Power Supply POP! Why?

[jocanon]

I have a question on a power supply I was testing. So I was testing this new power supply I got (well, new to me, but it was taken from a working computer server so it's really used). I plugged it in to my dummy load, cranked the amps up to about 18 amps and everything was going along just fine for about 10 min or so. Then, when I was handling it, with my ear up close to the fan to hear how loud the fan was, I just about went deaf when a very loud POP sounded off and everything just shut down. It tripped my house breaker. Once I turned the breaker back on, everything, except the power supply, still worked (i.e. my dummy load was not damaged). Now when I plug in the power supply that POPPED, the fan spins up but there is no DC power, so it's dead.

Any ideas what might have gone wrong? My theory is maybe one of the caps shorted to ground by touching the metal case because it happened when I was handling the case and putting a slight bit of pressure on the case with my hand so I may have made it touch accidentally. I did not get any electric shock when it happened because the case is grounded to AC ground, just an ear full. I would like to know what went wrong and if the problem can be corrected because I have 200 more of the same model of power supply. If it was a cap shorting to ground, how can I test and find which one so that I can add more insulation at that point??? Thanks in advance for any help!

 

[JimB]

At the risk of stating the obvious, if you think that something shorted to the case, simply remove the case and look at the inside surfaces of the case.

If it went with a bit of a bang as you suggest, then there will be a silvery black "splash mark" on the inside surface of the case where the live circuit made contact with the case.

JimB

 

[alec_t]

Have you checked for a blown internal fuse?

 

[jocanon]

I did open it right after it happened, expecting to see black marks, but everything looked normal. A lot of the power supply is buried inside and I can't see it without taking parts out. I didn't have time to do that yesterday, but if it was part that's buried then that would negate the theory of it being a part that touched the side of the metal case...just thinking out loud.

As far as the fuse, I just did a cursory look the insides of the power supply and nothing looked out of the ordinary. I will take a closer look tonight, but I don't think there was a blown fuse. Are there fuses that might not be in obvious places that I should be on the look out for?

 

[cowboybob]

Any chance you have, or could get, a schematic of the PS?

What output voltage were you using with the dummy load: 5 or 12VDC? Are ALL the outputs of the PS dead? What was the dummy load?

Since the fan still runs, can you determine if the fan is AC or DC?. What is its voltage rating?

Generally speaking, from your description of the circumstances and result, it has all the earmarks of a major filtering cap blowing (shorting, plus whatever else might have subsequently died) as the result of an extended overload.

 

[tvtech]

I did open it right after it happened, expecting to see black marks, but everything looked normal. A lot of the power supply is buried inside and I can't see it without taking parts out. I didn't have time to do that yesterday, but if it was part that's buried then that would negate the theory of it being a part that touched the side of the metal case...just thinking out loud.

As far as the fuse, I just did a cursory look the insides of the power supply and nothing looked out of the ordinary. I will take a closer look tonight, but I don't think there was a blown fuse. Are there fuses that might not be in obvious places that I should be on the look out for?

I really feel for you. I have seen it many times here where I am...

From working to Dead in the blink of an Eye......SMPS don't announce a potential problem.....they just die.

Unlike Linear that gives you a bit of time...anyway, going off Topic here.

I understand you have around 200 of these that were used in Servers......so according to my tvtech logic...it was working and then you killed it by bending it's Enclosure........ logically then I would say look for maybe Dry Joints. You have isolated a few things...

1. No Shorts on the bottom of the board or black marks either..
2. Cooling fan still runs....that means is the Main Fuse is not blown either....

A process of elimination...was working happily....twisted or bent it...bang. So hard to give advise...As much as I would like to.

SMPS supplies are very hard to repair if you are not in front of it and dealing with it....

My suggestion....leave the other 199 alone until more information is given from you and we can figure things out properly... of what maybe went wrong?

All the best,
tvtech

 

[jocanon]

I don't think I will be able to get the schematic. It is an HP DPS-1200FB. Here is a link to an eBay add that sells them. The specs are in one of the pictures.

**broken link removed**

I was using the 12VDC output voltage when it blew. Voltage rating of the fan is 12VDC. I don't know yet if all the outputs are fried, probably not now that I think about it...the fan is DC and it still runs. The dummy load is one that I built, basically it is a bunch of MOSFETs soldered to a water cooled copper pipe linearly controlled, the power is burned off in the MOSFETs.

One thing I didn't mention in the OP...and probably should have...I put it together quickly last night and did not take the time to insulate the positive and negative leads that I soldered on to it (they are female bullet type connectors on soldered onto 10AWG wire). I didn't mention it because I don't think that is what caused the problem for two reasons, one, I really don't think they ends connected, and two even if they did, the previous model (DPS-600PB) has short circuit protection built where the power supply simply shuts down when you short the leads and then starts back up again on power cycle. I would be really surprised if the new model does not have that. At any rate, I am going to purposely short the leads together on another working unit and see what happens when I get home tonight just to rule that out.

 

[cowboybob]

Kind of a radical test but, what the hey, since you've got a bunch of them...

Came across this info on the hot-swappable PS you've got:

"solder a 560 ohm resister jumping pins 33 and 36" which, I assume, is to fool the PS into thinking it's been plugged into the appropriate receptacle.

with this photo:

Don't see, right off, how you might have distorted the case to the point of shorting anything out.

 

[jocanon]

Yes, that's what I did...the resistor that is. Except I used a different value resistor which doesn't matter, it will turn on with anything between about 300 ohms and 1k.

 

[ronv]

Usually caps are the loudest, but FETs can make a pretty good pop. Look closely at anything mounted on a heat sink. Since it is like a 100 amp supply anyplace + & - touch will make a bang. Don't just touch the output wires together. It could spew metal.

 

[jocanon]

Usually caps are the loudest, but FETs can make a pretty good pop. Look closely at anything mounted on a heat sink. Since it is like a 100 amp supply anyplace + & - touch will make a bang. Don't just touch the output wires together. It could spew metal.

Yikes! Thanks for the warning, maybe in the other room with a broom stick... I want to see if they have short circuit protection built in anyway. Is there another way to tell? I can't find the spec sheet.

 

[jocanon]

I wonder if it would be ok to put the whole guts of the power supply in an ABS plastic enclosure? Does anybody see any potential problem with doing that? I could print it with a 3d printer and make compartments for volt meters, on/off switch, voltage adjustment switch, etc... Also, if something did short to the metal case, it would solve that problem.

looking at this one

https://www.studica.com/us/en/XYZpr...&kpid=737744&gclid=CITb-cHMvb0CFZSPfgodkX4ADw

 

[alec_t]

Does anybody see any potential problem with doing that?

Only if the metal case is being used as a part of the heatsinking system.

 

[jocanon]

Only if the metal case is being used as a part of the heatsinking system.

Yeah, that's a good point...I did consider that too. So how would one tell if it is being used as part of the heat sinking system. I mean, there is nothing connected directly to the metal case except for the screws holding the down the PCB and the AC ground wire, BUT when it is running the case does heat up. So does the fact that the case heats up mean it is being used as part of the heat sink, or is that just a function of it's proximity to the heat, in other words, take away the metal case and it would still keep cool just as well?

 

[jocanon]

Oh, and by the way, if I do get the 3d printer, I decided to wait at least until v2 they are coming out with soon because it can print in 2 different colors at the same time.

 

[skimask87]

Doubt it highly. Air is a terrible conductor of heat, which is why you bolt small stuff onto bigger stuff...more thermal conductivity from a smaller surface to a larger surface of worse conductivity.

As far as a 3D printer for cases, I wouldn't. I've got a Makerbot Rep 2X. I used to use it to print out small cases for LCDs/buttons for various projects. No more. Takes hours to get a decent case printed, not to mention it's usually the 3rd try that finally fits. You'd have to print hundreds, maybe thousands of cases, to get your money back. I switched back to surfing the enclosures listed on Digikey and finding one that I can grind out with a dremel.
The 3D printer sounded like a really good idea for cases when I got it...not so much any more.

 

[jocanon]

Thanks for the input skimask87. If I go the 3d printer route, I will be doing hundreds of them. I know it will be slow printing, I envision having it running during the day while I am at work, with my wife (who stays at home) clicking "print" ever so often. I like the idea of being able to customize it, even including my logo on the case. So you said it took 3 tries to get it to fit right, but that is because you had to tweak the design, right? I mean once you got the perfect design then you just click print and it should work every time, right? Or does the quality of ABS plastic not always hold up? I imagine there are differing grades of filament you can put in it with better or worse results too...you probably get what you pay for in the quality of ABS plastic I would imagine. I am not dead set on a 3d printer, so if it's a bad idea I want to know.

 

[alec_t]

there is nothing connected directly to the metal case except for the screws holding the down the PCB and the AC ground wire

Then the case isn't being used as a heatsink. If it were, you'd have FETs or whatever bolted to the case.

 

[skimask87]

Slow doesn't describe it. Like watching paint dry or grass grow (hey, that's Astroturf!). Last week I printed out 18 gas can caps. Took 9 hours. 2 of them were messed up. (why gas can caps? because the new plastic gas cans come with those things on the end that you have to push to get any gas out and all my gas cans are the older style and the caps keep getting lost. Yes, the gas fumes will eventually eat away at the ABS cap, which is why I printed 18 of them)

You don't HAVE to watch the thing run all the time, but I do. I'm using mine correctly, using the right filament, good temperature control in the shop, correct settings, etc, and still, occasionally, a print goes bad, piece of filament has some dust on it and doesn't stick, ruins the whole thing and have to start over. Or something stupid like running out of filament 1/2 way thru a print.
I don't care how good a company says their machines are, they aren't 100% 'click and print' all the time every time just yet. Like I said, I've got a Makerbot, supposedly one of the higher quality machines out there and I still get fails here and there.

If you've got the $$$ to blow on a machine, sure, go for it. Fun to play with. I got mine to kinda help offset some of my business taxes last year, and I can print stuff like replacement knobs or levers or whatever ('cause my business is fixing stuff). But, if you're on the border because of $$$, I wouldn't...

 

[jocanon]

hm, something to think about...sounds like they are not quite up to snuff yet...I say give it a few years and I think they will be a household item. Coupled with a 3d scanner, now that would be cool!