New Transfer Cable Protocol

This is a preliminary protocol definition to be used with my Sega CD (and soon Sega Saturn) transfer cable. Not all features are likely to be implemented immediately (if at all), but I wanted to think ahead. It's not 100% complete (some commands need further definition, a section on timeouts needs to be added and initial sync needs to be dealt with), but it's pretty close to final. I'm thinking about making checksums on data packets optional for speed reasons or maybe even completely eliminating them. Generally speaking, once synchronized I haven't had many problems with data integrity, though that might change now that I'm using more efficient handshaking (thus speeding transfers).

I'd like to get some input from other developers on what they would like to see in the protocol. I'd like to make the protocol more accessible to developers and structured so that different developers can make interoperating utilities and generally make the associated tools more useful.

 

The transfer rate using the controller port will be higher than using serial and the serial port requires more parts. Seems like a no-brainer to me.

That said, if someone wants to make their own Saturn serial cable I could probably be convinced to make a modified version of the software/protocol to support it.

You could have done this before witht the old transfer software (well at least the version I had before the hard drive wipe). The only thing preventing it then was the 64KB transfer limit (which was eliminated in my unreleased version). The virtual drive functionality is primarily there for a sort of Sega CD browser functionality.

Well it's not too useful if you're not a developer. Basically it's a description of a new protocol to use with my Sega CD transfer/dumping cable. The new protocol makes it easier to do some interesting stuff (make a version of SLO that can read files off the host computer, make a Sega CD webbrowser, etc.).

 

Well the 64KB limitation was a limitation in the transfer software not a limitation of the hardware. I had been lazy and used a dbra instruction for the receive loop which limited me to 65,535 iterations. I basically just made a version that did multiple 64KB transfers until the whole thing was done (the cart dumping software already did something like this).