I am trying to find info on the BBN C/70, a 1970s minicomputer (roughly equivalent to a high-end PDP-11) which played a role in the early days of the ARPANET (predecessor or Internet)

The thing that intrigues me about this machine is that its word length was 20 bits. Since I am in the process of building a 20-bit wide homebrew TTL machine myself, I would be interested to know more about its BBN forebear.

Any manuals/pictures/articles (especially h/w related) would be most welcome.

2 Answers 2


The C/70 is a follow up to he C/30 which was designed as a replacement/emulation of the Honeywell x16 series used as IMP.

The C/30 is a 20 bit implementation using what BBN called their 'Microprogramable building blocks', emulating the Honeywell 16 bit CPU. So a good start might be looking at C/30 manuals, specific the Native Mode Firmware System Programmer's Reference Manual. It includes detailed information of the ISA, including operation encoding.

As so often Bitsavers might provide additional sniplets, like in the


a) Resistor arrays were standard in the 1970,

b) C was considered weird stuff for system programming. Real men used Assembler. Including for the BBN CPUs :)

  • 1
    Thanks a lot. I know about resistor arrays and C ... I just have an idealized vision of those years. Please don't rob me of my dreams.
    – fi11222
    Sep 6, 2022 at 16:35
  • @fi11222 Sorry for having lived thru them working on these machines ... (BTW, Boards were as well not green, but 'nature coloured', so grey or yellow-brownish-grey:))
    – Raffzahn
    Sep 6, 2022 at 16:43
  • 1
    @Raffzahn I forget. Did they switch from phenolic resin to fiberglass laminates before or after they started using solder mask? The timelines I've seen have never gone into that much detail.
    – ssokolow
    Sep 6, 2022 at 18:35
  • I think this IMP software is more relevant to the C/30: walden-family.com/impcode/… Feb 7 at 6:51
  • It's my understanding that the C/70 instruction set was not related to the C/30, so there's probably not much help to be had anyway. Feb 7 at 6:51

It wasn't a follow-up, it was an alternate microcode load. (I used one for a summer as an intern, supporting the PSN group) The instruction set was evidently modeled after the PDP 11/70, and had very little in common with the C-30, which did not support C. (e.g. I don't think the C-30 didn't have instruction support for stack-relative addressing, and constants had to be loaded from the local code page of 512 20-bit words. They were talking about a C compiler for it in 1986, but I think they moved to a 68K system before that happened)

The C/70 ran a version of Unix that seemed ancient at the time (RFC 801 says it was UNIX Version 7), and programming in C was a little bit weird because bytes were 10 bits.

For a bit more info see https://elists.isoc.org/pipermail/internet-history/2017-October/004324.html Note that it looks like they might not have ever documented the machine language, as its only use would have been deep in the kernel.

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .