I'll ask the obvious, are you sure you are wiring the speaker properly?

This is from the manual:

Are you sure your speakers/buzzers work properly? Did you try them with other computers?

There's also the possibility that the cables could have gotten broken or desoldered.

The speakers and beepers are fine and work in another pc.

I've bought myself a multimeter today and beeped the board out partially.

Pin 1 is connected to Q3 via R33. Looking at the connections it should be a NPN transistor. Using the diode test on this cheapo multimeter, there's no voltage drop between the base and the emitter. So now I'm pretty sure that that transistor is the problem.

I soldered in a new smd transistor last night and it didn't make any difference

irisgrunn wrote on 2024-02-21, 09:39:

I soldered in a new smd transistor last night and it didn't make any difference

Do you have a scope? See if that transistor base is getting a signal at all. You can also use an LED with some 1k resistor (so pick a bright one) and it should light up somewhat when the PC speaker is being used. Just don't connect any speakers directly to the transistor base.

I actually have a scope, just no probes or BNC connectors or anything, I was already pondering about how I could rig something up. An LED seems like a good idea to figure out if it's even getting a signal.
I've attached what I measured, unfortunately after that the trace disappears behind one of the ISA slots and I can't figure out where it's going. I guess the best place to put the 1k resistor and the LED would be before the 4k7 resistor?

irisgrunn wrote on 2024-02-21, 12:32:

I guess the best place to put the 1k resistor and the LED would be before the 4k7 resistor?

Since you located that resistor, you can connect the LED directly to it and skip the 1k resistor (but that will require temporary removal of the transistor). Keeping it is safer (in case of picking a wrong solder point) and the added resistance is not so great that the LED will not work. If you don't want to mess with the soldered transistor anymore then connect the LED to before the 4k7 and then obviously the 1k resistor is mandatory. Since it's an NPN, connect LED cathode to GND and the anode (via 1k) to the signal source. This way the LED should not be on with no signal - if it's always on then either there is a problem with the signal driver or the transistor was not NPN but perhaps PNP.

LED isn't doing anything, so it's not getting a signal at all. Really weird because otherwise the motherboard works fine.

Do ISA cards with a PC speaker exists? Maybe I could bypass all of this.

If you have a meter with a good, fast beep on short function you could try to find out where the signal comes from by sweeping over the chipset pins with the meter probe. I did a lot of that for various mobos and never damaged anything. With the 4k7 resistor in there I don't see how any decent output could get damaged, unless somebody created a short before the resistor somehow. Did you check that BTW? Is that side of resistor showing any resistance to GND or 5V or 3V3? Pretty much all CMOS chips have protection diodes on their pins so you should be able to detect such a diode - otherwise I would suspect a broken trace.

As for speaker on ISA, the problem could be the chipset is not passing the I/O to the PC speaker related ports (PIT, gate mask) to the outside world. It's often the case these functions are integrated into the chipset and only happen there - so the ISA card will not see any I/O traffic and simply sit there and do nothing.

Deunan wrote on 2024-02-21, 14:26:

If you have a meter with a good, fast beep on short function you could try to find out where the signal comes from by sweeping over the chipset pins with the meter probe. I did a lot of that for various mobos and never damaged anything. With the 4k7 resistor in there I don't see how any decent output could get damaged, unless somebody created a short before the resistor somehow. Did you check that BTW? Is that side of resistor showing any resistance to GND or 5V or 3V3? Pretty much all CMOS chips have protection diodes on their pins so you should be able to detect such a diode - otherwise I would suspect a broken trace.

I actually managed to follow the trace until it goes underneath the chipset IC (BGA unfortunately), trace seems fine. It measured like 80k to ground, to me that seems fine too.

Well, I was rather hoping the PC speaker output would be handled by something ISA-related, like the UM8670F chip. Or at least I think this one is some sort of Super I/O and ISA bridge, there isn't much info about it. But if the problem is with the main BGA chip then I'd say you are out of luck.

One last idea I have is to try adding a pull-up of 4k7 to 3V3 for this signal (3k3 or 5k1 will work too, it's not super critical, just keep the current low). Add it before the base resistor. This is to test if perhaps the output on the chipset is a totem-pole type and the upper transistor got blown but the lower one still can drive the line to GND. In this case the pull-up will supply the necessary power. Never had much luck with such experiments but it's worth trying.

Deunan wrote on 2024-02-21, 15:52:

Well, I was rather hoping the PC speaker output would be handled by something ISA-related, like the UM8670F chip. Or at least I think this one is some sort of Super I/O and ISA bridge, there isn't much info about it. But if the problem is with the main BGA chip then I'd say you are out of luck.

I've actually found the SiS5598 documentation and has a speaker out on A20. So either something is wrong with the chip, or underneath the chip. In both cases it's out of my league.

Deunan wrote on 2024-02-21, 15:52:

One last idea I have is to try adding a pull-up of 4k7 to 3V3 for this signal (3k3 or 5k1 will work too, it's not super critical, just keep the current low). Add it before the base resistor. This is to test if perhaps the output on the chipset is a totem-pole type and the upper transistor got blown but the lower one still can drive the line to GND. In this case the pull-up will supply the necessary power. Never had much luck with such experiments but it's worth trying.

Tried that and it didn't help. I guess this board is foobar, thanks for your suggestions though!

I remember there is at lest one isa sound chip snooping on ISA bus traffic and emulating pc speaker internally