Prior work K8758 was used to test radar systems. Its detection of Royal Navy ships in bad weather was the harbinger of the destruction of the German U-boat force. The seminal
Daventry Experiment of 1935 proved the basic concept of radar was feasible and led to the rapid formation of the
Air Ministry Experimental Station (AMES) at
Bawdsey Manor to develop them. The AMES team's primary concern was the development and deployment of the
Chain Home (CH) system, providing
early warning for raids approaching the UK. As the team grew, the work diversified, and by 1938 there were a number of teams working on other projects as well. One of the first of these side efforts came about due to
Henry Tizard's concerns about the potential effectiveness of Chain Home. He believed that the
Luftwaffe would suffer so stiffly at the hands of the RAF's
ground-controlled intercept system that they would switch to the night bombing role. At night, a fighter pilot could see a target at perhaps , an accuracy Chain Home and its associated
Dowding system could not provide. Tizard's concerns were later brought up by
Robert Watson-Watt at a round-table meeting at the Crown and Castle pub.
"Taffy" Bowen offered to take up development of a new system able to be installed in aircraft to close the distance between CH direction and visual range at night. Due to the physics of radio transmission, antennas have to be about as long as the wavelength of the radio signal to achieve reasonable
gain. The
half-wave dipole, with two poles each about one-quarter the length of the signal, is a particularly common solution. CH operated at anywhere from 10 m to 50 m depending on the version, meaning antennas would have to be at least long, which made it totally impractical for use on an aircraft. Bowen started development of a new system operating at shorter wavelengths, first at 6.7 m following work by the
British Army, and then finally settling on 1.5 m, the practical limit of available technology. This became known as
aircraft interception (AI) radar, and was the primary focus of Bowen's work from 1936 to 1940. While testing an early 1.5 m set the team failed to detect any aircraft, but easily picked out large objects like cranes and ships at nearby wharves. Further experiments demonstrated the ability to pick up ships at sea, leading to a live demonstration where the team was able to track down
Royal Navy capital ships in terrible weather. This led to immediate interest on behalf of
RAF Coastal Command which saw this as a way to find enemy ships and
U-boats, and by the
British Army, which was interested in using the radars to
direct fire against shipping in the
English Channel. Work on the system for AI use largely ended.
AI development equipped with Mk. IV radar was the world's first truly effective night fighter.
"Paddy" Green shot down seven Axis aircraft over three nights in this example. It was not until 1939, with war clearly looming, that the team once again returned to AI work. Compared to the successful and rapid development of the anti-shipping radars, the team found themselves facing a continual stream of problems in air-to-air settings. There were two primary problems, a lack of maximum range that made finding the targets difficult, and a lack of minimum range that made it difficult for the pilot to see the target before it became invisible to the radar. Like Chain Home, the AI radar sent out a powerful pulse semi-directionally, lighting up the entire sky in front of it. Echoes from aircraft would be received on multiple directional antennas, and by comparing the signal strength from each one, the direction of the target could be determined. However, this also meant that the signal reached the ground and reflected off it, producing a return so powerful that it overwhelmed the receiver no matter where the antenna was positioned. Since this signal had to travel to the ground and back, it produced a line on the display at an indicated range equal to the aircraft's altitude. Flying at , a typical altitude for German bombers, meant anything beyond about was invisible in the noise. This left little range to detect the target. A more difficult problem was the inability to detect targets at short range. The transmitter signal was difficult to cut off sharply and was still broadcasting a small signal when the returns from nearby targets started to be received. Moreover, the powerful signal tended to bleed through to the receiver, causing it to oscillate for a time, blanking out nearby targets. These effects limited the minimum range to in the best case, just at the very limit of the pilot's eyesight at night. Attempts had been made to address this problem, and Bowen and Hanbury Brown were convinced they had a workable solution. However, the
Air Ministry was so desperate to get AI into service that they had used the team as a production facility, having them hand-fit aircraft with the prototype Mk. III units that were nowhere ready for operational use. While these sets were rushed to the squadrons, further work on developing solutions to the "great minimum range controversy" ended.
Arthur Tedder would later admit this was a "fatal mistake".
Early microwave work The Airborne Group had been experimenting with microwave systems as early as 1938 after discovering that a suitable arrangement of the
RCA acorn tubes could be operated at wavelengths as low as 30 cm. However, these had very low output, and on top of this, the receiver's electronics were not very sensitive at these frequencies. This resulted in very short detection ranges, essentially useless. The group gave up on further development for the time being, and Bowen described the topic being pooh-poohed by the engineers for some time. Nevertheless, pressure from the
Admiralty kept microwaves in everyone's mind. While the 1.5 m sets were fine for detecting larger ships, they could not effectively see smaller objects, like U-boat
conning towers. This was for the same reason that antennas need to be roughly the size of the wavelength; to provide a reasonable reflection, the objects need to be at several times larger than the wavelength. The Admiralty had the advantage of administering the UK's
vacuum tube development efforts, under the Communication Valve Development Committee (CVD), and were able to continue development of suitable tubes. Bowen and his counterpart at the
Admiralty Signals Establishment (ASE), Canadian polymath
Charles Wright, met at Bawdsey in the spring or summer of 1939 and considered the issue of a microwave aircraft interception radar. Bowen agreed that the main problem with the range limits of the AI sets was the floodlight-like transmissions and that the easy way to fix this would be to narrow the beam, focussing the power into a smaller area. He concluded that a 10-degree beamwidth would do the trick. Considering that the nose of an aircraft could hold a radar antenna about across, an antenna with poles shorter than 15 cm was desirable, and if that antenna had to move within the nose for tracking, 10 cm (~3 GHz) would be ideal. This strongly agreed with Wright's requirements for a shipboard system able to detect U-boats while having an antenna small enough to be mounted on small escort vessels. With both forces desiring a 10 cm system, Tizard visited the
General Electric Company's (GEC)
Hirst Research Centre in
Wembley in November 1939 to discuss the issue. Watt followed up with a personal visit sometime later, leading to a 29 December 1939 contract for a microwave AI radar set. This was followed by the CVD placing a contract for suitable valves with the
Birmingham University. Bowen arranged a January meeting between GEC and EMI to coordinate AI work, which led to further collaboration. The Birmingham group was led by
Mark Oliphant, formerly of the
Cavendish Laboratory at
Cambridge University but recently moved to Birmingham to set up the
Nuffield Laboratory. The team decided to base their development efforts on the
klystron concept. The klystron had been introduced by the
Varian brothers at
Stanford University in 1936, but produced a relatively low power output. Oliphant's team began applying new tube making techniques and by the end of 1939 they had a tube capable of delivering 400 watts.
AIS begins Watt moved to Air Ministry headquarters in London and
Albert Percival Rowe took over management of the radar teams at Bawdsey. He had a troubled relationship with Bowen and many others at AMES. At the opening of the war, the entire AMES establishment was moved from Bawdsey to a pre-arranged location at Dundee. The choice of Dundee was largely due to the university being Watt's
alma mater. He had made little effort to prepare the university for use by AMES and the rector was surprised when they arrived one day out of the blue. Almost no room was available as the students and professors had returned from summer holidays. The AI team was sent to a small airfield at Perth that was miles away and quite small. Both locations were totally unsuitable for the work and the teams constantly complained. In February 1940 Rowe began to organise a new AI team led by
Herbert Skinner. Skinner had
Bernard Lovell and
Alan Lloyd Hodgkin begin considering the issue of antenna designs for microwave radars. On 5 March they were invited to the GEC labs to view their progress on a radar based on VT90 tubes, which had by this time been pushed to useful power levels at 50 cm wavelengths. Provided with a low power klystron as a microwave source, Lovell and Hodgkin began experimenting with
horn antennas that would offer significantly higher angular accuracy than the
Yagi antennas used on the Mk. IV. Instead of broadcasting the radar signal across the entire forward hemisphere of the aircraft and listening to echoes from everywhere in that volume, this system would allow the radar to be used like a
flashlight, pointed in the direction of observation. This would also have the side-effect of allowing the radar to avoid ground reflections simply by pointing the antenna away from the ground. With a beamwidth of 10 degrees, a horizontal antenna would still create some downward-pointed signal, about 5 degrees in this case. If the aircraft were flying at , the beam would not strike the ground until about in front of the aircraft, leaving some room for detection against even the lowest flying targets. Lovell was able to build horns with the required 10 degree accuracy, but they were over long, making them unsuitable for installation in a fighter. At the suggestion of Skinner, they experimented with a parabolic dish reflector behind a
dipole antenna on 11 June 1940. They found it offered similar accuracy, but was only deep, easily able to fit within a fighter's nose area. The next day Lovell experimented with moving the dipole back and forth in front of the reflector, and found that it caused the beam to move as much as 8 degrees for 5 cm movement, at which point Lovell regarded "the aerial problem as 75 per cent solved." Follow-up experiments with a production antenna dish from the London Aluminium Company demonstrated the ability to move the beam as much as 25 degrees before it became distorted. After several months at Dundee, Rowe finally agreed that the accommodations were unsuitable and began plans to move to a new location on the south coast near
Worth Matravers. In May 1940, shortly after the breakup of the original AI team, Skinner moved along with a number of scientists from Dundee, as well as former AI team members Lovell and Hodgkin. They settled in huts at
St Alban's Head, outside Worth Matravers.
Cavity magnetron While Oliphant's group struggled to raise the power of their klystrons, they also looked at alternate arrangements of the device. Two researchers in the team,
John Randall and
Harry Boot, had been given the task of making one such adaptation, but it quickly became clear it did not help matters. They were left with little to do and decided to consider alternative approaches to the problem. All microwave generators of the era worked along similar principles; electrons were pulled off a
cathode towards an
anode at the far end of a tube. Along the way, they passed one or more
resonators, essentially hollow copper rings with a slit cut along the inner edge. As the electrons passed the slit, they caused the interior of the ring to begin to resonate at radio frequencies, which could be tapped off as a signal. The frequency could be adjusted by controlling the speed of the electrons (via the applied
voltage) or by changing the dimensions of the resonator. The problem with this approach was producing enough energy in the resonators. As the electron passed the opening in the resonator they deposited some of their energy as radio waves, but only a small amount. To generate useful amounts of radio energy, the electrons either had to pass the resonators multiple times to deposit more energy in total, or huge electron currents had to be used. Single-chamber klystrons, like those being used at the time, had to go the latter route, and were difficult to make in a form with useful output given a reasonable input power. Randall and Boot began to consider solutions with multiple resonators, but this resulted in very long and entirely impractical tubes. One then recalled that loops of wire with a gap in them would also resonate in the same fashion, an effect first noticed in the very earliest experiments by
Heinrich Hertz. Using such loops, one could make a resonator that sat beside the electron stream, instead of being wrapped around it. If the electron beam was then modified to travel in a circle instead of a straight line, it could pass by a series of such loops repeatedly. This would cause much more energy to be deposited in the cavities, while still being relatively compact. To produce the circular motion, they used another concept known as the magnetron. The magnetron is essentially a
diode that uses a magnetic field to control the path of the electrons from cathode to anode instead of the more common solution of an electrically charged grid. This was initially invented as a way to avoid patents on grid-based tubes but proved to be impractical in that role. Follow-up studies had noted the ability of the magnetron to create small levels of microwaves under certain conditions, but only halting development had taken place along these lines. By combining the magnetron concept with resonator loops created by drilling holes in solid copper, an idea from
W. W. Hansen's work on klystrons, the two constructed a model version of what they called the resonant cavity magnetron. They placed it inside a glass enclosure evacuated with an external
vacuum pump, and placed the entire assembly between the poles of a powerful
horseshoe magnet, which caused the electrons to bend around into a circular path. Trying it out for the first time on 21 February 1940, the tube immediately began producing 400 W of 10 cm (3 GHz) microwaves. Within days they noticed that it was causing
fluorescent tubes to light up across the room. Quick calculations showed this meant the tube was creating about 500 W, already beating the klystrons. They pushed this over 1,000 W within weeks. The main Birmingham team gave up on the klystron and began work on this new cavity magnetron, and by the summer had examples producing 15 kW. In April, GEC was told of their work and asked if they could further improve the design.
First magnetron radar . On 22 May,
Philip Dee travelled to visit the magnetron lab, but was forbidden to tell anyone else in the AIS group about it. He simply wrote that he had seen the lab's klystron and magnetrons, but failed to detail that the magnetron was an entirely new design. He did provide Lovell with a much more powerful water-cooled klystron to use as a test source for the antenna work, which took place in ramshackle conditions. This was a problematic device because the filaments warming the cathode tended to burn out continually, requiring the system to be disconnected from the water supply, unsealed, repaired, and then re-assembled. Dee's 13 June description notes: Skinner also gave Dee fits with his unusual method of testing that the klystron was working properly, by using the output lead to light his cigarettes. GEC was working on producing a fully sealed version of the magnetron, as opposed to one that used an external vacuum pump. After inventing a new sealing method using gold wire and adapting the chamber of a Colt revolver as a drilling template, they produced the E1188 in early July 1940. This produced the same amount of power as the original Randall-Boot model, about 1 kW at around 10 cm. Within a few weeks they had made two improvements, moving from six to eight resonators, and replacing the cathode with an oxide-coated version. The resulting E1189 was capable of generating 10 kW of power at 9.1 cm, an order of magnitude better than any existing microwave device. The second E1189 was sent to the AMRE lab, which received it on 19 July. The first E1189 would end up travelling to the US in August as part of the
Tizard Mission. By the spring of 1940, Bowen was being increasingly sidelined in the AI field due to his continuing battles with Rowe. Watt, responding to these problems, announced a reorganization of the AI teams, with Bowen left off the roster. Bowen then joined the Tizard Mission, secretly carrying the E1189 in a lockbox until he presented it to great acclaim by the US delegates who had nothing like it. This ultimately caused some confusion, as the supposedly matching blueprints were actually for the original six-chamber version. Lovell continued his work on the production antenna design using klystrons and wrapped up this work on 22 July. The team then began adapting the various bits of equipment to work together as a single radar unit based on the magnetron. J. R. Atkinson and W. E. Burcham, both sent to the AIS team from the
Cavendish Laboratory at
University of Cambridge, produced a
pulsed power source, and Skinner and A. G. Ward, also from Cavendish, worked on a receiver. At the time the team had no solution to switching the antenna from transmit to receive, so they initially used two antennas side-by-side, one on the transmitter and one on the receiver. On 8 August they were experimenting with this setup when they received a signal from a nearby fishing hut. With the antenna still pointed in the same direction, they accidentally detected an aircraft which flew by the site at 6 pm on 12 August. The next day, Dee, Watt and Rowe were on hand, but with no convenient aircraft available the team instead demonstrated the system by detecting the returns from a tin sheet being held by Reg Batt bicycling across a nearby cliff. With this demonstration of the radar's ability to reject ground returns and detect targets at basically zero altitude, interest in the 1.5 m systems began to wane. At some point in July or August, Dee was placed in charge of developing a practical 10 cm set, which was now known under the name AIS, S for
sentimetric. Dee began complaining to everyone that would listen about the fact that both his team and GEC were developing what was essentially the same solution, AIS using a 10 cm magnetron, and GEC using
Micropup tubes which had now been improved to the point where operation at 25 cm was possible. On 22 August 1940 a team from GEC visited the AIS lab, where the AIS team demonstrated the system by detecting a
Fairey Battle light bomber at a range of in spite of it being tail-on to the radar. This was far better than the GEC set. Soon after, Rowe received orders from Watt's office telling him to place all AIS development in Dee's hands.
GL sideline At this point the AI team was moved from their location at St. Alban's to a new one at a former girls' school,
Leeson House, outside
Langton Matravers. A new lab had to be constructed on-site causing further delays, but by the late summer of 1940 the magnetron system was effectively operational at the new site. Meanwhile, the Army had been very impressed with the performance of the 25 cm experimental sets and became interested in using it as a range finder in a
Gun Laying radar. Operators would point the radar at targets indicated to them by search radars, and from then on the radar information would be fed to the analogue computers that aimed the guns. The limited power of the 25 cm set was not a serious concern in this case as the range would be relatively short and the targets relatively large. The Air Defence Experimental Establishment (ADEE) of the Army was working on this using the klystron design from Birmingham and
British Thomson-Houston (BTH) as their industrial partner. According to Dee, in September 1940 when Rowe heard of this he attempted to take over the project. After a 22 September meeting with
Philip Joubert de la Ferte, Rowe built a GL team under the direction of D. M. Robinson using several members of the AIS team, telling them that they would have to focus on the GL problem for the next month or two. This led to increasing friction between Dee and Rowe, and especially Rowe's right-hand-man, Lewis. Dee claimed that Rowe was "seizing this opportunity to try and filch the GL problem from the ADEE" and that "only Hodgkin is carrying on undisturbed with AIS, and Lovell and Ward are fortunately engaged upon basic work with aerials and receivers and are therefore relatively undisturbed by this new flap." According to Lovell this did not represent as much of a disruption as Dee believed; to some extent, the klystron work at Birmingham had been instigated by the Army for GL purposes, so it was not entirely fair to complain that they were now using it for precisely that purpose. Lovell's primary task during this period was developing a
conical scanning system that improved the accuracy of the radar beam many times, enough to allow it to be used directly to lay the guns (that is, about the same accuracy as optical instruments). This did not really require much effort, and would be useful for any centimetric radar, including AIS. Shortly thereafter, on 21 October,
Edgar Ludlow-Hewitt, Inspector General for the RAF, visited the team. After the visit, Rowe told the team that a complete GL set had to be ready for fitting to a gun in two weeks. By 6 November Robinson had assembled a prototype system, but by 25 November he sent a memo to Rowe and Lewis stating that in the last 19 days, the system had only worked for two days due to a wide variety of problems. In December he was told to take the work completed so far to BTH for development into a deployable system. On 30 December 1940, Dee commented in his diary that: Although the project was soon out of the AMRE's hands, development at BTH continued. The
Ministry of Supply changed the specification to a magnetron in January 1941, requiring further development but producing a version of much greater range and utility. It was not until 31 May that the first set was delivered for testing, at which point information on the system was turned over to Canadian and US firms for building. The Canadian versions were eventually deployed as the
GL Mk. III radar, while the US team at the
Radiation Laboratory added an automatic scanning feature to their version to produce the superb
SCR-584 radar.
Scanning As the AIS team once again returned their attention full-time to the aircraft interception task, they had by this time produced what was a complete radar system. However, the system could only be used in the manner of a flashlight, pointed in the direction of its target. This was fine for gun-laying, but in order to be useful in the interception role, the system had to be able to find the target anywhere in front of the fighter. The team began to consider different ways to scan the radar beam to produce a search function. The team first considered spinning the radar dish around a vertical axis and then angling the dish up and down a few degrees with each complete circuit. The vertical motion could be smoothed out by moving continually rather than in steps, producing a helix pattern. However, this helical-scan solution had two disadvantages; one was that the dish spent half of its time pointed backwards, limiting the amount of energy broadcast forward, and the other was that it required the microwave energy to somehow be sent to the antenna through a rotating feed. At a 25 October all-hands meeting attended by Dee, Hodgkin and members of the GEC group at GEC's labs, the decision was made to proceed with the helical-scan solution in spite of these issues. GEC solved the problem of having the signal turned off half the time by using two dishes mounted back-to-back and switching the output of the magnetron to the one facing forward at that instant. They initially suggested that the system would be available by December 1940, but as work progressed it became clear that it would take much longer. As chance would have it, in July 1940 Hodgkin had been introduced to A.W. Whitaker of
Nash & Thompson, best known for their work on powered gun turrets. They began to talk about the scanning problem, and Hodgkin described their current solution of moving the dipole at the center of the parabola up and down while moving the parabola itself right and left. Hodgkin was not convinced this was a good solution and was proven correct when Whitaker built their first version of such a system in November. They found that the two motions combined to cause enormous vibrations in the entire system. Lovell and Hodgkin considered the problem and came up with the idea of causing the parabolic reflector to rotate around the axis extending from the nose of the aircraft, tracing out circles. By smoothly increasing the angle of the reflector compared to the forward axis while the circular motion continued, the net effect was a spiral scanning pattern. Whitaker was able to rapidly build such a system, scanning out a cone-shaped area 45 degrees on either side of the nose. The spiral and helical scan systems produced very different displays from the same basic data. With the helical-scan system, the radar dish was moving horizontally, producing a series of stripes across the screen while scanning up and down so subsequent lines were above or below the last pass. This created a
raster scan display, not unlike a television. Echoes caused the signal to brighten, producing a spot, or
blip on the display. The blip's location indicated the direction to the target relative the nose of the fighter, represented by the centre point of the display. The further the blip was from the centre of the screen, the further away the target was from the centreline. The range was not directly indicated in this sort of display. In contrast, the spiral-can system was essentially a rotating version of a conventional A-scope display. In the A-scope, a
time base generator pulls the CRT beam across the screen horizontally, and blips indicate the range to the target along the line the radar is currently pointed. For spiral-scan the only difference was that the line was no longer always horizontal, but was rotating around the face of the display at the same speed as the dish. Blips on the screen now indicated two values, the angle of the target relative to the centreline, and the range to the target represented by the distance from the centre. What was lost in this display was a direct indication of magnitude of the angle from the centre; a blip to the upper right indicated the target was in that direction but did not directly indicate if it was five, ten or twenty degrees off. It was later realised that the spiral-scan did provide angle-off information, through simple geometry and timing. Since the radar beam had a finite width, about five degrees, it would see some return even when the target was not centred in the beam. A target far from the centreline would only be illuminated when the dish was pointed in that direction while rotating rapidly away from it. The result as a short arc on the display about 10 degrees long. A target closer to the centre, say five degrees to port, would be illuminated strongly when the dish was pointed to the left, but still receive a small signal even when it was pointed to the right. That meant it produced a varying return almost through the entire rotation, creating a much longer arc, or a complete circle if the target was dead ahead.
Continued development Awaiting the arrival of a scanner, in the autumn of 1940 the AMRE had ordered the delivery of an aircraft with some sort of radio-transparent nose. The Indestructo Glass company proposed using thick
Perspex, while the AMRE team preferred a composite material of
polystyrene fabric and Egyptian cotton bound with
phenol formaldehyde resin (the glue used in
Bakelite), or a similar paper-based resin composite. The Perspex solution was chosen, and in December 1940
Bristol Blenheim N3522, a night fighter adaptation of the Blenheim V, arrived at
RAF Christchurch, the nearest suitable airfield. A number of attempts had to be made to successfully mount the nose to their test aircraft. It was not until the spring of 1941 that Indestructo delivered suitable radomes and the mounting issues were wholly solved. While this work progressed, the teams continued development of the basic system. Burcham and Atkinson continued their development of the transmitter section, attempting to generate very short pulses of power to feed the magnetron. They finally settled on a solution using two tubes, a
thyratron and a
pentode, which produced 1 μs pulses at 15 kW. GEC preferred a design using a single thyratron, but this was eventually abandoned in favour of the AMRE design. Further work pushed this system to 50 kW, producing 10 kW of microwaves at a
pulse repetition frequency of 2500 cycles per second. Skinner took up the task of developing a suitable
crystal detector, which essentially consisted of endless trials of different crystals; Lovell noted that "an abiding memory of the days at Worth and Leeson is of Skinner, cigarette drooping from his mouth, totally absorbed in the endless tapping of a crystal with his finger until the whisker found the sensitive spot giving the best characteristics." This led to the use of a tungsten whisker on silicon glass, sealed into a wax-filled glass tube. Oliphant's team in Birmingham continued these experiments and developed a capsule-sealed version. The radio receiver turned out to be a more difficult problem. Early on they decided to use the same basic receiver system as the earlier Mk. IV radar. This had originally been a television receiver designed by
Pye Ltd. to pick up
BBC transmissions on 45 MHz. It was adapted to the MK. IV's ~200 MHz by using it as the
intermediate frequency stage of a
superheterodyne system. To do this, they had added another tube that stepped down the frequency from the radar's 193 MHz to 45 MHz. In theory this should be just as easily adapted to the AIS's 3 GHz, using a similar solution. The problem was that the magnetron's frequency tended to drift, in small amounts pulse-to-pulse, and much greater amounts as it heated and cooled. Any sort of fixed-frequency step-down like the one used in the Mk. IV wouldn't work. After trying a variety of designs based on klystrons and older-style magnetrons, they eventually gave up. The solution was provided by well-known tube expert
Robert W. Sutton at the Admiralty Signals Establishment. He designed a new tube for this purpose, today known as the
Sutton tube but at that time more widely known as a reflex klystron. This was essentially a conventional two-cavity klystron with one cavity removed. The remaining cavity was fed a tiny amount of the output from the magnetron, causing the electrons passing by it to take up the pattern of the radio signal (this is the basis of all klystrons). Normally this would then pass the second resonator where the output would be tapped, but in the Sutton tube, the electrons instead approached a high-voltage plate that reflected them back towards their source. By carefully controlling the voltage of the reflector, the electrons would arrive having gained or lost a controlled amount of velocity, thus inducing a different frequency signal in the cavity as they passed it the second time. The combination of the original and new frequency produced a new signal that was sent to the conventional receiver. Sutton delivered an example producing 300 mW in October 1940. One problem now remained, the need for two antennas for broadcast and reception. Lovell had attempted a solution using two dipoles in front of a common parabolic reflector, separated by a metal disk, but found that enough signal leaked through to cause the crystal detectors in the receivers to burn out. On 30 December 1940 Dee noted that no solution had been found along these lines and that in spite of best efforts the crystals still lasted only a few hours. Another solution was suggested by Epsley of GEC, who used a tuned circuit of two spark gap tubes and dummy loads to switch off the receiver's input using the magnetron's own signal as the switching signal. This worked, but ¾ of the output signal was lost into the switch. In spite of this problem, the team decided to adopt it for the Blenheim in February 1941.
Flight testing In January 1941 scanner units from both
GEC and
Nash & Thompson had arrived at Leeson for testing. The aircraft was still being fitted with the
radome, so the team took the time to test both units head to head and see if one had a clear advantage in terms of interpreting the display. On the bench, watching the operation of the spiral scanner produced various results of awe in the team. Dee later wrote: By March 1941 the first AIS unit was ready for flight testing. This was fitted to Blenheim
N3522 under an early model radome with a wooden reinforcing band. Hodgkin and Edwards took it up for its first flight on 10 March, and after minor trouble with fuses, they were able to detect their target aircraft at about at about altitude, an altitude where the Mk. IV would have a range of only 2,500 feet. Using the Battle as a target, they soon reached . Tests of the prototype continued through October with a continual parade of high-ranking civilians and military observers examining it. At first the minimum range was over against an RAF requirement of . Two members of the AIS team, Edwards and Downing, worked on this problem for over six months before reliably reducing this to around . This represented a significant advance over AI Mk. IV, which was still around 800 feet or more. By this time the Air Ministry had decided to order the system into production in August 1941 as AIS Mk. I, later being renamed AI Mk. VII. The team had originally predicted that the system would have a practical detection range on the order of , but never managed to stretch this much beyond 3 miles. Much of this was due to the inefficient system being used to blank out the receiver during the transmission pulse, which wasted most of the radio energy. This final piece of the puzzle was provided by Arthur Cooke, who suggested using the Sutton tube filled with a dilute gas as a switch, replacing the spark gap system. During transmission, the power of the magnetron would cause the gas to ionise, presenting an almost perfect radio mirror that would prevent the signal from reaching the output. When the pulse ended the gas would rapidly de-ionise, allowing signals to flow across (or around) the cavity and reach the output. Skinner took up development of the concept with Ward and Starr, initially trying helium and hydrogen, but eventually settling on a tiny amount of water vapour and argon. The resulting design, known as a
soft Sutton tube, went into production as the CV43 and the first examples arrived in the summer of 1941. This testing also demonstrated two unexpected and ultimately very useful features of the spiral scan system. The first was that since the scanning pattern crossed the ground when the antenna was pointed down, the ground returns produced a series of curved stripes along the lower portion of the display. This formed an analogue of an
artificial horizon, one that radar operators found extremely useful in combat because they could immediately see if the pilot was responding correctly to their commands. Various members of the team record having been surprised by this outcome, noting that the effect was obvious in retrospect and should have been predicted. The other surprise was that ground returns caused a false signal that always appeared at the same range as the aircraft's current altitude, no matter where the dish was pointed. This was in much the same fashion as the Mk. IV, but in this case, the signal was much smaller whenever the dish was not pointed down. Instead of a wall of noise at the range of the aircraft's altitude, the signal caused a faint ring, leaving targets on either side visible. The ring was initially very wide, caused by returns not only directly under the aircraft but further away as well. After several months of work, Hodgkin and Edwards managed to provide a tuning control that muted down the weaker signals, leaving a sharp ring indicating the aircraft altitude. This too was a useful indicator for the operators, as they could see they were at the same altitude as their target when the ring overlapped the target blip. Finally, the team noticed that the system would often create false echoes during heavy rainstorms, and the potential for using this as a weather system was immediately seen. However, they were sure that shorter wavelengths like those in the
X band being experimented with would have a greater interaction, and this was not considered further at the time.
Further development Over the summer, the original experimental set was used in a series of experiments against submarines. The first took place on 30 April 1941 against
HMS Sea Lion, and a second on 10–12 August against
ORP Sokół. These clearly demonstrated that the AIS could indeed detect the submarines with only the conning tower exposed, just as the Admiralty had hoped. This led to orders for
air-to-surface-vessel (ASV) radars based on the AIS electronics. A second Blenheim,
V6000, became available for additional testing. The team began to use this aircraft as a testbed for alternate scanning solutions, leaving the original
N3522 with the spiral-scan system. One of the first tests was to use a manual scanning system in place of the spiral or helical systems, allowing the operator to scan the sky using controls on his receiver sets. Once a target was found, they could flip a switch and the system would track that target automatically from that point. After considerable effort, they decided this concept simply didn't work, and that the mechanical scanning systems were a better solution. The team then began to compare the performance and ease-of-use of the helical vs. spiral scanners, with the GEC helical system being mounted in
V6000. After extensive tests by George Edwards and O'Kane of GEC they had made no firm conclusions which system was better. Further work on these systems ended as the pressure to install the Mk. VII units, now improving in quantity, became pressing. This also seems to be the reason that US versions, known as the SCR-520, were largely ignored after having been developed with extreme speed over the winter. Bowen, who had returned from the US by this point, notes the confusion during the rush to install.
Mk. VII With the return of better weather during the spring of 1941, the
Luftwaffe began to ramp up their night bombing campaign, the
Blitz. By this time a number of changes in the night fighter groups were poised to greatly improve the performance of the defence. Along with increasing numbers of Beaufighters with Mk. IV, the first
ground controlled intercept radars were becoming available, which greatly improved the efficiency of arranging an interception. Losses to the night fighter forces continued to mount throughout the spring, roughly doubling every month until the
Luftwaffe called off The Blitz at the end of May. During this period the Germans noticed that aircraft dropping mines into ports and rivers almost always returned successfully. These aircraft flew at low altitudes throughout their missions, generally under . They soon began to take advantage of this, selecting targets near the coast and flying the entire mission at low altitudes. The reason for their success was due primarily to the fact that the CH radar's lowest detection angle was about 1.5 degrees above the horizon, which meant aircraft could approach quite closely before being detected, leaving little or no time to arrange an interception. Watt was able to rapidly respond to this threat by taking over deliveries of a British Army radar originally developed to detect ships in the English Channel, mounting them on tall masts to provide a long horizon, and renaming them
Chain Home Low (CHL). CHL was effective down to about . While CHL provided detection of a raid, the Mk. IV equipped night fighters were powerless to stop them. Under altitude the chance of seeing the target was basically zero. The AIS sets were perfectly suited to closing this gap, which led to a rush program to get them into service as rapidly as possible. A contract for 100 hand-built prototypes was ordered from GEC in May 1941 and given the name AI Mk. VII. At the end of July,
Sholto Douglas ordered four sets to be fitted with all speed to provide operational test units. By this point Dee had begun efforts to mount the system to its intended platform, the
Bristol Beaufighter. Hodgkin was put in charge of getting Bristol to provide an example with the radome fit, but he found that the engineer in charge of the workshop was reluctant to do so. High-level pressure from Dee and others followed, and
X7579 was quickly adapted, arriving at Christchurch in September 1941. At the time the Mk. VII consisted of a large number of fairly large equipment boxes that were entirely unsuitable for production use, and Hodgkin expressed his surprise at how well the work progressed in spite of this. The aircraft was ready for testing on 2 October.
American competition D,
DZ203, was used extensively during the war to test US radar systems in the UK. Bowen remained in the US after the Tizard mission, and had been instrumental in the creation of the MIT Radiation Laboratory, whose progress by November 1940 he described as "remarkable". Bowen began work with the RadLab on what became known as Project 1, the development of a magnetron-based AI radar similar to the prototype AIS. Their first system, generally similar to the GEC helical-scan unit, was ready for testing in February 1941, and fitted to the nose of a
Douglas B-18 Bolo bomber. It took flight for the first time on 10 March, the same day that the first AIS set flew in the UK. During this flight Bowen estimated the maximum range to be 10 miles, and on their return flight they flew past the naval yards at
New London, Connecticut and detected a surfaced submarine at about . Having heard of this performance,
Hugh Dowding, who was visiting the US at the time, pressed to see it for himself. On 29 April, after detecting a target aircraft at about Dowding once again asked Bowen about the minimum range, which they demonstrated to be about . Dowding was impressed, and before leaving to return to the UK, met with his counterpart,
James E. Chaney, telling him about the system's performance and pressing for its immediate development for purchase by the RAF.
Western Electric was given the contract to deliver five more units with all haste, under the name AI-10. One of these would be kept by Western Electric, another by Bell Telephone, one would replace the original lash-up in the B-18, another sent to the
National Research Council (NRC) in Canada and the final one sent to the UK. Originally the UK copy was to be installed in either a
Douglas A-20 Havoc or the RAF model known as the Boston, but neither of these aircraft were available. Instead, the Canadian NRC supplied a
Boeing 247 airliner, and after a test fit, it was disassembled and shipped to the UK. It arrived at RAF Ford and was re-assembled as
DZ203 on 14 August and widely tested, largely to everyone's satisfaction. AI-10 was similar in performance to the AIS systems of the same vintage, but Bowen found no strong desire on the part of the RAF to buy the device. This has been attributed to a number of factors including overwork by the AMRE team fitting their own equipment, as well as
not invented here syndrome. However, two technical issues appear to be the main reason. One was that the system did not display range directly, and had to be switched to a separate display mode that was described as basically useless. Moreover, the set was far too large to easily fit into a Beaufighter, having been designed for the much larger Havoc (P-70) or even larger
Northrop P-61 Black Widow. The US continued work on the AI-10, and put it into production as the SCR-520. The SCR-520-B, used in the P-70, weighed spread over six units, the largest of which was about a on a side. Efforts to develop a smaller version led to the slightly smaller SCR-720-A, and then to the definitive SCR-720, otherwise similar in performance to the 520 but much smaller and reduced to only .
Mk. VII into service As Mk. VIIs arrived through October and November 1941, aircraft were fitted at Christchurch and then sent to the
Fighter Interception Unit (FIU). The FIU was taking over the duties of a number of scattered experimental units and centralizing all test flight activities for Fighter Command. This process eventually reached SD flight and they moved to
RAF Ford on 10 November, at which point Christchurch returned to being a satellite field for
RAF Hurn. The newly organised FIU flew
X7579 with the prototype AIS for the first time on 30 November, with tests continuing until 14 December. During one test flight on 12 December, the operators came across a
Junkers Ju 88 bomber on a mine-laying patrol over the Thames Estuary. The crew decided to press an attack, damaging the Ju 88 and causing oil from their target's engines to spray across their windscreen. They landed without problem, and celebrated the first success of AIS. The total for these prototype sets stood at seven destroyed and many damaged by 15 May. Mk. VII's arrived in limited numbers over time. Even in experimental service, the sets proved to be excellent systems. A report compiled by the FIU noted that they gave considerably less trouble that earlier versions of Mk. IV at the same stage of development. They pressed for two squadrons to be completed as soon as possible. FIU had its first success with a production Mk. VII on the night of 5/6 June 1942, when a Beaufighter caught a
Dornier Do 217 over the Thames Estuary and shot it down. Generally, however, the introduction of the Mk. VII coincided with a decrease in
Luftwaffe activity, but the systems continued to score the odd victories against low-flying aircraft. Eventually, Mk. VII's operating over the UK and in the Mediterranean would claim 100 victories, one for every set manufactured.
Mk. VIII By the time the experimental Mk. VII units were beginning to arrive, the definitive Mk. VIII production version was being explored. One of the most pressing problems was the need to greatly reduce the size and complexity of the radar packaging, which almost completely filled the Beaufighter's rear section. Another issue was the desire to start using the new Sutton tubes for switching, which was expected to greatly increase the range of the system. Also desired was some way to use
IFF and
radio beacons with the AIS systems, as previous
transponders had been deliberately designed to listen and respond on the original AI Mk. IV frequencies around 193 MHz. The transponder problem had been growing before the introduction of AIS. IFF worked on the basis of a small receiver/transmitter set that listened for pulses from a radar and produced a low-power pulse broadcast on the same frequency but slightly delayed. The signal returned to the radar-equipped aircraft along with the original radar signal. When the two were amplified and displayed, the IFF signal caused the blip seen on the radar screen to stretch out. The original 1.5 m radar system had by this time been adapted to a wide range of roles including AI, ASV and acting as the basis for both the CHL and the new
AMES Type 7 GCI radars. To avoid interference problems, each of these operated on slightly different frequencies, from about 180 to 210 MHz. The Navy and Army added their own variations. The
IFF Mk. II, originally designed to respond to the Mk. IV, had to be repeatedly modified to respond to new radar frequencies, and none of the many models was able to respond to all of these. The solution was to choose a single frequency for all of the IFF transponders to operate on, no matter what the radar system's natural frequency might be. The selected frequency was 180 MHz, a little under the lowest of the existing 1.5 m radars. The transponder radio was tuned only to this frequency, not the radar itself. The radar system also added a separate radio system for transmitting and receiving these pulses, the
interrogator. When the radar operator pressed a button on their console, the interrogator began sending out pulses synchronised with those of the radar unit. The IFF unit in the target aircraft then responded with pulses with the same timing. The output of the interrogator's receiver was mixed with the radar's, causing the blip to extend as before. When this was added to the spiral scan display, instead of stretching the blip, the IFF signal appeared as a series of short line segments extending outward from the middle of the display, the
sunrise pattern. For unknown reasons, the team did not decide to use the same system for radio beacon use, as they had under Mk. IV. Instead, at meetings on the 13th and 14 July 1941, Hodgkin and Clegg decided to use the radar's own frequency for this role. This would require new transponders on the ground to support the AIS-equipped night fighters. The radar was adapted too, adding a switch that changed the pulse repetition frequency from 2,500 to 930 Hz, stretching the maximum range to . To offset the fact that fewer pulses were being sent, the pulse width was lengthened and two pulses were sent back-to-back, so the total radiated power did not change. Additionally, during this period the magnetron team at Birmingham had made a breakthrough. One of the problems with the magnetron was that every pulse caused slightly different oscillations within each cavity, sometimes interfering with each other. With some patterns, particularly the
pi mode, the signals added up and the tube was much more efficient.
James Sayers had discovered that if a strap of metal was run between alternating lobes of the magnetron's cavities, the pi mode was strongly favoured. This allowed power levels to be greatly increased, and GEC began producing the new CV64, designed to operate at as much as 50 kW. These were known as
strapped magnetrons. Finally, by this time the UK electronics establishment had developed means to produce low-power pulses of extremely short duration, which were used to produce electronic scales on the same displays. As these scale lines were drawn using the same signals as the main radar pulses, they were always perfectly in synchronicity with the radar, offering accurate distance measurements without the need to calibrate an external mechanical scale. The system adopted for Mk. VIII drew circles every to a maximum of . A new display mode was introduced for late stages of the interception, increasing the PRF and expanding the display to , with the scale generating circles at intervals.
Production plan With the success of AIS and Mk. VII, plans emerged to re-equip the entire night fighter force with Mk. VIII. A three-stage plan was put in place. In the first stage GEC would build 500 sets to the interim Mk. VIIIA standard, for delivery at the end of 1942. These would be able to be used with centimetric beacons designed for them, but did not include an IFF system. An order for 1,500 sets from a new production line was sent to
EKCO, working in any changes as needed to address problems found during the Mk. VIIIA production and use, as well as IFF support. Finally, the last version would be the Mk. VIIIB, which included a wider variety of beacon modes and IFF, which would work into the production line as soon as these were ready. Unfortunately, as Hodgkin noted: The first hand-built Mk. VIIIA arrived at Christchurch in March 1942, but does not appear to have been passed to the FIU. At this point the entire centimetric radar development became embroiled in new concerns about the increasing effectiveness of the
Luftwaffe signals intelligence and night fighter defences. In June 1942 the first evidence that the Germans were jamming the 1.5 m radars was seen, and this led to calls for the AIS team to assist bringing the Mk. VIIIA into service as soon as possible, thereby once again delaying development of improved versions.
Another move In February 1942 the German battleships
Scharnhorst and
Gneisenau escaped from
Brest, France in the
Channel Dash, undetected until they were well into the English Channel. German ground forces had gradually increased the jamming of British radar over a period of weeks, and British operators had not realised this was happening. In the aftermath,
Lord Mountbatten and
Winston Churchill approved plans for a raid on the German radar station at
Bruneval, near
Le Havre. The
Biting raid captured a German
Wurzburg radar system and a radar operator. During the weeks that followed, the British authorities became concerned that the Germans would retaliate in kind. When intelligence reported the arrival of a German
paratroop battalion across the Channel, Rowe was given orders to move the unit with all haste. The task of finding a suitable site eventually fell to Spencer Freeman of the Emergency Service Organisation. Freeman began scouring lists of schools and partially completed hospitals by the Ministry of Works and Buildings, but none seemed suitable. While waiting out an air raid in Bristol, Freeman recalled someone having mentioned
Malvern College. This had originally been set aside for the use of the Admiralty in case they were forced to leave London, but by this time the threat of invasion no longer seemed immediate and the site was no longer needed for their use. When the team visited the school in April they found it empty, to their delight. However, this was only because the students were on Easter holidays and soon returned. H. Gaunt, the headmaster, was concerned about the mysterious arrival of numerous government inspectors on 25 April, who left without telling them anything. When he contacted the
Ministry of Works and Planning he was informed that a government department would be moving into the school, forcing him to move the students for the second time in two years. ADRDE, the Army group developing gun laying and truck-mounted early warning radars, moved to the site in May, and was renamed the
Radar Research and Development Establishment (RRDE) in the process. They were soon joined by elements of the AMRE, who had also been renamed to become the
Telecommunications Research Establishment (TRE). After arriving, the teams developed a plan to install the first six AI sets at nearby
RAF Defford under the supervision of RAF fitters, at which point the aircraft would be flown to two operational fitting stations to serve as pattern aircraft for new sets as they arrived. This system ultimately proved very successful, with 80 aircraft a month being delivered at the peak.
Window during a raid on
Duisburg. At the same time, a fight between
Fighter Command and
Bomber Command was brewing. Bomber Command was ramping up its campaign, but was suffering mounting losses at the hands of
Josef Kammhuber's increasingly effective defences. They began pressing for permission to use
chaff, known in the UK under the code-name
window, which in testing had demonstrated its ability to blind radar systems. Air Chief
Charles Frederick Algenon Porter ordered Bomber Command to begin using window on 4 April 1942, but he rescinded that command on 5 May under pressure from Sholto Douglas. Douglas pointed out that the Germans would be able to copy window the first time they saw it, and it was unwise to use it until its effect on the UK's own radars was better understood. Under the direction of
Frederick Lindemann, an extensive series of studies were carried out by
Derek Jackson at
RAF Coltishall. Starting in September, aircraft with Mk. IV and Mk. VII were tested against window in a series of 30 flights. Much to everyone's consternation, Jackson concluded that the Mk. VII's spiral-scan display proved to be affected by window more than the simpler display of the Mk. IV. When he learned the results, Douglas wrote a memo to the Air Ministry asking that window be held back until new radars could be developed that were not as susceptible to its effects. One of the interesting coincidences of the war was that the Germans had already developed their own version of chaff under the code-name
Düppel, and had tested it near Berlin and over the Baltic. However,
Hermann Göring was worried that if they used
Düppel over the UK, the RAF would quickly copy the concept and use it against them. As Bomber Command's fleet was rapidly growing, the results would likely be greatly in the RAF's favour. Learning from past mistakes when older material had leaked, Göring had most of the paperwork on
Düppel destroyed. ==Operational service==