The Ohio Scientific 300 Trainer

In the late 1970s there were a host of companies that dominated the computer market before the introduction of the IBM PC. One of these was Ohio Scientific or OSI. [BradH] has an OSI Model 300 trainer — their first major product — and gives us a peek at it along with some history of the company.

Companies like OSI, Southwest Technical Products, Osborne, Northstar, and PolyMorphic were the second wave after the likes of MITS and IMSAI had opened the personal computer market. Only a few companies like Apple hung on and made it work over the long haul.

If the history lesson isn’t for you, the technical talk starts at 4 minutes into the video below the break. This is a 6502 with 128 bytes of RAM. Not 128 megabytes or even kilobytes. 128 bytes. There’s a pretty traditional front panel with switches and LEDs.

We were impressed the board is still working. The typewritten manual looks odd today, and the board reminded us of a precursor of machines such as the KIM-1 and even the old Apple computer — the one that was also a small 6502 board and is now retconned as the Apple I.

We love these little peeks at old hardware. This would be a fun one to reproduce on a breadboard. The hand-drawn PC board shows there isn’t much to connect compared to almost anything you’d build today.

Watching [BradH] flip the switches, we couldn’t help but remember our own tour of the PDP-8. That one was an emulator, but we love [Oscar’s] recreation for doing real switch flipping.

47 thoughts on “The Ohio Scientific 300 Trainer

    1. Yes lovely. I made PCBs like that as a teenager (but only once double-sided) but to save my Ferric Chloride I tried not to have large etched areas – I used my mother’s nail polish as etch resist to fill them in. Nowadays I’d connect them to ground and call it a ground plane! :-D

    2. Not drawn. Laid out with tape and adhesive pads, often at 4 times magnification. The tape can be lifted and put down many times so you adjust the pretty curves as needed. Thus the term “rip-up” for making a change to a PCB layout.

      1. I ended up with a lot of the tape, someone brought it home from work.

        But I just applied it directly to tge circuit. Fine if single boards like mine, coming letely useless if you need multiple boards.

  1. I can’t believe the 128 bytes of RAM. That is only a half “page”.

    On the RAM chip I do count 24 pins.

    8 pins for data
    2 pins for power
    1 pin Chip Enable
    1 pin Read/Write

    makes a total of 12 pins in use

    add 7 pins to address 128 byte

    leaves 5 pins unassigned. Unlikely.

    (did a websearch)

    However also says 128.

    So now I have the challenge of accepting what I couldn’t believe :-)

    At 15:00 in the video is revealed how to have a reset vector when not having ROM.

      1. The extra chip select lines allow you to decode address *without* requiring an extra external address decoder, so making easier to have multiple RAM chips or smaller granularity decoding. There are standard package sizes for DIP, so might as well make the extra pins do something useful.

        It evolves over time as it is a useful feature. The 28 pins 8kBx8 chip has an extra active high chip select. Smaller I2C EEPROM have multiple address pin for address ranges. Some of the modern SoC/processors have chip selects with programmable address ranges and timing.

        Most engineers think at deeper level.

      2. The 6 CS lines are because, back in the day, the IO hung off the data bus and was memory mapped.
        The CS’s could be programmed to certain ranges of memory and the lower address bits selected various peripheral registers.

        God, I’m old..

    1. The Motorola 6810 (if you pause the video when he holds the board close to the camera) has like five and a half chip selects, some of which are low active, some others are high active.

  2. I can’t help but think that entering data and addresses with those slide switches might be a bit painful after a while.

    I think I would have swapped them out for mini-toggle or better, rocker switches pretty quickly.


  3. It was a trainer, something to play with and learn, not a oard that could be expanded needlessly. If I recall OSI had a progrwm where you coukd trade in the trainer if you bought an OSIcomputer.

    Not much preceded the KIM-1, but OSI was one of those companies that existed before making a splash, so you can go back and see there ads but didn’t notice them before they hadbetter computers.

    The fact that this uses the 6810 as other commenters state is an indicator of earlier than if it had been a 6502.

    Soon after the Altair hit there were third parties selling improvements and expansion and soon whole compters. So I wouldn’t group IMSAI with MITS and then separate the others. It may have seemed slow at the time but things really changed fast in restrospect.

    And the Apple wasn’t a “single board” computer in the sense of this trainer or the KIM-1. It was a full computer, albeit limited. It had video output, a keyboard interface and 4K of ram. It could run BASIC.

    1. The 6810 is the 128-byte RAM. The bigger chip is a 6502.

      What’s interesting is the lack of ROM. It’s a classic “front panel” LOAD/RUN design like the COSMAC ELF that uses simple hardware to load RAM while the CPU is halted. But I’m not sure if it even bothers to halt the CPU.

      1. Yes, I was thinking the 6810. was a 6800 with 128 bytes of ram built in. It’s been a long time.

        What’s interesting sting is the idea of rom-less front panel-less comouters existed, but it never seemed mainstream. Someone had some notes in Byte suggesting using NOPs to advance the Program Counter so you could load the RAM directly. There was a full article in Byte about 1980, someone tring to build a simple comouter as cheap as he could, using the NOP technique, and a string of LEDs for readouts and pads of circuit board with a stylus. Later in Dr. Dobbs sipomeone showed a 68000 using the same technique.

        I don’t think I noticed until much later. Oddly, Don Lancaster’s Cheap Video Cookbook used NOP, to advance the program counter and hence the address bus. I don’t know where the idea originated.

        1. Also the 8080 handled interrupts by jamming an instruction onto the bus. I’m sure someone must have found a way to abuse that for a front panel.

          I’m also impressed that it only used 5 TTL chips: a 2-input NOR (used as an inverter and a cheap flip-flop for audio), and a bunch of non-inverting Schmitt buffers. And it looks like they could have ditched one of the 7417s too!

  4. I would have love to have one of those back when I was young even if it was already kind of obsolete by then, At the time my only exposure to programming was a friend’s TI-57 calculator. I did teach myself 6502 nachine language programming when my parents bought a VIC-20 for us kids one Christmas, most of my ML programs were shorter than 128 bytes back then.

  5. I did some searching and found a reference to the Feb 1976 issue of Byte. It’s there a small crude ad at the bottom of the page. Only $99. They seem to be offering boards for fancier computers.

    I checked the Jan 1976 issue, OSI isn’t listed in the advertiser’s index. Someone else can check earlier issues if they desire.

    Unless an earlier ad shows up, they basically started n Feb 1976. Without an ad, they couldn’t have existed as much of a company before an ad at the time.

  6. I bought an OSI system about 2 years before the IBM PC came out. C1P (I think) with OS-65D to have a disk based system.. My friend that I worked with at IBM disassembled and commented OS-65D while watching a football season on TV. I modified the hardware and he added driver software to run an 8-bit parallel impact printer. Then we modified HW and software to support 80-character line CRT. I believe we indirectly caused IBM to print the BIOS listing in the IBM PC Techref, because guys we talked to (“Don’t tell us you’re disassembling some other companies software, even at home!! What’s it like??”) convinced Boca Raton that “Some guy is just gonna disassemble INTEL processor code, anyway..”.

    Ancient History Lesson…


  7. I remember Ohio Scientific ads always being on the back (or inside back) cover of Byte Magazine. I desired the C4P for quite some time, but never got one. We did get an Apple ][, though, which I spent countless hours on.

        1. I didn’t get mine until the fall of 1981, by that time it was the II. The onmy difference was a software switch to allow for 80 columns of text. I can’t remember the number of columns. But the user had to supply the software.

          I did splurge on the extra 4K of RAM, though buying from the distributor tyay sold me the Superboard. I think the 4K was about fifty dollars extra.

          Never had an expansion board, and never put it n a case. I do still have it, but the last time I ran it it was 1984.

  8. Ah yes, the OSI-300: one of the few 6502 computers with a front panel known to exist, and one of only two commercial products sporting that feature. An inspirational little demonstration of minimalism I know I’d love to get my hands on some day. There were some modern mini replicas produced with 4K of RAM, but they’re no longer available.

    1. Someone described a 6502 based computer in Byte, maybe Dec 1976 or 77, which had a front panel.

      But the Altair was almost an exception. Soon after it came out the shift was to a monitor in ROM, either with an ASCII keyboard and video nterface, or a calculator style keyboard and readout.The KIM-1 was a prime example, the “KIM” stood for Keyboard Input Module.

      I think the Altair wanted to be like how people perceived computers, the flashing lights, but once it was out the rest were more reasonable. You needed some sort of keyboard and display for practical use, so soon the front panel served little purpose.

      1. The death of the front panel came in 1977 with the big three home microcomputers: TRS-80 Model I, Apple II, Commodore PET 2001-4 & -8. Altair was not an exception based on when it came out, there was also the IMSAI, many of the Cosmac ELF variations, and the remnants of the minicomputer ecosystem still looming. Teletypes and glass terminals were still coming down in price, and these machines made computer interaction inexpensive. And that doesn’t even cover the hex keypad designs like the KIM, SYM, and countless other small trainers — all designed to be inexpensive teaching tools with some potential for expansion to MAYBE run BASIC over serial.

        As for 6502 front panels of the era, I’m aware of four. The OSI-300 is one, and the CGRS Microtech System 6000 was the other. The latter was an S-100 machine with a handful of variations, three of which included a front panel. There were also 2 more suggested designs that were not commercial products. One of which was the one from Byte magazine, from Nov ’77 called the Kompuutar, but who knows if anyone besides the creator made one. The last one is a suggestion in the OSI-400 manuals for a do-it-yourself add-on. The circuit design is very much an extension of the 300’s but for a full 16-bit address space to accommodate the far more complex features of the 400 series ecosystem.

        Since then there have been a handful of modern 6502 front panel projects, but those are all homebrew designs.

        1. I may exaggerate, but not much.

          The Altair’s front panel was complicated, and useless if you put a different CPU in the computer.

          A front panel meant hardware, to load the RAM directly. A great thing for bootstrapping, a liabilty once you got further along. Hardly something to look back with pleasure.

          When the SWTP 6800 came out, no front panel. As soon as people could add a monitor to an S-100 computer, they would. No more loading a program by hand to load a program from cassette or paper tape.

          The KIM-1 and many other single board computers did not have a front panel, they had a monitor in ROM, the keyboard and readout were just to make sales to people without terminals. It was way easier than a bunch of switches and LEDs.

          The 1802 har a mode to load data into the RAM, which made the Cosmac Elf popular since it was easy to build and no rom.

          If you think 1977 was a dividing line, what about the Sphere all in one, the Processor Technology SOL-20, the Apple I? There were probably S-100 computers witbout a front panel but I can’t remember dates.

    2. That may have been due to Motorola’s support of 6xxx systems with products like Mikbug ROMs which meant that less finger ffff..licking of switches was necessary to bring a system up.

  9. OSI never did use solder mask or even tin plating. However, the Challengers had a massive back-plane with cheap Molex pin headers and 16 or 18 slots. Plus memory mapped I/O on the 6502 was easy, which made them very expandable with custom hardware.

    They pinned their hopes on small business computing with the wrong platform at the wrong time. Stsyems with a Z80 anda big hard drive, like this and the same thing built into a desk. I have only seen one of these – ever.

    1. But OSI also had a computer that could take multiple CPUs, maybe even at the same time. 6502, 6800, 8089 or Z80. The only other one was The Digital Group, until places like Godbout started making CPU boards with 8088 and 8085 on board.

      OSI was trying to streamline expenses. Their floppy controller used a 6800 series USART and 6522, rather than a standard floppy controller IC.

      They got a good deal on Microsoft BASIC in ROM, and kept using it in various comouters even when abug or two was noted.

      Even the monitor in ROM for some reason had no way of writing to cassette if I remember right. Leading was easy, it just switched from the keyboard to the cassette interface, so it was like fast typing. But to save, you had to enter a program, then run it to save it. That’s when I started buying 10 minute cassettes .

      The monitor even coukd be used with a serial keyboad and readout. If I remember, it was all there in the ROM, but an address line determined which was used. So the ROMs stayed constant over multiple computers, maybe the life of the company.

      1. Back in 1982 as a senior in high school, I worked for a place that had a C3P rack tower. They had two 48K RAM boards in it to run two users in 6502 MS-BASIC. I think it also had the hard disk. As far as I know, their Z80 and 6809 never even got warm. I just checked, and that model did include all three CPUs, it wasn’t “choose your CPU when ordering”. The problem with having all three CPUs is that you couldn’t (as far as I know) use them at the same time, so you were paying extra to not use the other two.

        On the other hand, by then the CPU chips were probably in the $20-$40 range, so that plus the TTL glue probably didn’t add as much to the cost as it would seem.

      2. The one I linked to was a 3 CPU but I never heard of anyone using other than the 6502 or Z80 on those. I had a 6502 CPU board (The 500 board?), 48K of RAM, and the dual 8 inch floppies to run a chemistry instrument. I added floating point hardware and stepper drivers and 12bit fast A/D and D/A. We used BASIC at first then I found a FIG-Forth for it, which I knew pretty well. Forth was a good 20 times faster than BASIC and much better on memory, and when I wrote a library to use the math hardware (AMD9511 did FP and integer 32 bit math) it was faster than using the school’s VAX. One of the desk models with the hard drive was on eBay a few years ago. It looks really nice. It was a local pickup only item too far away.

  10. Hand keying the code in – brings back memories I don’t really want to re-live!
    Watching you, I remembered those times when I managed to get a large chunk in, then realised I was one address ‘off’… go back and check where you skipped or added a byte then start keying again !
    It wasn’t in the least way rewarding compared to coding today, but back then it was magic!

    1. QUOTE “Hand keying the code in – brings back memories ”

      Actually in brings back a GOOD one for me as my son and I typed in the Motorola Data format data from Tom Pittman’s TINY BASIC to the 6502 machine we wirewrapped back in 1975 I think. We had put the MOS Technology TIM (Teletype Interface Monitor) ROM on one of the cards we made. (THis was all IBM Prototype “SLT module” hardware and backplane. )

      Pittman mailed us a folded paper tape with TINY BASIC. We had a model 33 teletype I traded from the United Press teletype repair man for a nice tape recorder. But no tape reader / punch. So we took the paper tape down to the local community college, ran the tape and printed out 3 or 4? pages of the motorola format data. Sat togthe for ?hours and typed it in, saved it on 300 baud cassette tape, Loaded it to memory, hand-hex-switched in the Reset Vector, and.. and.. BASIC PROMPT!!!

      I’ll show the front panel here soon…

      1. Dr. Dobbs ntended TinyBasic to be free, but Tom Pittman jumped in with a commercial variant. But it was only $5 a copy, and he sold quite a few.
        He seemed to fill in the gaps, offering for CPUs like the 6502 and the 1802.

        His website is still up,, there’s a link there to his TinyBasic page.

        1. No EDIT here?? AnyWay: You can see the computer, the front panel, and the Teletype Model 33. My son Barry is the one who typed in TINY BASIC with me. This is actually late 1975 I think.. The little girl helping play HANGMAN is my daughter Megan, who is now a BioChemistry Professor at YALE and has 4 kids of her own :-) Wow.. (The sign on the top of the computer says “A Busy Computer Is A Happy Computer”.. quote from my friend Dave Millman at IBM )..

  11. Cool, I always wondered how those front-panel systems worked; never thunk they load straight to the working-RAM.
    But, I couldn’t imagine designing a board like this without at least putting pads for a “bus”! Could be as simple as 40 pins 2-row IDC straight to the processor.
    Otherwise, who’s the market for this thing?

Leave a Reply

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