Oddest Behavior.

[jarrod0987]

Hi folks. Working on a Corsair PSU. I am semi new at SMPS but I have around 1500 pro hours working on 230/400 in the combat zone and some lesser amount on the side with 120/220 state side. I also have about 8000 hours pro time as a bench tech. So I feel like I know how to do this kind of work safely. I have been reading up on SMPS prints and watching video's for about 1.5 years now. Worked on one already which I had to change about 10 components. Cheap POS amplifier.

Anyways....Back to the weirdo Corsair issue. Fuse was blown when I got it. Did the light bulb trick and it stays on for any length of time. I noticed with the other PSU I worked on the bulb never really went out but dimmed some within the first few seconds usually. This one seems to bright to look at. So I assume a short.

Just a quick Q...Is this light bulb trick 100%? Does it always go dim or off or blink then off when things are good? Is there any scenario on any model where it might stay bright and not go out?

I ask because here is what happened. I have one of those laser temp probes so I let it run for at least 5 minutes and check all the major suspects for heat. Found nothing except 2 resistors that were about 80F (Ambient was 70F). They had that damn glue on em so I scraped it off after I discharged everything.
I disco'd and discharged everything but the caps 0v anyways. I checked out of curiosity. So I started probing around. Too my mind if something is shorted it will be shorted in the board. So If I'm looking for bad MOSFET etc, it should show as shorted. Take it out and retest. I checked the fets for the PFC, The fets for the main transformer, the MOV, even the caps. Nothing came up shorted. I even Ohmed out the AC and DC side of the rectifier sandwich. AC was infinite and DC was about 2K.
Normally I start pulling stuff out at this point but my 25 year old solder vac finally gave up the ghost. I have another 1 on order but that will take a while.

So I just could not accept that the light was on and I could not find a short. Seriously. I have a clamp on am meter and I was drawing about 750mA while the bulb was on. I decided to put on the rubber gloves and get serious. The bulb was dropping 116v and only 3v at the AC side of the rectifier and 1.5 volts at the DC side. Powered it off. SO by my calculations the bulb was only providing around 155 ohms. Not really all that much for 116v. But yet that bulb is hogging all the volts so there must be very few ohms in the rest of the circuit. A short. Probed and probed and couldn't find it. I even check the ohms on the outputs and the lowest reading I got was 300 ish ohms. Seems enough.
I had almost given up but then something happened. I was making one more time around with the diode tester and I found a almost dead short on one of the main filter caps. Here is the weird part. They are both in parallel across the + and - Pads. But the other one reads infinite. It's seriously on 1 inch away. As I creep the probe closer to the other end of the pad I go from infinite to 100 ohms to almost a dead short. At first I thought maybe there was some coating on the traces messing me up but I have sharp points and I left marks in the solder now ffs. I'm definitely pulling that cap when my new solder vac comes in but WTF. All I can think of is there is something strange going on inside the board. It has a front and back so maybe an inner layer too that's messed up. So odd.
Anyone have any insight?

[petehall347]

Re: Oddest Behavior.

cracked board trace .or dry solder joint

[jarrod0987]

Re: Oddest Behavior.

That's what I thought but then how is it varied across the same pad? How is it drawing all that power with a open circuit. A open circuit should produce no light at all. I have bright light which means a short somehow which I did find. I did confirm there are some inner layers to this board. Hard to see under the electrolytic but it is there. There is definitely some bad soldering inside this thing. Not terrible but I see some obviously cold joints on the thicker wires.

[momaka]

Re: Oddest Behavior.

Incandescent bulb trick always works. (Though on some PSUs with APFC, it can take a few good seconds before it goes dim until the main cap charges.)

As for one of the caps reading shorted while the other reading open-circuit, with the two caps in parallel.... the only explanation I can think of that can give you this result is residue flux on the board. Sometimes, you don't see it on the solder joints at all, but it is not until you poke around a little with your multimeter probes that they break through the flux layer and start conducting to the joints. Hence the "mismatched" readings.

It would be interesting to find that one of the main capacitors are shorted. I just had a PSU with similar failure mode: ThermalTake Smart M Series M850W. I posted about it in this thread:
https://www.badcaps.net/forum/showthread.php?t=62832
New primary cap seems to have fixed it for now, thought I've barely tested it so far.

[jarrod0987]

Re: Oddest Behavior.

Worked on this some more today. Determined the main problem is a blown mosfet on the APFC circuit. Obviously smoked and I missed it during my visual. Hate that. Also, it tests good from the Peak Tester. Even though half the writing burned off. The interesting thing momaka said came up When I tried to put one of the caps back in there was one leg that just would not play so I had to clean the crap out of it and it's still kinda sketchy.

I did find the caps were not charging before with the light bulb trick but they are now. 150v on both which sounds right for 120AC to me.

So far I cannot source the PFC MOSFET and I am way over the worth on hours anyways but really it was just for learning and to see if fixing the name brand 800W plus PSU's was worth it. As there are some huge lugs I would need a 300$ desoldering gun and the time it took for disassembly, analysis, and part hunting, I would say not. It was a blast being able to find the problem though.

[momaka]

Re: Oddest Behavior.

Good to see you found the problem.

So are you still going to fix it?

If not, then you may want to put it on eBay for sale for parts or repair. Or even here on BCN. Seems like a decent PSU to fix. You shouldn't need a $300 desoldering gun to remove parts. A well-designed soldering station rated for 70 Watts or more will do fine, but it has to be something that uses tips with a built-in heating element. Stations like that can be found for under $100 (I have a Circuit Specialits / Aoyue CSI 2900 70 Watt station that still sells for $60 last time I checked). There are also the portable versions of those, and they cost even less than $50. With those stations, you should be able to remove pretty much anything. I use my CSI 2900 to remove SMD TO-252 MOSFETs all the time (with a small help from a 30 Watt RS iron to desolder the MOSFET's legs).

[Quaddro]

Re: Oddest Behavior.

Light bulb trick is always 100% accurate.
When your your bub is light on when you put your psu switch on (from switch behind your psu), there must be shorted.

And dead fuse almost mean shorted mosfet..