The commonly overlooked weak points of cheap ATX PSUs

[Th3_uN1Qu3]

I've noticed more folks rebuilding crap ATX PSUs lately, and it doesn't seem like much work because they're rebuilding something that was working in the first place. Tackling a blown one can be a bit trickier...

So here we go. I won't mention the rectifiers, caps in 5vsb primary and things like that as those are already common knowledge.

1. Driver transistors. These are TO-92 types, located on the secondary side, somewhere around the base driver transformer. Follow the traces coming from the transformer to find them. I've blown more of them than i can remember. A primary short can take the driver transistors out as well, and when only one of them is shorted you get pretty interesting results, which may lead you to search for the failure in a totally unrelated area.
   
   
2. 1N4148 diodes in base driving circuit. Found in the primary near the main switchers. When these go short, they will make the primary switchers fail open because of reverse voltage over the base-emitter junctions. If you want to replace them with something else, you must use an ultrafast diode with very low recovery time.
   
   
3. Base resistors. Found in the primary, in the base of the main switchers. Sometimes these are omitted entirely, and replaced with jumper links. Many times they are too low in value, around 1 ohm. That might work fine for the tiny capacitors they put on the output, but when you've got a significant capacitance to charge at startup, too low value base resistors will blow the switching transistors' bases open because of too high current. Particularly if the NTC thermistor in the primary is too small or missing. Had it happen one too many times - a power supply that works just fine under load, and then you turn it off, and when you turn it back on... it doesn't.
   These must be at least 2.2 ohms, 4.7 ohms preferred. This takes a little bite of efficiency, but it's insignificant, and certainly better than having your primary transistors die.

That concludes it for today.

[momaka]

Re: The commonly overlooked weak points of cheap ATX PSUs

Interesting stuff.
I do have a few half-bridge design PSUs that I'm planning on rebuilding. Will definitely make sure to check this stuff out, even though they work right.

[Th3_uN1Qu3]

Re: The commonly overlooked weak points of cheap ATX PSUs

I cannot stress point 1. enough. A failed driver will cause all sorts of weird issues, and i'm going to tell you another tale.

The unregulated +/-35v and +/-20v PSU i have made for powering an audio amp (the one i blew in a spectacular way here) had a weird startup sequence (took a while to start, and started with a whine coming out of the speakers till it settled), and would cut out under high load, then come right back up when the volume was lowered. Like it stopped oscillating.

Took the PSU home for scoping. The waveform in the main transformer looked okay on one half cycle, but on the other it was sine-ish and slightly lower in amplitude - the PSU was self-oscillating on each half cycle! Weird waveforms in the primary as well, but it kinda had me stumped, as on each transistor one cycle was okay, the other one was buggered. All primary parts tested good.

Scoped the collectors of the driver transistors, they looked more or less okay, just that one of them didn't have the characteristic spike at turn-on, and the waveform on the middle pin of the driver trafo primary was all messed up. The difference between the two drivers' waveforms was small tho, an untrained eye wouldn't have noticed it, especially on an old analog scope like i have.

I pulled out a working PSU to get a reference of how the waveforms should look like. I then decided to pull the drivers again (even tho they tested good previously). I tried them on the diode check of my DMM, and one read 0.65v or something collector to emitter, the other one 0.9... and after a second the meter showed OL. Ah, gotcha. So there is such a thing as a partially open transistor. You learn something new every day. Swapped the drivers and all is good now.

PS. Primary transistors (generally all big bipolars) have a high Vce(sat) so they will show open c-e on the diode check even when they're good. A little jig with a switch, base resistor and LED will tell you whether they're in fact working or failed open.

Now that i remember, i did find the 494 dead, and it's impossible to have a dead 494 due to primary failure unless one of the drivers blew. That wraps up today's job, the power supply is working great and i'm sure it will for a long time.

[budm]

Re: The commonly overlooked weak points of cheap ATX PSUs

I never really trust using DMM Transistor checker, one thing is that the transistors or Diode will see such higher voltage than the meter can put out. I use variable dc power supply (0~100V) with current limitter and reistor for testing transistors and Diode. Iused to have curve tracer which was great for testing unknown transistors and Diodes.