IMSAI 8080 Resurrection

76 Comments

When MITS introduced the Altair 8800–about 43 years ago–it spawned the first personal computer clone: the IMSAI 8080. The clone had several improvements and MITS had difficulty filling orders for real Altairs, so they sold pretty well. [IMSAI Guy] has one of these vintage computers that has been in storage for over 30 years. He’s restoring the thing and there are 26 (and counting) videos of his progress. You can see the second video below, but be sure to check out the others, too.

The IMSAI is famous for being in the movie Wargames. We miss computers with switches and LEDs on a working front panel.

As you might expect, with over two dozen videos, there’s a lot of detail. The computer has quite a few boards.[IMSAI Guy] has to break out the scope, but he eventually gets CP/M and Adventure running.

Although IMSAI went bankrupt in 1979, the name was bought and there is still a presence online today. There is talk of a “series two” that looks like a combination recreation and PC case, but the last update was a few years ago, so we doubt these ever actually shipped.

The IMSAI was more or less an Altair and if you want the flavor of having your own without the expense, you can fire up a browser. We’ve seen other vintage computers made into PC cases before, too.

76 thoughts on “IMSAI 8080 Resurrection”

  1. If you really miss them, I’ll sell you mine. Original IMSAI 8080, 22-slot motherboard, 40 amp supply, Tarbell DSDD controller with 3 Qume Datatrak-8 drives, 4 port serial card, 64K static memory card, 256K dynamic memory card (runs as a disk), Northstar Z-80 CPU card at 8MHz (front panel only works up to 4MHz, however. Stupid 74LS123 one-shots…), IMSAI GPIO card, extension card, and several hundred diskettes. I’d love to bring it back up, but I just don’t have the time or drive to do so.

    Report comment
    Reply

    1. Oh man, you’re killin’ me. I’ve wanted one since approximately the beginning of time, but I have $2.79 in my Paypal and forty-four cents in my bank account. I’m dyin’ here.

      Report comment
      Reply

        4. Sorry, but unless it’s going to a home for old computers, it’s not going anywhere. I’m not sure why you’d think it would be cool to smash vintage computers for the sake of a few YouTube hits. That just seems pretty moronic.

          Report comment
          Reply

    2. 40amp supply? Kids today don’t know what that means. A very big and heavy transformer, with electrolytic capacitors the size of Coke cans. Hefty diodes on heatsinks. And that’s before the voltage regulators, which were on each card as needed.

      Not of that light as can be switching supply stuff. Thy weren’t a new thing, but it was “revolutionary” in home computers when the Apple II came out with a switching supply.

      Michael

      Report comment
      Reply

      1. Yes, and that 40Amps was at least 200 Watts so there was thermal management issues as well. In the pictures you can see thin metal rails being used like voltage bus bars (instead of PCB traces) to prevent voltage drop across the circuit board.

        These old computers could also run from button battery for 0.176 of a peco-second.

        Report comment
        Reply

          2. In all seriousness, you could probably run a system for a couple hours off a pair of car batteries, assuming you bypassed the power supply. Few systems actually needed 30 amps of continuous power, and I’ve rarely seen systems with more than 6 cards in them. If you’re using static memory instead of dynamic, that helps a lot. Many people replaced the IMSAI dynamic memory cards because they weren’t the most stable things in the world. 64K static cards were become pretty affordable around the time the IBM PC was rearing it’s ugly face.

            Biggest issue would be lack of AC power for the drive motors. And I can’t remember if the case fans were originally 12VDC or 120VAC. You could, of course, run an inverter, but you’d need a true sine inverter to make the transformer happier, and you’re losing 10% to 20% in inverter inefficiencies.

            Report comment

          3. @[Ken]

            It’s not o much about the math. [jcwren] and I were just pointing out that these old computers draw so much power that battery operation was impractical.

            Report comment

          4. Um… average small car battery – 55Ah. 2, in parallel – 110Ah.
            That is about 110*12 = 1320 Watt hours.
            If the PSU was 40A, 5V – top load was 200 Watts – so 6.6 hours till fully discarged (with 100% efficient power converter, before de-rating for high discharge current)

            Report comment

          6. Doesn’t really matter much – the watt-hours are the same if you take them in parallel or in series, if you load the batteries equally.

            Well – -5V rail is very low power, you can just add +5v to -5V converter, it was used, if my memory serves me right – for bias power on the CPU, etc + for serial ports.
            +12V should be stable if you are using anything mechanical, again, if my memory serves me right – this is the power rail for the floppy drives mechanics + again – the CPU.
            Main power hog is +5v – old TTL was not very power efficient. the only thing worse was ECL.

            In any case – even with the power efficiency of, say 50% – the run time would be few hours at full load.

            Report comment

          7. @[fhunter]

            Well that depend on the context of time. Sure you could do these things now but back then there was only linear regulators.

            The context presented is that battery operation was not at all practical. Someone mentioned that the floppy motor needed mains AC as well. It was probably an AC synchronous motor.

            While your point are true today, the meaning of the words “practical” and “possible” are different.

            Report comment

          8. @ [RÖB] Ok, if we keep to the period of time – then I’d use motor-generator unit (24 or 12 volt DC motor on the battery side, AC generator on the AC side, and run it all from the original power supply. Efficiency would be lowish (I’d make a guess at 50% total – from battery to PSU output, but even this would give practical time from batteries. But the whole setup would be quite unwieldy. And probably costly (flywheel, DC motor, AC generator + some mechanical means to stabilize frequency/voltage)

            Report comment

          9. @[fhunter]

            LOL they actually did things like that back then.

            The beeps on your phone for a busy signal might not seem like much power but when there are 10,000 beeping phones connected to a phone exchange then you would need a huge current driver.

            Instead what they did was to have a mechanically governed motor drive a sort of alternator that output the frequency of the tone and this had switches on it to interrupt the tones.

            Report comment

    5. I’m assuming this is a “if you have to ask you can’t afford it” situation but how much would you want for it? I have wanted an S100 machine but eBay prices are pretty absurd for the gear. Would be cheaper just to build one in most cases.

      Report comment
      Reply

    7. I worked on Univac mk 152 computers on tartar missiles siystems in the 70’s it had 18 bits of core memory 32k running the missle system at 1 mhz. I later got hold of imsai schematics and wire wrapped on with a cassete interface and it worked pretty good. then I bought an imsai in 78 and kicked myself ever since for getting rid of it.
      larry

      Report comment
      Reply

  2. Jeez this brings back memories. In college we had a lab full of these machines with the 8″ floppies and all, Used CPM and assembly language to program EEPROMs for various 8 bit micros of the day. Once the IBM PC came into existence these things were boat anchors! I even had to program by hand using the switches a time or two when first starting up.

    Report comment
    Reply

  3. “We miss computers with switches and LEDs on a working front panel.”

    Older computers were slow enough that the LEDs could actually serve a purpose – I question both the usefulness or the entertainment value of data and address line LEDs on a 3+ GHz processor…

    And I certainly don’t want to step back to single megahertz clockspeed computers anytime soon.

    Report comment
    Reply

    1. These days, it would be useful to have a indicator showing the amount of internet traffic in/out of your PC. Similar to the HD indicator of old, it could be a useful hint to let you know when your PC is doing something it shouldn’t be.

      Report comment
      Reply

    2. When running, you can see if there is a halt or stuck in a loop. But the lights are for programming and trouble shooting. You enter address with switches then data. You can also halt/run and read the address of the code that is taking all the time.

      I have one. I don’t know why all the reconditioning. Mine just works and always has. A couple 8 inch and 5 inch floppy drives. The only thing I would do is remove the 2.5 ton power supply (save it) and put a Switcher from Meanwell in its place. It will pay definitely pay in the savings on post-lifting hernia surgery.

      http://i1091.photobucket.com/albums/i382/Comedicles/BitsnPieces/LandOfLostProjects/IMSAI_zpsbs8g0nb3.jpg

      Report comment
      Reply

      1. “When running, you can see if there is a halt or stuck in a loop.”
        Prey tell how would you detect a pattern in an LED flashing on and off 3 BILLION times a second?

        How often are halt conditions a problem with current day computers?

        Do you really want to single-step through the Windows 10 core/Linux Kernel overhead just to watch your Visual Basic program execute?

        The LEDs were useful because the only thing running on the computer was your own program – to ‘debug’ a windows program you’ll be single-stepping through device drivers, OS code, etc…

        Report comment
        Reply

        1. > How often are halt conditions a problem with current day computers?

          Multiple times daily, if you are working on device drivers or OS kernel parts that is.

          > The LEDs were useful because the only thing running on the computer was your own program – to ‘debug’
          > a windows program you’ll be single-stepping through device drivers, OS code, etc…

          That ends in the moment you run CP/M or IMDOS etc. on the machine. Then you single step
          through OS layers for terminal I/O, filesystem I/O, interrupt drivers etc., great fun ;-)

          Even more fun to single step a PDP-11 running RSX-11 or UNIX, the software is a bit more
          complex than the early simple 8080 stuff.

          Report comment
          Reply

          1. PS:

            For getting Cromix booting on an emulated Cromemco Z-1 the frontpanel was invaluable to understand some details what Cromemco did inside there, not all the nasty details are well documented.

            Report comment

        2. When I was fresh out of school I worked at Boeing as a site manager for PDP-11 and VAX systems. Each group on the 7N7 had one, Propulsion, Aero, Sound, Structures, Avionics, Flight Test, etc. They were developing all the tools that became standard later on, like the splines and such for defining surfaces.

          An ongoing problem on those multi-user systems was someone running a piece of code that “froze” everyone else. And the person who was running the task with the problem could not tell either. The managers would eventually reset and then everyone – up to 32 engineers, would start yelling and cursing as they lost all their work.

          I had done a little with PDP11 in school and I could see that when there was a “freeze” some panel address lights were much brighter from being on more than the others. The OS was still functioning and undamaged so I flipped the run/halt switch until it stopped in USER mode – a few times at the most due to the resource hog user program. I entered a zero for data I think, and flipped to RUN. Whatever I did, it was a halt instruction. One engineer complained while the rest were happy, and we had our culprit. This would work today for a similar situation. It is a “Force Quit” for when your terminal won’t work.

          This was written up as company wide way to handle this and eventually also by DEC, an incredibly unresponsive company with the worst documentation of all time. Why didn’t they think of this?

          Report comment
          Reply

          1. My experience with the PDP-11’s was different, but I had intensive training for 11’s system programming at DEC Munich. Almost all 11’s other than the Micro-PDP had a front panel, either a binary one like Altair/IMSIA/Cromemco …, or one with 7 segment display and keypad for entering octal values. The instruction set is absolutely symmetric so that no disassembler is needed, you can read all and any instructions from the octal value. With these machines it was possible to locate almost every problem from the front panel by singe stepping and looking into memory locations.
            And the all DEC OS’s for the 11’s came as source, which you configured for your needs and assembled it with listing to the teletype. So you had a listing with addresses and op-codes including a symbol table, wich enabled you to debug all and anything. Sure, this work is tedious, but always successful.
            On later systems you use a kernel debugger compiled into the OS via some communication port, which allows source code debugging, much more comfortable than a front panel of course. Looking with a front panel into the bits and bytes still is fascinating for me, same as 40 years ago, so I do that in my spare time, because my idea of entertainment is kinda different ;-)

            Report comment

          3. No sorry, decades ago and I can’t remember the names of the DEC guys anymore :( All I can say is that the training we received was excellent, so that we got the jobs done perfect. The DEC staff really was very competent and I still have my 11 manuals we got. They even traded me in the DECUS C compiler for the RATFOR compiler port I did on RSX11M, very nice of them. Around that time I always was traveling with a RL-02 plater for exchanging software with others, thanks god for the USB flash devices nowadays.

            Report comment

  4. I had two friends with S100 backplane computers back in the day-
    They spent a lot of time keeping the thing running, as all those S100 card slots became unreliable due to oxidation on the contacts. Also, boot up was problematic- either you coded in about 250 instructions to start the computer so it could read in a program from a cassette tape or a paper tape reader, which contained another program that could read data from an 8″ floppy disk that held a whopping 300kB of data, then you could either have CP/M OS running, which would allow accessing a second 8″ floppy drive, or to access a hard disk and a teletype terminal.

    If this sounds like fun, then you could emulate this with higher performance with a micro controller SBC and have a more reliable processor that can run something like Linux.

    While the vintage computer scene has some allure, I’d think you’d grow tired of the pains experienced with getting it up and running each time you wanted to do something valuable/meaningful with it.

    Report comment
    Reply

    2. The smart hobbyist put those 200 or so instructions on a small EPROM, then all theyy needed do was set the starting address with the panel switches, kicking off the casette or paper tape reader…

      Report comment
      Reply

    3. The boot loaders for paper tape/cassette tape for this kind of machine are 20-40 instructions, you key the boot code into the machine in 2-4 minutes. Of course it gets old soon to enter the bootstrap code via front panel, so it was put into ROM. Floppy disk systems use boot code in ROM of course, so you just switch the machine on and it boots, same as any modern system.

      Report comment
      Reply

  5. I agree with the statement “We miss computers with switches and LEDs on a working front panel.”. I used to work on DEC minicomputers back in the 70’s and 80’s. Loved the twinkling lights and toggling a bootstrap loader into the front panel switches. Then we got a VAX 11/780 with ONE power light and TWO switches.

    Report comment
    Reply

    1. I tossed out an old Data General front panel a friend gave me because one switch was broken. Now that I have access to 3d printers, I almost daily kick myself in the butt.

      Report comment
      Reply

  6. @Ken,

    I used one of these in 1977 at my University- I do t know if the IMSAI 8080 we had was confidured with PROM banks for user programs, or not, but I know we had to flip the switches for the data byte and address for that data, press the LOAD toggle, and advance to the next address, until the boot loader was ready to execute. Maybe the prof wanted to use this method as a learning experience to make us students appreciate not having to do this on more advanced computers. I know this novelty wore thin after entering the boot loaded by hand three or four times. Our prof, BTW, was a double PhD…

    Report comment
    Reply

  8. It’s interesting how many different markets these machines had homes in. I know there was one at Georgia Tech’s WREK radio station back in the mid-1980’s that automated the tape carts. Another was installed in a carpet factory in Dalton, GA for controlling 128 dye nozzles on the pattern printer. Yet another ran a printing press for a local company that printed local editions of the TV Guide (for you youngsters, that was a printed booklet that had TV channel and program lineups).

    Report comment
    Reply

  9. I was in love with the IMSAI 8080 long before Wargames. It was and still is the most beautiful computer I’ve ever seen. I’d love to have just the front panel and blue cover. Maybe put an Arduino in there just to drive the blinkenlights. Or put a Raspberry Pi inside to run an S100/CPM emulator. I check eBay every now and then, but the offerings are way too expensive, even when they’re damaged or not working. *sigh*

    Report comment
    Reply

      1. Yeah, It would be easy to emulate with an AVR. The expensive/hard part would be the switches and front panel cutout. Perhaps a 3D printed guide and actuator that fits tight onto a normal toggle.

        Report comment
        Reply

    1. Unless you like to try to keep this running, you might want to find a different computer system. The S100 backplane was prone to getting intermittent contacts and had to be periodically disassembled and cleaned. I had several friends who had these at the time. I also used one for some of my electrical engineering labs to write machine code.

      You could get CPM to run on a Rpi and have a much faster, more reliable machine on which to develop.

      Those S100 chassis computers are akin to a vintage Ferrari- it takes a lot of tweaking and cajoling to keep them running. If that’s your thing, then more power to you. I’d rather use the computer as an appliance and accomplish things with it.
      All that being said, if you are in a major metropolitan area, check out ham radio swap meets and computer swap meets. It seems someone almost always has an Masai 8080 or Altair chassis collecting dust. I wish you well with your quest!

      Report comment
      Reply

      1. thanks for your post still set on getting an imsai 8080 i was hoping to build some new hardware for it before pushing for the latest designs using fpga and quad core arm cortex but its almost looking like that will be necessary these days with cheapening prices for 32 or even 64 bit arm cortex.

        Report comment
        Reply

  14. Say folks, not sure if my last post made it in. I’m looking for an IMSAI 8080 …. to buy. It will have a good home with a knowledgeable IT professional wanting to fulfill a dream AND teach the next generation about the basics. If you know of one for sale please call JimO 270-898-7651

    Report comment
    Reply

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.