The following article has been compiled from chapters of the book ‘Blind Landing – A History’ by R Bruce Lumsden  who was a scientist at the Blind Landing Experimental Unit (BLEU) at the Royal Aircraft Establishment at Bedford from 1966 to 1974 and continued in Government service until 2009. The book is available from the author at firstname.lastname@example.org or via the web site www.waverleyaerospacepublications.co.uk, price £30 plus £5 postage (UK), a snip for 540 pages.
This page is about the development of the system, have a look here for more about how the autoland system was used by the pilots.
Vickers VC10 and the Quest for Automatic Landing
VC10/Elliott’s Flight Control System 
Elliott Brothers (London) Ltd. had licensing agreements with Bendix in the U.S.A. and the VC10 auto-pilot system incorporated Bendix components and techniques associated with the PB20 autopilot, 100 series, flight director instruments and Polar Path compass. But while this meant that the Elliott system was to some extent based on well-proven Bendix components, the VC10 did incorporate several new features which had been designed by Elliotts.
The company considered that it would be expedient to approach fully automatic landing in two phases. To gain experience of the problems of automatic control during the landing phase, and to gain pilot confidence in the manoeuvre, automatic control to touchdown would at first be applied only in pitch below the normal break-off height, with the pilot retaining control in azimuth. Only when the experience gained in achieving acceptable auto-flare performance had been achieved, and the azimuth ground guidance issue resolved, was it envisaged that fully automatic landing would be incorporated on the VC10.
To meet the Air Registration Board safety requirements for automatic landing, the risk of a fatal accident during the automatic landing manoeuvre being less than once in ten million landings, Elliott Brothers had adopted the principle of a monitored duplicate control system (fig 1) for the automatic flight control system (AFCS) of the Vickers VC10. It had been shown that by adopting this system the mean time between failures required of one autopilot reached practical levels. In addition, attention had been given to the isolation between the two systems to minimise the risk of a common fault affecting both channels at the same time.
Among the new features introduced were the whole dual autopilot concept, the engage circuitry and error disconnect system, automatic change-over between autopilots, the input system to the aircraft powered controls, the air data system, dual throttle system, dual auto-trim system, pilot's control panels, equipment cases, junction boxes, power supplies and the method of electrical isolation of paired outputs.
The technique of building up electronic and electro-mechanical units in plug-in card modules, a feature of the PB20, was retained and a large proportion of the original Bendix cards were used, but Elliotts also designed many of their own. About 25 different cards were used. Additional advantages of the Bendix liaison were the prospect of at least partial inter-changeability of spares between the VC10 and the Bendix-equipped Boeing 707s already operated by B.O.A.C., and the possibility of reciprocal licensing to Bendix of Elliott automatic landing equipment.
Figure 2 shows the layout of ‘black boxes’ associated with the VC10 autopilot and the equipment by which each of two separate autopilots were individually monitored. The separation of the two channels was maintained right through to the individual inputs to the powered control units and there were individual circuits from the pilot's control panels and individual power supplies to both main and comparison channels. Monitoring was applied so that each important element throughout the chain was covered in such a way that disengagement automatically followed a discrepancy between any significant components in the system.
The two autopilots were normally energized as soon as the aircraft power supply was switched on, but only one autopilot at a time was used during cruising flight. Only during the automatic flare-out, and later during automatic landing, was the automatic change-over facility available. In that case monitoring was in full operation on both autopilots and the second autopilot was electro-mechanically slaved to the one which had control authority.
Almost all the mechanical actions normally required for changing from one to the other were already performed. When the automatic change-over was required, only two mechanical actions were performed, and these were automatic and mechanically simple. The monitored duplicate arrangement required the automatic change-over to operate on average only for half the total number of faults, because half of them occur in the stand-by system.
The system was of similar vintage to that of the Hawker-Siddeley Trident. Originally, it was thought that the ‘Duplicate monitored’ system would be a cheaper and lighter solution than a triplex system as in the Trident aircraft but, in the event, there was little difference.
Vickers VC10 Autoland Programme
The concept and the approach used by BAC and Elliott Automation on the VC10 automatic approach and landing was reflected in the papers presented at conferences in Los Angeles and Lucerne in 1961 and 1963 respectively [3,4] and in the aviation press (Flight International, October 9, 1960, and February 22, 1962).
The hydraulic flying controls were basically quadruplex and the 1 and 2 channels of the autopilot drove each inboard flying control surface, with one autopilot channel or the other acting as master, and the other three primary control units (PCUs) following up. During an autoland, in the event of a failure being detected in channel 1, there was an automatic changeover to channel 2. In practice, however, if the first channel disconnected, very often the second channel was not quite in synchronisation and the resultant ‘manoeuvre’ on change-over caused the second channel to disconnect.
The VC10 autopilot consisted of the dynamic vertical sensor, the three-axis rate control, the vertical gyro, the comparison vertical gyro, the autopilot dual controller, the power junction box, air data sensor, comparison monitor computer, longitudinal amplifier and computer and throttle actuator (figure 3). The VC10 flying controls test rig at Weybridge played an important part in the development of the standard duplicated autopilot, yaw damper and auto-throttle system, which entered service with BOAC in 1964.
Although BOAC never had the same commitment to automatic landing that BEA had, they had interest in operating to limits at the top end of Cat III. In 1965 the corporation had stated that its 17 Super VC10s would be fitted with the BAC/Elliott duplicate-monitored automatic landing system and G-ASGG, which first flew on 17 May 1965, was used as a flight development aircraft during the winter of 1966-67 (Flight. February 16. 1967) and in 1967, a BOAC VC10 achieved its first scheduled automatic landing.
Over 150 automatic touchdowns were made in this first phase of the trials programme. This was designed to prove the auto-flare capability, and a higher standard of azimuth equipment was then installed to bring the aircraft up to Cat IIIA standard .
The programme was progressing well until the flight tests on the third aircraft which was found to have unsatisfactorily long flares which resulted in soft touchdowns well down the runway. Unfortunately, this came just at the time when the aircraft was required by BOAC for its summer operations, and flight testing was held up until the winter of 1968-69 because of the shortage of aircraft. In the meantime, BAC and Elliotts reprogrammed the flare computer so that it would accept wider tolerances in individual aircraft aerodynamics. The new computer was flight-tested by BAC in winter 1968 and found to be satisfactory.
In 1968 and 1969 intensive rig and simulation programmes were set up by BAC and Elliott to support the Super VC10 flight clearance programme for the Cat II automatic landing programme. Measurements were taken of the actual aircraft dynamic control and autopilot characteristics by means of open- and closed-loop frequency response measurement in the hangar at Wisley.
The large number of runs required to test the system changed the management of the development and flight-test task. Instead of developing the system in flight, techniques were worked out to enable the test flying to validate the predicted performance. This reduced the test flight time and the commercial risks of this part of the programme. To achieve the best matching of the simulation and flight performance records required the mathematical modelling of the atmospheric turbulence and ILS beam noise characteristics to be improved. These techniques later proved invaluable in improving the performance of the BAC 1-11’s automatic landing system to a level approaching that of the Super VC10.
Of the 17 Super VC10s which BOAC operated, six were fully cleared for automatic landings in Cat I conditions, four were waiting for a short ground check by BAC followed by a flight test, a further four needed additional work on them before testing, and three aircraft had no modifications at all. BOAC was, however, expecting to receive ARB clearance for making automatic landings in Cat I conditions or better in late 1969.
The aircraft was certificated for Cat II conditions in 1969  and demonstrated to touchdown for future Cat IIIA certification. The corporation then restarted its low weather minima training programme for VC10 crews; the course covered both the operating side of automatic landings and the philosophy and technique of Cat II operations.
From that point BOAC carried out a programme of automatic landings in airline service, evidence which was expected to lead to the airline operating to Cat II weather minima in the winter of 1970/71.
The Super VC10 aircraft with the BAC-Elliott duplicate-monitored automatic landing system were cleared to operate in 25kn winds which covered more than 98 per cent of all winds encountered at London Heathrow Airport) and were also cleared with a l0kn tail wind component and for use at airfields up to 6,000ft altitude. Table 1 shows the performance achieved during the certification programme, in conditions sometimes much more than those in the flight manual. For example, successful landings were made in 30kn headwinds, 28kn crosswinds and 12kn tailwinds because of the excellent margins of performance designed into the system. For such operations, the required degree of reliability of the autopilot was achieved by the single-monitored configuration which gave it a 'fail soft' capability by preventing a sustained incorrect autopilot output appearing at the control surfaces. For Cat III conditions, the duplicate monitored system was installed, which provided failure survival by detecting the failure in lane one and switching control authority to the second standby lane.
Several additions were incorporated in G-ASGG. These were the self-monitored auto-flare computers which provided guidance in pitch during the flare-out and touchdown, the self-monitored lateral approach couplers which provided the precision azimuth guidance during final approach and touchdown, the auto-changeover relay which gave the duplicated system the failure-survival capability and the various improvements in control and self-monitoring which had been incorporated in the existing autopilot computers. Operation of the Elliott system was by selecting the required mode of autopilot on a rotary selector switch which was arranged so that the approach and landing positions followed on in a natural sequence from just after take-off until touchdown with the minimum of mode re-selection.
Approval from the Board of Trade
In 1969, the evidence which BOAC were gathering on automatic landings in service was to form part of its submission to the Board of Trade (BoT) for approval to operate to Cat II weather minima using suitably equipped aircraft and suitably trained pilots. It was for the BoT to give the final approval for operations, technically by means of an agreed amendment to the flight operations manual for the aircraft type.
The BOAC target was to achieve approval to use VC10s in Cat II minima by January 1, 1970, when all but two of the aircraft were planned to be equipped and half the captains and all the co-pilots trained.
BOAC was also aiming to gain approval to go to Cat IIIA limits by January 1971 although it was considered that this was likely to be delayed more by the lack of suitable ILS equipment than by aircraft or crew performance. Certification to Cat III had been agreed between the airline and the manufacturers.
At this stage BOAC was very happy with the performance of the Elliott autopilots in the VC10; accuracy at the 150ft ‘gate’ in both height and horizontal displacement was measured in inches rather than feet. However, like the SEP5 system in the Trident, nuisance disconnects were still a problem although progress was being steadily made to eradicate the problem. Many modifications were embodied to cure these nuisance disconnects, but the disconnect rate never proved acceptable for Category III. This was even although the VC10 had achieved Cat II certification, and it took too long to check whether the design improvements or maintenance action had been effective. In 1973, therefore, British Airways decided that the programme was not viable, and it was decided to delete the autoland facility and revert to Cat I operating limits. One of the main reasons for this decision was almost certainly the low utilisation of autoland on the route structure on which BOAC was operating the VC10.
BOAC/British Airways VC10 Aircraft
BOAC's successor British Airways (BA) began retiring their Super VC10s from trans-Atlantic flights in 1974, mainly due to the 1973 oil crisis, and used them to displace the standard VC10s in service. Ten of the eleven surviving standard models were retired in 1974/75. Of these, five were leased to Gulf Air until 1977/78, then purchased by the RAF.
Retirement of BA's Super VC10 fleet began in April 1980 and was completed the following year. After failing to sell them to other operators, British Airways sold 14 of the 15 survivors to the RAF in May 1981. The VC10 served its intended market for only one decade and a half. Written down and amortised by the 1970s, the type could have continued in airline service much longer despite its high fuel consumption, but high noise levels sealed their fate.
Autoland in operation with BOAC
From 1968 on the system was certified and ready for use. At the start not many runways were equipped for use with autoland, but this changed in due course. Around the same time BEA was certifying their HS Trident for autoland for many of the same reasons that BOAC was busy with the Super VC10. As both these systems became operational one thing became clear though: when compared to BEA the number of times that BOAC was forced to resort to an autoland was quite small. BEA's Tridents flew to many of Europe's cities where weather could be problematic for large parts of the year while BOAC flew to Africa and Asia where fog was not a major issue on most days. Some destinations in the US and of course the UK benefitted from the system, but it was not a significant competitive issue on the transatlantic routes.
Initially reactions from the pilots were not always very positive. At first BOAC requested pilots to use the system in good weather conditions to build confidence and experience but inevitably there were occasions where the system did not perform as expected, with erratic behaviour, disconnects and other issues.
There is an explanation on how to use the system on this page.
In his article on ‘Radio Development on the VC10’, Chris Mitchell explained that the autoland system was difficult to keep operational. The system was a combination of evolving technologies; like the SEP5 the basic PB20 autopilot used high power magnetic servo amplifiers and first-generation transistorised ac computing, but the Elliott elements used operational amplifiers based on second generation discrete components designed to perform more precise dc computing functions. As he explained many schemes were raised to improve the system but in the end an evaluation of the expected reliability made it clear that the system was too costly to continue with. As a result, the system was removed from the Super VC10s.
RAF VC10 Aircraft 
In 1960, the RAF issued Specification 239 for a strategic transport, which resulted in an order being placed by the Air Ministry with Vickers in September 1961 for five VC10s. The order was increased by an additional six in August 1962, with a further three aircraft cancelled by BOAC added in July 1964. The military version (Type 1106) was a combination of the Standard combi airframe with the more powerful engines and fin fuel tank of the Super VC10. It also had a detachable in-flight refuelling nose probe and an auxiliary power unit in the tail cone. Another difference from the civil specification was that all the passenger seats faced backwards for safety reasons. They were delivered with the standard VC10 Cat I AFCS without automatic landing.
The RAF aircraft were later augmented by additional examples (of both the VC10 and Super VC10) purchased from civil operators, the type being re-fitted to support air-to-air refuelling operations as the VC10 K.2 and K.3. The VC10 tankers continued to serve the RAF operating from Brize Norton until 20th September 2013 although the last recorded flight was that of ZA147 which landed at Bruntingthorpe on 25th September 2013, ending 51 years of VC10 service.