On 9/19/18 9:03 AM, Alexandre Souza wrote: yes, or at least get all the 25-pin parallel keyboard protocols documented before a keyboard in thousands of dollars instead of hundreds. Bob Rosenbloom let me borrow a 3178 keyboard to document. I have also been thinking about something like the 3megabit ethernet cape that Ken Schrriff did https://github.com/shirriff/alto-ethernet-interface except it would replace a 3174. 3174-23R with ethernet option is complete overkill for a situation where you want to attach one coax terminal to ethernet, like most people wanting a real console on Hercules want to do.

On 9/19/18 10:02 AM, Guy Sotomayor Jr wrote: 3278 Henk's page on an adapter, which describes the interface http://www.ibmsystem3.nl/IBM3278/

typo, meant 3278

On 09/19/2018 12:08 PM, Grant Taylor via cctalk wrote: You jostled my brain enough that a random thought came rattling out. There are some Multicians working on 3270 support for for a virtual FNP. This means that they already have some code to interpret (TN)3270 data streams. I think they were connecting a TN3270 client to the virtual FNP that was talking to the Multics machine. I don't know how much help that might be, but it's some existing code related to pretending to be a (TN)3270 source that is consumed by TN3270 terminals. -- Grant. . . . unix || die

On 9/19/18 12:04 PM, Chuck Guzis via cctalk wrote: There hasn't ever been much talk on cctlk about non-ASCII terminals. It's one of those things that never crossed my mind before having to deal with that world in the context of working with the Museum's holdings. Burroughs and Sperry had their own incompatible polling terminals too. Come to think of it, HP did in a big way as well, they just spoke ASCII.

On 9/19/18 11:08 AM, Grant Taylor via cctalk wrote: correct. non-trivial project, but something that needs to be documented. coax and s/3x twinax terminals are similar conceptually. There were companies like Telex/Memorex/MTX/Visara, Decision Data and Lee Data that sold terminals in both flavors The ones for coax talked to 'establishment controllers' that were the actual intelligence of the terminal. The terminals themselves were essentially polled keyboards and a (80 x N line) alphanumeric display. Later multiple simultanious mainframe sessions were added to the terminals that you could switch between. Coax uses a point-point connection between controller and terminal, twinax is multi-drop through T connectors. Well, sort of. Eventually both systems migrated to phone style CAT cable with coax to RJ baluns and patch panels. There were other companies that sold serial async converters that would take the polled protocol and could turn it into something an ASCII serial terminal could deal with. Then PCs got things like "IRMA" cards that could pretend to be one of the polled terminals, buffering the alphanumeric screen image on a local RAM buffer the PC could communicate with. That evolved into hardware that could support multiple sessions as well. Same thing happened with twinax, with different vendors. 'IRMA' was a DCA/Attachmate/Microfocus thing. The hardware got more and more integrated over time. The Telex 277 has three boards of TTL in the terminal. 276 generation used two 2901s. The last of the MTX terminals were pretty much just one National DP83445 ASIC, or one 317744 CIP ASIC in the case of Attachmate. If you don't want real coax hardware, you just run TN3270. I'd like to have something runnable at the museum on period hardware, though, that might be possible to maintain, which was why I picked up the Telex.