Locutus wrote on 2026-06-05, 18:43:

and the POST card immediately jumps to "FF"

with no previous codes? not even when you remove KBC? thats really weird

Hi,

It is definitely weird for the Diagnostic ROM to say Protected Mode is available once then not, but that shouldn't really matter when booting as the whole process should only use real mode in a motherboard of this era.

One thing that could be tried is that this MB has a jumper to switch CPU speed modes J5(https://theretroweb.com/motherboard/manual/31607.pdf), try to put the jumper on the other position than where it is now, not a high chance of working, but if you already have the MB in an accessible position and not in a case it can be checked very quickly.

EDIT: One more thing as the diagnostic ROM seems to work, but the normal ROM doesn't, have you programmed a new ROM chip when trying out the different ROM version or reprogrammed the original one? Maybe the ROM chip has a slight fault and a new ROM chip would help.

rasz_pl wrote on 2026-06-06, 05:35:

Locutus wrote on 2026-06-05, 18:43:

and the POST card immediately jumps to "FF"

with no previous codes? not even when you remove KBC? thats really weird

Yup... weird:

Dead DALLAS:
https://youtube.com/shorts/RS4x3QH9QHM

KBC removed:
https://youtube.com/shorts/YJsC0EbStYU

With 'good' DALLAS
https://youtube.com/shorts/AV44UwH05Dc

MagefromAntares wrote on 2026-06-06, 07:05:

Hi, It is definitely weird for the Diagnostic ROM to say Protected Mode is available once then not, but that shouldn't really ma […]
Show full quote

Hi,
It is definitely weird for the Diagnostic ROM to say Protected Mode is available once then not, but that shouldn't really matter when booting as the whole process should only use real mode in a motherboard of this era.
One thing that could be tried is that this MB has a jumper to switch CPU speed modes J5(https://theretroweb.com/motherboard/manual/31607.pdf), try to put the jumper on the other position than where it is now, not a high chance of working, but if you already have the MB in an accessible position and not in a case it can be checked very quickly.
EDIT: One more thing as the diagnostic ROM seems to work, but the normal ROM doesn't, have you programmed a new ROM chip when trying out the different ROM version or reprogrammed the original one? Maybe the ROM chip has a slight fault and a new ROM chip would help.

Hi,
I'm almost sure I've tried every possible combination, including all cache-related configurations, both with the cache chips installed and removed, but without any success.
I've programmed a new EPROM to rule out the possibility of a faulty original ROM (it reads back correctly in my programmer).
I've also tested with a different CPU (an AMD running at 40 MHz with an 80 MHz oscillator).
I checked the ISA bus resistances and they are all seem within spec.
I also inspected the board with a thermal camera and everything looks normal.

¯\_(ツ)_/¯

Locutus wrote on 2026-06-06, 11:50:

Hi, I'm almost sure I've tried every possible combination, including all cache-related configurations, both with the cache chips […]
Show full quote

MagefromAntares wrote on 2026-06-06, 07:05:

Hi, It is definitely weird for the Diagnostic ROM to say Protected Mode is available once then not, but that shouldn't really ma […]
Show full quote

Hi,
It is definitely weird for the Diagnostic ROM to say Protected Mode is available once then not, but that shouldn't really matter when booting as the whole process should only use real mode in a motherboard of this era.
One thing that could be tried is that this MB has a jumper to switch CPU speed modes J5(https://theretroweb.com/motherboard/manual/31607.pdf), try to put the jumper on the other position than where it is now, not a high chance of working, but if you already have the MB in an accessible position and not in a case it can be checked very quickly.
EDIT: One more thing as the diagnostic ROM seems to work, but the normal ROM doesn't, have you programmed a new ROM chip when trying out the different ROM version or reprogrammed the original one? Maybe the ROM chip has a slight fault and a new ROM chip would help.

Hi,
I'm almost sure I've tried every possible combination, including all cache-related configurations, both with the cache chips installed and removed, but without any success.
I've programmed a new EPROM to rule out the possibility of a faulty original ROM (it reads back correctly in my programmer).
I've also tested with a different CPU (an AMD running at 40 MHz with an 80 MHz oscillator).
I checked the ISA bus resistances and they are all seem within spec.
I also inspected the board with a thermal camera and everything looks normal.

¯\_(ツ)_/¯

Hmm, this seems to be a "hard nut to crack", it is still weird for the diagnostic ROM to work but the normal isn't, I can only purely hypothesise at this point, maybe the diagnostic ROM doesn't touch the storage device interfaces and the normal BIOS tries to check to where to boot from early and the error occurs here?

I will check on this topic from time to time, in case I get a good idea, or if a solution is found because it is genuinely interesting what causes a diagnostic ROM to not only work properly but doesn't report any errors, while the normal one simply fails.

EDIT: Whoops, the diagnostic ROM does check the floppy interface, I haven't noticed it when looking at the picture at first, the read fails, but that should be also handled by the System ROM as well, as floppies were never a truly reliable storage medium, so it would be weird for a BIOS to hang because of that instead of printing an error message.

theretroweb is missing bios dump 🙁
Those shorts nicely show BIOS not outputting any post codes despite printing errors 😮 what kind of clown car bios is this 😐
edit: and not beeping!
not initializing VGA could be explained by failing very early combined with weak bios implementation, or by something being wrong with ISA bus for example failing pin B13 signal IO_WR . Without this signal ISA cards would not see any IO writes = cant initialize VGA while MDA is so basic and simple there is a high chance depending on implementation one will work from the power on in some default ok state.

Hi, I haven't had much time today, but I checked for the presence of clock signals and signals on the ISA bus with an oscilloscope.
I didn't observe anything obviously wrong.

I also carried out two experiments...

MagefromAntares wrote on 2026-06-06, 07:05:

It is definitely weird for the Diagnostic ROM to say Protected Mode is available once then not, but that shouldn't really matter when booting as the whole process should only use real mode in a motherboard of this era.

1. I booted DIFFERENT, known-good 386DX motherboard using the Supersoft DIAGROM.
The Protected Mode test produces the same result (FAIL/PASS) as on the DTK board,
which indicates a bug in the test itself.

rasz_pl wrote on 2026-06-06, 12:47:

Those shorts nicely show BIOS not outputting any post codes despite printing errors 😮 what kind of clown car bios is this 😐
edit: and not beeping!

2. I installed the DTK BIOS in a DIFFERENT, known-good 386DX motherboard and, unfortunately,
everything works as expected: beep codes are generated and POST codes appear on the diagnostic card.
https://youtube.com/shorts/AnR3EOl5sUU

This being a very late 486 chipset everything is integrated. Looking at other boards using this chipset speaker should be wired to SL82C362 directly, maybe thru resistor at most. Since you arent even getting beeps I would poke all the chipset pins with a toothpick under magnification.
This would explain all tests passing while nothing outside gets signals.

Can you post bios dump?

This Symphony 461 / 362 chipset was used on FIC mainboards a lot and I have seen dozens of failing chips over the years. Their MTTF must be the lowest of all 486 chipsets.

rasz_pl wrote on 2026-06-07, 01:33:

This being a very late 486 chipset everything is integrated. Looking at other boards using this chipset speaker should be wired to SL82C362 directly, maybe thru resistor at most. Since you arent even getting beeps I would poke all the chipset pins with a toothpick under magnification.
This would explain all tests passing while nothing outside gets signals.
Can you post bios dump?

Hi,
Checking the pins on the "spider chips" was the first thing I did during the mechanical inspection, and they all look fine.
Overall, the solder joints on this board are in excellent condition: bright, shiny silver, with no signs of oxidation or crystallization.

The speaker is working properly and produces the expected beep during the DiagROM test.

Locutus wrote on 2026-06-05, 18:43:

One step of the diagnostic tests produces inconsistent results.: "Protected mode CPU" (FAIL & PASS) :
https://youtu.be/tdcFMrdUAlw

I also found three additional BIOS versions online and I'm currently testing them.
Once I've checked them thoroughly, I'll upload everything to this thread.

majestyk wrote on 2026-06-07, 04:25:

This Symphony 461 / 362 chipset was used on FIC mainboards a lot and I have seen dozens of failing chips over the years. Their MTTF must be the lowest of all 486 chipsets.

Of course, a chipset failure cannot be completely ruled out, but I find it rather unlikely...
IMHO too many functions are working correctly under DiagROM for that to be the case.

So its only the original clown Ami bios that doesnt output POST codes and doesnt beep in early stages?

OK, I have three BIOS versions dedicated for PEM-0036Y:
- AMI
- two versions of DTK (different releases).
With all of them, the DTK motherboard shows exactly the same symptoms as with the original BIOS.

So, I put on my Sherlock Holmes hat and ran a few experiments.

I installed the BIOSes intended for the PEM-0036Y in another fully functional 386 motherboard and ran a series of tests.
The "test motherboard" works correctly with all Y0038 BIOS versions and displays POST codes on the diagnostic card.
While using DTK BIOS:
- I stopped the RTC clock by loading the crystal with a resistor. The BIOS displayed a "Check battery" message and reported an error code on the POST card. NO BEEPS.
- I removed the KBC. The BIOS displayed an "8742 error" message and reported an error code on the POST card. NO BEEPS.
- I removed the cache memory. The motherboard booted normally.
- I removed all RAM. The POST card displayed a memory error code, there were NO BEEPS, and the MDA monitor showed only a blinking cursor...

Come on... could it really be something this trivial...???
Could it be that DTK motherboard is that RAM picky?
Unfortunately, I didn't have any other 30-pin SIMMs to test with.
I've used this memory in dozens of other 286/386 motherboards, and none of them had any problems with it.
.. to be continued....

Hmm I'm watching this topic because of the case of this motherboard are genuinely interesting, and reading your last post made me remember something.

Once I had a 386 motherboard were the RAM banks were labelled in the wrong order Bank 0 was 1 and Bank 1 was 0, that MB beeped when that happened so it was not hard to figure it out, but if this MB doesn't beep when there is no memory, it might have the same mislabelled RAM bank issue that old motherboard had.

Extremely unlikely, but I think the most likely possibilities are already exhausted, so I think there is no harm in trying to move the RAM sticks into the other bank.

Locutus wrote on 2026-06-08, 10:01:

- I removed all RAM. The POST card displayed a memory error code, there were NO BEEPS, and the MDA monitor showed only a blinking cursor...

Come on... could it really be something this trivial...???
Could it be that DTK motherboard is that RAM picky?

no, because in other mobo you still had POST codes

It looks like the “mystery problem” has finally revealed itself.

As it turns out, the motherboard isn’t especially fussy about RAM after all (it happily accepts 8 of 10 quad-SIMM kits I have on hand).
It just happened to dislike the particular memory kit I use for testing.
Once a different set of SIMMs was installed, the board sailed through POST as if nothing had ever been wrong.

But before we start praising it too much, let’s see what the POST diagnostic card has to say:
https://youtube.com/shorts/c60hyn8x7V8

Meanwhile, the DataExpert OPTI-386WB motherboard with DTK BIOS:
https://www.youtube.com/shorts/HTgVkCe1B5I

Hmm sounds like the DTK board is not allowing codes on i/o port 80 to get to the Diag card even though it's bios does push it (as seen on the DataExpert)...

POST card always in same ISA slot?
I still dont understand how it wasnt beeping before with removed Dallas, but was with Diag rom, but now it is again beeping with normal bios? 😮 does it just not beep on fundamental errors like missing Dallas chip?

rasz_pl wrote on Yesterday, 04:32:

POST card always in same ISA slot?

"No, the card was inserted into different slots to rule out a faulty connector/socket.

rasz_pl wrote on Yesterday, 04:32:

I still dont understand how it wasnt beeping before with removed Dallas, but was with Diag rom, but now it is again beeping with normal bios? 😮 does it just not beep on fundamental errors like missing Dallas chip?

It appears to beep only twice:
once after the KBC, Dallas, and RAM tests complete successfully, and once again when the operating system starts loading.