kotel wrote on 2024-08-25, 12:33:
... an blown resistor which measured 105 ohms in circuit but when I pulled it out it measured only 8.7 ohms?? (replaced it with an beefy 100 ohm resistor, which measures 100 ohms in circuit)).
Resistors *don't* go low-resistance when they fail. They go either high resistance or open-circuit (infinite resistance, so to speak.)
So if you measured 105 Ohms in circuit but 8.7 Ohms out of circuit, that means your measurement was not done correctly somewhere. Your multimeter should always show the lowest possible resistance of the circuit.
Depending on the circuit, it may actually be normal for a resistor to measure lower resistance in-circuit (due to other resistances and/or alternative parallel paths in the circuit.)
The only time you can definitely say a resistor is bad while doing an in-circuit reading is if the resistance reads much higher than the rated resistance of the resistor (i.e. if you have a 100 Ohm resistor and it measures 200 Ohms in-circuit, it's probably bad... unless there are batteries or charged capacitors in the circuit, which can skew those readings.)
kotel wrote on 2024-08-25, 12:33:
I know there are more crappy caps + these PSU's are trash due to their awful regulation (atleast from what I was told on TRW community server), but I'd like to atleast try to make it work.
FSP PSUs have pretty decent regulation, so not sure what the TRW community is raving all about. Perhaps compared to a modern PSU, sure. But even then, these old FSP PSUs are still well within ATX spec.
As a side note, this is why I stopped reading PSU reviews online a long time ago (when there were still any.) They would often berate (sometimes too harshly) PSUs with average regulation, yet completely overlook shitty capacitor brands in "top-performing" PSUs (and even praise them) so long as they had very tight regulation and ultra-low ripple. I get that good regulation and very low ripple is important, but I think longevity / long-tern reliability is also quite important. It's even worse when they overlook cheap caps on the primary side of a PSU with APFC circuit - those regularly get trashed by the APFC after some years of use and eventually make the PSU fail in a spectacular fashion.
/ PSU reviews rant
kotel wrote on 2024-08-25, 12:33:
The PSU doesn't wanna start when I short PS_ON with GND too. Any ideas what might be the culprit?
No point in shorting PS_ON to ground if 5VSB is not there.
1st test on an ATX PSU is to always check for 5VSB.
... and not only without a load but also with a small load too (~100-250 mA would be a good start, 250 mA to 1 Amp would be better.)
I keep a few car/auto 12V lamps for this. One is a 12V 20W and the other is a 12V 5W. The 5W one draws about 180 mA and the 20W one about 1 Amp. I've soldered wires onto the 5W one so that it plugs easily into the 20/24 pin ATX connector between the 5VSB wire and ground. Shove it in there and give AC power to the PSU. If the 5VSB comes up, I can see the bulb light up nicely. I also check the voltage under this load too. Should still be 5V or close. I ran through quite a few PSUs in the past that would output 5V "fine" on the 5VSB without any load, yet completely die and output almost nothing (or only partially output) even with the 5W bulb.
So anyways... that's test #1.
If test #1 fails, then the most common failure would be that either one of the small "startup" electrolytic caps on the primary side that is bad or the output caps on the 5VSB are bad - at least if the PSU has non-Japanese caps. If the PSU uses high-quality Japanese caps, then the issue with the 5VSB would be elsewhere (bad secondary rectifier, damaged driver/generator IC if any, open strap supply resistor, bad primary auxiliary winding rectifier, and etc.)
lti wrote on 2024-08-25, 17:34:
I thought those were decent power supplies, but they always have that same resistor cook. I don't think I've seen one with an intact resistor.
If I remember correctly from a somewhat similar FSP PSU I recapped many years ago, that 10 Ohm resistor is a "dummy load" resistor on the 5V rail so that the PSU can start if there is no load (or not significant load) attached to it, particularly on the 5V rail.
When O/P changed that 10-Ohm resistor with a 100-Ohm one, that could be why the PSU is not starting anymore... at least without a load.
On my FSP PSU, I also removed that resistor, though. Can't remember what I replaced it with, but it was probably either 2x 100 Ohm in parallel (for 50 Ohms equivalent) or something along those lines. I don't test my PSUs without any load, so I often remove those dummy load resistors or replace them with higher values to cut down on wasted heat (and energy) of the PSU. A 10 Ohm on the 5V rail would be dissipating 2.5 Watts, which is quite a lot and why it always looks so burned up. My goal is to stay around or under 1/2 Watt of wasted power per rail on load resistors. And depending on the PSU design and/or application, sometimes I completely remove some of the load resistors.
That said, it's usually a good idea to keep something on each rail for these older PSU designs like this FSP.
lti wrote on 2024-08-25, 17:34:
They don't have to bulge to be bad, and I see the dreaded Fuhjyyu.
Exactly.
I would pull every big(ish) cap over 100 uF out of the PSU and/or at least everything connected to a voltage rail output and the startup caps.
Chances are, there's going to be more bad caps in there.
kotel wrote on 2024-08-25, 18:08:
So I did what any reasonable repairman would do - start connecting and disconnecting PS_ON to GND very fast. That ended me up with bad smell from the PSU (don't know what burned since it still has live voltage inside).
🤣
I'm guilty of doing this a few times as well.
This testing "methodology" rarely resolves anything or brings any meaningful results, so consider reserving it only for "testing" (i.e. torturing) cheapo gutless PSUs that you actually want to see go ka-pow.
Oh, and I forgot to add...
I see tan glue on the secondary side in the provided pictures.
This glue is known to be problematic and often go slightly conductive over time, which can cause circuit to malfunction sometimes.
It's not too big a of a deal if you leave it on the secondary side, but it's usually a good idea to remove it from most places (that you can) on the primary side - especially high-voltage traces and/or sensitive circuits, like MOSFET Gate drives and etc. And definitely remove any that has turned brown or black - that's when this glue can actually start to go conductive. When/where it's tan, it's usually still OK.
