Retrotechtacular: TOPS Runs The 1970s British Railroad

How do you make the trains run on time? British Rail adopted TOPS, a computer system born of IBM’s SAGE defense project, along with work from Standford and Southern Pacific Railroad. Before TOPS, running the railroad took paper. Lots of paper, ranging from a train’s history, assignments, and all the other bits of data required to keep the trains moving. TOPS kept this data in real-time on computer screens all across the system. While British Rail wasn’t the only company to deploy TOPS, they were certainly proud of it and produced the video you can see below about how the system worked.

There are a lot of pictures of old big iron and the narrator says it has an “immense storage capacity.”  The actual computers in question were a pair of IBM System/370 mainframes that each had 4 MB of RAM. There were also banks of 3330 disk drives that used removable disk packs of — gasp — between 100 and 200 MB per pack.

As primitive and large as those disk drives were, they pioneered many familiar-sounding technologies. For example, they used voice coils, servo tracking, MFM encoding, and error-correcting encoding.

The software was written in BAL, the IBM assembly language, although there were a set of macros called TOPSTRAN to make it slightly easier. Originally, each depot was going to get an IBM card reader and punch machine, but these proved to be unreliable in the rugged environment. Instead, each depot had an emulated card reader and punch using a Datapoint 2200 — the famous computer that didn’t use the Intel 8008, but that CPU was made for use in this computer.

In the video, you can see some Datapoint 2200s and card readers in use back at the data center. They even take the cover off one of the Datapoints around the 3-minute mark. The machines had 12K of RAM (on three circuit cards) and two tape drives. Around the 24-minute mark you get a look at a 600 baud, although the railroad apparently only used 200 baud for reliability. They also show a 2,400 baud modem that, we are pretty sure, had to be tuned before use.

The video can’t seem to decide if it is for general audiences or technical people. For example, it describes the tones from the modem and shows block diagrams of many of the systems. There are even some fake oscilloscope traces of modem outputs.

As far as we can tell, some of TOPS is still in use today. We hope some of it has been modernized, though. If you like 1970s mainframes, we’ll go ahead and waste the rest of your day. No kidding. The video doesn’t embed, but you can play it by clicking the picture below.



29 thoughts on “Retrotechtacular: TOPS Runs The 1970s British Railroad

    1. The Dutch signalling training institute have set up a small-ish model railroad (In a tree armed star shape, each with about 10 switches and 40 signals), where each arm is controlled by a generation of real life signalling equipment. Especially the arm with 10 floor-to-ceiling cabinets of clicking relays is impressive. The one small cabinet with 40 or so plc cards less, and the third arm with an even newer solution that I forgot even less (although that had more “volume” as there were more translators from fiber optic cable to the local switches/signals which would be at each individual signal/switch).

  1. “We hope some of it has been modernized, though”

    Why? To help out Nvidia? If the mainframe stuff still works, then it still works. That’s why IBM still sells mainframes.

    1. Because it’s over 50 years old and failures would likely cause significant outages if spares were unavailable or found to be faulty.

      I know of and have worked on equipment that runs other British infrastructure on very similar ancient tech and there is considerable concern about what could happen if a failure isn’t repairable because it would affect the *whole* country.

      So there was a major project a few years ago to renew and replace the control systems where it’s not possible to decommission the infrastructure itself (we are talking *BILLIONS* and significant timescales to replace and decom)

      1. Makes sense. On the other hand, how to know that modern tech will run as long as the previous did?
        There are long-term studies yet for modern tech.

        If reliability has priority, wouldn’t it be wise to rather produce replacement hardware for the old technology, so that the existing architecture (software, hardware) can continue to be in charge ?

        I know this sounds archaic, but these dinosaurs simply have been proven by time, which can’t be said about modern technology.

        In an ideal world, or so I think, a new technology would begin to start working alongside the previous one.
        The previous one would be around to supervise or assist the new one and serve as a backup system.

        1. We can order industrial temp grade microprocessors, usually ARM based. It’s not that big of a deal to design something that has to work in a factory, car, or outdoors for 20+ years. It’s just another day at work for some engineers.

          1. It’s not hard to design. It’s hard to prove that it does. That’s why time tested designs are valuable.

            See the CD-R problem: you can “prove” that an individual disc can theoretically hold on to data for a hundred years under some particular conditions, but if you make a hundred thousand of them, will any of them actually last in the real world?

        2. The U.S. versions if TOPS used IBM 1050 (Card Reader/Punch/Printer and teletype terminals. These could be These were on analog (telephone) circuits and “could be” connected point-to-point to allow a railroad yard to send train consists and such to the next station should TOPS go down. I do know this was never used in my time at SP (1971 – 1994), but it was a design consideration of the project. TOPS started online in 1965, the project, itself, in 1960.

    2. Modern systems can, in some cases, do more than old systems.

      In the Netherlands, all kinds of modern passenger information systems are built on top of the 1950’s-era railway information systems. That works very well, you can even see on what part of the platform your train will stop. One part can not be included and that is the way around the train is, so you don’t know where 1st class is. Doing that requires a complete redesign of the train information system.
      Because it is a safety-critical system, that is not something you do with a weekly update.

  2. I worked for IBM at that time and did some work on TOPS. We all had to be trained on how to cross the tracks as a lot of the equipment was in the middle of hump yards. I never knew what the roots of the system was but it work well. I was IBM 370/158 trained too but got bored with how little real engineering knowledge was required to fix the systems – just a lot of card swapping. I moved into software expecting more of a challenge and the same thing happened there with a system called RETAIN. No dump debugging was required; just scope the problem and apply latest upgrade patches. Those 3330 disk drives were very heavy compared to the 2314’s and the drive was fascinating as it used voice coil technology.

  3. I was one of the guys in the computer room from the start of TOPS on BR. To be fair I don’t think I would recognise it now but the last time I saw the box running TOPS it was a hell of a lot smaller than the two mainframes we had in the 70s, that was in the later 90s.

  4. I was one of the first guys working in the BR TOPS computer room, starting in late 1971 before there was even a computer installed. Yes, TOPS still runs today but I wouldn’t recognise any of it now. The last time I saw the hardware it was the size of a four draw filing cabinet. When I worked in the computer room (71 to 80) there were two machines and associated disk drives, etc, in a room 50 feet by 150 feet.
    The list of CPUs that I had my hands on: 360/50, 370/145, 370/155 x2, 370/168 x2, 370/150 briefly.

  5. Who else cringed at 2:50 onwards as he turned the power off then immediately ran a finger around the CRT high-voltage power supply?!

    But I didn’t know wide-screen CRT monitors even existed back then – I thought wide screen was a relatively modern thing.

    1. Sure, with monochrome screens it’s no problem (no mask).
      Picture itself can be adjusted to any size and aspect, too.
      It’s just beam size and lines and timings and stuff.

      Generally speaking, the film gives not a correct representation of technology of that day, I think.
      Most of the devices are unnecessarily bulky, to match industrial look and expectations.
      The open chassises visible in the footage tell the true story, rather.

      The modem for example, isn’t much more than a variation of an 300 baud acoustic coupler that people had used in late 70s with their Apple IIs.

      It’s just much more “bloated”, with extra shielding, filtering circuits and so on.
      The basic principle, though, isn’t much different to the VHS sized peripherals of your typical home computer.

      In labs and data centers, there had been much much more sophisticated computer systems already.
      On a rail road system, this was neither needed nor wanted at the time.
      It all was about reliability and functionality, rather.

      To give an idea of technical possibilities, please watch the “mother of all demos”. :)

  6. Interesting use of km at around 13:30 to 13:50. There was a real push for metrification in the late 60s and early 70s as primary (5-11) and secondary (11-16, or 18) schools like the ones I went to taught and used only metric in the 70s and 80s. It’s depressing in the extreme that metric adoption is so much on the back foot now, with people frequently reducing the discussion to a question of patriotism of all things. Having said that, the UK government tried to make metric optional last year, but in the public consultation, 99% of respondents said their either wanted to keep things as they were or go more metric!

    You’d think that would send a strong message to the UK government, but instead they decided to introduce imperial pints of wine [facepalm].

  7. Yet another comment from me. The testing system described at about 20:00 is pretty amazing. It looks like they used a spare frequency band at around 3.8kHz to send 5-bit relay activation codes to any of the stations and then remotely measure frequency attenuation for any unit out in the field!

  8. There is an excellent (122-page) project report on British Rail’s TOPS, by the project manager in 1979 here:
    (Page 29-30 of that report detail the equipment used at various stages in the 1970s)

    TOPS was developed by Southern Pacific Railroad (with IBM and Stanford U.) in the USA during the 1960s, later also installed in Canada and British Rail in the UK during the 1970s. Huge assembly-language mainframe software projects, with many “man-years” of programming, and further years of implementation, updates and upgrades.

    There are several more videos, printed articles, and a write-up of the Canadian implementation history — all posted online.

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.