On 12/3/15 8:27 AM, Kyle Owen wrote:
A colleague and I are working on getting our respective Fabritek MP-12s
working. The MP-12 is an industrial-looking computer with a very limited
front panel (deposit doesn't increment PC...gah!) but does emulate most
respects of a PDP-8. So far, we've been able to find the device codes for
updating the 7-segment LED displays on the front, turning on and off
relays, and setting some arbitrary open-collector outputs. It appears as
though there is no serial port option on ours, which is unfortunate.
There are three 512-by-4 bipolar (configured as 512-by-12) PROMs that seem
to override the core memory at the top of the 4k field. If the PROM's value
is 7777 octal, the core memory is accessible. Otherwise, you're stuck with
ROM, best I can tell.
We've dumped the ROMs to verify, but here's the code I've backed out of
them, disassembled and labeled where appropriate:
*7756
TEMP, 7755 /NOT IN ROM
CLA HLT /7602
ADDR, 7755 /7755 STARTING ADDRESS?
START, TAD ADDR /1360 GET STARTING ADDRESS?
DCA TEMP /3356 SAVE TEMPORARILY
LOOP, RRB /6012 GET CHAR
CLL RTL /7106
RTL /7006
RTL /7006 ROTATE SO BIT 0 IS IN BIT 7, BIT 4 IN BIT 11, ETC.
6015 /6015 SKIP ON FLAG?
JMP .-1 /5367
SNL /7420 SKIP IF LINK IS SET (BIT 6 OF PAPER TAPE IS SET)
JMP LOOP /5363
DCA I TEMP /3756
ISZ TEMP /2356
JMP 7755 /5355
7776 /NOT IN ROM
JMP START /5361
$
The "NOT IN ROM" indicates that the ROM's value is unprogrammed, so that
you can in fact access those core locations. The values I've provided just
happen to be what's in my machine's memory at this time.
It looks a lot like a RIM loader, except I can't figure out for the life of
me what the format should look like. Best I can tell, if RRB ORs the read
buffer with the accumulator, you'll never be able to send anything but
patterns matching (data & 7737). At least, that's what I've simulated. Due
to the fact that they're clearing the link after the second go around,
you'd be losing one bit of data.
So, maybe this isn't really a program loading routine, but rather just
something to store 8-bit values in core? Any other ideas?
Thanks,
Kyle
I have what sounds like the same machine outfitted with the same
interface (basically it looks like a synchronous serial port). I don't
have the paper tape reader, though. I appear to have the same PROMs in
mine; here's my disassembly from a couple of years back:
Addr Data
7757 7602 CLA, HLT / Probably data
7760 7755 / Also data
7761 1360 TAD 7660 / Read starting address
7762 3356 DCA 7756 / Deposit AC (7755) in 7756
7763 6012 / Start high-speed reader operation
7764 7106 CLL, RTL / Clear link, rotate left twice
7765 7006 RTL / twice more
7766 7006 RTL / and again (rotated left 6 bits, bit 7 is in
link)
7767 6015 / according to MP-12 docs, a command of "5"
indicates "skip if device ready and transfer data"
7770 5367 JMP 7767 / loop until data ready
7771 7420 SNL / skip on link set - end of word?
7772 5363 JMP 7763 / read next if link not set
7773 3756 DCA I 7756 / deposit at address in 7756
7774 2356 ISZ 7756 / increment address, skip if zero
7775 5355 JMP 7755 / jump to instruction loaded at 7755 from tape?
7776 2175 ISZ 175 / data (not in PROM) -- overwritten by
tape (JMP to start of routine)?
7777 5361 JMP 7761 / jump to start of routine
I haven't spent much more time with it and I've probably screwed up, but
to me it looks like it loads in two 6-bit quantities off tape, expecting
the 2nd of the two to have bit 7 set; this word is then stored in
memory, the address incremented and the next pair loaded. It increments
until the write address falls off the top of memory and then (i assume)
executes the data at 7776, which I assume would be a JMP to the start of
the code loaded from tape...
My MP-12 also lacks an async serial interface, I was thinking of
building some glue logic to convert from the sync (paper-tape) interface
to RS232 (probably with a small microcontroller) so I could load code
in, but I haven't gotten around to it yet...
- Josh
