Half dozen psu repair challenge

[pdavid]

Last week I got my scope back plus I managed to salvage a huge isolation transformer. 1kW. I just need to find a place for it.
So, one coworker wants me to repair a bunch of psus he piled up in his shop over the time.
Since he doesn't have much equipment over there we are going to haul all the units to me so I can do it more properly.
There are 6 units in total: Fsp 350-60hlc; 350W and 400W -80gln; 600W Epsilon, Macron MPT-5002p and a Chieftec CTG-450.
Only the Chieftec works but the pc restarts and goes bsod after a few hours (works fine with an old GPS-400AA) The rest won't even start, one of the 350W Fortons has two broken fan blades too.
I'm eager to get started Expect lots of pictures soon.

[pdavid]

Re: Half dozen psu repair challenge

I opened all the units up for visual inspection. All except the Chieftec has buldging caps. I'm affraid some of the fortons may even have bad primary caps.
I decided to start with the Macron, I haven't meet the insides of this psu so far.
Standby voltage is 3.4V closing pson ground: unit doesnt start.

Passive pfc
8A Greetz
2x1000uF 200V primary capacitors
2 x 2sc2765 fets, 3843pwm
5N60G standby fet, driven by a some sot-23 pwm
...Fuhjyyu caps everywhere.
WT7525 supervisor
smd 339
1040CT for standby
2 x 20100 for 12V; 2 x 3040 for 5V; 4045 for 3.3V schottkies

Fuse, transistors and diodes test good. I replaced all capacitors except the primaries, measured 950- 960uF on them...
Still no go. The fan doesn't even twitch, only standby voltage is present. I'm going to look for shorts and measure supervisor and other ic tomorrow.

[c_hegge]

Re: Half dozen psu repair challenge

Could be a bad transformer too.

[Khron]

Re: Half dozen psu repair challenge

First clue: 5Vsb is nowhere near 5V - start your investigations there.

Prime suspect: those two small cap in the 5Vsb circuit, between those two (nearly) upright diodes. Replace those with some brand-name trustable ones, and see if it works right after that.

[momaka]

Re: Half dozen psu repair challenge

There's also that stupid voltage boosting circuit on the primary (the 2 small electrolytic caps next to the primary caps, along with the 2 diodes next to them). Not that that's the cause of the problem, but if you do end up fixing the Macron, you should remove/disable that circuit - it generates a lot of heat and really servers no practical purpose.

[c_hegge]

Re: Half dozen psu repair challenge

^
I doubt it's causing the problem. Usually, the caps on those circuits fail, but all that happens is that the voltage boost circuit itself stops working. The PSU still runs without it, but it's just somewhat less efficient.

[tom66]

Re: Half dozen psu repair challenge

That circuit reduces efficiency... it increases power factor. It's a valley-fill circuit. Disadvantage is ripple is about 50% of line voltage (stretches transformer design) but PF is about 0.8~0.9 ideally.

[pdavid]

Re: Half dozen psu repair challenge

Ah, yes I have replaced all the small lytic caps too. Standby got back to normal right after I replaced the caps.
The pfc voltage doubler circuit is gone. I removed it before I reassambled the psu. Interestingly the pcb wasn't fried there at all.
Only standby voltage is present, it won't start.

Also here are some shots of the 600W Forton. Turning on the unit it's buzzing and gives 1,5-3V standby. I took it apart and I keep finding bad components. Glimm lamp isn't lit when AC is applied, it has popped at some point. Capxon everywhere. The Capxon LP primary cap has 252uF (real parameters: 420uF 420V ) capacity and it looks corroded
One of the 12V diode is shorted. Just great...
I hope the other fsp units aren't like this.

[Wester547]

Re: Half dozen psu repair challenge

That circuit reduces efficiency... it increases power factor. It's a valley-fill circuit. Disadvantage is ripple is about 50% of line voltage (stretches transformer design) but PF is about 0.8~0.9 ideally.

Wouldn't 50% of the line voltage be a good thing, if you mean 50/60Hz ripple as opposed to 100Hz/120Hz ripple (unless you mean something else, like 25/30Hz ripple with half wave rectification and 50/60Hz ripple with full wave rectification with passive PFC involved)?

[momaka]

Re: Half dozen psu repair challenge

Ah, yes I have replaced all the small lytic caps too. Standby got back to normal right after I replaced the caps.

Check if PS-ON signal is present. Also check the standby auxiliary voltage rail on the secondary side (if there is one). If both of these are okay, check the auxiliary voltage on the primary side next.

[tom66]

Re: Half dozen psu repair challenge

Wouldn't 50% of the line voltage be a good thing, if you mean 50/60Hz ripple as opposed to 100Hz/120Hz ripple (unless you mean something else, like 25/30Hz ripple with half wave rectification and 50/60Hz ripple with full wave rectification with passive PFC involved)?

The main caps get 50% ripple, so around 320-160V (160Vp-p) operating range for the PSU, which means the PSU transformer and controller has to be designed to work over a wide voltage range, which is tricky... The ripple is still at 100Hz. Typical ripple is around 70Vp-p at the point where the control loop begins to struggle and 100Hz becomes dominant on the outputs.

[Wester547]

Re: Half dozen psu repair challenge

I thought half wave rectification (excepting the stress of APFC designs) meant a ripple frequency that is equal to that of the mains rather than double that of the mains... did I miss something? :o

[tom66]

Re: Half dozen psu repair challenge

It's not half wave rectification. It's full wave rectified.

Here's an example of what the valley fill circuit does. Efficiency of this stage is 98.5%, power factor is 0.95 for a 560W load. So it works fairly well, and it's efficient.

The resistor is for limiting inrush current, and may be removed and replaced with an inrush NTC on the input; it also helps reduce the THD.

[Wester547]

Re: Half dozen psu repair challenge

Oh, that's another type of PFC. I'm curious, though, does APFC use full wave rectification (in forward PSUs) since its ripple is essentially equal?

[tom66]

Re: Half dozen psu repair challenge

Yes, most (I don't know of ones that don't) APFC circuits use full wave rectification.

[Wester547]

Re: Half dozen psu repair challenge

So a passive PFC is a low line frequency (50/60Hz) inductor and Active PFC uses a higher switching frequency... but with half wave rectification, would a passive PFC (50/60Hz) cut the ripple frequency in half again (to 25/30Hz) or would it remain the same? And would a passive PFC in a bridge PSU mean half wave rectification?

[tom66]

Re: Half dozen psu repair challenge

No, most (all?) passive PFC systems use full wave rectification too...

[Wester547]

Re: Half dozen psu repair challenge

But passive PFC is a 50/60Hz inductor (at least referring to passive PFC in forward topology). Full wave rectification has a ripple frequency that is double that of the mains, right?

[tom66]

Re: Half dozen psu repair challenge

The passive PFC coil goes before the bridge rectifier in most cases, so it sees 50/60Hz.

[Wester547]

Re: Half dozen psu repair challenge

If it's before the bridge rectifier, it wouldn't change the type of rectification a forward or bridge PSU would use, would it (or would it have to be after the rectifying bridge to do that, even though if it was after it would be operating at a 50/60Hz line frequency)?