by Damien Burke
The investigation by BSEL found that it was in the power band in the region of 98 to 100 per cent that LP driveshaft problems were showing up, and even in this region the condition could not be easily replicated. If an engine was kept at or below 98 per cent there was a very good chance the problem would not show up at all. Even in the danger zone, up to 3hr of running could be expected on average before damage was apparent. The engines installed in XR219 had not been inspected to ascertain the condition of their LP shafts, but as one had run only 80hr and the other only 120hr, and the one with a strain gauge fitted had shown no detectable stress so far, BSEL was confident that both shafts were in good condition. Thus the first flight now became an option, if a risky one. It was feasible to take off with just 98 per cent power on the engines, but it opened a window of risk during the take-off run, during which there was a period of several seconds when an engine failure allied with a brake-parachute failure would mean the aircraft could not stop before entering the arrester barrier at the end of the runway. The speed region in which the engine failure would be critical would vary with aircraft weight, and Boscombe’s barrier had a 66kt (76mph; 122km/h) engagement limit for ‘safe’ use. To keep below that, the aircraft weight would have to be extremely limited, which meant carrying very little fuel. This was about as far from an ideal situation for a first flight that you could possibly get.
The initial engine runs of XR219 at Boscombe Down ran into various small problems before a more significant issue with the reheat fuel manifolds became apparent. In this shot the aircraft can be seen to be tied down for early engine runs, with instrumentation cabling and fuel and air supply lines also attached. BAE Systems
This dramatic shot gives some idea of the sheer power of the Olympus 320. Noise levels exceeded those of all previous engines, and triggered additional research into the health hazards of extremely high noise levels. BAE Systems via Warton Heritage Group
The take-off on Flight 1, with engines of frankly dubious reliability that were really far from being truly airworthy. BAE Systems via Warton Heritage Group
The decision to fly or not was left with BAC, which placed the decision in the hands of the men at the sharp end, the flight crew. The pressure to fly was immense and the risk was taken. In the event, the first flight proceeded without any engine problems. Henry Gardner of BAC later wrote: ‘I must confess to a greater feeling of relief after it had touched down than I have known for many years.’ With this flight successfully undertaken, it was thought prudent to not push their luck, and the aircraft was grounded for several months while the engines were changed and the basic fuel restrictor modifications introduced. The port engine of XR219 had had very little life left on it anyway, owing to limitations imposed by early-design turbine entry ducts, and the starboard reheat fuel pipe had been leaking.
With the LP shaft still considered a weak point, though, the same 98 per cent restriction would only allow a succession of very short flights limited by the tiny fuel load permitted, so the restriction had to be lifted so that 100 per cent power was available (for take-off only). In an attempt to give the crew some warning of impending LP shaft failure, the output from the strain gauges on the LP shaft was linked to an amplifier and the output sent to warning lights installed in the cockpit, with the amplification set to trigger the light if the stress on the shaft exceed 10 tons/sq in (1,575kg/sq cm); well below the breaking stress. If these lit up, the crew would know to throttle back immediately and ideally shut the engine down. Unfortunately a single-channel warning system based on amplifying a very-lowvoltage signal like the one from the strain gauge was inherently subject to false indications and could well indicate a problem where none existed. This is exactly what happened on Flight 3, resulting in the flight being aborted and an emergency landing. The warning system was later removed, and once the test flying was based at Warton after Flight 14, even longer-duration flights could take place, as the Warton barrier was rated to 100kt (115mph; 185km/h), permitting heavier fuel loads.
This shot taken during Flight 7 illustrates the other hazard of the Olympus 320: the environmental one. This was a particularly filthy engine, and redesigned combustion chambers would probably have been incorporated in production engines, as happened with the later Olympus 593 for the Concorde airliner. BAE Systems via Brooklands Museum
While XR219 initially flew with a single-stage fuel restrictor, a two-stage fuel restrictor (to widen the envelope within which reheat could be engaged) was fitted to the first two aircraft just before the project was cancelled. The fully variable flow restrictor that would not inflict any limitations on reheat operation at all while preventing any overfuelling was never fitted, and its effectiveness on test bench engines was only verified some two weeks after the project’s cancellation. The accessories-bay redesign for aircraft 10 to 14 was completed but never embodied in any aircraft. A BAC investigation into any means of improving engine change times was also terminated by the project’s cancellation.
Thus the Olympus 320 programme came to an end shortly after TSR2 itself was cancelled, but a lot of the lessons learnt along the way went into the Olympus 593, which powered the Concorde supersonic airliner. That turned out to be another reliable and sturdy version of the Olympus, a reputation that the 320 never achieved.
While most of the Olympus 320 variants produced up to the time of cancellation were scrapped, a handful survived. Several are now on display in museums, notably at the RAF Museum at Cosford, which displays one next to its surviving TSR2 airframe, XR220. The NGTE at Pyestock continued to expand and was the site for Olympus 593 testing in the years after the TSR2 was cancelled, and for many other engines until it was closed in 2000. Sadly, despite the site’s history of development and test work of international importance, it has lain mostly derelict since then, and is now scheduled to be entirely demolished.
Working hours for BSEL and BAC personnel made for a punishing schedule – engine runs for XR219 at Boscombe Down often continued late into the evening. BAE Systems via Warton Heritage Group
CHAPTER EIGHT
Electronic Systems
This chapter describes the various items of equipment fitted to the airframe that enabled it to be not just an aeroplane, but a complete weapons system. Although the aircraft was meant to be procured, designed and developed as a complete system, the realities of the procurement process made a mockery of this idea. Manufacturers of electrical and electronic equipment could not afford to sink large amounts of their own money into development of equipment unless they were sure a contract would result. The end result of this was that, while the project as a whole started on 1 January 1959, work on the various electronic systems was delayed until Vickers-Armstrongs had a firm signed contract, in October 1960. While preliminary work was done on some systems, the delay between the start of the TSR2 project and the first contract meant that much of this work resulted in systems that were nearly obsolete even before their development was completed. The MoA’s insistence on controlling many of these items of equipment also injected additional delays into an already over-optimistic schedule.
Navigation
Stable platform
The stable platform was the heart of the INS. A collection of accelerometers – horizontal, vertical and azimuth – detect accelerations and can thus be used to track the acceleration of the overall platform in particular directions. The ‘stable’ part came in because the platform had to be gyro-stabilized against the rolling, pitching and yawing of the carrying aircraft. With knowledge of the starting position, the stable platform provided the raw data for a navigation system that could fairly accurately track the progress of the aircraft, but only over a theoretically flat Earth with no wind. Correcting for the fact that the Earth is a sphere, and not even close to a perfect sphere at that, along with detecting displacement due to wind (which, displacing the air mass within which the aircraft is flying, produces no acceleration that can be detected by the accelerometers) was a job for the central computing system. The stable
platform was responsible for providing the central computing system (CCS) with outputs of elevation, bank angle, azimuth angle plus horizontal and vertical velocities.
A prototype version of the Ferranti stable platform. BAE Systems via Brooklands Museum
The stable platform; an exploded view. BAE Systems via Brooklands Museum
The stable platform was a Category 1 item, selected and controlled by the MoA rather than Vickers. The Ministry overruled Vickers’ (and English Electric’s) preference for a Honeywell platform incorporating miniature integrating gyros (to be produced under licence by English Electric in the UK), and instead gave the job to Ferranti, using Kearfott gyros and accelerometers despite predicted inferior accuracy, higher costs and the fact that the Honeywell platform was already well-proven and the Ferranti platform was only just beginning development. The decision was primarily a political one, taken by Duncan Sandys, to spread work among the equipment manufacturers. He did not believe that giving further work to English Electric when it already had half the TSR2 work was fair. The RAE’s belief that the existing Honey-well system was not up to scratch was just the sort of excuse Sandys needed to force the issue (ignoring the fact that the RAE actually preferred a Sperry system developed for the North American X-15).
The Decca ARI.23133 Doppler radar antenna. This was housed in a small bay within the lower central fuselage, ahead of the weapons bay. BAE Systems via Brooklands Museum
Doppler (ARI.23133)
To deal with the errors introduced to the INS by wind, input was also needed from a Doppler radar (produced by Decca). With a wind from the side, the aircraft’s track over the ground can vary significantly from its heading. Similarly, head- or tailwind components modify the aircraft’s velocity over the ground. A Doppler radar fires pulses at the ground both ahead of and behind the aircraft, and measures the Doppler shift in the returns (the same effect as the change in pitch of a train whistle as it approaches and then recedes from you) to ascertain the velocity of the aircraft along its track, rather than its heading. Limited stabilization was provided for the Doppler antenna, which was mounted in a bay within the lower fuselage aft of the nose gear bay, but if the aircraft were manoeuvred too violently, Doppler tracking would be lost and memorized values for drift were used until it was regained. The stable platform had to be capable of withstanding up to two minutes of constant violent manoeuvring without Doppler information. Development proceeded with less drama than with other aspects of the project, and flight tests of the early models began in November 1961 in de Havilland Comet XS235 at the A&AEE. These were encouraging despite the very early standard of the model being used, and progress on this aspect of the aircraft was fairly smooth right up to cancellation.
Doppler components and controls.
The waveform generator, transmitter/receiver, modulator and power unit for the SLR. BAE Systems via Brooklands Museum
With the corrected velocities provided by the combination of the stable platform and Doppler inputs, the CCS would be able to calculate where the aircraft was at any given moment. No system is perfect, however, and a variety of errors were constantly being introduced into the overall INS; gyro drift, variations in gravity, and so on. While the computer could correct some errors, such as handling the imperfect sphere that is the Earth, the accumulated hardware errors and small losses in accuracy over repeated computations would inevitably lead to a loss of accuracy in overall position. The longer the flight, the greater the error.
Sideways-looking radar (ARI.23130)
To guarantee that the aircraft, would arrive on the target with acceptable accuracy, some means of correcting the accumulated errors en route was necessary, and this was where the SLR, also known as the navigation fixing radar, came in. The state of the art in FLRs was not up to the job of accurate mapping, and the high power levels and dish sizes were impractical anyway. Consequently SLRs with large aerials were the only viable choice for high-resolution radar mapping. Most of the areas of the world over which the TSR2 was expected to fly were well mapped, so various ‘fix points’ could be nominated during mission planning, points that would be particularly easy to recognize on radar. On approach to a fix point the navigator would monitor his SLR display (which could be switched to display either the left or right returns, or both sides at once), and mark it when it was seen. With the SLR able to provide the exact distance and relative bearing to the fix point, the navigation system was then able to reset itself with an accurate position. The frequency of fix points would define just how accurate the navigation could be. Looked at in another way, the accuracy of the inertial/Doppler system decided just how often a fix point would be necessary.
An SLR aerial on the test bench. BAE Systems via Brooklands Museum
The SLR display is on the right-hand side in this photo, with magnifying lens overlaid. Below the display and X/Y cursor movement wheels are the fix controls and ground speed selector; at top left is the control and monitoring panel for the various navigational equipment including the SLR, reconnaissance pack SLR and linescan. BAE Systems via Brooklands Museum
Display of the SLR output also introduced challenges, as the SLR had a secondary task of radar reconnaissance. The output therefore needed to be recorded, and the solution to both recording and displaying this strip of radar data was one that now looks almost Heath Robinson. In basic terms, the SLR on each side displayed its output on a small 45mm-wide CRT buried within the aircraft, out of sight of the crew. A small camera was focused on this and constantly filmed the output, exposing a constantly running roll of 5in (12.7cm) wide photographic paper that then needed to be processed and fixed so that it could be viewed. A rapid processing unit (RPU) processed the paper, and it was then scrolled along the navigator’s SLR display unit. Paper transport and camera exposure (via aperture variation) had to be synchronized with aircraft speed; and processing of the paper had to be incredibly fast (less than two seconds) to provide anything like real-time display of the radar signal. The navigation system would inform the navigator of an approaching fix point within 10 miles (16km) by illuminating a warning lamp. Once the aircraft was within 5 miles (8km) the autopilot would ensure that the wings were kept level to stabilize roll and give a steady SLR display on the final run-up to the fix point. Markers printed on the SLR paper display would lead towards the theoretical location of the fix point, and the navigator would check this display against a preprepared folio of fix-point photographs to make sure he had the right position. A cursor overlay, moved by X/Y-axis wheel controls, enabled the navigator to mark both the predicted and real fix points and allow the CCS to get back on track, with cursor movement accurate enough to permit a fix point to be marked within 200yd (183m) of its actual location.
Even with a relatively large 5in-width paper, making out detail from the navigator’s seated position in an aircraft subject to low-level turbulence would be a challenge, so a large lens was attached to the cursor bar, giving a magnified view of the paper. This in itself was a compromise, as the large lens found to give the best view was so large that it impeded on the ejection seat’s exit path, so a smaller lens had to be used. The RPU was designed such that at low level both left and right SLR returns would be displayed, with scales of either 50,000:1 or 100,000:1. At higher altitudes, either left or right SLR returns would be displayed at scales of 200,000:1 or 500,000:1. Enough paper and developing and fixing solution would be provided for 6hr of coverage. While tests found that radiation doses in excess of 10 rads would fog the photographic paper and introduce navigational difficulties, to incur this sort of dose the aircraft would need to have accidentally flown through at least two mushroom clouds, which was considered unlikely, particularly as the greatest concentration of radioactivity in such a cloud was at high altitude. Therefore no radiation shielding was provided for the RPU.
The navigator also had a moving-map display that displayed map segments on the intended route, rotated so that aircraft track was always ‘up’ and giving a constant view of what was ah
ead on track. This further aided recognition of fix points, and a repeater moving-map display was present on the pilot’s panel. Of course, were the crew limited to navigating along a preset route using only preset fix points, they would be severely limited in the sort of missions they could undertake, so the navigator was able to miss out or reorder fix points, as well as make ad hoc visual and FLR fixes at unplanned points. Thus detours could still be made away from the intended route to avoid unexpected defences or attack targets of opportunity. Preparation of the navigator’s fix folio before a mission would have been a significant job, and plans were in hand to carry out a massive survey programme to choose and photograph suitable fix points all over the areas where TSR2 operations could be expected. Naturally provision of fix points deep within Soviet or other ‘enemy’ territory would have raised certain difficulties in acquiring the necessary coverage, and navigational accuracy could be expected to decrease gradually in relation to how far into ‘Indian territory’ the target was.
