That's a rather nice power supply made by LiteOn - a known good OEM manufacturer.
Topology is forward converter type with single transistor and probably a uc384x IC.
The caps are top quality Japanese brands. Probably won't be a bad cap issue, but let's not rule them out yet.

No worries about not having any electronics experiences. Just be extra careful when measuring voltages with the PSU open if it's plugged in the wall.

Regarding the 160V measurement - that doesn't quite sound right.... though it should be noted that measuring at the transformer isn't exactly the right place to measure either (it's high frequency AC, so you often won't get a meaningful number.)

Start by measuring the DC voltage at the two legs of each big capacitor on the primary side (these are the big, round cans that have "200V 470 uF" written on them.) Be careful not to touch any part of the metal probes on your multimeter or anything inside the case / on the PCB in that area, since the primary side is NOT isolated from the AC mains and thus can pose a shock hazard. You should get about 160-170V DC for each cap (total is 340V DC, since the two are in series.) If you seem to be getting about half of that voltage, check the voltage selector switch is correctly set for your country's AC voltage (115VAC or 230 VAC.)

Another thing to mention: some AT PSUs don't like running without a good load on the 5V rail. If you already tried the PSU in the PC and it doesn't start, you can use car/auto (12V) light bubs to put on the 5V rail. One rated for 20W or more would be preferable. MR16 halogen (12V bulbs) can also be used.

I don't see where the RIFA capacitor is (perhaps that one under the power receptacle socket?) Either way, if it says RIFA on it, you can remove it. No need to put anything in its place yet - it's an EMI/RFI filter cap for the AC mains. The PSU can work without it. It will just give off more noise on the AC lines, which can mess with AM radio (so advisable to get replaced, once the PSU is established working again.)

Anyways, report what voltages you get on the two big caps and we will go from there afterwards.

Thanks for the reply! Ok, measured voltages on those caps and confirmed 160V on each.

Check those power resistors in the bottom left. Those are probably used to bootstrap the supply and get it running. They often fail open after running hot for many years.

kingcake wrote on 2024-02-29, 04:28:

Check those power resistors in the bottom left. Those are probably used to bootstrap the supply and get it running. They often fail open after running hot for many years.

Aha! Found a .21 ohm one that didn't seem right, pulled it out and confirmed it's failed. Can't find any others that test bad - I'll add this to the Digikey order I've had on standby and hopefully this is the only thing wrong here. 🙂

Edit: or maybe I should go ahead and replace all similar value ones just to be safe?

ubiq wrote on 2024-03-01, 04:10:

kingcake wrote on 2024-02-29, 04:28:

Check those power resistors in the bottom left. Those are probably used to bootstrap the supply and get it running. They often fail open after running hot for many years.

Aha! Found a .21 ohm one that didn't seem right, pulled it out and confirmed it's failed. Can't find any others that test bad - I'll add this to the Digikey order I've had on standby and hopefully this is the only thing wrong here. 🙂

Edit: or maybe I should go ahead and replace all similar value ones just to be safe?

Did it read high resistance or low resistance? Resistors almost always fail high resistance. If you're reading low resistance out of circuitt it might be an inductor. Some look like resistors.

kingcake wrote on 2024-03-01, 04:13:

ubiq wrote on 2024-03-01, 04:10:

kingcake wrote on 2024-02-29, 04:28:

Check those power resistors in the bottom left. Those are probably used to bootstrap the supply and get it running. They often fail open after running hot for many years.

Aha! Found a .21 ohm one that didn't seem right, pulled it out and confirmed it's failed. Can't find any others that test bad - I'll add this to the Digikey order I've had on standby and hopefully this is the only thing wrong here. 🙂

Edit: or maybe I should go ahead and replace all similar value ones just to be safe?

Did it read high resistance or low resistance? Resistors almost always fail high resistance. If you're reading low resistance out of circuitt it might be an inductor. Some look like resistors.

Did not know that! And yeah, it failed low. 😑

Still, it's labeled "R202" on the PCB, so I still think it's a resistor (and I found a green guy labeled L301 with no resistance), . Thanks to your tip though, I think I've found another questionable one that's reading high.

ubiq wrote on 2024-03-01, 04:50:

Did not know that! And yeah, it failed low. 😑 IMG_0788.jpeg Still, it's labeled "R202" on the PCB, so I still think it's a resi […]
Show full quote

kingcake wrote on 2024-03-01, 04:13:

ubiq wrote on 2024-03-01, 04:10:

Aha! Found a .21 ohm one that didn't seem right, pulled it out and confirmed it's failed. Can't find any others that test bad - I'll add this to the Digikey order I've had on standby and hopefully this is the only thing wrong here. 🙂

Edit: or maybe I should go ahead and replace all similar value ones just to be safe?

Did it read high resistance or low resistance? Resistors almost always fail high resistance. If you're reading low resistance out of circuitt it might be an inductor. Some look like resistors.

Did not know that! And yeah, it failed low. 😑
IMG_0788.jpeg
Still, it's labeled "R202" on the PCB, so I still think it's a resistor (and I found a green guy labeled L301 with no resistance), . Thanks to your tip though, I think I've found another questionable one that's reading high.

The color codes say that is supposed to be 0.21R / 5%.

Green "resistors" are almost always inductors.

kingcake wrote on 2024-03-01, 04:54:

The color codes say that is supposed to be 0.21R / 5%.

Looks like 0.27 Ohm to me (red =2, purple =7, silver = 0.01 multiplier, gold =5%)... not that 0.06 Ohm difference would matter here that much.
This is very likely the Source-connected resistor for the switching MOSFET on the primary, as low resistance values are typical for this spot.
Also, these resistors are metal-oxide type (flameproof / flame resistant) and sometimes fusible, since they are used both for measuring the current going through the MOSFET and provide extra protection if the MOSFET shorts out and goes bad. Thus, if you ever have to replace one, make sure to get metal-oxide type. Carbon film may set itself on fire and arc over if there is a fault, which is why carbon film is not used for Source current detection.

The fact that this resistor is reading very low resistance means it's probably OK... though you would need a very good multimeter or one of those transistor / cap ESR testers to really check its resistance. Although very rare, I've seen these resistors go slightly higher in resistance (a few Ohms instead of 0.x Ohms), which can cause the PSU to not work. With cheaper multimeters, first short the two probes together on the lowest resistance setting to see how low of a resistance your multimeter can measure, then try measuring the resistor. Some won't go/show values properly below 2-3 Ohms. So if you have one like that, just see if the resistor shows the same resistance as if the probes were shorted together. If yes, the resistor is *likely* good. On the other hand, if the resistor seems to add a few Ohms more to the shorted probes reading, then it may not be OK.

In any case, at least the MOSFET is fine, otherwise that resistor would have blown open too. Actually, since you already mentioned that you get brief low output voltages on the PSU, this suggests the PSU is trying to turn On but likely is encountering a problem with regulation and turning Off as a result. So for this reason, I would suggest as a next step to get some 12V home or auto light bulbs (incandescent or halogen) to put as a load on the 5V rail of the PSU, then try plugging it in to see if it comes up. Again, some PSUs will refuse to turn on without a proper load (for PSUs year <~2003, that is often the 5V rail and not the 12V rail.)

If this fails, use your heat gun / hot air station (as you mentioned you have in the other thread) and try heating the little caps by the big yellow transformer. Don't over-heat them. Just somewhere around 40-50C would be ideal. Use your hot air station on the highest airflow setting and lowest temperature. Now see if the PSU comes up again. Oh and just as a warning again - be careful around that area and the big 200V caps. Since the PSU doesn't turn On, those caps might remain charged for a little longer than usual (a few minutes or more possibly.) There are bleeder resistors in parallel with them, so I don't imagine there would be any charge left in them after 5-10 minutes. But if you've just unplugged the PSU from the wall, be careful how you handle it - particularly around the components / areas mentioned above.

If heating the caps still doesn't change how the PSU behaves, turn off and disconnect PSU from the wall, the measure the following resistances:
- Black probe connected to ground (black wire) and red probe connected to 5V rail (red wire)
- Black probe connected to ground (black wire) and red probe connected to 12V rail (yellow wire)
- Black probe connected to ground (black wire) and red probe connected to -5 rail (white wire)
- Black probe connected to ground (black wire) and red probe connected to -12V rail (blue wire)
These tests will essentially look for any shorted diodes on the output rails. Report what resistances you get.

Following that would be to check the primary side auxiliary supply rail. But I'll get to that part in the next post after I see the result from the tests mentioned above.

The primary side aux. rail provides power to the PWM IC, which is basically the "heart" of the PSU. Part of this rail is supplied by two high resistance resistors from the B+ (340V DC bus on the primary) and put into a cap (one of the small ones around the 8-pin IC) so that the PWM IC on the primary can begin pulsing the MOSFET On and Off. Once it does that, the primary side aux. rail is supplied by a diode from the big transformer. Unfortunately, the top-side picture is a little blurry for me to be able to tell you exactly which components to check. If you could, try taking another picture of the components close to the big transformer and around the "KA38xx" 8-pin IC (that's the PWM controller here.)

kingcake wrote on 2024-03-01, 04:54:

Green "resistors" are almost always inductors.

If it's labeled as L-something on the PCB, it definitely should be. Otherwise, it could be a resistor... though I agree that green resistors are indeed rare.

momaka wrote on 2024-03-01, 14:20:

kingcake wrote on 2024-03-01, 04:54:

The color codes say that is supposed to be 0.21R / 5%.

Looks like 0.27 Ohm to me (red =2, purple =7, silver = 0.01 multiplier, gold =5%)... not that 0.06 Ohm difference would matter here that much.

Correct. Looked at it on a different monitor and it looks purple for sure. Seemed brown-ish on my phone.

momaka wrote on 2024-03-01, 14:20:

Looks like 0.27 Ohm to me (red =2, purple =7, silver = 0.01 multiplier, gold =5%)... not that 0.06 Ohm difference would matter h […]
Show full quote

kingcake wrote on 2024-03-01, 04:54:

The color codes say that is supposed to be 0.21R / 5%.

Looks like 0.27 Ohm to me (red =2, purple =7, silver = 0.01 multiplier, gold =5%)... not that 0.06 Ohm difference would matter here that much.
This is very likely the Source-connected resistor for the switching MOSFET on the primary, as low resistance values are typical for this spot.
Also, these resistors are metal-oxide type (flameproof / flame resistant) and sometimes fusible, since they are used both for measuring the current going through the MOSFET and provide extra protection if the MOSFET shorts out and goes bad. Thus, if you ever have to replace one, make sure to get metal-oxide type. Carbon film may set itself on fire and arc over if there is a fault, which is why carbon film is not used for Source current detection.

The fact that this resistor is reading very low resistance means it's probably OK... though you would need a very good multimeter or one of those transistor / cap ESR testers to really check its resistance. Although very rare, I've seen these resistors go slightly higher in resistance (a few Ohms instead of 0.x Ohms), which can cause the PSU to not work. With cheaper multimeters, first short the two probes together on the lowest resistance setting to see how low of a resistance your multimeter can measure, then try measuring the resistor. Some won't go/show values properly below 2-3 Ohms. So if you have one like that, just see if the resistor shows the same resistance as if the probes were shorted together. If yes, the resistor is *likely* good. On the other hand, if the resistor seems to add a few Ohms more to the shorted probes reading, then it may not be OK.

In any case, at least the MOSFET is fine, otherwise that resistor would have blown open too. Actually, since you already mentioned that you get brief low output voltages on the PSU, this suggests the PSU is trying to turn On but likely is encountering a problem with regulation and turning Off as a result. So for this reason, I would suggest as a next step to get some 12V home or auto light bulbs (incandescent or halogen) to put as a load on the 5V rail of the PSU, then try plugging it in to see if it comes up. Again, some PSUs will refuse to turn on without a proper load (for PSUs year <~2003, that is often the 5V rail and not the 12V rail.)

If this fails, use your heat gun / hot air station (as you mentioned you have in the other thread) and try heating the little caps by the big yellow transformer. Don't over-heat them. Just somewhere around 40-50C would be ideal. Use your hot air station on the highest airflow setting and lowest temperature. Now see if the PSU comes up again. Oh and just as a warning again - be careful around that area and the big 200V caps. Since the PSU doesn't turn On, those caps might remain charged for a little longer than usual (a few minutes or more possibly.) There are bleeder resistors in parallel with them, so I don't imagine there would be any charge left in them after 5-10 minutes. But if you've just unplugged the PSU from the wall, be careful how you handle it - particularly around the components / areas mentioned above.

If heating the caps still doesn't change how the PSU behaves, turn off and disconnect PSU from the wall, the measure the following resistances:
- Black probe connected to ground (black wire) and red probe connected to 5V rail (red wire)
- Black probe connected to ground (black wire) and red probe connected to 12V rail (yellow wire)
- Black probe connected to ground (black wire) and red probe connected to -5 rail (white wire)
- Black probe connected to ground (black wire) and red probe connected to -12V rail (blue wire)
These tests will essentially look for any shorted diodes on the output rails. Report what resistances you get.

Following that would be to check the primary side auxiliary supply rail. But I'll get to that part in the next post after I see the result from the tests mentioned above.

The primary side aux. rail provides power to the PWM IC, which is basically the "heart" of the PSU. Part of this rail is supplied by two high resistance resistors from the B+ (340V DC bus on the primary) and put into a cap (one of the small ones around the 8-pin IC) so that the PWM IC on the primary can begin pulsing the MOSFET On and Off. Once it does that, the primary side aux. rail is supplied by a diode from the big transformer. Unfortunately, the top-side picture is a little blurry for me to be able to tell you exactly which components to check. If you could, try taking another picture of the components close to the big transformer and around the "KA38xx" 8-pin IC (that's the PWM controller here.)

kingcake wrote on 2024-03-01, 04:54:

Green "resistors" are almost always inductors.

If it's labeled as L-something on the PCB, it definitely should be. Otherwise, it could be a resistor... though I agree that green resistors are indeed rare.

Ok!

My multimeter probes shorted give 0.2 ohms. I put that resistor back, but in-circuit it's reading 0.6 ohms.

Tried applying heat, no effect.

I didn't have anything around the house, so I picked up a couple auto light bulbs as suggested - 12V, 21W each. Hooked one up to the 5V rail (one should be good right?) and it didn't change anything. Btw, this PSU uses non-standard coloured wires, but I figured it out.

I tried to check the resistances on all the rails, but the only one I could get a good reading on was the -12V at 2.1 KΩ. The others would always give different readings and wouldn't settle, always climbing or falling, never consistently. Sometimes 15MΩ and falling, sometimes 4KΩ and rising, etc. Seemingly random.

I have another AT PSU that I was able to get good readings off, so I'm fairly confident I was doing things right.

(those values were:
5V: 180 Ω
12V: .700 KΩ
-5V: 200 Ω
-12V: 526 Ω
)

Here's that better close up you wanted:

Anyway, appreciate the help!

Oh, and I have LCR and ESR meters, if they would be any use here.

There's probably a TL431 lurking in there that should be tested, too.

Check out these videos, they are for ATX, but many things will apply here.

https://www.youtube.com/watch?v=GjZtZNssroE&l … index=7&t=1736s
https://www.youtube.com/watch?v=UyS9rCtgoXQ

ubiq wrote on 2024-03-02, 05:25:

My multimeter probes shorted give 0.2 ohms. I put that resistor back, but in-circuit it's reading 0.6 ohms.

OK, I think we can give that a pass for now.
Techincally, 0.2 Ohms from MM probes + 0.27 Ohms from the resistor should give about 0.5 Ohms... but at these low resistances, it's acceptable for the MM to show error like this.

ubiq wrote on 2024-03-02, 05:25:

I didn't have anything around the house, so I picked up a couple auto light bulbs as suggested - 12V, 21W each. Hooked one up to the 5V rail (one should be good right?) and it didn't change anything.

So you still got about 1.2V out on the 12V rail and 0.5V on the 5V even with that, or were these readings even lower?

ubiq wrote on 2024-03-02, 05:25:

Btw, this PSU uses non-standard coloured wires, but I figured it out.

Ah OK, thanks for noting that.
Looks like PG (Power Good) is white instead of orange, 12V rail is orange instead of yellow, and -5V is yellow instead of white.

ubiq wrote on 2024-03-02, 05:25:

I tried to check the resistances on all the rails, but the only one I could get a good reading on was the -12V at 2.1 KΩ. The others would always give different readings and wouldn't settle, always climbing or falling, never consistently. Sometimes 15MΩ and falling, sometimes 4KΩ and rising, etc. Seemingly random.

That's fine for now.
High resistances are OK, especially any climbing ones (it's the output caps charging.)
I was just looking to see if there are any low-resistances, particularly sub-30 Ohms, to check if there are any suspicious / potentially bad output rectifiers. That said, a good number of older PSUs - and especially cheaper ones - have low-resistance resistors on some or all of their output rails as "dummy" loads (to provide a minimum starting load for the PSU if there isn't one provided by the motherboard/hardware.) Just mentioning this, since a low resistance on the output doesn't always mean something is shorted. But if there is low resistance and there aren't low resistance resistors, that's when it's worth further investigating.

ubiq wrote on 2024-03-02, 05:25:

Here's that better close up you wanted:

Awesome, thank you, that really helps.

So it looks like the PWM IC is a KA3884... which is a pretty close cousin to UC384x series (actually, it might even be a pin-compatible / drop-in replacement.)
And what I find very interesting from this picture is that the PSU doesn't have a switch for the mains like most AT PSUs do. Instead, the switch appears to be connected to the VCC pin (pin #7) of the PWM IC, thus only enabling the IC when it is turned On. The switch appears to have black and white wires and connects to the board where it says "SW1".

With that said, use your multimeter and measure the resistance of the switch when it is in the Off position and also when it is On. In particular, it's the On resistance that's of interest here. You should get a nice low resistance, preferably under a few Ohms. Make sure the PSU is UNPLUGGED FROM THE WALL and the two big capacitors discharged (less than a few Volts across each) before touching anything with your multimeter.

While you have the PSU unplugged, also check the resistance of resistor R102. It appears to be a 270 KOhm resistor(?) It's located right in front of the leftmost big 200V capacitor shown in your top-side picture above. Also check the resistance of Zener diode ZD302 and report what resistance you get. It's close to R102 and the big 200V cap as well.

Also, while I don't think a "bouncy" switch should have any effect on the operation of the PWM IC, it still may be worth to perform the following test:
Put the switch in the ON position and then plug in the PSU to see if that changes anything with the PSU being able to turn On or not.

If not, do the the following test next: unplug PSU and have your multimeter set to measure the DC voltage between *primary* side ground and the white wire on the PSU switch that goes to the front of the case. Again, CAUTION white taking this measurement as to not touch any metal part of the MM probe, since this is the primary side (not isolated, hence shock hazard.) Primary ground is the negative (-) lead of the leftmost big 200V capacitor of the top-side picture you provided above (re-linking below for convenience):
download/file.php?id=186754&mode=view
With multimeter set and ready, plug in the PSU (or more conveniently, have it connected to a power strip with a switch) and note the voltage read by the multimeter. Report back what you get. The KA3884 IC needs at least 16V DC on its VCC pin to start switching. Since the white wire of the switch connects to the VCC pin, this will tell us if the PWM IC is getting proper VCC or not.

ubiq wrote on 2024-03-02, 05:25:

Oh, and I have LCR and ESR meters, if they would be any use here.

YES, those would actually be super useful!
I honestly wasn't expecting you to say you have such equipment.

Since you do, desolder and check the small capacitors near the KA3884 IC out of circuit. These are related to the startup and run function of the PWM IC, so a bad cap there can easily make the PSU not want to start (and in fact is a common failure on small power adapters, such as ones for routers and chargers.)

Last but not least, I see some tan glue. Just see if there is any of this glue anywhere on the PSU where it has turned brown or dark brown. This glue is known to go conductive when it does this. It was commonly used by many manufacturers back then.

I think that's about enough homework for you now. 😀 Hopefully it isn't too much and feel free to ask for more details if you need any.

kingcake wrote on 2024-03-02, 05:45:

There's probably a TL431 lurking in there that should be tested, too. […]
Show full quote

There's probably a TL431 lurking in there that should be tested, too.

Check out these videos, they are for ATX, but many things will apply here.

https://www.youtube.com/watch?v=GjZtZNssroE&l … index=7&t=1736s
https://www.youtube.com/watch?v=UyS9rCtgoXQ

I've watched this fella's vids on capacitors - I'll check these out, thanks!

momaka wrote on 2024-03-02, 10:19:

OK, I think we can give that a pass for now. Techincally, 0.2 Ohms from MM probes + 0.27 Ohms from the resistor should give abou […]
Show full quote

ubiq wrote on 2024-03-02, 05:25:

My multimeter probes shorted give 0.2 ohms. I put that resistor back, but in-circuit it's reading 0.6 ohms.

OK, I think we can give that a pass for now.
Techincally, 0.2 Ohms from MM probes + 0.27 Ohms from the resistor should give about 0.5 Ohms... but at these low resistances, it's acceptable for the MM to show error like this.

ubiq wrote on 2024-03-02, 05:25:

I didn't have anything around the house, so I picked up a couple auto light bulbs as suggested - 12V, 21W each. Hooked one up to the 5V rail (one should be good right?) and it didn't change anything.

So you still got about 1.2V out on the 12V rail and 0.5V on the 5V even with that, or were these readings even lower?

ubiq wrote on 2024-03-02, 05:25:

Btw, this PSU uses non-standard coloured wires, but I figured it out.

Ah OK, thanks for noting that.
Looks like PG (Power Good) is white instead of orange, 12V rail is orange instead of yellow, and -5V is yellow instead of white.

ubiq wrote on 2024-03-02, 05:25:

I tried to check the resistances on all the rails, but the only one I could get a good reading on was the -12V at 2.1 KΩ. The others would always give different readings and wouldn't settle, always climbing or falling, never consistently. Sometimes 15MΩ and falling, sometimes 4KΩ and rising, etc. Seemingly random.

That's fine for now.
High resistances are OK, especially any climbing ones (it's the output caps charging.)
I was just looking to see if there are any low-resistances, particularly sub-30 Ohms, to check if there are any suspicious / potentially bad output rectifiers. That said, a good number of older PSUs - and especially cheaper ones - have low-resistance resistors on some or all of their output rails as "dummy" loads (to provide a minimum starting load for the PSU if there isn't one provided by the motherboard/hardware.) Just mentioning this, since a low resistance on the output doesn't always mean something is shorted. But if there is low resistance and there aren't low resistance resistors, that's when it's worth further investigating.

ubiq wrote on 2024-03-02, 05:25:

Here's that better close up you wanted:

Awesome, thank you, that really helps.

So it looks like the PWM IC is a KA3884... which is a pretty close cousin to UC384x series (actually, it might even be a pin-compatible / drop-in replacement.)
And what I find very interesting from this picture is that the PSU doesn't have a switch for the mains like most AT PSUs do. Instead, the switch appears to be connected to the VCC pin (pin #7) of the PWM IC, thus only enabling the IC when it is turned On. The switch appears to have black and white wires and connects to the board where it says "SW1".

With that said, use your multimeter and measure the resistance of the switch when it is in the Off position and also when it is On. In particular, it's the On resistance that's of interest here. You should get a nice low resistance, preferably under a few Ohms. Make sure the PSU is UNPLUGGED FROM THE WALL and the two big capacitors discharged (less than a few Volts across each) before touching anything with your multimeter.

While you have the PSU unplugged, also check the resistance of resistor R102. It appears to be a 270 KOhm resistor(?) It's located right in front of the leftmost big 200V capacitor shown in your top-side picture above. Also check the resistance of Zener diode ZD302 and report what resistance you get. It's close to R102 and the big 200V cap as well.

Also, while I don't think a "bouncy" switch should have any effect on the operation of the PWM IC, it still may be worth to perform the following test:
Put the switch in the ON position and then plug in the PSU to see if that changes anything with the PSU being able to turn On or not.

If not, do the the following test next: unplug PSU and have your multimeter set to measure the DC voltage between *primary* side ground and the white wire on the PSU switch that goes to the front of the case. Again, CAUTION white taking this measurement as to not touch any metal part of the MM probe, since this is the primary side (not isolated, hence shock hazard.) Primary ground is the negative (-) lead of the leftmost big 200V capacitor of the top-side picture you provided above (re-linking below for convenience):
download/file.php?id=186754&mode=view
With multimeter set and ready, plug in the PSU (or more conveniently, have it connected to a power strip with a switch) and note the voltage read by the multimeter. Report back what you get. The KA3884 IC needs at least 16V DC on its VCC pin to start switching. Since the white wire of the switch connects to the VCC pin, this will tell us if the PWM IC is getting proper VCC or not.

ubiq wrote on 2024-03-02, 05:25:

Oh, and I have LCR and ESR meters, if they would be any use here.

YES, those would actually be super useful!
I honestly wasn't expecting you to say you have such equipment.

Since you do, desolder and check the small capacitors near the KA3884 IC out of circuit. These are related to the startup and run function of the PWM IC, so a bad cap there can easily make the PSU not want to start (and in fact is a common failure on small power adapters, such as ones for routers and chargers.)

Last but not least, I see some tan glue. Just see if there is any of this glue anywhere on the PSU where it has turned brown or dark brown. This glue is known to go conductive when it does this. It was commonly used by many manufacturers back then.

I think that's about enough homework for you now. 😀 Hopefully it isn't too much and feel free to ask for more details if you need any.

Ok, just rechecked:
5V rail spikes to 1.6V on power-up with no load, and roughly 0.5V with the load connected
12V rail hits about 3.6V with no load, and about 1.3V with a load

Resistance across the power switch: wouldn't really settle with it off, but around 80-100 KΩ, and 0.3 Ω in the on position. (or beeeep in continuity mode)

R102: had to lift a leg to get a good reading, but it's showing 100 KΩ (I see brown-black-yellow-silver on it, so that checks out)
ZD302: reading about 20-40 KΩ (and rising)

As expected, no change in behaviour plugging it in with the power switch on.

Hmm, looking for 16V where you directed, and only showing 12-13V. (I can hear a quiet little *click* from the PSU when I flip the switch)

I took out the 3 caps near that IC, they all checked out, so I cleaned up the yellow gunk as best I could and put them back. There's a bunch more gunk near where all the power rails are soldered in (and where I took out the fan leads) - I'll clean that up too.

Feel like whether or not I get this thing running, I'll have learned a lot! (Like, working on this is the first time in my life I've ever cared to look up resistor colour codes)

ubiq wrote on 2024-03-03, 00:36:

Ok, just rechecked:
5V rail spikes to 1.6V on power-up with no load, and roughly 0.5V with the load connected
12V rail hits about 3.6V with no load, and about 1.3V with a load

OK, so PSU really is trying to turn On, but something is stopping it.
The output rectifiers on 5V and 12V don't seem to be open-circuit, since you are getting some voltage output. However, before we fully conclude that, do a thorough visual check on them. The output rectifiers for the 5V and 12V rail will be parts screwed onto the secondary heatsink (the upper one if you are using this picture as a reference.) In particular, look carefully if the outer leads of each rectifier (rectifier: typically a 3-pin device in either case TO-220 or TO-247... google those case names to see what they look like) are not pulled out from the case. This can sometimes happen if the PSU or entire computer was dropped hard. An easier way to see if the leads are still attached is to try to lean the heatsink back and forth and look for the leads detaching. Or if you like / don't mind doing more desoldering, remove the screws at the bottom of the PCB that holds the secondary heatsink, then desolder the two rectifiers and remove the heatsink with them still attached. Should be fairly easy to do if you have a solder sucker / desoldering bulb. There may also be two other TO-220 components by the looks of the underside picture - possibly a 7905 and/or 7912 linear voltage regulators for the -5V and -12V rail respectively. With the secondary heatsink removed, you will be able to test the output rectifiers with diode test on your multimeter. Each 3-pin output rectifier is basically 2 diodes in one package, with the two outer pins being the anodes and the middle pin being a common cathode to both diodes. If you know (or google) how to check a diode, that's pretty much all you have to do. Then this will be more conclusive if the rectifiers are good or not.

Also check the -5V and -12V rail rectifiers / diodes. These will be regular single diodes, typically close to or under the secondary heatsink and connected directly to the transformer output pins. (For PSUs that use STF topology like this one, -5V and -12V rails will have their own pins from the transformer... or in more modern PSUs, there's no transformer pin for the -12V rail and instead the -12V rail is generated through the output toroid - that bigger round "thingy" with different colored wires wound around it.)

ubiq wrote on 2024-03-03, 00:36:

Resistance across the power switch: wouldn't really settle with it off, but around 80-100 KΩ, and 0.3 Ω in the on position. (or beeeep in continuity mode)

R102: had to lift a leg to get a good reading, but it's showing 100 KΩ (I see brown-black-yellow-silver on it, so that checks out)
ZD302: reading about 20-40 KΩ (and rising)

OK, good. So switch is not the issue, and looks like R102 and ZD302 are not faulty / pulling the startup VCC for the KA3884 IC down.

ubiq wrote on 2024-03-03, 00:36:

Hmm, looking for 16V where you directed, and only showing 12-13V. (I can hear a quiet little *click* from the PSU when I flip the switch)

Welp, that's a bit of an interesting result. According to KA3884 datasheet, VCC pin needs to reach at least 16V to start-up and then needs to be maintained to at least 10V for the IC to keep running. Anything under 10V (or 9V at the lowest tolerance chips) will trigger UVLO (under-voltage lock-out) condition, which happens when the IC sees a problem. But 12-13V suggests either the IC is not seeing a fault -OR- the voltage on Vcc pin never goes above 16V, so the IC is not able to start.
Either way, this another lead/path worth following.
First, check the output side of both optocouplers. These are small rectangular 4-pin components, labeled with part number "LTV817" on your PSU (right next to the big transformer.) (The generic part number here is "817", and LTV is just specific to the manufacturer who made it.) The output side is the one connected to the primary - i.e. the side towards the small resistor R610, ZD601, and jumper wire J6. To check the output side of the optocouplers, use resistance test and check across the two pins. Out of circuit, these two pins will show open-circuit. In circuit, I *think* you should see a high-ish resistance regardless of the orientation of the MM probes, depending on what components are connected to them downstream. One of these optocouplers is used for disabling the PSU in case of fault and the other is for feedback (voltage regulation of the PSU.) Can't tell which one is for what at this point, since the two heatsinks cover a lot of components in the PSU for me to be able to trace the circuit (plus those two heatshrunk Y2 safety caps on the bottom of the PSU, which cover a lot of traces in that area.) But what we are looking for is any low(ish) resistance that will stop the KA3884 IC from regulating any higher.

Further, as a general guidance, also test all of the resistors and diodes attached to those optocouplers on the primary side (components seen in this picture.)
For resistors: if resistor measures same or lower resistance than what its color code says, it's probably OK. Only if it measure over 5% higher resistance than its color code, it should be removed and re-checked out of circuit. Of course, if you have doubt about any measurement, unsolder and test out of circuit.
For diodes: if it shows normal diode voltage drop one way and not the other, the diode is likely OK. If it shows a diode reading both ways, probably best to remove or lift one side of the diode and re-test. Resistance for any removed diodes should always test really high or open (though that may depend a bit on MM used.)

ubiq wrote on 2024-03-03, 00:36:

I took out the 3 caps near that IC, they all checked out,

So no abnormal ESR or capacitance over 20% tolerance, correct?

ubiq wrote on 2024-03-03, 00:36:

so I cleaned up the yellow gunk as best I could and put them back. There's a bunch more gunk near where all the power rails are soldered in (and where I took out the fan leads) - I'll clean that up too.

Excellent!

ubiq wrote on 2024-03-03, 00:36:

Feel like whether or not I get this thing running, I'll have learned a lot! (Like, working on this is the first time in my life I've ever cared to look up resistor colour codes)

Yup, that's the idea here - even if we don't succeed in fixing the PSU (which I find unlikely, as these PSUs aren't too complex to fix, once you get the basics of how to test for faulty components and look up datasheets), you'll probably gain some good electronics troubleshooting knowledge.

Just be careful, it's a slipper slope. 😉 ATX power supplies is pretty much what got me deep into electronics repair.

momaka wrote on 2024-03-05, 10:20:

Welp, that's a bit of an interesting result. According to KA3884 datasheet, VCC pin needs to reach at least 16V to start-up and then needs to be maintained to at least 10V for the IC to keep running. Anything under 10V (or 9V at the lowest tolerance chips) will trigger UVLO (under-voltage lock-out) condition, which happens when the IC sees a problem. But 12-13V suggests either the IC is not seeing a fault -OR- the voltage on Vcc pin never goes above 16V, so the IC is not able to start.
Either way, this another lead/path worth following.

If the 16V supply is insufficient, the switchmode chip can operate in hickup mode: The bootstrap circuit charges the chip supply capacitor to 17V, then the chip starts and uses more power than the bootstrap circuit can provide. As soon as the capacitor voltage drops below 10V, the switchmode chip enters UVLO and waits for the cap to be recharged. If that cap has extremely high ESR, the voltage may drop just due to the current consumed by the switch-mode chip, so the hick-up frequency might be high enough for a meter to just read an average voltage. The 16V supply capacitor is a primary suspect in supplies that do not start, especially if they make some continous ticking noise.

On the other hand, if the run-time supply of the switch-mode chip is designed to be just 12V, the observed voltage could be normal: You don't need more than 12-13V to keep the chip happy after it started at 16V. My experience is that the switchmode chip supply circuits are typically designed to feed around 15V into the chip. As this supply circuit depends on the power supply running, we can't be sure whether the low output causes the supply to not operate properly, or the improper operation of the supply causes the 16V supply to drop down.

A lot to read and take in here! I'll probably get back to this tomorrow, just wanted to say thanks for putting in the time to help me out. 🙂

mkarcher wrote on 2024-03-05, 13:00:

If the 16V supply is insufficient, the switchmode chip can operate in hickup mode: The bootstrap circuit charges the chip supply capacitor to 17V, then the chip starts and uses more power than the bootstrap circuit can provide. As soon as the capacitor voltage drops below 10V, the switchmode chip enters UVLO and waits for the cap to be recharged. If that cap has extremely high ESR, the voltage may drop just due to the current consumed by the switch-mode chip, so the hick-up frequency might be high enough for a meter to just read an average voltage. The 16V supply capacitor is a primary suspect in supplies that do not start, especially if they make some continous ticking noise.

Right, exactly.
Which is why I suggested ubiq to check the ESR of those small caps. One of them (I suspect the Rubycon close to the transformer - 50V 47?? uF) is likely part of the bootstrap circuit.
Since that was said to be OK, the only other possibility is the main supply of the bootstrap circuit, as you suggested. But I'm having a hard time tracing that one, due to the black heatshrunk Y2 caps soldered on the bottom and also a few "hidden" components close to the transformer's primary side pins that are obstructed by the small caps in the image. Normally, there's a diode connected on one the primary side pins on the transformer that then dumps current into the bootstrap / "startup" cap. Here, I can't distinguish it as easily... though upon a 2nd closer look, I think it might be the diode close to the transformer's corner and the black wire for SW1. If that's the case, I also see a green resistor than then appears to connect to the black wire SW1. If this green resistor is open, that would stop the PSU from starting. So I guess that's more homework for ubiq now 😁 : check that green resistor.

mkarcher wrote on 2024-03-05, 13:00:

On the other hand, if the run-time supply of the switch-mode chip is designed to be just 12V, the observed voltage could be normal: You don't need more than 12-13V to keep the chip happy after it started at 16V. My experience is that the switchmode chip supply circuits are typically designed to feed around 15V into the chip. As this supply circuit depends on the power supply running, we can't be sure whether the low output causes the supply to not operate properly, or the improper operation of the supply causes the 16V supply to drop down.

Yes, correct... hence the suggestions to check the output rectifiers or look for open-circuit / bad connection on the wires of the output toroid. A shorted optocoupler on the output pins could also trigger crowbar / shut-down (since wide-open = max. duty cycle and fully-closed = min. duty cycle / disable PWM IC.) And we haven't even looked at the feedback circuit on the secondary side yet (looks like it's done by an LM393 / KA393 comparator rather than a TL431)... though I find that rarely to be the problem.

ubiq wrote on 2024-03-06, 03:45:

A lot to read and take in here! I'll probably get back to this tomorrow, just wanted to say thanks for putting in the time to help me out. 🙂

No worries.
I still enjoy fixing PSUs after all these years. So if there's someone willing to try/learn repair one, I always chime in. 😉

Got as far as taking the rectifiers out. Don't mind if it was necessary or not, it was fun. 🙂

Used diode mode on my MM. Voltage drop across the big fella was 0.16V (that might be bad?), and the little one on the left was 0.42V. Checked other diodes I saw on the board and they seemed fine. Gonna poke at it some more tomorrow.

Samsung and ST parts. Well made supply.

The big guy is a dual Schottky diode so it will have lower Vf.

Datasheet shows Vf should be around 0.15 at low current (like your multimeter uses) so it's fine.