Army Cell The first mention of radar in the UK was a 1930 suggestion made by
W. A. S. Butement and P. E. Pollard of the Army
War Office's
Signals Experimental Establishment (SEE). They proposed building a radar system for detecting ships to be used with shore batteries, and went so far as to build a low-power
breadboard prototype using pulses at 50 cm
wavelength (600 MHz). The War Office was uninterested and did not provide funding for further development. The matter was mentioned in the January 1931 issue of the
Inventions Book of the
Royal Engineers. With the
Air Ministry's successful demonstration of radar and rapid progress on the system that would become
Chain Home (CH) in 1936, the Army suddenly became interested in the topic and visited the CH radar team at their new headquarters at
Bawdsey Manor. Here they were introduced to smaller versions of the CH system intended for semi-mobile deployments. This appeared to have possible uses in Army roles, leading to the 16 October 1936 formation of the Military Applications Section, referred to universally as the Army Cell. This group was given room at Bawdsey, and included Butement and Pollard from the SEE. The Cell was initially given the task of improving anti-aircraft fire, and was told that the main problem to address was the accurate measurement of range. Optical instruments were used to detect aircraft and accurately determine their
bearing and
elevation, but
rangefinding through optical means remained difficult, slow and open to simple errors in procedure. A radar system that could provide accurate and rapid rangefinding would greatly improve the chances of successfully engaging an aircraft. They were given the goal of producing a range measure accurate to within at a range of . In keeping with Army nomenclature, the concept was given the name "gun laying radar" due to its role in aiding
gun laying (GL) of the
anti-aircraft artillery. That same year, an Airborne Group had been spun off from the main CH development team to develop a much smaller radar system suitable for mounting in large aircraft. This would become the
aircraft interception radar role (AI), the intention being to detect bombers at night and allow the
heavy fighters to find and attack them with their own radar. When these sets demonstrated the ability to easily pick up ships in the
English Channel, the Army Cell started a second group to adopt these systems to the
Coast Defence role (CD), providing both range and angle measurements with enough accuracy to blind-fire their
shore batteries. This team was led by Butement, leaving Pollard as the primary developer of the gun laying systems.
Mk. I development The GL effort was started very early during Chain Home development, and was deliberately based on as much of its technology as possible. Unlike Butement's earlier experimental systems, CH was based on existing electronics from commercial
shortwave radio systems, as these were among the highest power radio systems. The downside of this approach was that radio
antennas generally have to be a significant fraction of the radio signal's wavelength to work with reasonable
gain, the
half-wave dipole being a common type. For the 50 metre wavelengths initially used by CH, antennas on the order of would be needed. This was not practical for any sort of mobile system, but as newer electronics arrived through the late 1930s, the wavelengths being used by the radar systems continued to drop. By the time GL was ready to begin testing, the system was able to operate at wavelengths between reducing the antenna size to a more manageable several-metre length. Similar changes in electronics also produced smaller versions of CH, the
Mobile Radio Units or MRU's, which were used both as a mobile early-warning system and a backup service in case a main CH station was knocked out.
CH-type radar displays use a
time base generator to produce a
sawtooth wave voltage that is fed to one of the inputs of a
cathode ray tube (CRT). The time base is calibrated to move the CRT dot across the screen in the same time that echoes would be returned from objects at the radar's maximum range. The dot moves so rapidly that it looks like a solid line. The return signal is amplified and then sent into the CRT's other channel, typically the Y-axis, causing the spot to deflect away from the straight line being created by the time base. For small objects, like aircraft, the deflection causes a small
blip to appear on the display. The range to the target can be measured by comparing the location of the blip to a calibrated scale on the display. The accuracy of such a display is relative to the size of the tube and the range of the radar. If one might be expected to measure the blip to an accuracy of 1 mm on the scale along a typical CRT, and that radar has a maximum range of 14,000 yards, then that 1 mm represents, just over . This was far less accuracy than desired, which was about . To provide a system able to make such an accurate measurement, and do so continually, Pollard developed a system that used the entire CRT display to provide a measurement showing only ranges a short distance on either side of a pre-selected range setting. The system worked by charging a
capacitor at a known rate until it reached a threshold that triggered the time base. The time base was set to move across the screen in a time that represented less than a kilometre. A large
potentiometer was used to control the charging rate, which provided a range offset. The range to the target was measured by using the potentiometer to move the blip until it was in the middle of the display, and then reading the range from a scale on the potentiometer. The basic system developed rapidly, and a test system was providing accuracy for aircraft between and by the summer of 1937. By the end of the year, this had improved to as accurate as . As the original requirement for the system was to provide additional information to optical instruments, accurate bearing measurements were not required. The system did need some way to ensure that the target being ranged was the one being tracked optically, and not another nearby target. For this role, the system used two receiver antennas mounted about one wavelength apart vertically, so that when they were pointed directly at the target the received signals would cancel out and produce a
null on the display, where the signal disappeared. This was sent to a second display, whose operator attempted to keep the antennas pointed at the target. The transmitter, which produced short pulses of about 20 kW, was mounted in a large rectangular wooden cabin on a wheeled trailer. The single half-wave
dipole antenna was mounted on a short vertical extension at one end of the cabin, with the "line-of-shoot" along the long axis. The antenna was only marginally directional, sending the signal out in a fan shape about 60 degrees on either side. The receiver was considerably more complex. The operator's cabin was somewhat smaller than the transmitter, and mounted on the AA
gun carriage bearing system which allowed the entire cabin to be rotated around the vertical axis. A short distance above the roof was a rectangular metal framework roughly matching the outline of the cabin. Three antennas were mounted in a line down one of the long sides of the framework; range measurements were taken off the antenna in the middle, and directionally by comparing the signal on the two antennas at the ends. Behind the two bearing antennas were reflectors mounted about a wavelength away, which had the effect of narrowing their reception angle. In the field, the transmitter would be aimed in the expected direction of attacks, and the receiver placed some distance away to help protect it from the signal being reflected off local sources.
Initial deployment By 1939 the team was happy enough with the state of the equipment that production contracts were sent out.
Metropolitan-Vickers won the contract for the transmitter, and
A. C. Cossor the receiver. Mass-producing the GL set did not prove particularly difficult, and by the end of 1939, 59 complete systems had been delivered. Another 344 would be completed during 1940. The system did exactly what had been asked of it; it provided range measurements with accuracy on the order of 50 yards. In the field it became clear that this was not enough. By late 1939 the spectre of
night bombing was a major concern, and as the GL system could not provide accurate bearing information, and no elevation, it was unable to direct the guns at night. Instead the
World War I style of operation was used, with
searchlights hunting for targets largely at random, and conventional optical instruments being used to determine bearing and elevation once a target was lit up. In practice this proved just as ineffective as it had during World War I. Despite spending considerable time, effort and money on the GL system, when
The Blitz opened the entire Army air defence system proved to be ineffective. General
Frederick Pile, commander of the Army's
Anti-Aircraft Command, put it this way: For detecting the targets, GL was largely ineffective. From a mechanical standpoint, the need to swing the entire system around for tracking presented a major problem. A more serious limitation was the displays, which showed only a small portion of the sky in the range display, and a single on-target/off-target indication in bearing. Although it might be possible to swing the antenna in bearing to find a target, the direction was accurate to only 20 degrees, enough to keep the antennas aligned with the target, but of little use directing optical instruments onto a target, especially at night. Additionally, the bearing display only showed whether the antennas were aligned or not, but not to which side the target lay if they were mis-aligned, so more work was required to determine which direction to turn the antenna for tracking. In addition to these problems, the wide fan-shaped signal presented serious problems when more than one aircraft entered the beam. In this case, multiple blips would appear on both the range and bearing displays and it was impossible to tell which was from which target. Even the most experienced crews were unable to satisfactorily track a target in these conditions.
Radar at Dunkirk GL Mk. I sets were deployed with the
British Expeditionary Force, along with the
MRU systems which provided
early warning. Following the collapse of the defences and the eventual
Dunkirk evacuation, these sets had to be abandoned in France. There were enough parts left behind for
Wolfgang Martini's radar team to piece together the design and determine the basic operational capabilities of the systems. What they found did not impress them.
Luftwaffe radars for both early warning (
Freya) and gun-laying (
Würzburg) were significantly more advanced than their British counterparts at that time, operating on much shorter wavelengths around 50 cm. This evaluation, combined with the failure of a mission of the German airship
LZ 130 Graf Zeppelin to detect British radars in August 1939, appears to have led to a general underestimation of the usefulness of the British radar systems. Despite being aware of Chain Home, German reports on the state of the
Royal Air Force written just before the
Battle of Britain did not even mention radar at all. Other reports mention it, but do not consider it to be very important. Other sections of the
Luftwaffe appear to have been dismissive of the system as a whole.
Mk. II development The GL team had already started plans for a greatly improved version of the system that could also provide accurate bearing and elevation information. They had always wanted the GL system to be able to direct the guns in all measurements, but the pressing need to get the system into the field as soon as possible precluded this. To add this capability, they adapted a concept from the Coast Defence radars being developed by Butement. The idea was to use two antennas aimed in slightly different directions, with their sensitive areas, or
lobes, slightly overlapping down the centreline of the two. The result is a reception pattern where each of the antennas produces a maximum signal when the target is slightly to one side of the centreline, while a target located exactly in the middle produces a slightly smaller but equal signal on both antennas. A switch is used to alternate the signals between the two antennas, sending them to the same receiver, amplifier and CRT. One of the signals is also sent through a delay, so its blip is drawn slightly offset. The result is a display similar to CH, showing the range to targets within view, but with each of the targets producing two closely spaced blips. By comparing the length of the blips, the operator can tell which antenna is more directly pointed at the target. By rotating the antennas towards the stronger signal, the longer blip, the target will be centred and the two blips will become equal length. Even with the relatively long wavelengths used, accuracies on the order of ½ a degree could be attained with these
lobe switching systems.
Mk. I* As Mk. I arrived in the field, several improvements in the basic electronics were introduced. These were collected together to form the Mk. I* version. The differences between the Mk. I and Mk. I* were primarily in details. In Mk. I, the displays in the receiver van were triggered by the reception of the transmitted signal on a small antenna. It was found that in certain orientations of the transmitter and receiver, the antenna would receive too little signal to work. This was replaced by a cable between the two cabins, which was known as
cable locking. Certain details of the RF stages on the receiver improved
signal-to-noise ratio, a
voltage regulator was added to correct for differences in generators, and a new system was introduced that replaced the complex grounding system for the potentiometer with an electronic version. A more major change was the introduction of anti-
jamming features.
Bedford Attachment By late 1939, it became clear that the Mk. I in its current form would not be entirely useful in the field, especially at night, and that it would be until at least early 1941 before the Mk. II was available. Leslie Bedford had formed a radar development department at Cossor to produce CH receivers and was well acquainted with both the desires of the AA gunners as well as the possibilities inherent to the radar systems. He suggested that it would be relatively easy to adapt the antenna and display systems from the Mk. II to the Mk. I system, providing many of the same advantages. The result was the GL/EF, for Gun Laying/Elevation Finder, although it was referred to almost universally as the "Bedford Attachment". This modification added a set of vertical antennas and a new elevation-measuring CRT to read them, along with a
radiogoniometer that allowed the vertical angle to be accurately measured. Mk. I*'s with GL/EF began to deploy in early 1941, just as The Blitz was reaching a crescendo. With the Bedford Attachment, the system provided all of the information needed to aim the guns based on the radar alone. As all three axes could be read continually, the predictors could be fed information directly from the radar with no optical inputs needed. Likewise, the guns themselves were either automatically driven from the predictor, or 'laid', or alternately required the layers to follow mechanical pointers to match the predictor output, a concept known as 'laying needle on needle'. Even the fuse settings were automatically set from the range values coming from the radar. The entire gunnery problem was now highly automated.
Calibration problems . The ramp and platform at the centre are prominent. It was at this point that serious problems with calibration appeared. After considerable study, using reflectors hung from balloons and testing against the occasional aircraft, it became clear that the main problem was the levelling of the ground around the station. The long wavelengths used in these early radars strongly reflected off the ground and back into the sky. These reflected signals sometimes reached the targets and were returned to the receiver, along with the signal directly from the transmitter. Interference between the two caused nulls to appear in the reception pattern, which today is known as
multipath propagation. This made it difficult to find the target as the antennas rotated to follow a target, which caused it to appear and disappear on the displays. At first, it was believed that this would not be a serious problem and that it could be addressed by developing a calibration table for each site. For instance, along one direction the ground might be slightly lower and create a particular pattern of nulls in the air when pointed in that direction, while in another direction the ground might be higher and make a different pattern. In theory, one could measure this pattern and produce a table for any azimuth angle. Unfortunately, even the very first tests demonstrated that the calibration changed with wavelength, which happened when one radar underwent maintenance or was replaced by a backup. This meant that they would either have to make multiple calibration tables, one for each radar, or that if a single table of corrections for different bearings was desired, the antennas would have to be moved vertically as the wavelength was changed to adjust for this. Neither was practical. Once again, it was Bedford who suggested a solution; instead of calibrating the radar, he suggested calibrating the ground itself, flattening the area around the station through the use of a metal wire mat. This way, the pattern would remain the same as the radar rotated or changed frequencies. Designing such a system fell to
Nevill Mott, a physicist who had recently joined the Army Cell. The proper dimensions were ultimately found to be a diameter octagon of square wire mesh. This was supported in the air by hundreds of tensioned wires running over wooden stakes about in the air. To get the proper clearance between the antenna and the wire ground mat, the radar system had to be raised into the air on blocks, and was accessed via a wooden catwalk above the mat. The effort to equip UK-based GL sets with these ground mats was enormous. Each mat consumed 230 rolls of wire mesh, each one wide by long. In total they covered an area of about and used up of wire – along with another of heavier wire used in the support structure below the mesh. They initially planned to install the mats at 101 sites immediately, but by December 1940 they had consumed over of galvanized wire, using up the entire nation's supply of the material and causing a countrywide shortage of
chicken wire. Construction of the mat took about 50 men four weeks to complete. By the end of January 1941 only 10 sites had been upgraded, and all the while new AA emplacements were being set up so that the number of prospective sites was increasing more rapidly than they could be completed. By April, Pile had concluded that 95% of the AA sites would need the mats, and they expected 600 sites to be operational by March 1942. The program ultimately ran on for years, petering out as new systems were introduced that did not require the mats. The mat program formally ended in March 1943. Another problem, never wholly solved, was that any
balloon barrage in the area would form a powerful reflector rendering anything behind it invisible. This was particularly annoying as the balloons were often placed near the AA guns as the two systems were used together to protect high-value targets. A solution was considered in the form of a system that would allow low-lying reflections to be eliminated, but this was not fully developed.
Dramatic results In addition to the continued technological advancement of the GL systems, Pile greatly improved the overall state of AA starting in September 1940 by appointing a scientific advisor to the highest echelon of the AA command. For this role he chose
Patrick Blackett, who had World War I experience in the
Royal Navy and had since demonstrated considerable mathematical ability. Blackett planned to study the AA problem from a purely mathematical standpoint, a concept that proved extremely valuable in other areas of air defence, and would ultimately develop into the general field of
operational research. Blackett formed a study group known as the Anti-Aircraft Command Research Group, but universally referred to as "Blackett's Circus". Blackett deliberately chose members from different backgrounds, including physiologists
David Keynes Hill,
Andrew Huxley and
Leonard Ernest Bayliss, mathematical physicists
Arthur Porter and
Frank Nabarro, astrophysicist
Hugh Ernest Butler, surveyor G. Raybould, physicist I. Evans and mathematicians A. J. Skinner and M. Keast, the only woman on the team. Their goals were neatly summed up by Blackett: Meanwhile, in November 1940,
John Ashworth Ratcliffe was moved from the Air Ministry side of Bawdsey to start an AA gunnery school at
Petersham on the west side of London. One problem that became immediately evident was that the inputs to the predictors, the
analog computers that handled
ballistics calculations, were very easy to get wrong. This information was fed back through the Army hierarchy, and again it was Leslie Bedford who produced the solution. This resulted in the building of several Trainers that were used at the AA school, allowing operators to hone their skills. The Circus soon added a fourth trailer to some AA sites in the
London area, dedicated solely to recording the inputs to the predictors, the numbers of rounds fired, and the results. These numbers were fed back through the AA command structure to look for any chance of improvement. The official history, published just after the war, noted that between September and October 1940, 260,000 AA rounds had been fired with the result of 14 aircraft destroyed, a rate of 18,500 rounds-per-kill. This was already a great improvement over pre-radar statistics which were 41,000 rounds per kill. But with the addition of GL/EF, GL mats and better doctrine, this fell to 4,100 rounds per kill by 1941. Pile commented on the improvements by noting:
Mk. II arrives Production of the Mk. II was by the
Gramophone Company and Cossor. Prototype Mk. II sets began to appear as early as June 1940, but considerable changes were worked into the design as more information from the Mk. I sets flowed in. The final design began to arrive in production quantities in early 1941. Displays were located in a wooden cabin below the receiver array, including separate CRTs for range, bearing and elevation, allowing continual tracking throughout the engagement. The transmitter antenna now came in two versions, one with a wide angle beam for initially picking up the target or searching for it, and another with a much narrower beam that was used while tracking a single target. Although this introduced complexity, it also greatly reduced the problem of more than one target appearing on the displays. The Mk. II also included a new transmitter, which had increased in power by a factor of three, from 50 to 150 kW. This extra power offered somewhat better range, but more importantly it allowed the
pulse width to be significantly reduced while offering the same range. The sharpness of the echo is a function of the pulse width, so by reducing it the system became more accurate. The Mk. II could offer bearing measurements as accurate as ½ degree, about twice as accurate as the Mk. I*, and just within the range needed to directly aim the guns. The Mk. II had largely replaced the Mk. I* by mid-1942 and remained in service until 1943. An analysis demonstrated that the Mk. II improved the rounds-per-kill to 2,750, another significant advance. 1,679 GL Mark II sets were produced between June 1940 and August 1943.
Mk. III development The introduction of the
cavity magnetron in 1940 allowed radars to operate effectively at much shorter
microwave wavelengths than was possible with earlier vacuum tube designs. The early magnetrons operated at a wavelength of around , which reduced the dipole antennas to only a few centimetres long. The antennas were so short that they could be placed in front of
parabolic reflectors, which focused the signal into a very tight beam. Instead of the broadcast pattern being as much as 150 degrees wide, typical microwave designs might have a beam width of perhaps 5 degrees. Using a technique known as
conical scanning, a rotating version of lobe switching, this could be further reduced to well under ½ a degree, more than enough to directly lay the guns. In late 1940 the Army was well into an effort to build a microwave-frequency GL radar system, and by 1942 had already sent the plans to companies in the UK for production. Work also began in Canada in 1940 on an entirely Canadian-designed and built version with production starting in September 1942, and deliveries arriving in the UK starting in November 1942, as the
GL Mk. IIIC, with British units arriving the next month as the Mk. IIIB. These were dramatically more mobile than the earlier Mk. I and Mk. II designs, consisting of two-wheeled trailers and a generator set. Because the antennas were so much more directional than the wide, fan-shaped beams of the earlier systems, the entire problem with ground reflections could be avoided by ensuring the antennas were always pointed a few degrees above the horizon. This meant none of the signal bounced off the ground on transmission, and that any nearby reflections of the returned signal would also not be seen. The need for the wire ground mat of the earlier models was eliminated, and sites could be unlimbered and fully operational in hours. The new microwave sets began replacing the Mk. II during 1943, but deliveries were not particularly fast and these sets were often sent to new units as opposed to replacing Mk. II's in the field. The 1944 arrival of the US
SCR-584 radar was the catalyst for the rapid replacement of all of these sets, as it combined scanning and tracking into a single unit with an internal generator set. In the immediate post-war era, these were in turn replaced by the smaller and lighter
AA No. 3 Mk. 7 radar, which remained in use until AA guns were removed from service in the late 1950s. ==Description==