Debugging a Skynet SNP-Z101 110W PSU

[eccerr0r]

I was looking at all these small approx 4x6 open frame PSUs that typically are around 40W. Then I came across this "monster" 110W (130W max) unit! It was rated 15A at 5V, 5A at 12V, and a half amp on the -12V rail. Looking at its specs it should be 80% efficient unlike the other 4x6 where average in the 70% range.

However, it failed. Not surprisingly I suppose, though, it's too dense. Anyway it won't start properly. It comes up for a fraction of a second but goes dead to 0. It tries starting again two seconds later, and goes out. So, it's the dreaded blip blip blip type failure.

It has pretty much all Chemicon LXP capacitors on it except the bulk supply capacitor, which is a Chemicon KMG capacitor. What's surprising is that it's a 420V 47uF capacitor...

47uF?!?!?

Seems a bit low, as most of the other smaller openframe caps I've seen 100uF or more, and compared to the switched range PSUs that have two caps, it's 1/((1/470)+(1/470)) = around 235uF for a 200W PSU...

It uses a dreaded 8-pin SK-8120 PWM controller which I can't find much info about unlike the ubiquitous UC3842-series.

Anyone worked on these before?

[eccerr0r]

Re: Debugging a Skynet SNP-Z101 110W PSU

I just noticed the resistance from the -12V output to GND is 0 ohms...
Hmm.

[kaboom]

Re: Debugging a Skynet SNP-Z101 110W PSU

What's surprising is that it's a 420V 47uF capacitor...

47uF?!?!?

No, that's fine. There's a PFC booster feeding that cap.

Check for a shorted -12 rectifier.

[eccerr0r]

Re: Debugging a Skynet SNP-Z101 110W PSU

I destroyed the vias on the board trying to pull that #%*($@$ing rectifier. Yup, it shorted out. It was a 31DQ10. I stuck in an 1N5822 and it fried too. Crap.

Nevertheless ... WEIRD.

Perhaps a capacitor shorted out or something but it doesn't seem shorted out anymore once the diode was pulled. And the load seemed to be around 4.1K ohms or so...

So much for "short circuit protection"?

Oh well, running the PSU without -12V, at least the PSU powers up now.

[japlytic]

Re: Debugging a Skynet SNP-Z101 110W PSU

Never heard of Chemi-Con LXP Series before - was that a misread?
It seems that these units are Chemi-Con LXY Series.

[eccerr0r]

Re: Debugging a Skynet SNP-Z101 110W PSU

Yeah, looks like LXY, I think most of the caps are fine, except perhaps what's going on with the -12V.

I wonder if this computer can be booted without the -12... It's an old Watchguard Firebox X1000 rackmount firewall/router... The machine at least beeps, but I suspect the serial port may have problems working without -12, unless it has some max232 somewhere...

[kaboom]

Re: Debugging a Skynet SNP-Z101 110W PSU

It was a 31DQ10. I stuck in an 1N5822 and it fried too. Crap.

Nevertheless ... WEIRD.

Not at all- you replaced a 100 PIV part with a 40 PIV one.
There should be a small ceramic cap, in series with a resistor, near that diode, either across the diode, or from the secondary (anode side) to low-side common. That's a snubber. The cap may be anything from .01u to .068u, smaller sizes more common. That resistor will be from 3.3 to a few tens of ohms.

Even if the snubber's OK, you've probably got spikes or ringing beyond the capabilities of the '5822, especially with low or no load on -12.

Try an SB5100 or SB5H5100.

The same thing happend to me when I thought I could "get away with" a lower rated part. It was a Panasonic "inverter" supply for a microwave. It seemed to need 10-15kV devices. The strings of UF4004s I subbed added up to 6 or 8kV (can't remember) and the unit ran into current limit, with the diode strings avalanching and smoking.

[eccerr0r]

Re: Debugging a Skynet SNP-Z101 110W PSU

I ended up sticking in a 31DF4 as that's all I had, and stuck in a 100 ohm resistor in series. This time it didn't fry, and I got only -8V on that rail. Not great but good enough, not sure what this thing is used for on this board anyway. Good enough.

There appears to be no snubber across the diode though on this design. The weird thing is that this output is rated for -12V at 0.5A. Not sure why there is so much margin on this output...

[kc8adu]

Re: Debugging a Skynet SNP-Z101 110W PSU

lose the 100ohm and try it.
btw i rebuilt one of these in a cisco gigabit switch a while back.it was mainly the small caps that went.would not start up after power failed.
this one was like 11a at 12v

[eccerr0r]

Re: Debugging a Skynet SNP-Z101 110W PSU

I removed the resistor, it still worked...Good.
I wonder why the original schottky diode died, nevermind the 1N5822. This fast recovery regular diode seems to be holding up at least...

This should be an 80+ supply it seems, so I think I lost a bit of efficiency going down to a FR diode...

[eccerr0r]

Re: Debugging a Skynet SNP-Z101 110W PSU

LOL

The PSU failed again

Different failure mode. The PSU overheated due to a failed fan

What's funny: the label on the 47uF 420V cap browned. Also its ESR is now greater than 5 ohms, and it looks like it spilled its guts... ugh. It's a Chemicon KMG, and it's dead...

I don't have any suitable spares, sigh. Wonder if I should either trying a 10uF to see if it's the problem, or two 470uF 200V in series. Neither of which are permanent in any sense...

[eccerr0r]

Re: Debugging a Skynet SNP-Z101 110W PSU

Okay, best reasonable test spare is now two "470uF" caps that test as 348uF and a "330uF" capacitor, all three in series... this should be able to handle 400V...

Oh wait, that crappy JEE "330uF" measures at 220uF... Okay... I'll hook that "330uF" with a Panasonic 220uF (measures at 203uF). Now this should be near 100uF which is much closer to the 47uF...

The things one has to resort to when not having the right parts to test... and make sure they don't blow up because of mismatches...

---

Well, looks like the two capacitors in series did it... Power supply 12V came up and strong unlike before with the dead 47uF cap. I guess I need to put a 47uF 420V cap on my purchase order.

For the heck of it I measured the voltage across the capacitors... 400V (ouch!) I may have to get an even higher voltage capacitor for design margin? The original was 420V which is only 5% margin...

[RJARRRPCGP]

Re: Debugging a Skynet SNP-Z101 110W PSU

LOL @ Skynet

[eccerr0r]

Re: Debugging a Skynet SNP-Z101 110W PSU

Yeah, it got terminated.