SSI power supply help

[Photoman]

Hi All,

I am trying to repair an SSI PFQ355 power supply.  Its not going well...does it ever?
It has 5 output voltages, +5v, +24v, +15, +15, +12.  The +5 I think is 7-10 Amps.  The others are probably closer to 1-2 amps.

Ive done the usual check, the caps all check out within spec and ESR.
Checked for any shorted or exploded components, none that I could find on a first pass.

The power supply just blows the input fuse, a 7 amp AGC type.  Its spectacular and instant.
So I think something is definitely shorted but where?

I did pull the full bridge rectifier and up to that point, it wont blow the fuse.  The bridge does seem to check out as well.
I did try to just power the board with my Rigol DP832 30v 3a (max one channel) and I didn’t get anywhere, no current draw at all.
I placed the DC + on the DC output of the bridge and - on the -.  I could parallel up the channels?
My thought was I could at least see if something like a diode or transistor was getting hot.


2 things in the pictures, That resistor must have gotten really hot, I peg it as a 470 ohm (Yellow, Purple, Brown, Gold).
Does check out surprisingly at 460 ohm.  But I can see the resistance wire burned in the outer casing.
That resistor was across the +5V output (between + and -).  That gave the 5v output about a 24 ohm between +5v and Negative.
With it removed, its open i.e. no resistance between the + and - but it will still blow the fuse.
I don't think that has anything to do with it, it should at max draw about 208 mA (24 ohms 5v).  I think its just a minimum load resistor.

Also the NTC has a chunk missing and yet also measures good.  Resistance at room temp about 14 ohms, data sheet said 10 ohms.
I replaced it with an NTC I had that was close to the same size.  Though it made no difference.


Im kinda at a loss as to where I should go from here.  Should I just pull all the transistors, dual diodes and regulators?

Any thoughts would be great!
Thanks!!

[TheMG]

More often than not, when a switching power supply instantly blows the fuse, one or more of the main switching transistors are shorted.

[Photoman]

Thanks.  Tomorrow ill pull them all and see if that’s the case.

[coromonadalix]

More often than not, when a switching power supply instantly blows the fuse, one or more of the main switching transistors are shorted.

Due to the frequency smps work(s)  sometime in the 100khz to 200khz (more or less)  the fuse is the last thing to blow...  you have to take them out of the pcb, you may have coupling transformers or other parts who may interfere with your readings / tests

And without a schematic, you may embarq on a long journey.

[Photoman]

After a lot of desoldering, I think I finally found the culprit.
A BUF420A that shorted on all 3 terminals.  I was kinda hoping it would have blown a crater but it looks like it died silently.
Also a couple of H15NA50FL's that read marginal on my transistor tester.
Im placing an order from Utsource for the parts.  They're obsolete (of course) but they have them.

With a little luck, those will be the only problems (famous last words I know...im buying 10 just in case).

[TheMG]

Be sure to check other components particularly those driving the base of failed transistor. Often when a switching transistor fails and shorts between all three terminals, it will also take out one or more of the components responsible for driving that transistor.

Also a technique I use when repairing SMPS is to apply voltage externally from a bench power supply to the switch control ICs (no mains AC input connected!!!). Looks like yours has two: a UC3854N which is for the PFC pre-regulator and a NE5560N for the main converter. Which of the two circuits is the shorted transistor associated with? I'm guessing the main converter?

What you want to do is refer to the applicable datasheets for the correct pin and voltage to apply (normally this will be a relatively low voltage), and use an oscilloscope to check for a pulse train at the base/gate of the switching transistors. This helps to determine if the drive circuits seem to be operating correctly or not, and hopefully avoid everything going "bang" the moment you power it up with replaced components in place.

[Photoman]

Thanks for the tip!  I already bought the replacements for those 2 chips just in case.
I pretty sure the NE5560 also died.  I had it powered the first time before it was blowing fuses (no output) and I was probing the NE5560.
I barely saw a waveform on the scope and then it blew the fuse.  I think it was running maybe a couple of minutes or so.

On a side note, I think the failure mode of these supplies is lack of cooling.  Its fanless and the heat sinking is questionable.
The paste that was used is kinda crumbly at this point.  That’s one thing ill address when I get the new parts.
Also, to me the sil pads don't feel like they're making good contact.  I think ill goop both sides to be sure.

[Photoman]

Encouraging news, the replacement parts arrived today and after some three handed soldering its working again!!
Only 2 issues, the 24v (+v2) is low at +21v.  Also the 24 ohm resistor I replaced is getting hot and stinky.
See the pic (ignore the flux, will clean later and goo near the caps, its glue).

It’s a 3W 24 ohm.  It started as a light purple and its more yellow now.
I really want to move it away from the caps but its way too tight and the leads aren’t long enough.

Its across the +5 and 5v ground.  I think its either a minimum load resistor or maybe to bleed the caps?
Should I just parallel up a couple or just wait till burns off the stinky coating?
Or a higher value resistor, 240 ohm?


The +24v supply doesn’t have an adjustment pot.  Only the 5v, 15v, 15v, and 12v have one.
I think the 24v is derived somehow from the 5v.  The other 3 are LDO's (LT1083).
More research to figure out how to bring that up.

Thanks!

[duak]

I would bet the 24R is a minimum load.  With any luck, the 24 V will come up to spec when the resistor is replaced.

[Photoman]

So good news.  I loaded down the 5v supply on my dc load, the 24v came good.
Makes sense now that I think about it.

So im just left with the hot resistor which was new from DigiKey.
I wonder if more current is going through it when its not loaded (from the dc load)?

The other thing I was thinking about was proper heat transfer.
I cant modify the supply or how the parts are mounted (it fits into something else).
So I was thinking of added heat sink compound between the sil pad and the metal case and between the sil pad and the transistor.
The way they have it mounted, I doubt adequate pressure can be achieved without it.  Especially considering the age of the plastic retainer.

I have Wakefield 120 (silicone) and 126 (non silicone).  Honestly don't know which would be better.
I don't think any part has silicone in it so melting shouldn’t be an issue.

[TheMG]

Was there any evidence whatsoever that the resistor had been previously replaced, or did it look factory? If it doesn't look like it had been tampered with, I would replace it with another 470 ohm resistor.

I can't explain the difference in in-circuit resistance. Maybe a manufacturing defect in the original resistor that caused it to be too low value, but somehow rectified itself from the thermal and mechanical stresses of desoldering it? Who knows, but I've got doubts that 24 ohms is the intended value.

While the resistor you put in is operating well within its ratings as far as power dissipation (a touch over 1 watt), it may be overheating due to lack of airflow and close proximity to other components which doesn't allow natural convection to cool it.

Anyways, try a 470 ohm resistor. If the power supply starts up and produces stable output on the 5V with no external load, you should be good to go.

[Photoman]

TheMG, thanks for the help!  I feel really dumb, I put the 24 ohm instead of the 470 ohm!
I have a distinct memory of holding the 470 in my hand and thinking why did I buy this and then putting the 24 ohm in.
That explains why its cooking