Yes, I only have an 100W bulb. APFC gets charged (indicated by the bulb going on then off, and staying off). And then after 2-3 seconds the PSU starts doing the same noise and the bulb lights up. When it does that I imminently cut the power to save myself from a potentional hazard (as it isn't a hazard already 😅).

I have checked the 5VSB and it generates properly (5.10v).

As for "5v shorted with GND", I meant at the secondary transistor for 5v there were stray wires shorting cathode and anode, which were connecting to the 5v rail and GND directly or by some inductors.

Got enough courage to leave it on for longer, and the light bulb starts flashing after that 2-3 seconds, indicating the APFC being unhappy due to the dim bulb tester.
I think I could safely place in the E13007's, or should I fiddle with putting two 100w bulbs in series?
What do you think?

Excellent!

I think that pretty much confirms my suspicion now: it's the APFC that's causing the weird noise because of the 100W lamp. Since the 5VSB is coming up at 5V, it looks like the APFC circuit might be OK.
Try stepping up the single 100W to a 3x 100W bulbs or something like a 200-400W halogen lamp (those long tubes)... or if you have a heating element in that range (or even up to 500W), that should do as well. Before powering up the PSU with that, connect your black MM to measure DC voltage (250V scale or higher) across one of the two big caps and monitor what happens to it when you give power to the PSU. This should hopefully tell us if the APFC circuit is working properly or not. That said, note the voltage rating of the two big caps. If it's 200V, check that the voltage does not go over 200V on your multimeter. If the big caps are rated for 250V, I still don't think the voltage should go more than 200V on each cap. APFC circuits typically boost up to around 380-390V DC... and in rare cases with higher-voltage caps, up to 400V DC. Since you have 2 big 200/250V caps wired in series in this PSU, the voltage should be approximately about half of that (i.e. 190-195V DC, usually.)
Worth noting that without APFC, the voltage should be about 230 * 1.4142 = 325V total across the two caps, or about 162V across each big cap,
Careful when probing across these caps due to the high voltage. It won't jump out and grab you and shock you... but just be careful not to touch the board with your bare hands/fingers, obviously. Do that, and you'll be safe.
If you need more courage, watch Electroboom's Youtube channel.
.

Not yet.
First let's really confirm the APFC circuit is boosting up to the proper DC voltage... and also how many incandescent bulbs it would take to make it happy. Otherwise, we won't be able to test the rest of the PSU properly. Moreover, if the APFC is going over-voltage, that could result in your "new" 13007's to short out again or possibly for the primary caps to cook and eventually let out the magic smoke (and boy, do they hold lots of magic smoke! )​

I think I could also solder on some wires to the pins of the primary caps and then measure from them using my MM. And yes, I need to make sure they won't short against the case.
What do you think about this idea?

I just found out that the bulbs arent 100w. One is an 60w while the other looks to be an 45w (markings on it are gone).
I have put them in parallel but stil, the apfc is unhappy. Now I'll solder some probing cables to the caps and measure the voltages.

Both caps have more or less 154v DC on them.
Any ideas if its ok to solder in the bjt's now? Boths caps are 680uF 200V.

It also looks like I don't own any bulbs that are 100w or more and heating elements. But we can be 75% sure this PSU won't trip breakers.

Since the voltages were fairly in spec of what you've wrote, I decided to solder in the transistors, and..... IT LIVES! Well, sorta. 3.3v is still screwed (2.28v no load) and 5v is a lil too high (5.3v no load). Maybe the 5v is due to APFC struggling thanks to my protection bulbs? When the PSU is close to discharging its caps the 3.3v actually goes up to 2.89v, but then the caps discharge and everything goes to 0v. All caps on 3.3v were replaced with low ESR nichicons IIRC.
Aby ideas on how to fix the 3.3v momaka?

Actually, rereading what i wrote about 5v, it doesn't make any sense.
With an 7v 35w bulb as load the voltages go down to 5.1v, which is in spec.

As for 3.3v rail, right before the caps discharge the 3.3v goes up to 2.39v, not 2.89v. All smaller caps were either in spec of their ESR or were replaced with generic nichicons (I don't even think there are any 4.7uF and 1uF low ESR caps).
Any ideas now?

Hey, that's good news to hear.

So you managed to get the PSU running with the dim bulb still used? Or did you connect the PSU directly to the AC line?
If it's still running on the bulbs, see my suggestions further below about what to use in place of the dim bulb.

In any case, I don't think the APFC not running can affect the output voltages in this way (99% sure in this case.) Usually, when the voltage on the primary caps drops too low, either all of the voltage lines drop together or the PSU triggers its (crude) over-power protection (this being an old half-bridge platform.) Since the 5V rail is actually high with no load (and OK even with some load, as you tested with the 7V 35W bulb), there's no excuse for the 3.3V rail to be staying low... so most likely, there's a problem with one of its voltage-regulating components.

Now, if I remember correctly, my ePower PSU had a buck-regulated 3.3V rail that's directly generated from the 5V rail via a single-ended buck-converter (comprised of a small toroid, a single MOSFET, and one free-wheeling diode rectifier.) I think this is the same for your PSU, based on what I could trace from the pictures you posted in post # 5 on the previous page. See my annotated image below and confirm your components match the descriptions I put. If your 3.3V rail circuit seems to match this, then make sure not only the 3.3V rail caps are good, but also the 5V rail caps. In particular, there are 3 caps on the 5V rail: two as main filters and one after a "PI" (rod-core) inductor that's used as a pre-filter (which I annotated on the picture) for the buck-converter circuit that generates the 3.3V line. If any of these caps are bad (despite the 5V rail looking "OK"), that could be why it's not generating a proper voltage.

Also, can you read the part numbers that appear on your 8-pin PWM controller. IIRC, I couldn't find any info on mine, which was "KEL231" from ST-Micro, if I remember properly. But write down all of the numbers you see on yours, just so we can compare and so that I can look for a datasheet.

Here is the annotated picture:
https://www.badcaps.net/filedata/fet...0539&type=full

Yes, 100W is not really enough for many APFCs. Some will eventually "catch up" to the right voltage and then the bulb will dim down again in 3-5 seconds.

154V is a bit on the low side... but perhaps because your AC line is a little lower too (220V AC)?

I would have waited with the new BJTs until the APFC situation is sorted, just in case... even though that doesn't seem to be the problem here anymore.
In general though, when troubleshooting electronics, it's best to tackle circuits one at a time. That makes it less likely for one part to affect another part of the circuit and possibly create confusion.

No hair dryer? Hot air / heat gun? Old toaster? Smaller space heater (preferably 1 kW or 1.5 kW max).
Most hair dryers and heat guns are usually rated for more than 1 kW of power (1.2 to 1.6 kW being typical)... but on the "low" setting, they should be equivalent to around 600-900 Watts... which comes out to around 2 to 4 Amps AC - a still reasonable short-circuit current limit, I'd say.
Toasters are typically in the 700-1000 Watt range... so the allowable short-circuit current will be a little higher at around 3 to 5 Amps, which again is still reasonable.
Same goes for smaller space heaters rated for 1 kW (1000 Watts) or less. With anything more, particularly above 1.5 kW, the allowable short-circuit current can be 7 Amps or more, which at that point may be high enough to destroy weaker parts, like 5VSB MOSFETs, smaller NTCs, under-sized bridge rectifiers... and even the fuse in some case! But since there is an actual limit on the current, the destruction likely won't happen fast or be as "violent" as connecting the PSU / device under test directly to the AC line.

So if you have any of the above listed appliances, consider using those any time the dim bulb tester isn't enough.

The 8 pin IC you've marked is an AMC34063AN (bottom markings) and with an 0407J top marking (manufacturing date?).
Those "3.3v rail filter caps" you've marked are actually on the 3.3v rail, although no idea why they are 10v instead of 6.3v (cost cutting maybe?). The 5v pre-filter is an 2200uF 16v which I have replaced with an 1000uF 16v low ESR nichicon for testing (an 2200uF 16v ltec cap with low esr also landed me the same results). As for the diode its an SBL2040CT schottky diode rectifier (I think) and the mosfet is an P80NE03L from ST micro electronics (W1Z116, MOROCCO (?), P80NE03L full markings from top-bottom). 3.3v buck converter inductor is on the 3.3v rail.
As for the low primary cap voltage, I think the cause could be the fact ac voltage is kinda low here. I think it was around 220-210AC (electrical company here in some seasons had a crap ton of power delivery problems )


Another idea would be to just ditch the 3.3v entirely and drop in an buck converter on the 5v rail, but lets be honest, this would be an worse fire hazard than the excellent power, especially since this PSU is packed to the brim with components. Actually, there's no room for an extra one.

Okay so from the TRW discord I got info that this PSU's manufactuer should be Top Power and the 3.3v has its own transformator. So now I have no idea where to go next besides putting test leads on the cathode of the diode and drain/source of the MOSFET and seeing if the 3.3v is present there...
The gate of the MOSFET seems to connect directly to the cathode of the SBL2040CT diode.

Ah, finally! I couldn't find a datasheet for mine based on its markings, but based on this, it looks like the 3.3V rail PWM controller is an IL34063A. Datasheet attached at end of this post.

OK, that confirms it then that it's the same circuit as in my ePower PSU.
I just had you do this because I used the solder-side picture of my ePower PSU to trace the parts, since yours wasn't clear enough to trace (hint: you could use more light in your shop / work area, I think. )

Right, but since your PSU has APFC, the line voltage becomes irrelevant, because the voltage should then be boosted to around 370-390V DC.
So again, when you tried to test the PSU, was this still with the series dim bulb(s) connected or without? And if the dim bulb is connected, how does the APFC deal with it?

That's exactly what you have there right now. The components I annotated are all part of that circuit. Instead of replacing it with a new one, let's try to troubleshoot this one.
First, check all of the resistors, diodes, and small caps directly connected to the 3.3V rail PWM controller - i.e. the IL34063A.
Also check the two small TO-92 transistors that drive the 3.3V rail MOSFET. These 2 transistors are located where the 2 small transformers are in the middle of the PSU. There are actually 4 small TO-92 transistors total in that area - two closer to the edge of the board, separated by three diodes between them (the BJT drivers for the main power supply) and the two transistors closer to the big main transformer (right above where I put the text "5V rail pre-filte cap"). The ones that drive the 3V3 rail MOSFET are the latter ones - i.e. the two closer to the main transformer. Pull them out and check them out of circuit. Also check any resistors connected to them. And very important: pull out and check the three small electrolytic caps near those transistors.
Next, power up the PSU and check the voltage on pin # 1, 6, 7, and 8 of the IL34063 PWM IC. Pins 1, 6, and 7 are all tied together, so should measure the same voltage. Pin #8 I believe is pulled up to the same voltage as pins 1, 6, and 7 through a resistor... but still check its voltage and report all of those back here.
Also check the voltage on pin #5. This is the feedback pin that senses the voltage on the 3.3V rail. When the 3.3V rail is at 3.3V, you should see about 1.25V on it.

From what I've been able to trace so far... well, this 3.3V rail buck-regulator circuit is quite weird.
If you look at page 7 of the attached IL34063A datasheet, it shows a sample step-down buck regulator circuit, but with "external NPN switch" rather than a MOSFET. So the circuit in this PSU will be quite similar to that, but there are some notable differences - one quite obvious that the "NPN switch" is a MOSFET in this PSU. And the MOSFET Gate is not driven directly from pin # 2 on the IC, but rather through a pair of BJT transistors, as noted above. There's also no equivalent of the Rsc resistor in the circuit of this PSU (as evident from pin # 6 being tied to pin # 7 and 8 of the IC), implying that I_peak current detection is not used here.
Of most interest, though, is the feedback on pin # 5. I haven't been able to trace it out completely (and I've spent almost 1 hour on this now), but it looks like part of the primary side current-detection circuit for the main PS via the small middle transformer also feeds back some of its signal into the feedback pin of the IL34063A. This MAY BE why the 3.3V rail is low and why it goes higher once you remove power from the PSU. And of course, this could be due to the APFC circuit not running... so we still need to figure that out too.

Anyways, do the above component tests I mentioned and report back the voltages I asked so that hopefully we can get a fuller picture of this circuit.

And if bad comes to worse and there's no easier way, it may be possible to convert the 3.3V rail buck-regulator circuit to a linear regulated one, since you already have the MOSFET there. But let's leave that for now - I'm just jotting it down here as a thought/idea, in case I forget it later.

Correct, this is indeed a Topower PSU, as is my ePower PSU.
No, the 3.3V rail doesn't have its own transformer. It just has a step-down (DC-DC) buck-converter circuit, which takes power from the 5V rail in order to generate the 3.3V rail. This PSU is from the days before DC-DC module boards were popular, hence the weird implementation of the 3.3V rail buck-converter circuit with the IL34063A PWM IC.

No, not quite.
5V is present on the Drain of the MOSFET (middle pin). The Source is the output pin and connected to the common Cathode of the SBL2040CT diode and also to one side of the 3.3V rail toroid. The Anodes of the SBL2040CT are tied to Ground.
How it works: MOSFET turns on (allows current to pass from Drain to Source) for a short moment and current starts passing through the 3.3V rail toroid. The MOSFET then closes / turns Off. Because inductors don't like sudden changes in the current passing through them, the 3.3V rail toroid then pulls current up through the SBL2040CT diode from ground. Yes, I know that sounds very strange, but that's how all step-down buck-regulator circuits work. If you think of the 3.3V rail toroid as a heavy flywheel that is turned when current passes through it from the MOSFET, then when the MOSFET turns off, the flywheel will still want to spin and creates "suction" (due to its rotational energy) which can suck up current from... anywhere. With the MOSFET turned off, the only place where the inductor can take current from is the SBL2040CT diode. This is why the SBL2040CT in this circuit is called the "free-wheeling" diode/rectifier.
What's also interesting about buck-circuits is that we input square pulses of current through them (when the MOSFET turns On and Off sharply), but the output at the other end of the inductor is a triangular wave (with DC offset)... which with the help of the filtering caps is smoothed out to become nearly DC voltage.
And that's how buck-regulator circuits work, more or less.​

Those 2 small transistors near the big transformer (both Q5) are good. One has 898mV and 375hFE while the other (c945p45c) has 733mV and 247hFE.
All of the resistors measure okay which are connected to those transistors. The three caps were replace with generic purpose caps (47uF 50v with sme 6(k)1f with 4.5ohm esr and the 1uF 50v with nichicon VX(M) with 3.5ohm esr).
PIN 1 of pwm ic has 12v, PIN 8 has 8.27v, PIN 5 0.79v, PIN 6 11.56v and PIN 7 11.51v. I am suprised at how low pin 8 is
And now I run this psu through an 750-800w toaster I completely forgot about so APFC shouldn't be an issue now

I also cracked D25 in half but it still measures fine in circuit and I don't know if it has two black bands or are they 4148 markings so I left it as is. Besides before I cracked it I got the same results on the 3.3v rail. It also hasn't change after adding the toaster.

I saw something weird while I was double checking the caps and another 4148 diode cracked in half. This time near the 3.3v PWM IC. Replaced it and the other one which cracked at the begining and.... 3.3V IS FINALLY 3.4V!
Man was I wating for this moment. I'll put some load on the rails and double check if everything is okay and report back the results. I just hope no other line screwed itself up now.

12v seems to be off under load.... Normally its 11.50v but with an 12v 35w bulb it goes down to 10.8v. Any ideas why is that?

Well, if it's cracked, better replace it just in case. We don't want to introduce new problems.
Any diodes on the board without specific markings to show they are Zener diodes (e.g. ZD***), then they are more than likely regular in4148 diodes.

*EDIT*
I see you got to the bottom of the problem, finally. Nice!
This is why you don't ignore bad or even "questionable" parts.

What are you using as a load on the 5V rail?
This is an old design PSU that still favors a semi-large load on the 5V rail for this to regulate.

Also, are you still testing with the toaster in series on the AC line? If so, that might be because the AC line is dropping too low when you use the 35W bulb.
Or it could be because the APFC is not running... though I think that's less likely to be the case here, as this PSU appears to be exactly built like mine, so probably shouldn't care if there is APFC or not. Of course, we can never truly really know, since only the main transformer could have been changed with one with higher primary turns ratio to accommodate for the higher voltage of the APFC.

So one way or another, I think we will need to have that APFC circuit up and running.
While still using the toaster, what voltage do you get across both primary caps (you can either sum the total of the two, or attach red probe to the + on the "upper" primary cap and black probe on the negative of the "lower" primary cap, since they are in series, so that will give you the total voltage in one measurement.) To be more specific, you need to measure this voltage when testing the PSU with the 35W bulb on the 12V rail, to see what's happening.

On the 5v rail currently it's nothing. I was doing a quick test with that bulb on each of the rails too see if they are in spec.
I am currently testing it without the toaster since it looks to be interfering when I stress test the lines. It powers on slightly when I put load on the rails (I can hear it "hum").

Do I still need to do the testing on primary caps without the protection?

Scratch that I used an real load (lga775 shitbox mainboard + two sata 3.5" drives) and the votlages are within spec. I guess I'll do some more test just as a sanity check before calling this repair done.

No problem, glad to hear it's up and running.

So looks like really the 3.3V rail was the main and only problem here the whole time.

BTW, 1n4148 is *not* a Zener diode. It's just a regular small signal diode.

So that's 1 out of 3 for the count... with the FSP being almost there.
.