mbalmer wrote on 2024-09-20, 01:53:

Horun wrote on 2022-08-31, 23:39:

I have a giant DTK PEM-3301 386. It works well but is huge :p

Sorry for the necro here, but I thought I would ask: I just got my hands on one of these boards with the original DTK BIOS, but I'm hoping to see if there's a working BIOS to compare mine to as for some reason, the BIOS process stops just as it's about to count RAM. Do you happen to have a dump of the ROMs for your board? For reference, mine are labeled as version 4.26.

Yes I am sure I still have it. Will take me a bit to dig it out and read the roms. The board is very ram picky, here is the original manual if you do not have it. There is also and Epson Apex 386 (KEEN 3304) manual that has the board....
Will add the roms as soon as I dig out the board.

Update: Ok read the roms with TL866. They are two ST M27c256b. Bios version is DTK 386 4.26 (A8994), attached the Hi and Lo plus a merged (romwak) if needed. The board only supports 9 chip 30 pin of 256k or 1Mb size, 3 chip do not work AFAIK (did not for me).
Ram can be 70nS or 80nS. Mine have 9 MT4c1024DJ chips on each, and booted fine with 4 or 8MB (in 2020, have not tested since, picture from back then with 4Mb, Dallas battery was dead)...

Horun wrote on 2024-09-20, 04:06:

Update: Ok read the roms with TL866. They are two ST M27c256b. Bios version is DTK 386 4.26 (A8994), attached the Hi and Lo plus a merged (romwak) if needed. The board only supports 9 chip 30 pin of 256k or 1Mb size, 3 chip do not work AFAIK (did not for me).
Ram can be 70nS or 80nS. Mine have 9 MT4c1024DJ chips on each, and booted fine with 4 or 8MB (in 2020, have not tested since, picture from back then with 4Mb, Dallas battery was dead)...

Awesome. Thank you, I will compare the two later this evening when I'm not at work. I agree about the board being pretty picky about RAM; it seems like the board refuses to boot entirely if the cache is bad (it doesn't appear that you can disable it).

I'd pulled and tested all of the cache SRAM and while the CY7C164s all tested OK, the two QS8888s (the tag RAM) both tested bad. When I pulled them, I tried powering the board and it got through the POST sequence to the point where it would tell me about the lack of an FDC and such, but it wouldn't produce the "press F1 to run SETUP" prompt that should've shown up when the RTC battery is dead, and the keyboard would just freeze.

I'm assuming the DTK bios requires cache to boot.

Highly unlikely a bios would require cache

There is such a large 386 motherboard, some time ago I increased the amount of RAM to 8MB (72 memory chips)

https://www.phantom.sannata.org/viewtopic.php … =756723#p756723

The computer worked at a flour mill in the USSR in 1990-1991. I got the motherboard in 2005 from an employee who worked there. The board was defective, I managed to fix it.

Horun wrote on 2024-09-20, 04:06:

Update: Ok read the roms with TL866. They are two ST M27c256b. Bios version is DTK 386 4.26 (A8994), attached the Hi and Lo plus a merged (romwak) if needed. The board only supports 9 chip 30 pin of 256k or 1Mb size, 3 chip do not work AFAIK (did not for me).
Ram can be 70nS or 80nS. Mine have 9 MT4c1024DJ chips on each, and booted fine with 4 or 8MB (in 2020, have not tested since, picture from back then with 4Mb, Dallas battery was dead)...

OK, wow, there's some substantial difference here. They are both the same "revision" but date-code wise they're almost a full year apart and when viewing them with HxD they are VERY different.

The datecode on my ROM reads 11/03/89 while the datecode on yours reads 09/13/90. I'll have to flash yours to a couple of 27C256s and see if it results in any different behavior.

Interestingly, the PEM-3301 appears to take the MR BIOS designed for a discrete "generic" 386 just fine.

Actually the DTK release date is 12/05/1990 (is at end of bios and part of the # string). Makes me wonder if the 36 is a internal code # for the board....
The internal DTK bios utilities (calculator, head park, etc) is dated 09/13/90.

Now I'm even more curious. Does your board boot all the way to DOS without any cache installed?

Hmm if weather changes will boot the board with a pulled TAG cache chip (simulating a bad one), is very dry - low humidity now.
I do not want to pull all the cache chips since they were working last boot, not so easy to replace all these old chips.
As rasz said it is unlikely not to boot with bad cache, more likely with an error but guess that would depend on the bios code.