Tyzack 64-M Revisited

In other words: great to have you 'round here, Unsound. Tell us all about the Tyzack!

Since the 8280 was suggested in the older thread, I can add that the digitizer was a home grown one. We didn't use a digitizing MSX (which were prohibitively expensive at the time, to us at least). The 64M started out with a Sony HB-F9P (main) and a Toshiba HX-10 (sampler subsystem). The sampling hardware was partly inspired by the Doepfer/Elektor sampler, though much less sophisticated. It could produce plenty of bass though.

The VSD incorporated a new 'main' in the form of a Philips box (I think it was the VG8235) which we had upgraded (more or less) to MSX2+ by a company which I remember we were very, very annoyed with. Later on this also included a self-designed 256kB (yeah!) memory mapper, ran MSX-DOS2 and included the above mentioned SCSI disk with controller.

The big black box in the centre of the VSD picture is a Philips P2000-P. Yes, portable! If you had done some serious weightlifting before, that is. We used this dual Z80 machine for sampling since it had its own 5.25" floppy disks. This machine ran CP/M.

The 8086 (or to be more precize, NEC V30 clone) machines were Tulip System I's (MS-DOS 2), which I mainly used for word processing (Wordstar ruled!) until TED came along for the MSX. The 6502 was an Acorn Atom, which was also used in the infamous Casia LI-series of light controllers.

The speech processor was a GI SPO256-AL2. We added two Philips SAA1099 synth chips as well.

All system interconnections were by means of bi-directional 8-bit wide 16 word FIFOs (built from 40105s), which meant the different computers didn't need to wait to communicate. They could just throw out messages and continue, as long as the messages fit the 16-byte FIFOs.

What MSX ports did you use for the communication?

None. The FIFO chips were straight on the bus on their own I/O ports, as was all the other additional hardware.

What software did you run on the 'slaves'? Just slave specific software, which would just act on some commands that come over the line?

Indeed. The slaves had a small ROM in them that was automatically started on power up (much like a game ROM, really) and that contained a tiny program that listened to the FIFO ports for incoming data. That data was copied into program memory and executed. That way, we could develop all software on the main computer without the need to burn ROMs all the time (the MSX slaves were diskless).

Reading and writing the FIFOs was easy:

PDIGFIFO    EX   AF,AF'
PDIGLOOP    IN         A,(68)
            AND        1
            JR         NZ,PDIGLOOP
            EX         AF,AF'
            OUT        (64),A
            RET

GDIGFIFO    IN         A,(68)
            AND        2
            JR         Z,GDIGFIFO
            IN         A,(64)
            RET

You see we had two flags on port 68, one which indicated whether an output fifo was full or not, and one to indicate data was ready. These correspond 1-to-1 with pins 2 and 14 of the 40105's. We clocked shift-in and shift-out from the RD and WR pulses of the Z80. Note that machines could hang quite dramatically on this code. It's not something I would write again today!

BTW, there are four ports between the fifo and the status because the status was a combination of more than one fifo queue. Above is the digitizer fifo. The sampler was on port 66, with status in bits 4 and 5 of port 68. 48T keyboard was on port 67, with bits 6 and 7 of port 68 as status bits, etc.

Tyzack has come out of the cellar (and there was much rejoicing...)

New pictures and documentation is up at: http://www.tyzack.nl/
Much more to come...

transputer I believe it was called, I remember the article also.
there was even some logo with checkered 3d style backdrop for it right ?