A question about IBM 2780/3780 over TCP/IP for systems under simulation.

On Mon, Apr 4, 2011 at 4:50 PM, Michael Gemeny <mgemeny at pgcps.org> wrote: OK. That makes sense. We used to routinely patch different machines in the office together with our own boards and a shelf of sync modem eliminators. It's cheap when you make your own hardware and can press repaired or older rev boards into service (and don't have to pay yourself the license for the software). I found this as the first hit googling for "3780 encapsulation over tcp" http://www.advancedrelay.com/ci/solutions/bi-mono Looks like they are just taking the bisync packet and sending it with its length prepended as a 16-bit quantity, with the leading two syncs and trailing CRCs being optional (with the choice being both or neither). Obviously this technique wouldn't care if the payload is 3780 or HASP, and the emulation layer could know when the emulated system was sending out the CRCs by inspection (i.e., calculate what the CRC needs to be as each byte accumulates in the buffer, then if you get a match, send the buffer as a TCP packet). It's a bit CPU intensive but it's payload-protocol agnostic. If you _know_ the protocol (typically HASP or 3780), you can parse it and be expecting a CRC when it makes sense that the packet is finished. -ethan