the scans were ‘clear’ of any signs of a serious problem like thrombosis or fracture. So although the full notes of the tests hadn’t appeared they agreed it was probably fine for me to go home again. Thus after more than 9 hours of what seems like a story from a comedy play about the NHS – much of the time with me on a trolley – we went home again.
During the following months I became accustomed to using a walking stick. Even walking from one room to another was difficult for some weeks, so my lectures had to be postponed. By April, though, I was going back to work each day. I still needed the stick to walk, and was occasionally tempted to put a stuffed parrot on my shoulder and do ‘Long John Silver’ impersonations when lecturing. It was about six months before I was able to completely stop using the walking stick. And at times since then I’ve had pains in that knee. So, yes, my initial reaction that “I’ll feel that later” turned out to be accurate!
New Scientist 1st February 1992 - Spy’s Guide to Chaos.
In their February 1992 issue New Scientist published another feature article on our work on ‘Chaos’. In this case it was specifically on the possibility of using ‘Semi-Chaotic’ behaviour as a system to hide information and make it look like ‘noise’ to unwary eavesdroppers. These days people are familiar with encryption as an everyday process. When you view a webpage via an address that starts ‘https:’ the communications between your browser and the source of the webpage are encrypted. People are less familiar with ‘steganography’. This is the science (or art) of hiding the fact that any information conveying message was even sent or received!
The idea I had was that mm-wave sources which operated in ‘noise’, chaotic, or semi-chaotic modes could spread the resulting output over a very wide bandwidth – typically well over 1,000 MHz. This meant that the resulting pattern could have a fairly low power spectral density (amount of power in each Hz of bandwidth). So would appear to the unwary as ‘background noise’. However if we consider again the way a nonlinear system can undergo bifurcations in its operation as we adjust its operating conditions we can see that it is possible for it to be switched in a controlled manner. So, for example, it might be operated in a mode which has been set to be such that having, say, 25 bifurcations, represents a binary ‘1’ and, say, 26 bifurcations represents a binary ‘0’. (I’m choosing these values simply for the sake of an example.)
After 25 bifurcations the output pattern only repeats after more than 33 million cycles of its basic oscillation, or about 67 million for when in the 26 bifurcations mode. However if the user changes the mode in less time than this takes, the actual observed patterns aren’t given time to show that they repeat. In effect, we are transmitting what looks like ‘noise’. Since the details of the pattern depend on the precise nonlinearity of the device being used, that device becomes a sort of ‘key’. Using mm-wave devices this process can potentially be very fast, and requires a similar ‘key’ device to be detected and demodulated. Otherwise the transmitted signal may simply be regarded as background noise. My initial interest in this area was because the devices could generate noise, but almost naturally, that then extended into novel uses for the semi-chaotic behaviour that exploited the fact that although it looked a lot like noise, it could actually be a very complex, but structured, pattern.
In March 1992 I send a letter to Martyn Chamberlain at Nottingham giving him some of the concluding details of the work we’d been doing. This was for the final report he had to write as the main Grant Holder for the work on their quantum well devices. I apologised that I’d not done this sooner, but explained I’d been off work due to illness for a time.The grant had now reached the end of its term. At about the same time I received a copy of a new Grant request proposal which Nottingham had submitted to the SERC for funding. This put me in a dilemma because of our experience working with them. Their proposal seemed to me to fail to tackle the main difficulties in what they said they wished to achieve, and their stated goals were actually of a lower level of performance than I knew others elsewhere had already obtained by simpler and more appropriate methods.
I was strongly tempted to decline to comment, but I felt this would be unfair on others seeking SERC grants. And in itself would have seemed odd to the SERC. So I wrote a long and detailed response for the relevant SERC committee. In order to explain my doubts I had to explain what had happened during the grant which was coming to and end so they could see the basis of my concerns. This meant they could take into account deciding if I was unfairly biassed, or not. That sent, I don’t know if the proposal received grant support or not. I accompanied the comments with a letter, explaining that I’d been reluctant to be frank about my experiences and pointed out that I’d avoided being involved with actually writing up the previous joint grant. From my point of view I’d have been content for the matter to be “dead and buried” as I put it in the letter. However having been sent the new proposal to comment upon I felt I should be honest, but stress that what I said was just my opinion and that others might take a more positive view.
Presentation of National Metrology Prize Award at the NPL on 22nd May 1992.
At the start of April 1992 I received a letter from Peter Christmas of National Physical Laboratory. This told me that I had been awarded another National Metrology Prize. In this case I’d submitted a proposal to develop a mm-wave chaotic oscillator as a calibration ‘noise’ source, and this idea had been chosen for the award. The photo above shows my being given the prize at the NPL on the 22nd of May. As with the previous award the NPL also then paid the mm-wave group to carry out research to develop suitable sources. This was really excellent news because it would be accompanied with some funding to build a suitable instrument, this bringing in more support for Graham Smith and the mm-wave group.
A few days later I got a fax from someone at the SERC telling me that they had not yet received the final report from Nottingham regarding the joint grant that had recently ended, and that it was now overdue. I explained that given the circumstances of that grant and how it had proceeded we had agreed to leave writing this report to Nottingham University. The ‘no show’ was a problem for us as well as Nottingham. I had also submitted a proposal on a different topic, but the SERC rules blocked them considering proposals from someone when a final report on an earlier grant was overdue. To clear the blockage so far as we were concerned I then wrote a ‘partial’ final report of my own, dealing with the minor fraction of the work we’d been responsible for. This then resolved the issue for us. But it meant that I had to go though explaining the difficulties we’d had, yet again, which I’d hoped to avoid. So overall, not a happy experience for me, nor – I assume – for Nottingham.
By this time I’d established that whenever the mm-wave group was given a large contract, grant, or prize, I’d take a bottle of champagne into the lab. This was then cooled using either ice from an ice-making machine, or ice made via the liberal use of liquid nitrogen. We would liberate some plastic cups from the coffee machine in the foyer and use these to toast the event! On occasions we’d also have a group meal. Hence, following the Metrology Prize award we organised a party at the ‘Balaka’ restaurant in St Andrews. The above photos identify the ‘cast list’ of the group at the time, and you can see that both Duncan and Malcolm Robertson appear...
26th Jan 2019