root42 wrote on 2020-07-08, 09:18:

If you leave the LM393s like that you will have a bunch of baby op amps very soon... same goes for the 74LS74 and a bunch of others...

Maybe I should leave them like this overnight? Might be profitable business!

I am sad to report that after replacing each and every one of these ICs, the distortion on digital noise remains. Which narrows my troubles down to the two things I still have not changed:

1. The DSP chip: What kind of a flashing tool do I need to reflash this?
2. C65/C68 for which I have 1000V rated capacitors instead of 50V: I know that normally higher voltage capacitors should work but 1KV vs 50V is a bit much.. I wish I had thought to order these along with the ICs, now I need to wait a couple of weeks before I can order and test replacements..

Or there is an issue with my soldering that I have not found yet. Or the board is damaged somehow.

*SIGH*

I'd measure reference voltage on pin 14 of U12 (DAC). If it is not 5V then the clipping may occur. If it is not - check R38 and voltage drop circuit ZD1, C46-C48, R21. And make sure R22 and R25 are 27k.

LABS wrote on 2020-07-08, 11:09:

I'd measure reference voltage on pin 14 of U12 (DAC). If it is not 5V then the clipping may occur. If it is not - check R38 and voltage drop circuit ZD1, C46-C48, R21. And make sure R22 and R25 are 27k.

LABS, You may be onto something! I have absolutely 0V on Pin 14 and Pin 15 (Vref+ and Vref-)! Pin 13 (Vcc) is 4.92V and Pin 16 (Compensation) is -8.30V!. Now, by 'check ....' do you mean check that they are the correct parts, or do you want me to check the voltage across them?

appiah4 wrote on 2020-07-08, 12:10:

LABS wrote on 2020-07-08, 11:09:

I'd measure reference voltage on pin 14 of U12 (DAC). If it is not 5V then the clipping may occur. If it is not - check R38 and voltage drop circuit ZD1, C46-C48, R21. And make sure R22 and R25 are 27k.

LABS, You may be onto something! I have absolutely 0V on Pin 14 and Pin 15 (Vref+ and Vref-)! Pin 13 (Vcc) is 4.92V and Pin 16 (Compensation) is -8.30V!. Now, by 'check ....' do you mean check that they are the correct parts, or do you want me to check the voltage across them?

Check if R38 is measuring correctly. Is there a short to ground? Or maybe a broken solder joint? Is it the correct value? Check the ZD1, e.g. measuring its forward voltage. If it's broken it should read something funny. Is the Zener installed the correct way around?

root42 wrote on 2020-07-08, 12:32:

appiah4 wrote on 2020-07-08, 12:10:

LABS wrote on 2020-07-08, 11:09:

I'd measure reference voltage on pin 14 of U12 (DAC). If it is not 5V then the clipping may occur. If it is not - check R38 and voltage drop circuit ZD1, C46-C48, R21. And make sure R22 and R25 are 27k.

LABS, You may be onto something! I have absolutely 0V on Pin 14 and Pin 15 (Vref+ and Vref-)! Pin 13 (Vcc) is 4.92V and Pin 16 (Compensation) is -8.30V!. Now, by 'check ....' do you mean check that they are the correct parts, or do you want me to check the voltage across them?

Check if R38 is measuring correctly. Is there a short to ground? Or maybe a broken solder joint? Is it the correct value? Check the ZD1, e.g. measuring its forward voltage. If it's broken it should read something funny. Is the Zener installed the correct way around?

I measured R38 in circuit with a multimeter and it is a perfect 2.7K Ohms. It has continuity to Pin 14 of DAC on one end, and to the cathode (-) of the zener diode on the other. The zener diode is installed correctly according to the silkscreen and the stock image.

Pin 14 of DAC is not shorted to Ground in circuit.

Voltage across R38 is 5.25V.

Voltage between Pin14 of U12 and Ground is 0V.

Voltage across the Zener Diode is 5.25V one way, -5.25V the other way. Maybe I am measuring it wrong?

P.S. Thanks to everyone bearing with an electronics noob like me and still helping 😁

Try using the diode mode of your meter on the powered off card. It should show the forward voltage of the diode, something between 0.3 and 0.7V. Not sure where Zeners are, but I guess 0.7V.

EDIT: https://www.fluke.com/en-us/learn/best-practi … ital-multimeter

root42 wrote on 2020-07-08, 13:31:

Try using the diode mode of your meter on the powered off card. It should show the forward voltage of the diode, something between 0.3 and 0.7V. Not sure where Zeners are, but I guess 0.7V.

EDIT: https://www.fluke.com/en-us/learn/best-practi … ital-multimeter

ZD1 Forward Bias: 0.692V
ZD3 Forward Bias: 0.544V
ZD2 Forward Bias: 0V <- There is continuity between the two ends: Failed short? It is in parallel with C44 so I guess either ZD2 or C44 is faulty? Could this be the cause? They are on the -5VA lane and that is actually used in the Current to Voltage Converter and Antialias Filter, right?

Measure ZD2 out of circuit, e.g. by desoldering one leg. Might be that it failed short. Cap failing is also possible, but more unlikely I would say

root42 wrote on 2020-07-08, 13:58:

Measure ZD2 out of circuit, e.g. by desoldering one leg. Might be that it failed short. Cap failing is also possible, but more unlikely I would say

I checked Pin 11 of U11 and there is no -5V there, so the -5V circuit has failed indeed. Which is strange because I am 100% certain there was -5V there when I was troubleshooting the noise on the left channel. Regardless, I completely desoldered the diode on ZD2 and it has failed short indeed. I ordered a few replacements tomorrow, it should take a few days to get here I'm afraid 🙁 will buy some replacements tomorrow and check if that solves the issues.

Question is why did the diode fail. Was it broken? Did it get destroyed by some overvoltage? I have no clue how the 5V gets created, but a shorted diode is definitely one part of the whole problem. But are there more problems lurking...?

root42 wrote on 2020-07-08, 14:57:

Question is why did the diode fail. Was it broken? Did it get destroyed by some overvoltage? I have no clue how the 5V gets created, but a shorted diode is definitely one part of the whole problem. But are there more problems lurking...?

ZD2 is on a very simple circuit and it's connected to some very simple passive components, -12V and U11 only as far as I can see. I have a few theories:

1) I remember inserting the DAC the wrong way while randomly replacing ICs at some point, which caused the DAC and DSP to overheat in an ungodly way within seconds. the DAC is connected to -12V which is also connected to the -5V circuitry so..
b) I also caused the PSU to shut down due to a short while checking voltages once, so..
c) I know that U5 was bad at first, and I replaced it later, and U5 is connected to -5V, so..

Who knows. I'll replace the Zener Diode tomorrow and we'll see if that fixes everything 😀

Going back to my +5V measurement on U12 - What am I doing wrong? There CERTAINLY is 5.25V on R38, and R38 is CERTAINLY connected to Pin 14 (Vref+), but there is 0V between Pin 14 and case Ground. Dead DAC? Not working because of no -5V Vref- on Pin15?

Another question: If ZD2 failed short, doesn't that mean -12V on the ISA slot got shorted to Ground as well, according to the schematic? Wouldn't that cause issues on the -12V rail and components that use it (U11 for example) to malfunction as well? Actually, shouldn't that have caused a more major systemwide issue?

Vref- = 0V is good as it is GND.
Firstly you should check the power rails of your machine with SnarkBarker removed. You need +12, -12 and +5 rails. -5 is not used. If they look good - remove the DAC (U12) from SnarkBarker, install the card and measure all local supply points, i.e. outputs of ZD1-ZD3 circuits. If they are ok - measure pin 14 to ground (without DAC installed). If it shows stable enough voltage around 5V then it is the faulty DAC.
Anyway you should maintain all power supplies first and do not hurry:)

EDIT: Also check your soldering iron's temperature, it may be too hot and you burned the diode.

LABS wrote on 2020-07-08, 15:38:

Vref- = 0V is good as it is GND.
Firstly you should check the power rails of your machine with SnarkBarker removed. You need +12, -12 and +5 rails. -5 is not used. If they look good - remove the DAC (U12) from SnarkBarker, install the card and measure all local supply points, i.e. outputs of ZD1-ZD3 circuits. If they are ok - measure pin 14 to ground (without DAC installed). If it shows stable enough voltage around 5V then it is the faulty DAC.
Anyway you should maintain all power supplies first and do not hurry:)

I have ZD2 removed at the moment, is this still safe to do?

Remove SnarkBarker from the machine and check PC's power rails first. On white 4-pin Molex connector black wires are GND, red is +5 and yellow is +12. Google your motherboard and find where you can measure -12V as it is absent on Molex. When, after you install new ZD2, install SnarkBarker and check local supplies on SnarkBarker with DAC removed.

LABS wrote on 2020-07-08, 15:48:

Remove SnarkBarker from the machine and check PC's power rails first. On white 4-pin Molex connector black wires are GND, red is +5 and yellow is +12. Google your motherboard and find where you can measure -12V as it is absent on Molex. When, after you install new ZD2, install SnarkBarker and check local supplies on SnarkBarker with DAC removed.

+5.1V on 5V Rail
+12.2V on 12V Rail
-11.65V on -12V Rail (I checked on the B7 pin of the ISA slot)

I will check the local supplies on the card tomorrow, thanks LABS, I really appreciate the help!

Looks good. I'd also check the DAC (when removed) in continuity mode. Pins 3, 13, 14 and 15 are in question and should not buzz against pin 2 (GND) and between each other in both ways (switching probes).

LABS wrote on 2020-07-08, 17:15:

Looks good. I'd also check the DAC (when removed) in continuity mode. Pins 3, 13, 14 and 15 are in question and should not buzz against pin 2 (GND) and between each other in both ways (switching probes).

I checked out of circuit and there is no short between any of the pins 2/3/13/14/15 both ways.

OK, ZD2 is installed. When I was shopping for a 500mW diode, the seller gave me some that looked suspiciously large, so I bought some 250mW ones as well. I came home and checked the markings vs datasheets online and the alleger 500mW ones were 1W, and the alleged 250mW ones were actually 500mW. It always pays to check, I suppose. I installed the correct diode, patting myself on the back all the way. The forward voltage drop on the installed diode was o.63V and now the short in the -5V circuit is fixed. I will check the voltages LABS asked me to later this afternoon.

Great progress! I think we are getting close! 😀

Two steps forward, two steps back.

There is certainly +5V and -12V elsewhere on the card, I can measure them on Pin 4 and Pin 11 of U11, for example. However, there is still no +5V on Pin 14 of U12. There is +5V on the +5V side of R38, but the side that is connected to Pin14 shows no voltage, it is shorted. When I remove the DAC at U12, I can measure +5V at Pin 14 of the socket.

While taking these measurements, I managed to short something. I was poking at the DAC and there was a brief and bright spark, and the computer shut down. I had to pull the plug from the wall and reinsert it before it fired back up, and when it did, the DAC was completely dead and I could not get a DSP version from the card. Replacing it with the other faulty DAC brought it back to life, but that DAC has no +5V on Pin 14 either, and the same distortion.

I also replaced the motherboard yesterday (the old one was failing and having memory issues) and after the spark incident now I notice there is an ungodly amount of interference on the left channel from every component on the motherboard. Even the DOS menu timer ticking can be heard on the left channel, I don't know why there is so much noise now - is it the shitty PCChips M560TG board I am using or did I fry something on the card? Is it a coincidence that the crackling I once had was on the left channel (I got rid of it by replacing U2) and now I have left channel issues again (time to order another IC for U2 I guess?)

Also, two different DACs from two different sources failed the same way - coincidence? Is something else on the card killing them? Could the socket be faulty?