Scali wrote:

Jo22 wrote:

Well, I'm certainly not going to disagree with you, but he hasn't got an ordinary 386.
It's one of the early models and *might* be incompatible with the VGA BIOS.

Well, the only flaw I know of in early 386 models is that the 32-bit multiply doesn't work properly, and they should only be used with 16-bit software.
The double-sigma marking should actually mean that these were tested and confirmed not to have the 32-bit flaw. Else they were marked with '16 bit S/W only':

Ah, sorry, my bad. 😅 I started with the later ones and never really cared for this issue. My point was that these chips belong to the
older generation of chips which were based off the original 386 design. Hence the double-sigma markings,
which were removed again from the later chips (like the loadall 386 instruction?).

Scali wrote:

That shouldn't be an issue for a VGA BIOS, as they're generally 286-compatible, and most of them are even 8088-compatible (some Tridents and other cards have a jumper to select 8-bit BIOS for 8088 compatibility).

Yes, I used to think the same, but for some reason the TVGA9000 cards never worked for me in 286 machines.
I remember this quite well, because I tested them in all my other PCs were they worked just fine (386/486/586).
The cause for this still bothers my as of today. The ISA timing was also normal, as far as I recall.

Scali wrote:

I'm not saying it's impossible that this VGA card doesn't work, but as I say, I think it's more likely that it's something else, such as some broken or badly seated chips on the motherboard.
That happened to my 386SX-16... It worked fine when I put it in storage. Pulled it out some years later, and it wouldn't work at all. Turned out that the first 1 MB on the motherboard was faulty. After I removed it, and ran only on the 4x1 MB SIMMs that were installed, it worked fine.

I totally agree. It's not the first time that something trivial is the cause for a big problem.
That's why I replied to this thread and mentioned the possibility of an incompatible BIOS.
But maybe it's just something else. I heard that some people successfully "fixed" their mainboards by washing them in hot(?) water. 😳
I never tried this myself, though. There are numerous videos about such tricks on youtube (some people even bake their graphics cards in the oven!)

The Trident 8900c came today.
Unfortunately the seller put sticky tape on it and I couldn't get it all off.
Any idea on how I can remove that residue on the Quadtel chips?
Or should I just leave it?
I'm attaching a picture.

The card works, but not with the 386.
When jumpered to color, the 386 doesn't beep with the 8900c installed, but nothing is displayed on the monitor.
When jumpered to monochrome, the 386 has the continuous beep. Again nothing is displayed on the monitor.
We tried the 8900c in both the 16-bit slots and the 8-bit slots.
So it looks like this 386 board has other problems -- maybe one or more of the 72 memory chips.

There's nothing unusual about this board. All early 386 boards looked like that.

I don't see anything unusual either. If we only had a manual for this. 🙁
The last idea that comes to mind is the keylock (apart from a hardware damage or wrong config).
Some odd mainboards had it working in reverse. Without a keylock attached, it wouldn't power up/unlock.

Jo22 wrote:

The last idea that comes to mind is the keylock (apart from a hardware damage or wrong config).
Some odd mainboards had it working in reverse. Without a keylock attached, it wouldn't power up/unlock.

I think if it was a keylock issue it would at least try to POST and a "Keyboard locked" or similar message would appear, that's how it works on most of the boards that I have used with this feature.

skitters wrote:

So it looks like this 386 board has other problems -- maybe one or more of the 72 memory chips.

Now that you've kind of ruled out the video card as a problem, this seems the way to go I think. Try filling only the first bank, if it doesn't work try another set of chips (unless you have bad luck, they shouldn't all be bad).

If it's not a memory issue then I am also unfortunately out of suggestions 😢 Maybe try removing every chip that is not strictly necessary (i.e. math coprocessor), reseat BIOS and KB chips and check traces for continuity?

My brother located an old monochrome video card. Then he went crawling under the house and returned with an old monochrome monitor I that I don't think I've ever seen before. According to the manual, the jumpers on the video card are only related to the printer port memory addresses, not the video.

We didn't get a clear picture on the screen, but we did get something. The horizontal sync was off and couldn't be adjusted. The picture would shift every five seconds or so but never showed anything legible. We don't know if the problem is with the monitor or the card or the board. The PC speaker was sounding off the whole while we had it connected, so the 386 was still unhappy about something. We tried the video jumper on the 386 set to monochrome, color, and with no jumper, but the monochrome was the only mode that made any type of picture.

skitters wrote:

My brother located an old monochrome video card. Then he went crawling under the house and returned with an old monochrome monitor I that I don't think I've ever seen before. According to the manual, the jumpers on the video card are only related to the printer port memory addresses, not the video.

That monitor looks really cool! 😀
And that Hercules clone card looks also interesting!
Yes, mono cards do usually not require any configuration (except that mono switch on PC/XT mainboards or mono setting in the CMOS setup).
I assume the large chips at the end of the card are for the ~64KiB of RAM ? - I haven't ever seen such chips before.
And could that SIS 23C01 be an MGA Character Generator ?

skitters wrote:

We didn't get a clear picture on the screen, but we did get something. The horizontal sync was off and couldn't be adjusted. The picture would shift every five seconds or so but never showed anything legible. We don't know if the problem is with the monitor or the card or the board. The PC speaker was sounding off the whole while we had it connected, so the 386 was still unhappy about something. We tried the video jumper on the 386 set to monochrome, color, and with no jumper, but the monochrome was the only mode that made any type of picture.

This reminds be of something! I once had a similar issue when I made myself a custom cable for my Hercules compatible monitor.
Turned out I had mixed-up the pinout for h/v-sync (it later worked when I fixed that). But anyway, this was a problem caused by my clumsiness.. 😊
I guess you've got the original monitor cable and the issue is something else. Maybe you can test the monitor+card combo on another PC first to make
sure it's not the 386 mainboard that causes trouble (just an idea).

It's not the video card, you've led yourself up the garden path by assuming it has been this whole time.

Bad RAM is going to be the culprit, or possible one of the discrete logic chips.

Nah, I meant the "new" issue with the Hercules monitor+Hercules card. 😀 There's a lot which can be going wrong, like timing.
His card is using a 32MHz xtal, while my card has a 16MHz one (doesn't have to be an issue, frequencies can be divided).
If the 386 mainboard is having trouble it could be an issue, too. For example, some 8bit card used the PC xtal for timing (not sure about this card).

The VGA thing was another matter. I could also have kept my experience to my own, but I decided to mentioned it as a matter of politeness.
In case someone wants to read more about VGA chips, there's a an excellent website called s100computers.com.
They do have a VGA Board on their site which utilizes the Trident chip. It's quite fascinating seeing this on an S-100 sytem!
http://www.s100computers.com/My%20System%20Pa … VGA%20Board.htm

Edit: The most sensible would be to just use an POST card in this case.
Some models do also have a built-in PSU tester. This should clear up the mainboard issue once and for all.

I meant OP. He jumped to the conclusion of it being a video card problem (see thread title), while skipping basic mobo troubleshooting steps.

IMO it's been evident since day 1 that the mobo has other issues.

There's a reason bad RAM is almost always denoted by long solid beeps: without RAM, the CPU can't do much, but it can perform the 2-3 necessary ops to initialise the timer and set the timer-speaker gate to 'on'.

Ah, ok, never mind. 😀 Skitters has got an AMI BIOS, so it shouldn't be too hard to find the right POST codes for it.
Even with a lifeless board I was a few times able to get at least some hex codes on such an diagnostics card.
Perhaps they can also assist to figure out at which point memory is getting bad (providing the BIOS in question perfoms such a check during POST).
They can be ordered from China for as little as $5.. It's no suggestion, though. I was just curious about current models. 😅
(Sample: http://www.ebay.com/itm/331709141010)

Here's my card. 😀

Btw, Skitters, I'm really sorry about the 8900 not beeing helpful in your situation (hope it wasn't so expensive).