ATX Power Supply repair - Debugging/help needed (BQ SU7 - 700W) [FIXED]

[johnkeates]

Hello Everyone,

I am currently trying to figure out why one of the power supplies I have doesn't work the way it's supposed to.
It is a Be Quiet! System Power 7 model, for 700W. It's doing pretty much everything the way it's supposed to do, except for making Power_Good high. This obviously means it thinks something is wrong (voltage, ripple), or the method used to figure out of the power is actually good is broken.

When turning the PSU on (with no motherboard attached as it won't power on the PSU as long as Power Good isn't high) it runs just fine by shorting PS_On to GND. All voltages measure fine, well within tolerances of the ATX spec, most of them are pretty much spot on.

Since I have plenty of test parts lying around to test the PSU on, I thought I'd give forcing it on while shorting Power Good to +3v3 a go while connected to a motherboard, and it powers up fine, BIOS reads all power levels as fine, OS works, I can pretty much load it down with a bunch of peripherals to about 400 Watt (I don't have more power-hungry stuff to hang off of it, short of whipping up some sort of big resistor made out of wire in water) and it holds fine. Temperature doesn't seem to do anything strange (it simply stabilises after a few minutes of use and doesn't go up), fan is working properly, so it's basically working fine, if not for that power good signal.

The first thing I did was trace the power good pin on the ATX connector to the power monitor IC inside the PSU, connection is good so it's not like there is a break somewhere between the pin on the IC and the ATX connector. Next thing I did was measure the voltages inside the PSU, which are fine too, both in stand-by mode and running mode. (yes, I used isolated tools, and a pair of oxxa gloves to not electrocute myself)

Inspecting the PCB I only found a trace that was slightly lifted, but still connected securely, it seems to have been some factory problem because it was one of the pins of a DC module which was mounted slightly off-angle and then glued down, which might have forced the track to come off. There was solder on both sides of the track, and when the PSU came in, the warranty sticker was still good. I completely de-soldered that DC module, cleaned it up, and re-soldered it, just to be sure, but nothing changed.

The DC module is the only module with a chip that has markings on it that lead somewhere. It's a Weltrend WT7527, which is a "PC POWER SUPPLY SUPERVISOR" according to the data sheet (1st link on google, not sure if posting a link is allowed), and that is the chip where the Power_Good pin routes to. The only other components on that small daughterboard are 3 caps and a resistor, which measures 50 ohm, which is what it's coded for, and the caps are 22µF, electrolytic, and they measure fine as well. (I de-soldered them, and since I don't have a real ESR meter, measured them with an Extech EX330.

So for the next step, I'm a bit unsure on how to continue. The PSU works fine, the output is well within spec, there are no issues when powering equipment, but something must be broken, and I can't figure out where. Should I just go shotgun-troubleshooting and replace the chip? Should I just not care and tie power_good to 3v3 standby? (this is what many PSU suppliers seem to do on lower end supplies) Is there somewhere else I should be looking for the problem?

I don't have fancy measurement tools at hand, that EX330 is actually the only tool I have with me and until I have moved into my new house, I don't really have any option of going to a lab to measure more stuff like ripple.

It's not that the PSU *has* to be fixed, but I'm kind-of interested now to figure out what went wrong with the thing.

Any help or ideas would be appreciated

(Tiny edit: The only two DIPs in there are helpfully labeled "6600" and "6602", which is rather useless as far as my google-fu goes)

[johnkeates]

That's right, I'm replying to my own post. Because I solved it! :p

This was a rather stupid case: as it turns out, that WT7527 actually was being fed bad information! Something must have gone horribly wrong at some point, because there was a short on the FPOB pin and a track underneath the IC with PGO which pretty much made it constantly force PGO low. It appears there is a slight delay between PSONB being high and PGO being high, which causes any power-on pulse to never stick. It was on the same side as the gob of glue and the pulled PCB track, so I suspect that it might be some combination of unlucky placement and maybe someone dropping it. The track was slightly torn, but with a tiny bit of solder it works like a charm on 7/10 motherboards I tested it on. Seems to be a slight difference between the way they pull PS_ON high and pull on the Power_Good line that decides on whether it sticks now. Maybe the latching part inside the IC is a bit broken now, but hey, it works! And can be fixed back to it's normal state by replacing the chip. Better yet would be a total replacement of that small board, but it doesn't seem to have any reference on it that leads somewhere. One large number doesn't get any hits and seems like it's some sort of internal ID or serial number...  (3bs0326410gp)

The only reason I actually found out was because I de-soldered the chip to stick it on a breadboard because I wanted to test it outside of the PSU.

[Zucca]

Congratulations, maybe next time put some pictures.