ubiq wrote on 2024-03-07, 04:36:

Used diode mode on my MM. Voltage drop across the big fella was 0.16V (that might be bad?)

If it's only showing that reading in the forward direction, but in reverse it shows O/C (open-circuit), then it's likely fine and the low V_f is due to low test current from the MM, like kingcake suggested.

The KA7905 and KA7912 are your -5V and -12V rail regulators, just as I suspected. For these, just test resistance between all of the pins in all combinations to check for any low resistances (under ~100-200 Ohms.) If none show any suspicious low resistance, they should be fine too.

When you get to it, check that green resistor mentioned in my previous post, close to the black wire for SW1.

Took a break on this one for a few days. 😅

Anyway, since they were obviously very important to this and I didn’t trust myself, I tested those three caps around the SW1 again - still good.

That green resistor.. is an inductor, no? Anyway, it measures 24.4Ω

Looked at the optocouplers - a little googling told me I can only test the LED side with a multimeter alone, but they seemed fine.

Haven’t found any resistors or diodes on the board that seem to test bad.

Anyway, I put everything back where it should be and made sure it was still broken in the same way, and yes - I haven’t managed to break it any further, so that’s something. Oh, and to be clear, it doesn’t make a continuous ticking noise on power up, just a single tick.

ubiq wrote on 2024-03-11, 03:26:

Took a break on this one for a few days. 😅

Hey no worries.
This is actually a more effective way to troubleshoot stuff, since it allows you to think problems through in the back of your head. 😀

Also, I myself will be traveling a little this week and might not able to reply for a few days.

The green resistor may or may not be an inductor (most likely not, though.) What does the PCB designator say? If it's R-something, it's a resistor. And if it's L-something, then it's inductor. You can also verify if the resistance you measured across it corresponds to the color codes (probably going to be a 25 Ohm if it is a resistor.) If it's an inductor, it should show very low resistance.

ubiq wrote on 2024-03-11, 03:26:

Looked at the optocouplers - a little googling told me I can only test the LED side with a multimeter alone, but they seemed fine.

The output / driver side of the optocouplers can be tested too, but it's a bit tricky.
An optocoupler taken out of circuit should show open-circuit on the output / driver side.
Now, if you have a spare 470 - 1000 (1K) Ohm resistor, you can wire that in series with the input / LED side and connect this series circuit to a battery. This should drive the LED to turn On, which in turn should make the output / driver side resistance become low/lower. Of course, this test is another one that only works with the opto out of circuit.

ubiq wrote on 2024-03-11, 03:26:

Haven’t found any resistors or diodes on the board that seem to test bad.

Perhaps remove and test the caps on the output. I suspect they will be fine. But perhaps in the off-chance one of them has a broken lead or something more crazy / exotic like that. After all, if you have never had the PSU working and you don't have any history of what happened to it, there's no telling what could have caused the problem. So in such cases, you just have to keep your eyes and mind open before ruling anything out. Most likely, it will probably be something very simple that we either overlooked or didn't see. PSUs like these generally don't have issues.

I have a few more tests up my sleeves. I'll put more details when I get back... that is, if you're still interested to continue. I don't want to be pushing by any means. But if you're enjoying this, we can keep going.
These will involve voltage injection. So in the mean time, see if you have a 5V and a 12V power adapter, each rated anywhere from 0.75 to 3 Amps should do. Also a non-grounded (2-pin wall plug) adapter than can do more than 16V but less than 30V (ideally less than 24V).

ubiq wrote on 2024-03-11, 03:26:

Anyway, I put everything back where it should be and made sure it was still broken in the same way, and yes - I haven’t managed to break it any further, so that’s something.

That's always good. 😀

I didn't give up on this... Good news, it's working!

Unfortunately, I never found anything to fix - it just started working. Not a very satisfying end to all this troubleshooting. In a way, it makes sense - all the obvious things to investigate seemed to be fine. I dunno, guess it just spent enough time on the healing bench. Or maybe it will flake out again. I've got a Digikey order on the go, so I added the necessary caps to swap them out on this PSU just in case I need to at some point.

Anyway regardless, I learned a lot going through this process, and gained a lot of confidence in my ability to do future troubleshooting. Many thanks to momaka and kingcake putting in the effort to help me out!!

ubiq wrote on 2024-03-26, 23:10:

Unfortunately, I never found anything to fix - it just started working. Not a very satisfying end to all this troubleshooting. In a way, it makes sense - all the obvious things to investigate seemed to be fine. I dunno, guess it just spent enough time on the healing bench. Or maybe it will flake out again.

I had this happen to me when I fixed a SMPS in a Toshiba LCD TV. In that case, the supply immediately went into OVP (overvoltage protection mode) when you turned it on, and did not do anything until power cycled. I couldn't find an obvious fault, all the components checked out fine, and when I connected an DMM to the main feedback optocoupler output, it suddenly started working. (ATTENTION: Do not attach any measurement device to the mains side of that optocoupler unless you exactly know what it is doing, and are perfectly able to explain to an electronics professional why the way you are doing it and how you handle it is safe. Explaining how to avoid shorting something to PE (when using a scope) or getting shocked is beyond the scope of this post) It kept working even after disconnecting the clamp. I concluded that probably the solder joint I attached the clamp to was broken, and resoldered the optocoupler. The supply kept working for years after that.

The OVP circuit was operating perfectly fine: As the feedback didn't work, the primary side pushed "full power" to the secondary side, even after it reached the target voltage of 5V, and as soon as it approached 6V, the overvoltage circuit tripped a second optocoupler that killed off the complete switching circuit. I'm glad that supply had this extra line of defense, because it kept the supply from breaking while the primary regulation circuit was inoperative.

Nice!
Good work on making it work... without having to do anything to fix the problem?! 😁 😁

ubiq wrote on 2024-03-26, 23:10:

Unfortunately, I never found anything to fix - it just started working. Not a very satisfying end to all this troubleshooting. In a way, it makes sense - all the obvious things to investigate seemed to be fine. I dunno, guess it just spent enough time on the healing bench. Or maybe it will flake out again.

You know the saying: problems that go away by themselves sometimes also come back by themselves.

Well, it's still puzzling why the PSU started working all of a sudden by itself. The two theories I have:
1) Perhaps some of the bigger caps needed reforming?? Trying to turn the PSU On and Off enough times eventually got that done.
2) Bad solder joint somewhere that either got fixed or moved around enough after resoldering / putting everything back together.

Whatever the case, I'm glad to hear that at least you enjoyed the learning experience. Most PSUs are usually not more complex than this, especially the older stuff. In fact, many are just caps issues or solder joints issues or the occasional burned/shorted diode... and sometimes a standby-generating IC.

mkarcher wrote on 2024-04-02, 07:26:

The OVP circuit was operating perfectly fine: As the feedback didn't work, the primary side pushed "full power" to the secondary side, even after it reached the target voltage of 5V, and as soon as it approached 6V, the overvoltage circuit tripped a second optocoupler that killed off the complete switching circuit. I'm glad that supply had this extra line of defense, because it kept the supply from breaking while the primary regulation circuit was inoperative.

It's not so much an extra line of defense as it is just how the design works.
Usually, a PSU with 2 optocouplers = forward converter design with a current-mode PWM IC on the primary. These designs are generally safer than the older half-bridge designs that use voltage-mode PWM IC (*494, *7500, and the more integrated clones). With the forward converter designs (such as the PSU in this thread), one optocoupler is for enabling power to get to the PWM IC on the primary, and the 2nd opto is for regulation. The PWM IC also has pulse-by-pulse current limiting implemented via probing voltage levels from the MOSFET Source resistor. So even in the event of the 2nd opto getting stuck open and not providing any feedback, OVP protection will turn off the 1st opto and disable power to the PWM IC. Also, the pusle-by-pulse current limiting, if implemented properly, allows the primary side to self-limit the power (over-power protection) so the primary MOSFET doesn't blow out. With half-bridge designs, OPP is a bit tricky, since the middle (driver) transformer of the BJT transistors is also used as a feedback device. But often times, OPP on these designs is set either too high or just not implemented at all, leaving many half-bridge designs vulnerable to blow-out if the secondary-side monitoring protections don't catch an issue on the output.