Looking for information on Ruggedized Honeywell 516/ARPAnet's IMPs

I’m talking beyond what can be found in ‘where wizards stay up late’ or the wikipedia article (by the way good book, if you don’t have it. Get it.) Thing is, while I don’t currently have the money, I want to build at least a case that replicates the control panel possibly with a pi zero in because emulating the thing would have to be pretty simple to do even for a pi, and the inside serving as either a case for something more modern or…

I dunno. I just like the idea of something like this existing because we’re fast approaching the fifty year anniversary on ARPAnet’s first tests (August 30th I think.) And even if I can’t because lack of funds, someone doing something related to this hardware would be great. Trouble is I’m not seeing a lot of pictures out in the wild beyond a few bits of promotional or low resolution shots of the control panel (which to me looks like ‘yep those are dials and blinkenlights’)

So, anyone here have any information they’d be willing to share?

1 Like

I don’t have any information directly, but there are a couple places you might check. You may already be aware of them, but you didn’t mention who or where you have tried.
The first is ddp116.org. He has quite a bit of information on the series 16 machines, but he also has links under “resources.” You might contact him directly. He also mentions under “resources” at the link to the Computer History Museum that they have an IMP on display. I suspect they would have most of the information you need, as well as pictures. They might even be willing to take some pics for you if you tell them what you are doing. In any case, I would look at their site and probably contact them.
Hope this helps.



1 Like

Yea, that. Can’t really read all the text (though can make some educated guesses.) Measurements are a thing I’m gonna need.

Also anyone here have any experiance on this kinda hardware? On the one hand ‘why would anyone here have worked on a mini?’ On the other… I wouldn’t put it past some folk here.

Agree with @bdk6’s idea of contacting the CHM.

Certainly hooking up a Pi to drive a recreated front panel is possible: that’s what Oscar Vermeulen has done, twice, with the PiDP-8 and PiDP-11 kits. In both cases the simulation is done using SIMH, and SIMH can already simulate a 516. It’s still a lot of work, but it’s certainly possible.

This PDF is a nice 1970 overview of the IMP in the early Arpanet.

In a lightly-loaded network with a mixture of nearby and distant destinations, an example of heavy Host traffic into its IMP might be that of 20 links carrying ten single-word messages per second and four more links, each carrying one eight-packet message per second.

And there’s a 2013 retrospective and resurrection in this PDF:

However, getting decent OCR from the scan was not successful. Several different OCR programs were tried and none produced good results. Still Walden posted the scan of the listing on his website with other historical IMP system content [IMP73].

Hence, Charlie Neuhauser, who was a technical consultant to the law firm, and his colleague Tom Kilbourn hired a person to retype the entire listing, including the octal representation of the assembled program. Then they had other people proofread the entire listing — one person reading out loud the newly typed version and the other person checking what was heard against the scan of the listing. This caught about 40 errors in the retyped listing, and Charlie caught another 10 (e.g., the letter O substituted for zero, and the letter I for 1) while he was trying to
understand the code. The typing job was amazing.

All of the scans and listings mentioned above and below are available for the reader to peruse at walden-family.com/bbn/#ref-impsoftware

1 Like

I have poked at the IMP source code quite a bit when I made the host interface work on the SIMH emulator. I think some pretty good guesses can be made as to how the panel lights and switches work.

1 Like