ACBel ATX PSU Weird Behavior (power-cycling).

[Dannyx]

Good day folks. Pretty tough nut to crack here, so I'll try to keep it (as) brief (as possible): this ACBel power supply came out of a Hikvision NVR which was power-cycling. I fixed an identical one before and it was as straightforward as it gets: replaced the caps on the secondary and it was good to go - one of them was bulging at the top, making the problem even easier to spot. Job done.

THIS one though........

It had the same exact behavior and even the same bulging cap (a 2200uF one), but upon recapping it just like before, I threw it back into the NVR, powered it up and big surprise: same power-cycling as before, almost like I hadn't done anything to it !  Since time was short to get this thing back up and running again, we ordered a replacement SFX supply and called it a day, so the problem is not with the NVR. Had it not been for the less common form factor, I wouldn't have given it second thought and just binned it entirely, since there's nothing special about it and there's no reason to bother fixing ATX supplies these days, but I thought I'd be nice to get this one going as well and keep it around as a spare, as well as add to my own personal satisfaction.

With the supply on my bench, I plugged its ATX connector into one of those breakout tester boards with screw terminals to allow me to power it on without anything attached. By itself, the supply comes on just fine, the voltages appear stable (though I do not have an oscilloscope to properly confirm) and it appears to run normally. However, when I attempt to plug in a hard drive in one of the SATA connectors (which is the only thing I have on hand that represents a typical load for this kind of supply - the NVR has 4 of them, by the way), the hard drive simply refuses to spin up, despite the supply itself staying on. The voltages don't change in the slightest, at least on my DMM - a scope would probably pick up some very rapid droop which might explain the behaviour. It doesn't matter if the power supply is already on, or I switch it on at the mains switch, the drive never powers up.

However, one way I AM able to get a drive to spin up on this supply is like so: I chopped up a SATA power connector so I can hook up its wires individually: if I connect the 12v wire (yellow) AFTER the 5v wire (red) the drive spins up just fine......however, this is STILL hit and miss and doesn't always work. I tried tapping into the 12v rail before and after the "cut-off" transistor which switches off the 12v rail with the supply is in stand-by (the D-PAK transistor in the picture), but the results are too inconsistent to point me in the right direction. I even tried powering the 12v wire with the bench supply and a common GND, while the ATX supply fed the 5v rail and it worked. I even tried the opposite and powered the 5v rail externally, but it's the same inconsistency.

Yes, I DID try different junk drives I have lying around and are still functional - they all behave roughly the same.

I replaced every single cap I could find on the board, including the main one, but nothing changed.

The IC on the hot side is marked as 73k12a and although I could not find any part by this name, I came across a russian blog where another chap was asking about it, as it was apparently being used in a Philips stereo system, so I connected the dots, pulled up the service manual for that stereo and it appears to be an OB2273. The pinout seems to agree, though I'm still not 100% sure this is it, or if it's even the culprit to begin with. If it WERE the culprit, I assume the supply would shut down entirely or not turn on at all. Instead, everything I plug into it either doesn't turn on, or power-cycles in the case of a motherboard.

Running out of ideas......the only thing I can think of is that not having the properly-sized caps on hand when I replaced them and leaving the legs exposed like that somehow causes enough noise or interference pickup to make the rail unstable enough for devices to refuse to power on....

I took some pictures of various parts of the supply......anyone faced something like this before ?

[Dannyx]

Pictures wouldn't fit, so here's the other side. They're not the best, so if you'd like to see something more specific, I can focus on that part better....I just noticed the lighting is pretty sh!t...

[fzabkar]

I use Google Images or Aliexpress to verify the ICs.

https://www.aliexpress.com/w/wholesale-OB2273.html

https://www.google.com/search?q=OB2273&tbs=isz%3Am%2Citp%3Aphoto&udm=2

Could there be leaky rectifier diodes on the 12V side? Did you check the main 400V (?) bulk capacitor?

[Dannyx]

Could there be leaky rectifier diodes on the 12V side? Did you check the main 400V (?) bulk capacitor?

I haven't checked the diodes very thoroughly, since I dismissed them as being the problem, seeing how the supply technically works....might as well...

Yes, like I said, I replaced every cap, primary and secondary, including the main one (though I changed it back to the original one since the replacement didn't fit and I just ran some wires to it to confirm it's not it). I don't blame you for not reading that far - it's a wall of text that's likely to put people off....

It doesn't have PFC, so "B+" only goes up to 320-ish volts.

[fzabkar]

Did you check the main 400V (?) bulk capacitor?

Yes, like I said, I replaced every cap, primary and secondary, including the main one (though I changed it back to the original one since the replacement didn't fit and I just ran some wires to it to confirm it's not it). I don't blame you for not reading that far - it's a wall of text that's likely to put people off....

Sorry, I did read it all, but it didn't register in my brain.

BTW, when a HDD spins up, it draws a lot of power from the 12V rail. The 5V rail supplies the logic, after being down-converted. AISI, your positive result with an external 12V supply would point in this direction.

My examination of AT and ATX PSUs suggests that they typically regulate the output voltages by sensing the weighted average of the 5V and 12V rails. When the 12V rail dips, the regulation circuitry increases the 5V rail by a proportionate amount, and vice versa.

[Harry_22]

......anyone faced something like this before ?

Hi!
I had one case. The power supply did not work until I replaced one dry capacitor. Then all the voltages appeared on the test load. And then it suddenly exploded upon returning to the device.
The reason was that not all faulty capacitors were found.
Some hid under a thick layer of glue.

Conclusion: if replacing the capacitors did not help, then another one is hidden somewhere!
In fact, in such cases it is necessary to restore the key block diagrams. Only then will you understand what else was missed.
The truth is out there and satisfaction lies in finding it!

[Dannyx]

Sadly, I'm pretty sure I didn't miss any cap....MAYBE...I'll have to check again. If that's not it, I don't know what is....unless I got really unlucky and one of the "new" caps had high ESR out of the box...

[Harry_22]

Please clarify these three TO-220 elements and make a photo from this side.

[Dannyx]

Those are 3 transistors: 2 AP40T03GP and 1 P40nf03L. They most likely switch the 3.3v and 5v rails on/off. The 12v rail is switched by that D-pak in the corner.

I don't have the board in front of me right now to confirm, so I only have the photos to go by, but I'm not entirely sure how the 3.3v rail is generated. It seems to come out of the same diode which rectifies the 5v rail and they it's stepped down through one of these 3 FETs I mentioned - at least that's what I was able to determine by following the traces in the pictures...

In the last picture, the 2 AP40T03GP are the one in the middle of the heatsink (the middle arrow) and the one at the bottom, right where the date is hand-written on the top of the heatsink (bottom arrow). The P40nf03L is at the top.

[Harry_22]

I prepared the diagram. But unfortunately I cannot see elements from the top layer due to the big capacitors close them.

[Dannyx]

Thanks. There aren't too many components left underneath the caps. I'll probably mark some of them once I get back to working on this project. One thing I know for sure is the remaining IC on the bottom is an op-amp. There are also a couple of TO92 devices close to the other side of the heatsink, which we can identify on the bottom by their 3 legs side by side. If I remember correctly, those are 431 devices, but I have to check this as well.

[u666sa]

What it says on M12 16 leg chip? Because you diagnose by pulling datasheet on that chip and checking all the voltages that go onto it.

[Harry_22]

M12 is the EST.7627A PS monitor with Power Good and Fault Protection.

See also PCB drawing in Paint.NET.
https://drive.google.com/file/d/1ioSKdIbU5OzwVAya7Q1peoBTQO55Dylf/view?usp=sharing

[u666sa]

Excellent, on page 2 and 3 there are all the parameters and conditions to check. End result you will know what's making PSU shut down or not start.

[Dannyx]

Here's some pictures of the other side of the heatsink - the one opposite the capacitors. There are 3 431 devices here, as well as one BT169D SCR in the middle. You can also see one of the TO220 diodes there. There's a second one behind the transformer.

[Harry_22]

Please make a photo of output wire names on top PCB layer.
Could you make the best photo of bottom layer.