• Please review our updated Terms and Rules here

Porting CP/M-68K to a recycled board

Plasmo

Experienced Member
Joined
Aug 29, 2017
Messages
216
Location
New Mexico, USA
A decade ago I purchased a box of salvaged single board computers based on Motorola MC68302. They were made by ADC Communication as a controller for their Soneplex communication product line. I'd had no documentations but I figured out how to turn on the power and observed the outputs of console serial port to know these are likely good working boards.

I tried a couple times to upload my own 68000 executable files, but were not successful. Eventually I wrote a boot monitor, took over the boot flash and overwrote the existing software. In the process I also figured out most of the pc board wiring. Recently I have sufficient understanding of the board to port CP/M 68K to it.

The board has 3 banks of flash memories. The first bank is 256K bytes and hold the monitor. The other two banks are 1 megabytes each. Collectively they formed a read-only disk to hold the CP/M 68K ver 1.3 distribution files. The RAM is 1 megabyte with the top 256K backed up with a super capacitor. The backed up RAM may serve as a small RAM disk. CP/M 68k resides in $15000 - $1FFFF and the Transient Program Area is from $20000 to BFFFF, or 640K--which "ought to be enough for anybody". :rolleyes: I copied the CP/M distribution files into flash and configured the BIOS for Flash disk and RAM disk and this is what I have:
 

Attachments

  • DSC_29670906_10.jpg
    DSC_29670906_10.jpg
    106.9 KB · Views: 1
  • cpm_on_mpu.jpg
    cpm_on_mpu.jpg
    87.6 KB · Views: 1
Attached are schematics of the pc board. It took quite an effort to trace through the pc board because it is a 6-layer board with power/ground layers in the center. Fortunately the inner signal layers are clearly visible so when all components are removed, circuitry can be traced visually. To speed up the process, I took high-resolution pictures of component side and solder side and overlay them in Photoshop so I can follow signal traces between component side and solder side easily. In many instances it is still necessary to use cotinuity meter to verify the connections.

Schematics were drawn using a 20-year old WINDRAFT by IVEX. Now that's vintage software!
 

Attachments

  • ADC_IO_scm_PDF.zip
    49.7 KB · Views: 1
  • adc_mpu_scm_PDF.zip
    51.1 KB · Views: 1
  • ADC_RESET_scm_PDF.zip
    11.6 KB · Views: 1
Thank you. I have 6 of them for sale at RetroBrew. They are $30 per set (MPU board, daughter board, including the CF drive)
 
I followed the link in your signature and really enjoyed your story about Kermit. I have no idea that Columbia University has played such a major role in development, porting and maintenance of Kermit.

Kermit is THE file transfer tool between my PC and the repurposed CP/M 68K hardware. I use HyperTerminal on the PC with a transfer rate of 38400. I downloaded the Kermit for CP/M-68K (actual file name is gkermit) from Columbia U's website, ftp://kermit.columbia.edu/kermit/archives/gkermit68k.tar.gz
it contains a precompiled gkermit.rel that works without recompiling (other than converting gkermit.rel to gkermit.68k for faster execution).

Motivated by your experiences in recompiling Kermit, I take a closer look at the source codes. They are all in C, specifically the CP/M68K version of C, so they should compile correctly in the native CP/M68K environment without modifications. There are four C programs plus a header file:
GKERMIT.C
GPROTO.C
GUNIXIO.C
GCMDLINE.C
GKERMIT.H
The submit file CGK.SUB is for compiling each C program. When that's done, LOGK.SUB is used to link them together. It takes about 4 minutes to compile and link on my repurposed hardware. The resulting gkermit works like the original. Very cool!
 
Very nice follow up on the Kermit angle, thanks!
Yes you are on to a subclass here of vintage.
It takes a special type of mindset and tenacity to reverse engineer an old
undocumented board and make it functional. Perhaps it will catch on!
 
Back
Top