Need some help with a broken PSU

[bob808]

Hello everybody!

A bit of background for me. I've been in the electronics hobby for about two years and I started with tube audio amplifiers. Nothing fancy, just grasping the basics and doing some experiments with simple circuits. Also I've repaired some broken computer supplies that had obvious faults like blown caps. I only have 2 DMMs and my newly aquired oscilloscope is on it's way. I should have it in a week or so. Also I've just bought an ESR meter that I've used for the first time in this project that I'm about to tell you about.
I came across this power supply that's rated for 12V/84A output (about 1000W). The problem is that it's faulty. I would like to get it working again. It had one of the two fuses blown. Trying to get the fuse out I broke it and saw that the internal wire burnt at one end. I resoldered it on both ends so I can do some experiments. I did that as I didn't have any at hand (16A/250V). Since I learnt something from the tube world I decided to power the unit with a lightbulb in series (poor man's current limiter). I used a 75W bulb.
When I power the psu the bulb lights up and then slowly fades for about two seconds. Then the relay comes on and the bulb starts flashing brighly. The power supply has two status LEDs (DC Good and AC Good). At startup for the two seconds that the relay is still off the AC Good is lit up. After the lightbulb starts flashing the AC Good turns off and then I have no clue whats happening
I used the ESR meter on all the large electrolitics (there are a few anyway). I've also tested a few smd ones but it's not that easy since the whole board+components are covered in a clear silicone like substance. I have to scratch that off so I can get a good electrical connection with probes. Also I've tested the large silicon for shorts and there aren't any. I would have removed these components to test individually but it's a real pain to do that. The board is double sided and somehow there's solder on both sides for all of these parts. I guess I'd have to cut all the legs and replace with new parts if I ever have to take them out.
I figured that everything is ok until the large caps fill up and then the bulb flickering occurs. I measured 12V at the relay supply so that's ok. I traced that back to a board that is attached to the large pcb that's next to the two large electrolytic capacitors. Also there are some large diodes that tested ok for voltage drop and shorts and I also managed to remove the rectifier from the board and test it and that's ok as well.
Other than that I don't know what else to do. There isn't a single component that looks damaged, burnt or bad solder. I've also probed for temperature but there was nothing above 30 degrees C.
What would you recommend to do next? I'd really like to repair this PSU and is also a nice learning project. I attached some pictures with some notations for the large silicon and connectors etc. I can take more detailed pictures and more clear. I made these in a hurry with my cell phone.

http://imageshack.com/a/img199/2308/4cvl.jpg
http://imageshack.com/a/img14/2890/81el.jpg
http://imageshack.com/a/img547/9440/pc0n.jpg
http://imageshack.com/a/img593/8077/vws8.jpg
http://imageshack.com/a/img823/2926/bx5f.jpg
http://imageshack.com/a/img713/1597/iark.jpg
http://imageshack.com/a/img585/6972/ijlg.jpg
http://imageshack.com/a/img194/5665/0po7.jpg
http://imageshack.com/a/img845/1862/jv3m.jpg
http://imageshack.com/a/img853/4871/ztxt.jpg
http://imageshack.com/a/img202/9335/h8ca.jpg

[Stonent]

I hate to be "that guy" but it will be eventually said, shorting a fuse especially on something that already blew the fuse isn't good to do, now if there is a hazard it will not be able to safely fail. It might be worth it to get some resettable circuit breakers to wire to the fuse points.

[PedroDaGr8]

I hate to be "that guy" but it will be eventually said, shorting a fuse especially on something that already blew the fuse isn't good to do, now if there is a hazard it will not be able to safely fail. It might be worth it to get some resettable circuit breakers to wire to the fuse points.

The method he mentions is a common method for trouble shooting in the audio amp world. You put a light bulb in series with the AC line, then remove the fuse. Basically, it limits the current through the device decently well as too much current blows the bulb at worst. At best, the max current that can be drawn is the wattage of the bulb (as it is a resistor).

That being said, i would be checking your transformer. It definitely sounds like you have a short somewhere. First check the isolation between the primary and the secondary. Then check the mains filter caps, diodes, etc. Lastly, check stuff like the pass transistors (these are the ones mounted to the heat sink).

[bob808]

I hate to be "that guy" but it will be eventually said, shorting a fuse especially on something that already blew the fuse isn't good to do, now if there is a hazard it will not be able to safely fail. It might be worth it to get some resettable circuit breakers to wire to the fuse points.

I haven't shorted that fuse, I just re-soldered the inside wire that burnt off at one end. So the thickness of that conductor is the same, only that it isn't in a glass envelope. It still blows at the same rating (over 12A). But I also added the bulb in series with the psu so even if the psu is a dead short, that would only complete the circuit for the bulb. So the light bulb limits the current drawn by the PSU to 75 watts/230 volts = 0.32 amps.
There are three transformers on the PCB, two large ones and a smaller one. I will check those as well.

[fluxcapacitor]

whats on the other small pcb behind the 2 large caps .

[SeanB]

Try a 150W light bulb instead of the 75W one. It is likely that the resistance of the bulb is dropping the incoming mains down to the point that when the PSU tries to start up it only has enough charge to start then reaches the undervoltage level. Using a 150 or 200W lamp will still provide limiting but the supply should start up.

[oldway]

Hello,
It's likely a full bridge configuration.
The problem seems to be a too high no load consumption, not really a shorted power mosfet because when this happen, one or more of the others power mosfets use to really explode before de fuse blows.

It could also be a shorted secondary rectifier but you checked all of them and they are ok.

I think its possible you have a faulty drive of de Mosfet bridge, (one of the Mosfet is not conducting) and this produce a dc current who should saturate the transformer.

[bob808]

I will first try to add some more lightbulbs in parallel.  With a 75watt bulb I measured the ac on the first bridge rectifier and I get about 60VAC so that may be a step in the right direction. Also I will test the transformers, should be easier to remove than any mosfet. The heatsinks are inserted into a plastic frame that's screwed to the pcb an use some strong metal clamps to hold the mosfets in mechanical tension. There's some work to get those bastards out
Thanks you for the help! I hope I have some time to do these tests today and keep you updated.

[bob808]

It could also be a shorted secondary rectifier but you checked all of them and they are ok.

There are four STPS Schottky rectifiers and I measured a very low voltage drop on them and with dmm on resistance I get an increasing value from .2 ohms up to 115 ohms where it stops. I think I have to take them out from te circuit to test them properly. Also I don't know if I've correctly noted them in the picture. I have to double check their code.

[bob808]

Seems it's not that easy to test a smps transformer in circuit with a dmm 
Instead I upgraded my light-bulb current limiting contraption and I added two more sockets in parallel. With a total of 275W from bulbs that's 1.2 amps drawn by the psu. Bulbs still flicker but this time in about 5-7 seconds they suddenly go dark and I can hear a buzz from the output area.  Also the 12V output measures about 27V AC not DC during the flickering and when the bulbs go dark and the buzzing is stronger the output reads about 42V AC. There's little current drawn from the psu, the bulbs don't even light up a bit. When the lights go out, weirdly, the AC Good and DC Good LEDs light up.
On Thursday or Friday I should get my scope and I can look closely at what's happening inside.
I tried to look for transformer specs online but I couldn't find anything. There are two larger ones on the output that are noted with IA000261 Rev.A FEC 1207 and the smaller one is RHTA000267 Rev.C FEC 1207. I couldn't find anything online about them. They have solder on both sides of the PCB and removing them will certainly be a pain. Same goes for anything on this PCB. I may try to look at the STPS80H100CY schottky rectifier if I manage to get them out of the circuit. I presume the diode test should clear their status?

[bob808]

I forgot to mention that DC voltage ramps up to 380V on the large caps when it shuts down. Also I applied the AC voltage directly on the rectifier bridge and I get oscillations on the DC side, it bounces all over the place. After the supply goes in protection or whatever that is I only get about 240VDC on the bridge, this time steady.

[Alana]

I did few ATX PSUs only with DMM and this light bulb circuit you have there.
First inductor, next to a relay is probably active PFC and about 400V on main caps after such circuit is a possibility if not a normal thing.
Strange thing with DC output - as if there was a short in one of rectifiers and what you measure is AC voltage on ESR of output caps. In typical ATX PSU this shuts down the device. Here - dunno.
For rectifiers - expect as low as 100-150mV forward voltage drop on diode test in DMM - its rather normal for those devices.

[bob808]

I added two more 100 Watt bulbs and that was the problem with the psu start-up. It seems that with 300W limit the bulbs flicker for about 6 seconds then turn off completely. While this I unscrewed two bulbs and the remaining bulb lights up just a little bit so it draws some current with no load. I get 12V on the output. I also loaded with a 60W car bulb and all seems good. I think the problem was the fuse.
Now that's interesting since this equipment got replaced with a new one and that one worked ok. So probably no extra load has been applied to it when it blew the fuse.
Also this equipment worked in a hot environment, so is it possible that some component got crazy because of temperature and managed to blow the fuse?
I will try with extra load on this psu for a longer time so I can be sure that it's stable. I'm pretty bummed as I didn't have to replace anything as it would have been a good learning project but still I learnt something about PFC and SMPS. So I have one working 12V/84A SMPS unit  Thanks for your help and I'll update if anything else happens.

[Jonny]

I guess in hindsight you said the fuse was broken, not blown hence your ability to solder it back together and suggesting there was never any fault to start with. Fuses have been known to let go when they're old and tired. There's usually nothing left of the fuse wire when most devices fail.
One thing I think it's pointless trying to start a SMPS via a series bulb. They need to develop a full charge across the main filter cap so the rest of the circuit can operate correctly. Proved this on an LCD recently when my father chose to do this and thought it's bright pulsing meant heavy fault current (original issue was a blown main filter cap which took the fuse with it). I hooked it straight up after a quick check that revealed there were no obvious faults that could cause this and away it went.

[trackman44]

In the middle of this image (on the right side, just below the output connector) :-

http://imageshack.com/a/img593/8077/vws8.jpg

One of the capacitor leads has a ring around it on the pcb. Since the power supply was working in a hot environment, I figure that that cap started to seep electrolite on that lead and ate the solder in the via (thus the ring of solder residue), which in turn caused an intermittent open and probably blew the fuse. Replace that cap and make sure that you use plenty of solder on both leads, good for conductivity and heat dissipation.

Will

[bob808]

That's actually some layer that's covering the whole board, and it peeled off in that area. That thing gave me a headache trying to probe around the board. I had to scratch every electrical contact so I can probe them

[bob808]

I guess in hindsight you said the fuse was broken, not blown hence your ability to solder it back together and suggesting there was never any fault to start with. Fuses have been known to let go when they're old and tired. There's usually nothing left of the fuse wire when most devices fail.
One thing I think it's pointless trying to start a SMPS via a series bulb. They need to develop a full charge across the main filter cap so the rest of the circuit can operate correctly. Proved this on an LCD recently when my father chose to do this and thought it's bright pulsing meant heavy fault current (original issue was a blown main filter cap which took the fuse with it). I hooked it straight up after a quick check that revealed there were no obvious faults that could cause this and away it went.

Yes, I guess you are right. I broke the ceramic casing and found that the whole inside conductor was in one piece, only disconnected at one end, looked a bit melted but if it wasn't properly attached/soldered the high current may have taken it's toll on it.
Was a nice experience trying to figure out what was happening and seems like the poor man's current limiter has it's limits after all