Following National Service in the RAF and graduation from Cambridge University, I joined Elliott Brothers (London) Ltd. as a junior engineer in their Radar Research Laboratory at their Elstree Way, Borehamwood, Herts. establishment in September 1962. Prior to this my work experience had been limited to delivering newspapers during the school holidays, two 6 week periods in the Umbro Sportswear warehouse in Wilmslow, Cheshire; 6 weeks as a technical assistant in the Argos computer laboratory at Ferranti Ltd., Wythenshawe; a few weeks as a technical assistant at Hunts Capacitors in Cheadle, Cheshire and an 8 week formal mechanical workshop training period at Ferranti Ltd. This was part of my needed experience before commencing my Mechanical Sciences Tripos Pt.II year at Cambridge.

Elliott Brothers (London) Ltd.was part of Elliott Automation led by Sir Leon Bagrit, which later got absorbed in to English Electric, which company merged with GEC under Lord Weinstock, and eventually moved to BAE Systems. For much of my nearly four decades in the Company our trading name included Marconi, already part of English Electric; for example, Marconi Avionics, Marconi Elliott Avionic Systems Ltd., Marconi Defence Systems, GEC Marconi.

The history of Elliott Brothers (London) has now been written by Simon Lavington.

Radar Research Laboratory (RRL)

My first work in the RRL was familiarisation with rather smaller waveguides than those I'd encountered in the RAF, some aerial testing, transistor application and generally finding my feet and being guided in the ways of a small technical laboratory. Shortly after I'd settled in, the three new graduates were called in to the Laboratory Head's office and asked which of us had covered lasers on our courses. I was the only one, and therefore found myself being responsible for making a Helium Neon gas laser - invented just two years earlier in 1960 (when I had been reading physics) by Bennett, Javan and Herriott at Bell Telephone Labs. After a period of corporate review as to which divisions should go forward with laser work, my work did lead to a commercial range of lasers in a sister division. Looking back to this period, I'm amazed at how good our technical library was, and am now struck at the good quality of training we received, both from more experiened people in the laboratory, and the external educational courses which we were regularly sent on. These were a far cry from the "training" courses I had towards the end of my career which simply allowed my employers to tick some boxes.

Much of our work came from small research grants from the MoD funded Royal Radar Establishment (RRE) at Malvern, Worcestershire, and about a year after I joined, we received funding for investigating what was then a very novel idea indeed. A totally solid state radar. We already had a nearly all solid state 2-man transportable radar for the Army, but it still relied on a magnetron and klystron which required heaters, and wasn't as rugged as a pure solid state design should be. So I found myself working on a small project by the codename PERT - Power Economy Radar Techniques. (This acronym predated the much more common usage as a management tool -Project Evaluation & Review Technique.)

In general, magnetrons produce short but high peak power pulses of microwave energy. Using a purely solid state source, the high peak powers would not be available so our design was to investigate whether using a form of pulse compression and coherent detection, we could improve the efficiency sufficiently for eventual Army use. The PERT radar design investigated the use of transistorised microwave sources, bought in from Microwave Associates and a simple 7 bit Barker Code phase coded Pulse Compression scheme. The overal pulse length was 210ns and the elements were 30ns each, giving a spatial resolution of slightly better than 1 foot. On reception the signal was mixed down to 150MHz, and amplified in an IF amplifer before being phase detected in a ring modulator phase sensitive detector at 150MHz. Carefully cut lengths of low loss coax was used for the delay lines in the pulse compression detector, with each length having individually tweaked correction for the inevitable losses.

Trials during a rain shower

Trials in X-band were conducted using a standard dish aerial and the company's mobile laboratory pantechnicon in the South Mimms area where the road has now been swallowed up by the M25 and A1(M). The van was fitted with a noisy petrol/electric generator - a far cry from the modern device easily available from DIY outlets. The whole van shook when it was being started. Internal personnel moves meant that I became responsible for the project, and had to present the final report to representatives from RRE. I remember this as being a difficult meeting and it seems that our successful work had surprised some of the older establishment boffins.

At the request of RRE, we also investigated how technology could help the infantryman from being surprised by enemy incursions into the Forward Edge of Battle Area (FEBA). Infra-red beams using the very recently developed IR emitting diodes, ultra-sonic arrays and geophones (developed by the oil exploration industry) were all looked at. I investigated the infra-red beam system and decided that to comply with the very stringent battery requirement laid down by RRE, they simply weren't feasible. The problem wasn't the IR aspects, but heating the lenses so that condensation didn't occur and destroy the focusing of the beam - reducing the range to next to nothing. (Passive IR systems followed many years later.) What was successful was the use of geophones in an array or simply set into gravel. Simple signal recognition circuitry was devised, and the system went into Service, and a commercial variant was used to protect the Tutamkhamun Treasures when on display at the British Museum in 1972.

Display technology

Solid state light emitting diodes were still in their infancy, but it was already recognised that they were the way forward, and I spent some time trying to replace a fairly heavy and unique type of display used the company's infantryman radar. The existing display used a solid cube of transparent plastic rotating about its vertical axis, and through which the operator viewed the video outputs from the radar's range gates. The rotation spread the flickering signals into a horizontal line and helped the operator to decide whether the signals were from vehicles (regular blips), or from ground clutter (irregular blips). Various ideas were tried including colour coding the display to indicate speed, but what we really needed was an extended array of LEDs to simulate what was being done so easily by the rotating cube. Easy now, but totally uneconomic at the time; and the rotating cube with its noise, motors and power drain continued for some time.

QER (Q-band Experimental Radar)

At around this time there was increasing interest in exploring the use of shorter wavelengths for radar, and whether good quality mapping could be achieved. We received a research project to produce a ground-based radar for RRE which could be used to compare the quality of mapping of the Malvern Hills (from RRE) using radar at Q-band compared to what the OS maps showed. This required the development of a monopulse aerial system with associated transmitter, receiver etc. Some time into the project the requirements were changed, and we were asked to make the system fully airborne for trials in a Varsity aircraft. I was project leader, and gained invaluable experience during this period. Prior to the airborne trials we carried out trials from the upper floor of our building in Borehamwood, comparing our results against the OS maps. In one trial it looked as though there was an echo in the received signal, as the maps showed just one line of electricity pylons. We checked the circuitry carefully and could find no problems, so visited the area where the echo signals seemed to come from. Since the maps had been printed, another row of pylons had been built, and the finer resolution of our Q-Band aerial combined with a short pulse magnetron had resolved the two rows.


Aerial scanner mechanism on handling frame before delivery. Aerial with Cassegrain reflector mounted in BAC Varsity.

The transmitter and receiver (TR) unit for this radar was housed in a pressurised cylindrical container (pressurised so that the high voltages associated with pulsing the magnetron didn't cause damage by arcing at low pressures), and heat was removed by a forced air circulation system driving air through a heat exchanger in the double wall of the case. Additional cooling was via piped cooling to individual units in the TR unit. Novel zero displacement sealed fluid connectors were designed, and allowed the internal units to be removed for servicing without the need to release other units.


TR outer case, showing airway & heat exchanger coils. TR unit in its flight tray, showing the four Q-band monopulse inputs.

After bedding down trials, test flights were flown with the mapping output fed into a Ferranti moving map display, which I was privilged to see. This allowed the air-crew to see either the OS map or the radar map, or the radar map overlaid onto the OS map thereby confirming one's position.


The APS20 (Air Pulse Search 20) was an old radar (WWII) fitted to RN Fairey Gannets and RAF Shackletons, and used in the look-down AEW mode. Maintenance in the UK was handled for the RAF by Elliott Brothers at Borehamwood, and a new design of magnetron was fitted during one of the mid-life updates. This was a coaxial magnetron with a substantially more stable output, and consequently it was possible to now contemplate the addition of an airborne moving target discriminator, AMTI. This, if successful, would allow much better suppression of sea clutter and better visibility of aircraft targets, and was an early digital design. I was made Project Manager, and unusually for the period, the contract was a fixed fee, rather than Cost Plus. (So the quicker we did the work and spent less on salaries, the higher the percentage profit.) What a difference this made! For once my project had prioritiy in the DO, workshop etc. Flight tests over the North Sea were successful, and the records showed much enhanced performance.

Airborne Early Warning Radar

We were involved in experimental work in support of the Royal Radar Establishment and this lead to a major development contract but at a time when UK politics were in turmoil, and after a change in government, the contract was cancelled. Somehow we convinced RRE that a 6 months extention during which we would write up our progress would be sensible, and this was granted. A 16mm film was produced recording our progress, and this was shown to RRE and members of MoD.

Airborne Interception Radar for Tornado Fighter variant

Following the cancellation of the first AEW contract, I joined a small team of 12 charged with the task of bidding for the nose radar for the Tornado Fighter. This was a variant of the ground reconnaissance /interdiction/ low level strike version of the aircraft and would be equipped with different missiles, and would have a longer fuselage as a result. To our delight we won the bid, and were given overall charge of the development contract. I was made Equipment Manager, reporting to the Project Manager. The contract carried the proviso that the transmitter and scanner mechanism would be provided by Ferranti Ltd. in Edinburgh, using their design. As it was our aerial which was mounted on the Ferranti scanner, this was quite a difficult interface, and a challenging one to manage.

We developed two engineering prototypes. X1 was a simplified but fully airborne radar and this flew in a research aircraft to provide us with detailed statistics regarding ground clutter so that our signal processing could be optimised. X2 was built to explore the problems of packaging a complex radar in the confined nose of the Tornado fighter.

The above photograph shows the X2 experimental prototype for which as Equipment Manager I was responsible, and thought had been destroyed, but found on display it at Manchester Museum of Science and Industry in Feb. 2009. This was built essentially as a proof of concept to demonstrate that we could get fairly close to the packaging requirements of the nose of the Tornado, and that the cooling and electrical screening requirements would be satisfactory. However this model ignored the small but significant mounting slope in the Tornado (to bring the zero axis of the scanner in line with horizontal flight datum) and thus had the same depth top and bottom for the rear mounted peripheral mounted electronics units. It was also slightly larger in overall diameter than there was space for in the Tornado.

Re-engineering the configuration from X2 to those necessary for the Tornado was a complex exercise, and was done without the benefit of CAD - this was the early 1970s and CAD and similar computer aids were not available except as experimental and very primitive batch operated processes. (Hand in one's code, compile, and come back a week later to correct one's errors!)

Tornado Radar delays - The Blue Circle period

(Blue Circle was a well known brand of cement - used as ballast pending delivery of a working radar)

One of the common criticisms of the Foxhunter AI-24 radar for the ADV was that it was late on delivery, and as a result early versions of the ADV flew with ballast in the radar nose bay, and without any radar fitted. There a number of reasons why delivery was late.

Pre-Contract or Bid phase

Before the contract was let, and during the bid phase, two of us, myself the bid overall Hardware Design Engineer, and the bid Systems Designer visited RAF Fighter Command at their High Wycombe base. We met two RAF officers who were familiar with flying the RAF's version of the Phantom, and responsible for advising MoD on their needs, and we discussed with them what facilities they would like in a replacement radar system. Unfortunately, whilst we were very familiar with early computer games, at the time these were unknown to these key RAF aircrew, and we couldn't get them to understand how fast the world of digital development was advancing. Consequently, our bid was tailored to what they, the customer wanted, and clever processing was largely left out, with the flight crew carrying on much as in the Phantom. A few years later, when we were in the development stage, and I was being urged to harden the radar's physical design, everybody knew about computer games, and public perception had changed enormously. Consequently, at meetings between ourselves and MoD, fresh technical demands from the RAF were now coming through thick and fast, and we were struggling to accommodate them, without raising the component count, power consumed or physical size. This was the period when few microprocessors were approved for a military airborne environment, and most of our digital circuitry was built from relatively simple building blocks of military grade digital ICs. (Much later on, shortly before I retired, we were incorporating the latest in ASIC devices, but in the 1970s these hadn't been heard of.)

Post Bid phase

Once we had won the contract, we were keen to get on with the job as soon as possible, but cash constraints within MoD (and presumably budgetary realities) meant that our initial contracts were let in just 6 week durations, with I think an occasional 8 week one. Consequently my management (GEC-Marconi) wouldn't let me recruit any new development engineers, and during this period of hesitant commencement another major MoD contract, that for the JP233, was let to Huntings at Ampthill. They immediately started advertising for engineering staff in our local papers, and by the time we were able to advertise, about 6 months later, most engineers with itchy feet had already moved to Huntings, which wasn't that far from Borehamwood. In the meantime at Warton, BAC were well advanced on their modifications to the Tornado for the Air Defence Variant.

Development Phase Delays

I'm not sure whether we ever recovered from the hesitant start, but during the later development of the Foxhunter AI-24 radar some other serious delays crept in.

Computing Speed (or lack of)

In an FMICW radar such as AI-24 Foxhunter, the received signal has to be analysed to find what frequency of echo has been received. The frequency difference from what was transmitted will tell us how fast the target is moving. Range is worked out by linearly ramping the frequency transmitted, up and down with a constant frequency period as well. This means that return signals are composed of triplets, and in the receiver there's an ADC (Analogue to Digital Converter, which converts the signal to digits), and then a specialised hardware digital frequency analyser which works out the frequencies, followed by a digital correlator which works out which frequencies are triplets - and thus, what may be targets. So, lots of computing using early ICs and all of these somewhat behind the curve, needing to be approved to a full Military Specification. At the time of our bid, we knew that we couldn't buy ICs fast enough to do the job before echoes from the next radar pulse started to be received... We knew that the speed of electronics was increasing, so we plotted the clock-rate of suitable ICs against time, and we were confident that by the time that we got to the commissioning phase, that there would be ICs with sufficient speed. (Faster ICs using a different methodology were available but their consumption of power was beyond what the Tornado's engines could supply.)

The expected increase in clock rate didn't quite happen, and one day the Chief Commissioning Engineer walked in to my office, and told me the stark news - that we didn't have a working radar! Much worry and head scratching, but no offer from our suppliers that later batches of the early ICs would be sufficiently fast for our purposes. Fortunately one of our very bright mathematicians took apart our implementation of the Cooley-Tukey algorithm, and worked out that by reducing the number of coefficients used in the Fourier expansion the ICs would be able to complete the computing in time, and that the loss of accuracy was just acceptable. This meant a complete re-working of the multi-layered boards on which the ICs sat, much testing and discussion with RRE, and overall this caused about an 18 month delay.

TWT cathode vibration

To allow operation on a variety of channels, it was decided that the radar would have a travelling wave tube (TWT) for its power amplifier, with the various frequencies being generated by a solid state microwave frequency generator. A power klystron was considered but none was available, nor could we find a suitable TWT either in the UK or USA. (The McDonnell Douglas Phantom radar had used a klystron.) It was decided that a new TWT would be developed, and EEV at Chelmsford got the job. (On our exploratory visit to the USA, we were told in no uncertain terms that series production of such a TWT was asking for the impossible - by the head of the power klystron team at Varian. Coming from such a well known figure in the industry, this prediction haunted us during the TWT development.)

New developments of vacuum devices and similar for military purposes were handled by CVD, a branch of MoD, and the funding was separate to that for the rest of the radar. (See The National Archives for Admiralty and Ministry of Defence: Co-ordination of Valve Development Department: Reports and Registered Files (CVD and other Series))

As Equipment Manager I attended the development progress meetings at Chelmsford along with representatives from EEV; the Royal Radar Establishment (RRE); Ferranti, Edinburgh (who were developing the transmitter) and various representatives from MoD. Somewhere along the line in the two or three year development, it was decided that vibration tests on the TWT would not be performed. Ferrantis and ourselves objected, and I believe we were supported by RRE, but there was insufficient funding. In due course the TWTs were delivered to Ferrantis and incorporated in the radar transmitter successfully. Some time later, vibration tests were carried out on the transmitter, and microphony was found and traced to the lack of stiffness in the cathode support. This was disastrous for our purposes as it would decrease the purity of the transmitted signal, and it was essential that the TWT operated as a very clean "HiFi" amplifier. EEV strengthened the cathode supports which then led to another delay as the stiffer supports leaked heat away from the cathode, and a redesign of the cathode heater transformer was then necessary. This took time, and it too required to be vibration tested etc.

This strengthening of the cathode supports delayed transmitter delivery by around 18 months. Fortunately, much of this ran concurrently with the delays mentioned above.

In passing, it's worth mentioning that away from our specialist use of the TWT, EEV had commercial success with it in other fields.


RF Filters

One of the battles I lost was getting the funding for a full test of the radar in simulated near action conditions. This saved money at the time, but came back and bit us during the deployment of Tornado ADVs later on in live missions, where it was found that the radar interfered with other aircraft. Urgent measures fixed the problem, but not without a serious measure of embarrassment.


Move to Milton Keynes

As a result of needing to expand our workforce to meet our development programme, we ran out of space at Borehamwood and considered a variety of buildings in the northern Home Counties including a modern office style building at Aylesbury. The GEC surveyors however discovered that it was unsound (it was still standing last time I checked), and in early 1980 we were told that we were moving to what had been the Rank Xerox / Xerox Research site at Linford Wood, Milton Keynes.

Aerial view of the Linford Wood site when Xerox were in situ circa 1978. A422, H3, Monks Way, before dualling, is at the lower left. Linford Wood itself is off to the right of the picture.

Phase One - As can be seen the buildings were predominantly of the Terrapin type of construction, and there were also two round-shaped air-houses. The smaller was very dark inside, whilst the larger and much lighter one was the main staff restauarant with kitchen behind, and visitor's dining room to the upper right. There were three industrial steel framed buildings - that in the centre of the picture was filled with large steel cabins which we adapted for a variety of purposes, and the largest one at upper right we used for Site Services, mechanical engineering workshop with lathes, presses etc. and this was also where we housed the site main frame computer. (Remember, 1980 - PCs were still some years away.) We used the remaining steel framed building as a library/document store.

Phase Two - Whilst most of the development and planning work was at Milton Keynes, much of the work was still being done at Borehamwood with much time lost by staff travelling between the two sites, about 45 miles apart. Two new blocks, A Block, and the Pavilion were built on a site to the top of the picture above. A Block was single storey manufacturing support and around 50,000 square feet, whilst the Pavilion, also single storied, was a combined Kitchen, Sports Club, Dining Room, Bar and had additional conference room facilities. The Pavilion had a pitched roof, and still survives.

Phase Three - The portakabins remained unsatisfactory and work started on a three story brick faced building, B Block on the far side of A Block. Eventually this was completed, we moved in and the Portakabin style buildings were no longer needed.

Phase Four - The Berlin Wall was breached, and reviews of the future need for defence systems designed for the Cold War went ahead. Expected orders for enhancements to the radar were delayed, and our personnel numbers started falling. To save costs, it was decided that A Block was big enough to hold all of us, and that B Block would be sold. Before consolidating in A Block, we refurbished A Block to a higher standard than before with better lighting, carpeting and more air-conditioning.

Phase Five - Further consolidation in the industry led to the closure of A Block with transfer of staff to a variety of other sites, such as Luton and Edinburgh. A Block was demolished, and the site has now been redeveloped with Kyocera occupying tha main building.