Transit Satellites and the 1961 TRW-130 Minicomputer

This post explores a curious intersection of several UT computing stories during the sixties, all connected with providing periodic corrections to the inertial navigation systems onboard Polaris submarines. A number of methods were developed for acquiring position fixes on the submarines, but the leader was Transit, the first operational satellite-based navigation system. The system relied on measuring the radio signals transmitted by Transit satellites. Systems onboard the submarines extracted navigational fixes from the shapes of the received doppler curves. Processing this complex doppler data required digital computers, which were at the time still typically room-sized machines. To bring this capability onboard submarines, researchers designed a pioneering and influential minicomputer. Introduced in 1961 by Ramo-Wooldridge, the TRW-130 was designated as the AN/UYK-1 by the Navy. It is famous as the first computer small enough to fit through a submarine hatch, partly thanks to specially rounded corners. It could process the satellite's orbital parameters and doppler data to generate a position fix in about fifteen minutes. In size and shape it was quite similar to its contemporary the DEC PDP-1. The first photo here seems to be one of the only images easily available, and shows a programmer's card at the Smithsonian. There's no evidence of one of these machines being at UT, but there are other less direct connections. [1]

Los Angeles based TRW and its precursor Ramo-Wooldridge were where David Young was working on numerical computing when UT Austin hired him away in 1958 to become director of the Computation Center. Within the geographic corridor of Los Angeles, TRW and RAND functioned as a symbiotic intellectual relay. RAND utilized game theory and strategic forecasting to establish the mathematical requirements for nuclear deterrence, while TRW served as the technical architect responsible for translating those requirements into physical hardware. They shared DNA, geography, and a special relationship with the Air Force. Both acted more like universities than like traditional aerospace companies. When Bernard Schriever was deciding who should manage the ICBM program, he explicitly rejected using traditional airframe manufacturers, noting that existing industrial organizations "generally lack the across-the-board competence in the physical sciences". TRW’s headquarters in Inglewood was colloquially known as the Schoolhouse and provided the necessary cultural and technical infrastructure for the application of Young’s SOR method, treating the digital computer as a laboratory for physics simulations. This was the corporate culture that produced the TRW-130.

The long story of the Transit navigation satellite system was interwoven with the story of computional modeling and simulation at UT Austin, particularly the story of UT aerospace from around 1960 onward and what evolved into the Center for Space Research. An important focus of Byron Tapley and Bob Schutz’s research was the Transit system and the quality of the orbital information being uploaded to the Transit satellites. Schutz became one of Tapley’s earliest coworkers and was deeply involved with implementation of their research on the UT Computation Center CDC 6600, Cyber, and Cray. The 6600 arrived in 1966 while he was a grad student, and he certainly ranked as an expert user of the 60-bit systems at UT, with roughly twenty years of experience from the arrival of the 6600, through the follow-on Cyber era, and then the Cray in the eighties. He developed a special geopotential formulation and other algorithmic innovations to take advantage of the computing hardware for specific improvements in applied gravity field modeling.

Their work aimed to improve the system's accuracy and extend it for use in scientific and geodetic applications. The system relied on the satellites broadcasting their own known orbit, and the Navy had to continually track the satellites from the ground and upload fresh orbital ephemerides and clock corrections twice per day. This limitation is what ultimately drove the continuous push for better gravitational models and advanced orbit determination techniques. The upload data was generated by the Transit system’s ground stations. This ground segment was the brain that kept the Transit system accurate. Tracking stations like the one operated by UT Austin at McMurdo Station Antarctica, along with a similar station in Thule Greenland, were strategic assets. Because the Transit satellites flew in polar orbits, these high-latitude ground stations were able to observe and track every pass of every Transit satellite. Every time a Transit satellite passed overhead, the stations recorded the doppler shift of its signal. During the seventies, Station 019 at McMurdo was typically staffed by electrical engineering graduate and undergraduate students from UT Austin, who operated the equipment out of a small quonset hut. The station measured the doppler shifts, transferred the raw tracking data onto 5-hole paper tape, and then transmitted it via commercial and military teleprinter networks.

Los Angeles based TRW precursor Ramo-Wooldridge circa 1958. This was around the time of David Young's departure for UT Austin, and work on the TRW-130 minicomputer was underway. 

UT Austin's Transit Station 019 at McMurdo Antarctica in the seventies.

Station 019's control room, with the radios for measuring Transit satellite doppler data.

Bob Schutz and Byron Tapley, center left and right, at UT Austin. Almost certainly in WRW, the old aerospace building.

[1] Note from the Smithsonian at TRW-130 AN/UYK-1 Digital Computer Coding Card . This code card assisted with the programing of TRW-130 computers made by Thompson Ramo Wooldridge, Inc. It includes the octal and decimal equivalents for common logand codes. The word “logand” was a computer acronym for “logical command” the intermediate level programming language planned in the design of the TRW-130 computer. The donor, who programmed TRW-130 computers, shares the following: “Logan programming allowed assembly languages to be developed for specific applications. … . Most programmers never saw the logans because they used one of these specialized assembly languages. I did most of my programming in logans, however. The logan code was extremely difficult. One had to understand how the machine operated at the flip-flop and gate level. Each instruction was at the clock level and one had to understand what the hardware was doing at each clock cycle. If you made a mistake, the error was difficult to find. The machine basically went nuts. It started doing things that it was never supposed to do. Every bit configuration in the machine was executable. One could easily create a function which you had no idea what would happen. It was not unusual to make a mistake that caused the machine to branch into data, and since every bit configuration was executable, the machine would start executing data thinking it was instructions.”