[REPAIRED] Cheap 150W 48V SMPS repair (clicking noise)

[Richard Head]

Power the control IC from a separate bench power supply and measure its gate drive output.The bench PSU obviously has to be isolated! You will probably find that as soon as you provide external power to the chip that the PSU works perfectly. If that is the case then disconnect the external bench supply DC and measure the signal from the feedback winding when the unit is clicking. There should be a string of pulses from the feedback winding that is able to charge the bleed-on cap and power the chip.

[dominicM]

Did you also check the 48V output for a short circuit? There can be can be other reasons like an open circuit in the secondary circuit.

There are no shorts between any output terminals.

[dominicM]

It is also possible to have shorted windings in the transformer. This can be difficult to prove.

Seems kind of unlikely to me for it to happen to 2 psu's though, no?

[dominicM]

Replace D2

No change, what was the reason for this suggestion specifically?

Do not replace this one by a "normal" diode. You need to use a ultra fast recovery type in many circuit to work correctly. Also ensure is has a high enough reverse breakdown voltage. I had 2 SMPSes failing because the diode reverse breakdownvoltage was serverly limited (got to hot probably), but measured still ok with a multimeter.

Some more info which type to select can be found here (altough I just picked one I had lying around from a previous project):
https://www.maximintegrated.com/en/app-notes/index.mvp/id/849

I replaced it with a working one from a similar 24V psu and there was no change, 24v worked while 48. clicked as before with D2 swapped out.

[dominicM]

Is U2 LM431? Can you swap it with a good one?

U2 is a TL431A. After swapping it works! So what was the logic behind this suggestion? The question now is why did it fail, it happened to 2 psus so I doubt it's just chance.

Looking at the datasheet it seem like it's only rated up to 36V, am I missing something or is it a design flaw?

[strawberry]

OB2269

[kxenos]

 
No flaw. The 431 is probably in series with a resistor and optocoupler's LED.
You said that after removing power the PS continued to make the noise for a while so there was no short-circuit in the output. So the next reason for the PS not to be able to start is if there is a problem in the feedback circuit. I looked at the photo and I saw a TO92 in the secondary side. This is a classic configuration and a classic problem in cheap PSUs. Probably it will be OK if you install a genuine part. If there are problems in the future again you will have to add a snubber RC parallel to the 431.

[Bondguy]

@Richard Head: it is an interesting idea but what will be tested that way? feedback winding of primary transformer or control IC?
Am I correct to think that by feedback winding you mean one of the primary winding that provides voltage for control IC when the power supply capacitor of control IC is being discharged?

[dominicM]

No flaw. The 431 is probably in series with a resistor and optocoupler's LED.
You said that after removing power the PS continued to make the noise for a while so there was no short-circuit in the output. So the next reason for the PS not to be able to start is if there is a problem in the feedback circuit. I looked at the photo and I saw a TO92 in the secondary side. This is a classic configuration and a classic problem in cheap PSUs. Probably it will be OK if you install a genuine part. If there are problems in the future again you will have to add a snubber RC parallel to the 431.

Will get some from rs. Out of curiosity is it safe to assume that most parts like this one from china on ebay are fake?

Would modifying the 24v ps to output 5v be feasible? It seems like it should be possible by changing resistors close to the 431. Don't want to blow it up just after repair

[xavier60]

The output should have been pulsing up to about 1 volt during the clicks with the faulty TL431.
Reducing the regulation voltage would also reduce the supply voltage to the control IC which would cause it to repeatedly shut down after it starts in this case.
It would work in some designs where the IC is powered from somewhere else and not affected by the regulation voltage.

[kxenos]

I don't know about ebay parts because I have never bought any but I guess if it's too good to be true it probably isn't. When you repair or build something you try to minimize uncertainty factors, so buying parts from well-known distributors ends up being cheaper IMO.

About changing the output voltage, I'm sure you can. You will have to DaveCAD the secondary circuit around the 431 and replace a couple of resistors and maybe the output caps. The idea is that the output voltage controls the brightness of the optocoupler's LED. You will have to change the resistor values so that the LED has the same brightness (current through it)

[dominicM]

Well I did see a similar mod by someone somewhere and it worked fine, so it is feasible. In that case it was only maybe 10v lower than original supply so 24v to 5v may be bit too far. That being said 48v and 24v supplies look almost identical even most parts are identical or very close to.

[dominicM]

Got the parts but I think I may have gotten the wrong part. I ordered TL431C without really paying attention but the one in the PSU was TL431A. After replacing the clicking was still present though different sound. I can't really see the difference in the datasheet for these parts. What is the difference?

[kxenos]

The difference is precision 1% for the A part vs 2% for the part you've got and also smaller operating temp. I don't think it matters much. Can you check the resistors around the 431 and install back the 431A part from the other PS and see what happens? Also can you place a small load in the output? Something like a 330 resistor

[dominicM]

The difference is precision 1% for the A part vs 2% for the part you've got and also smaller operating temp. I don't think it matters much. Can you check the resistors around the 431 and install back the 431A part from the other PS and see what happens? Also can you place a small load in the output? Something like a 330 resistor

I replaced the TL431 with the original and it was the same thing. The new TL431 worked fine in 24v psu so the new part is fine. The 330 resistor made no difference. I remembered that I dropped one of the boards and after close examination 2 pins of D3 have snapped  An easy fix if nothing else is damaged.

[dominicM]

Just to say all the PSU's are repaired and 1 of them has been working for quite a while now so it seems the repair was completely successful!

[SoundTech-LG]

That LM341 voltage reference could also be a (cheaper) Korean version KIA2431. Just saw that one in my Denon AVR-3312 SMPS...