Sunday, October 9, 2022

A Technical Appendix to Postblogging Technology, June 1952: SAGE, Whirlwind, Simplicity


Wikipedia says that Kelly Johnson's current campaign for a "simple" fighter, inspired by "a series of interviews with Korean War fighter pilots," is going to lead directly to the F-104 Starfighter, which is going to gobble up a large proportion of American Mutual Defence aid at the expense of the Lightning, which was no big loss, and the Buccaneer, which was. In retrospect it seems absurd that the F-104 beat out the Blackburn product in the ground support role. Of course, it turns out that Lockheed edged out foreign orders in the fighter and turboprop transport sectors on the strength of massive bribes, and it is this overwrought demand for the Yankee dollar that the MSDAP was obliquely addressing in the first place. 

The question here is what "complicated" looks like, and the answer is the Starfighter's predecessor, the F-94C Starfire, a "first generation . . . all-weather, day/night interceptor," which renders into the English as "Oops." And I say that as a Canadian with a patriotic attachment to the CF-100, but there's a reason the pilots nicknamed it "the Clunk." 

319 Squadron USAF, flying F-94Cs, deployed to Suwon Japan in January, 1952, so Johnson would have had a chance to interview pilots and RIOs flying the latest Lockheed product. 

He would have  heard all about the basic problem with this generation of aircraft, which was their marginal uselessness. Instrument flying and radar interception require two crew, and extensive electronic impedimenta. This gave them marginal performance at interception altitude, particularly the F-94C, which was  heavily dependent on an inefficient afterburner to get the necessary performance boost. This meant that they have only a very short window to gain a firing solution before the pilot has to wrestle the plane into a not-stalling trajectory. That meant a "fire control system," which was not a novel concept at the time, and worth developing in its own right from an industrial strategy point of view, but leading to carrying even more weight, and also Fifties-era electronics, into the air. The Hughes E-1, and later E-5, which combined a radar, a computing gunsight, and, as we heard in the first installment of June techblogging, a crude heads up display for targeting. Clearly none of this would be practical in a high performance single seater, and the MiG-15 was doing fine in the air defence role by depending on GCI. The F-104 ended up with a spartan set of avionics by the standards of its competition, notably the Lightning's AIRPASS. It never mattered in the least on account of operators declining to fight any major wars with F-104s, but one has to wonder if it was the right decision.

It also, of course, places a heavy reliance on the ground side of "ground controlled interception," about which I am going to talk today. No history of Twentieth Century technology can ignore SAGE. 

Sage is the Semi-Automatic Ground Environment, one of those grandiose Fifties-era projects that it is hard to believe have come and gone. I've treated you to this American Air Filter ad partly because I finally have my hands on Fortune magazine again and partly because it speaks to the pictures of SAGE Direction Centres in the Wikipedia article, especially Acroterion's shot of the abandoned Centre on Stewart Air Force Base, New York. These four story, 3.5 acre concrete blockhouses evidently could not be squirreled away underground, so the Air Force settled for hardening and overpressuring them and giving them their own generators.  Twenty-six Direction Centres were built (but perhaps two never got their equipment?), including the one at North Bay, Ontario, and ten more were planned but never completed. IBM ultimately delivered 56 computers (two for each installation), and was paid a half-billion dollars for them at a time when a billion dollars was real money. 

The reason for all the real estate was that the GCI concept ultimately adopted by the USAF required radar signals from multiple stations to be resolved at central locations, with all incoming data arriving as analog signal over telephone lines, and all outgoing direction developed at the Centre. Resolving the output of multiple radar stations quickly enough to intercept possible low altitude intruders meant automatic processing of signal into track, and that required computing. Specifically, an   AN/FSQ-7 Combat Direction Central requiring 22,000 square feet of space, or an entire floor of the standard blockhouse, "the largest discrete computer system ever built . . . . weigh[ing] 250 tons . . . .[and] us[ing] a total of 60,000 vacuum tubes . . . and up to 3 megawatts of electricity." In some ways it is almost more amazing that only 46,000 of the vacuum tubes were in the computers themselves. The telephone and other support equipment took up fourteen thousand vacuum tubes. Compare that with the 600 that raise questions about the availability of Link's new B-47 trainer!

SAGE has been under development since the beginning of the Korean War panic on the basis of an equipment which MIT had begun developing as a universal flight simulator for the Office of Naval Research, the Whirlwind I.  The USN was already following the Royal Navy's Comprehensive Display System, which had been taken up for the Chain Home replacement, ROTOR, so everyone was current with the failure of the Elliott 152, the original, digital computing, element in the Comprehensive Display System. Undeterred,  and notwithstanding a University of Michigan Willow Run proposal to develop CDS as a SAGE alternative, the USAF elected to proceed with the gargantuan AN/FSQ-7 system. 

One of those modern historians of science at the intersection of "social studies of science" will probably remark on institutional imperatives and the emergence of the idea of SAGE as an actor in its own right in its social network, if I am remembering reviews of Aramis right. There's something about the fact that MIT has, as of June 1952, 1300 employees and a brand new building devoted to turning Whirlwind I into AN/FSQ-7. Michigan's contract would be terminated in April, 1953. By this time the concept had been refined. The AN/FSQ-7 would consist of two improved Whirlwind IIs, so that one could be down for maintenance at any given time. MIT needed a new tube from Sylvania with a high-purity tungsten filament, the 7AK7 "computer tube" of late Fifties advertising, pre-installation stress testing, and one involving tapping on the glass tube. "A thratron-triggered circuit was bit to automate this test," which sounds good until you realise that "automatic" had some interesting meanings in 1952.

Whatever. I'm sure that it would all have worked brilliantly in the event of a Soviet nuclear attack and that the American taxpayer was well served by all the money sprinkled on MIT and IBM. Certainly better than letting precious US dollars escape to foreign places like England or Michigan. 

No comments:

Post a Comment