I guess the first thing is check all voltages, see if there is a short to confirm your potential tantalum failure hypothesis.

If the voltages are OK I'd then be getting as many datasheets for the ICs as I could to see if you can probe the pins for sticking reset or using a logic analyzer (or better, an oscilloscope) to check that signals are as expected.

Once you've checked the basics on the card (correct voltages, no shorts) a good next step would be to determine if the video BIOS on the card is readable / intact. My go-to method for this is to just load up DEBUG.EXE and use the "D C000:0" command to dump the first 128 bytes of the VBIOS. I usually use an MDA card for this purpose, but you could also use a CGA card. (In this situation, configure the motherboard switches for MDA or CGA, depending on which card you have installed.)

When you configure the switches on a 5150 for EGA/VGA display, what you're actually doing is telling the BIOS there is no display attached. If for some reason the option ROM on the EGA card isn't readable, the video BIOS will never be run, and won't get a chance to POST and initialize the card. You won't get any error beeps in the scenario. If the option ROM is readable, the video BIOS will run a very thorough POST check of the EGA hardware and it *will* complain loudly if it finds anything wrong.

Very early revisions of the 5150 BIOS (the ones on the 16-64K system boards) did not support option ROMs at all but I think it's unlikely that your machine has one of these early revisions.

kdr wrote on 2023-05-18, 09:15:

Very early revisions of the 5150 BIOS (the ones on the 16-64K system boards) did not support option ROMs at all but I think it's unlikely that your machine has one of these early revisions.

In that case it would also be unlikely that the EGA card was in there, as it never would have worked in that configuration.
Unless the machine was just randomly thrown together from parts before selling.
But indeed, it doesn't hurt to check the BIOS date of the machine, and replace it with an updated one if required.

paradigital wrote on 2023-05-18, 07:22:

I guess the first thing is check all voltages, see if there is a short to confirm your potential tantalum failure hypothesis.

If the voltages are OK I'd then be getting as many datasheets for the ICs as I could to see if you can probe the pins for sticking reset or using a logic analyzer (or better, an oscilloscope) to check that signals are as expected.

That makes sense, yes.

kdr wrote on 2023-05-18, 09:15:

Very early revisions of the 5150 BIOS (the ones on the 16-64K system boards) did not support option ROMs at all but I think it's unlikely that your machine has one of these early revisions.

Scali wrote on 2023-05-18, 12:01:

In that case it would also be unlikely that the EGA card was in there, as it never would have worked in that configuration.

Yes, the machine has a 16-64K motherboard, but it also is equipped with the latest BIOS. It's clearly been upgraded significantly since it came with 640K of RAM through AST cards, the EGA, a 135 watt power supply, and a combo floppy/HDD controller.

One PCB trace on the back (behind U21) looks suspicious, test it for continuity. There are also some other scratches but those don't seem too bad.

The plot thickens. This card works perfectly fine in my IBM PC/AT. HWINFO16 reports "EGA Color" and 64 KB of RAM. A little test program I'm working on that uses graphics mode 0Dh works fine.

When in the 5150, and the jumper settings set according to the IBM EGA manual, I get the one long + two short beeps, indicating video BIOS ROM error. Not sure why I wasn't getting that before. I might have set the DIP switches wrong.

There are some switches you must set on both the PC & XT system board when using the EGA on the older systems.
Page 79 in the link below.

https://www.minuszerodegrees.net/oa/OA%20-%20 … s%20Adapter.pdf

(Memories...that EGA card was ordered with the first IBM/AT @6MHz , no less, that I got at work.
Except it was delivered 6 months after the AT itself, so I had to get a cheaper 3rd party board with 256kB RAM
meanwhile, and got Windows 1.0 bundled with it too.
The IBM board was eventually used to test some software gfx compatibility by another team, once it arrived.
As I recall, the IBM ram expansion boards for 64k->256k cost about 5000 SEK at that time, never got those.)

Rwolf wrote on 2023-06-02, 19:00:

There are some switches you must set on both the PC & XT system board when using the EGA on the older systems.
Page 79 in the link below.

https://www.minuszerodegrees.net/oa/OA%20-%20 … s%20Adapter.pdf

I do have those set. Without them, nothing happens, no beeps.

More progress.

The 5150 produces the one long + two short beeps with both the IBM EGA card and a different, known working clone EGA card.

When I remove my network card (a Racal Interlan NI5210C-10BT), the system boots and displays video with both EGA cards.

That network card is not configured to conflict with either the I/O address or the video RAM of the EGA.

I'm a little confused as to why I get that 1long+2short beep no matter what.

It could be the Network card has a simplified IO address decoder, there is sometimes not a full address decoder implemented in the hardware, so extra address ranges could be overlapping without being properly documented - I've seen that on some boards. Any known conflicts could perhaps be found on that website as well.

looking at the Racal board, you have both IO and memory mapping

I guess you have this manual too; if not, check page 2-4 regarding EGA card memory mapping conflict:

https://archive.org/details/racal-inerlan-10b … up?view=theater

ps. you've got a TP-adapter for that card, right? That cable connector is a bit unwieldy.

Possibly, yes. The network card also has a BIOS ROM on it. Maybe that is conflicting.

I have to wait until next week to work on the 5150 more, though, since I have some parts coming in the mail for it.

My previous post was unclear. Here are the failure states so far:

1. IBM EGA or clone EGA inserted, no network card, DIP switches set correctly on cards and motherboard: one long and two short beeps, system boots and works regardless.
2. IBM EGA or clone EGA inserted, network card inserted: one long and two short beeps. No video display.

Removing all of the other cards does nothing to fix the beeping. Both EGA cards work fine with no beeps or issues in my PC/AT 5170.

Using EGA on that PC requires a specific BIOS version; dated 10/27/82
There is a note about a bug in this BIOS with under-populated RAM banks which can cause the beeps, at the very bottom of the page:

https://www.minuszerodegrees.net/5150/misc/51 … ch_settings.htm

But I guess you know all this.

Yes, I have the correct BIOS version. Everything that's in the docs on minuszerodegrees I have followed as far as I am aware.

There is a simple fault that can be checked easily with a tester; the jumpers themselves can have poor connection due to oxidation or cracked solder joints / cut copper lines. If so, you could have an explanation to the beeps.

keenmaster486 wrote on 2023-06-02, 20:53:

Possibly, yes. The network card also has a BIOS ROM on it. Maybe that is conflicting.

That's exactly what is happening. 8 bit ethernet cards would often allow you to set the UMA window for the onboard boot ROM to anywhere in the upper address space (including C000 which is where all video option ROMs live) since they expected to be used along side MDA/CGA cards with no video BIOS

maxtherabbit wrote on 2023-06-04, 14:52:

That's exactly what is happening. 8 bit ethernet cards would often allow you to set the UMA window for the onboard boot ROM to anywhere in the upper address space (including C000 which is where all video option ROMs live) since they expected to be used along side MDA/CGA cards with no video BIOS

You may indeed be right. But I missed this in the Racal card's manual as posted by Rwolf earlier:

That was dumb of me.

I'll have to try this again.

keenmaster486 wrote on 2023-06-06, 19:31:

You may indeed be right. But I missed this in the Racal card's manual as posted by Rwolf earlier: Screenshot from 2023-06-06 13- […]
Show full quote

maxtherabbit wrote on 2023-06-04, 14:52:

That's exactly what is happening. 8 bit ethernet cards would often allow you to set the UMA window for the onboard boot ROM to anywhere in the upper address space (including C000 which is where all video option ROMs live) since they expected to be used along side MDA/CGA cards with no video BIOS

You may indeed be right. But I missed this in the Racal card's manual as posted by Rwolf earlier:
Screenshot from 2023-06-06 13-29-15.png
That was dumb of me.

I'll have to try this again.

Ah ok so it wasn't the boot ROM, but a memory mapped packet buffer. Same result in either case, I was still right about the cause being a C000 memory conflict 😀