Blown cap (plus blown FET?)

[pentium]

Knew it was there but after a power failure my system started to really show signs of a bad cap as it would constantly reboot itself before it could even POST or just lock up on POST.
Board is an MSI 865PE.



I'm replacing all four of the 16v 1500 KZG's. While I was preparing to remove them however I noticed there was dirt that I could not clean off one of the ferrite cores, one of the caps and one of the FET heatsinks. Upon closer inspection it looked like at least one leg of one FET was discolored and oxidized. Pulled off the heatsink and there was a square soot outline.
Yet with these evident signs I could not find either FET to be shorted. Gate to Source are infinite with both pins reading between drain and either gate or source. I do believe neither is blown open either, right?




System never unexpectedly turned itself off before. Only time it gave magic smoke was when a poly fuse on a USB card committed suicide.

[ben7]

Re: Blown cap (plus blown FET?)

Could be from the bad cap, or the MOSFET could just be partially broken. I know it happens with regular BJT transistors, usually the hfe goes down and they become noisy. Replace caps and test system, checking if the heatsink for the suspect MOSFET is hotter than the others.

[pentium]

Re: Blown cap (plus blown FET?)

Recapped the system and powered it up. System went through POST, crashed on boot, made it up in safe mode and almost booted on its own again before going down again. Possibly my OC settings. While in the BIOS the system locked up, I pressed restart and the PSU crowbarred. Now the PSU instantly crowbars if you reset the breaker and press the power button (well, you get a brief blip form the power LED because the system default is to power on when power is restored). Issue continues with drives and video card disconnected. Power supply does NOT crowbar if tested separate from system.

[trodas]

Re: Blown cap (plus blown FET?)

Looking at the PCB - it makes me wonder, if the missing phase could not be added back up...

BTW, the mosfets are looking like that they need to be replaced with some hi-quality components.

[momaka]

Re: Blown cap (plus blown FET?)

Sounds like it's time for new MOSFEs. Check resistance across the 12V CPU connector. If you see low resistance, it's definitely time for new MOSFETs.

[lti]

Re: Blown cap (plus blown FET?)

What breaker are you resetting? Do you have some weird PSU that has a circuit breaker?

[pentium]

Re: Blown cap (plus blown FET?)

What breaker are you resetting? Do you have some weird PSU that has a circuit breaker?

It's just the usual thing that trips for any short condition and resets if you remove power for five seconds.

Also, define a low resistance. Checking between 12V and GND on the four pin connector gives a value that wobbles between 30k and 1.3k, depending on how long you keep the probes attached (value decreasing with time).

[momaka]

Re: Blown cap (plus blown FET?)

Under 100 Ohms usually.
The resistance reading you got doesn't indicate any obvious problems. Did you perform this measurement with or without the CPU?

What you should try next is test the low-side and high-side MOSFETs in the VRM individually.

To test the low-side MOSFETs, remove CPU and measure resistance on the motherboard between CPU VCC (the VRM output) and ground. Hold the probes on there for a few moments so the caps have time to charge up (to get more accurate reading). Then post what resistance you get. If it's in the kOhm range, they are probably fine.

For the high-side MOSFETs, also remove the CPU and measure resistance on the motherboard between CPU VCC and the 12V points on the 12V CPU connector. Hold the probes for a few moments so that the caps have time to charge up. Then post what resistance you get. Again, if you get readings in the kOhm range, then those should be fine too.

[pentium]

Re: Blown cap (plus blown FET?)

The Cat Came Back.

In the end I sourced a replacement 865PE and transferred the four 16v 1500's over. One of the 6.3 1500's however was bulging as well but being UNC I thought it must of been a one-off as the rest were fine so I swapped out that one cap.

Some two years later I come home to a reset machine that would crash every time you hit the desktop. Safe Mode also seemed impossible to boot into. I suspected corrupt video drivers and when I was 75% through that process it suddenly started crashing at POST and I then knew exactly what was happening.

Opening the machine up revealed that excluding the one cap that I had replaced, All seven other UNC caps on the rail had now failed.
I guess it took that two years for them to all fail and cross the point where the remaining healthy cap could no longer save the system.
I do not have replacement caps on hand, however I still have my old board which used Panasonic caps instead of UNC's so I have carefully removed them and transplanted them over. The machine has yet to be tested, but these caps have a MUCH better ESR reading than the ones I just removed.

[Wester547]

Re: Blown cap (plus blown FET?)

KZGs, KZJs, and all their derivatives are never fine just because NCC makes them. They are faulty enough to bulge without having ever been used. See here:

https://www.badcaps.net/forum/showpo...5&postcount=21

[pentium]

Re: Blown cap (plus blown FET?)

Well the recap has apparently been successful so far as we can make it past POST, however while resetting and verifying my OC settings memtest indicated we were too late....

One of my matched Corsair XMS 1gb PC3200 sticks has a stuck bit. The whole DIMM is pretty much useless now. This was verified by putting it into another machine as the lone installed DIMM with no OC settings. Once removed however we're passing everything with flying colors.