Friday, October 9, 2020

A Technological Appendix to Postblogging Technology, June 1950: Pointing the Way

Pretty cool and historical that the USAF specified that its 1956 fighters should be capable of remote control via the Sperry Zero Reader directing the autopilot, right? Hopelessly precious, to be sure, but even that is grist for the historian's mill. Anyway. What's a Zero Reader?

Stop giving me those pitying looks. I figured it out.  The Sperry Zero Reader is just a flight director. I feel dumb not figuring it out, or at least not pursuing the question far enough to find someone to explain it lucidly. 

There's a little more to be said about its gyroscopic magic, and it wouldn't be Fifties-era militariana if it weren't a little sinister in a nuclear-holocaust-sort-of-way. I'm also going to touch base with a classic of history of technology, Donald MacKenzie's Inventing Accuracy. Looking back a generation later, it does seem to me that MacKenzie's pioneering work  ought to have been the starting point of a historiography rather than all the profession has written on the subject. There's a lot going on here that historians could pull into perspective so that we could understand this world of ours before we run head on into slow motion disasters like the 737Max grounding.

But what do I know? I'm just a historian. 

Here it is, the Sperry Zero Reader.  You use the little knob on the left to set a course, and afterwards, when the needle on the left points to the top of the dial,  you're on course, so all you have to do is steer the plane until the needle points to zero. The cross in the middle is because the zero reader is superimposed on the attitude indicator. Sperry did this to space on the dashboard, and reduce the burdens on the pilot's attention, but it also quickly proved to be an additional flying guide. 
Later flight directors abstract things a bit. The magenta little plane-like wedge in the middle are the flight command 

director bars. If they line up with the yellow bars that indicate the course, you are flying the right course. This permits the pilot to follow the ILS with such precision that the "decision point" where a pilot aborts a landing if the ground can't be seen is reduced to 100ft, something to which Sperry already aspired in 1950.

I've skipped over the two ways that a course can be set --from the compass and gyrocompass or from external "localiser" radio beacon signals. The problem of reconciling them is agonising Sperry in 1950. While my modern guides to flying with a flight director treat the problem as one solved by a judicious choice of black boxes, Sperry is talking about projecting the localiser on the Zero Reader display and minimising the distance between them. 

This is clearly something for the future. The brand new standardised BOAC/BEA dashboard has separate ILS and  Zero Reader dials that can't have been very easy for an old time commercial pilot to integrate into an understanding of the world outside, and a serious accident at the end of 1950 highlighted Sperry's nightmare scenario. 

On 31 October 1950, a BEA Vickers Viking, registration number G-AHPN will try to land at London Airport (or Heathrow!) in 40yds visibility under GCA. At a decision height of 140ft, the pilot declared an overshoot, but he was already descending quickly enough to strike ground a moment later, damaging both props. The plane struck ground again 3000ft further down the runway, tearing off a wing and skidding to a halt in the midst of some stored drain pipes. The plane caught fire, in spite of which the fog was thick enough that it took rescuers 17 minutes to find the plane, by which time all 28 souls still on board were long dead. (Two stewardesses had been sitting in the back and tumbled from the plane when the tail tore free.) 

The G-AHPN accident isn't quite what Sperry was worried about. GCA is radio talk down by  ground controllers with radars, not an automated ILS system. In theory there should be no problem reconciling the ground controllers' data with that of the pilot. The controllers are right. But, especially if the results of the accident inquiry are correct, same issues are at play. The controllers might know exactly where the plane was, but they didn't know visibility conditions at the same precision as the pilot, who seems to have taken the plane down below 140ft while beginning the overshoot manoeuvre, probably looking for the ground. By that time he gave up, he was too low to pull out of his descent without touching ground,. Attitude error? Speed? Elevation? We will never know, although the pilot may have been told that there was 400ft visibility, not 40. 

 The aim of the flight director is to prevent these errors, among other things, by measuring "control parameters as aircraft attitude, attitude rate, and deviation rate."  This is asking a great deal of a gyroscopic system, and as I pursue a connection between the Sperry work and the MIT Instrumentation Laboratory, later the Charles Stark Draper Laboratory, I find that by the mid 1950s the lab had added micromanufacturing techniques to its research as it sought smaller and more precise gyros. Unfortunately, the Lab's work only comes into focus with the Polaris and Apollo programmes, because otherwise I would really unload on "Charles Stark Draper."

That other thing is the reason that the USAF is so interested in putting a Zero Reader in the 1954 fighters --aside from the fact that even fighter pilots have to land safely. In December of last year, Dr. George Valley's Valley Committee reviewed the problem of intercepting Soviet atomic bombers and concluded that so many radar stations were needed, and that detection ranges would be so short, that only a computerised system would work. That is, interception courses for incoming bombers would be calculated automatically, much as the GPO computers in Chain Home Low stations had been doing by 1944. However, the number of tracks that would have to be computed was vastly greater than in the last war, which was mainly a problem for the long hairs, and the interception course relayed to the interceptor and instantaneously implemented by the pilot with no computation delay. On the ground side, the solution was to be SAGE, the Semi-Automatic Ground Environment ("Someone really, really wanted it to spell out 'Sage'"), although the full realisation of the Valley Committee recommendations were still some years away. 

In the air, the USAF looked to the Zero Reader. Here the issue is intercepting jet bombers accurately enough to shoot them with rockets, nuclear-armed rockets, and air-to-air missiles that don't exist yet. It seems as though the Air Force is looking towards a solution to this problem that cuts the pilot out entirely. This strikes me as a bit premature in 1950 and almost hilariously unaware of the electronic warfare environment. 

I obliquely referred to that last week with a reference to Edmond Hamilton's 1959 Starman Come Home, which imagined interstellar warfare in the mold of the Battle of Britain, with opposing squadrons of spaceships engaging with missiles and "jamming beams." At the time I had just been made aware that this charge had been levelled against SAC in the wake of Sky Shield, the 1960 air defence exercise. Electronic warfare became an international scandal when details about participating RAF bombers were leaked to the Daily Mail. The British aircraft evaded interception with offensive jamming, but this isn't necessarily a British triumph over the colonials. It appears that SAGE was overwhelmed by the 'attacking' bombers' use of electronic warfare in general, and not by British bombers in particular. 

In the end, it seems more likely that the unmanned fighters of 1956 were defeated by the inadequacies of existing control systems. This is something that you would expect that existing autopilot designers would have explained to the Air Force, but perhaps Sperry is overestimating likely progress over the next five years. 

Dude, where's my singularity?

1 comment:

  1. This was the Golden Age of military automation: Cruise missiles (like Loon, Regulus, Navajo, etc.) which required precision over immense distances, SAGE, and my favorite, the navy QH-50 DASH program. DASH crashed, a lot (about half of them were lost between IOS in 1963 and removal in 1969). Which (along with the increasing retirement of the WW2 era destroyers that served as the operating platforms) was why thy served such a short period.

    ASW seems like it would be ideal for drones: permissive environment and no EW (too easy to triangulate the jamming back to the source), but DASH proved to be too aggressive about its engineering as well. I know, the US military believing the salesmen on what was technically possible, when they should have listened to the engineers instead, what an unusual story.