Intel D945GTP (PLM?) dies when 4pin Molex connected

**c_hegge** · 02-14-2011, 03:44 AM

Re: Intel D945GTP (PLM?) dies when 4pin Molex connected

You say it works with the first PSU. That tells me it is a problem with the second PSU that can't handle the extra load put on it by the CPU. What brand and model are they?

If I read that wrong and it doesn't start with either PSU, then it sounds like a shorted MOSFET on the CPU VRM.

**mockingbird** · 02-14-2011, 03:49 AM

Re: Intel D945GTP (PLM?) dies when 4pin Molex connected

I have the same problem with an Asus P5B. With the 4 pin 12V connector, when the power is switched on, the fan twitches for an instant. Someone suggested that one of the mofsets are burnt. None of them show visible damage, but one of these days I will probably replace them to see if it helps.

**Barf** · 02-14-2011, 03:52 AM

Re: Intel D945GTP (PLM?) dies when 4pin Molex connected

Originally posted by c_hegge

You say it works with the first PSU. That tells me it is a problem with the second PSU that can't handle the extra load put on it by the CPU. What brand and model are they?

If I read that wrong and it doesn't start with either PSU, then it sounds like a shorted MOSFET on the CPU VRM.

Sorry, badly written. No, neither PSU work with it. I only have the 2nd one though - the first was swapped out back at the uni's workshop.

1st PSU - dead
2nd PSU - ran for a few minutes, now dead

A bit of further info - Just done a resistance test between the connector (on mobo) and ground, and its only 4ohm, that to me is far too low (incidentally - im an electronics engineer, but my field is radio comms not computers!) and suggests something not necessarily shorted out, but certainly very leaky

The nearest cap to the connector (and to which its connected!) is a michicon HD series

**Barf** · 02-14-2011, 03:55 AM

Re: Intel D945GTP (PLM?) dies when 4pin Molex connected

Im tempted to pull the first cap and see if i get anywhere with it out of circuit. Anyone see a good reason not to try this? ok so the smoothing will be out and probably reduce the stability of the rail, but hey its doing nowt anyway!

I can test the caps but only for value, ive no ESR or leakage meter

**c_hegge** · 02-14-2011, 04:41 AM

Re: Intel D945GTP (PLM?) dies when 4pin Molex connected

I don't think that the caps are the problem. What brand and series are they? As I said before, this sounds more like a shorted MOSFET.

**Barf** · 02-14-2011, 04:49 AM

Re: Intel D945GTP (PLM?) dies when 4pin Molex connected

Originally posted by c_hegge

I don't think that the caps are the problem. What brand and series are they? As I said before, this sounds more like a shorted MOSFET.

Ok, sounds plausible. Question is then - how to locate it?

In an ideal world (and as is the case with everything else i repair) i'd have a schematic of the circuit, so know where and what to trace. I dont have that in this case, and dont know the circuits in question.

Can you suggest to fault finding tips?

**Barf** · 02-14-2011, 05:03 AM

Re: Intel D945GTP (PLM?) dies when 4pin Molex connected

ok, i could be barking up the wrong tree here, but there are two SMT parts, marked 545 T40 N03G, that read short across their pins, located next to each other and very close to both the CPU and 4pin connector. It apears from a rudimentary search on google that these are N03G switching power MOSFETs.

They look like good candidates, provided of course their reading short from gate to source isnt as a result of something else in the vicinity.

I dont fancy placing an order for them if they are not likely to be at real fault. Can anyone suggest any further tests i can do? I could of course take them off board, but they are located in a bit of a tight spot and ive no heatgun for this job

**seanc** · 02-14-2011, 03:30 PM

Re: Intel D945GTP (PLM?) dies when 4pin Molex connected

Test them all with the diode function of your multimeter. Remove all shorted ones from circuit and test them again - if they're still shorted, replace them.

Next question, if they are shorted, why? Any blown caps? Have you checked the insides of the PSUs?

**Barf** · 02-19-2011, 04:48 PM

Re: Intel D945GTP (PLM?) dies when 4pin Molex connected

Seanc, no obviously blown caps, no obvious problems at all - unless you count the abortive attempt of mine to lift one of the fets for off-board testing! (i now definately need to replace at least one fet

)

As ive noted on the thread running about the PSU for this machine - it looks like im at risk of getting into a fault loop, with each device killing the replacement!

Im now on the second PSU (both same type) and have a spare mobo (same type but thoroughly soak tested!) being shipped to me. Were working on the principle that a failed mobo took out the PSU. This second PSU has been with the same mobo, but not for long, and was pulled as soon as problems arose. With luck its survived the stress. Im going to load test it on the bench, and if alls well then try it with the replacement mobo. I dont want to put it straight on in case the PSU is faulty and it kills the board!

**Barf** · 02-19-2011, 04:53 PM

Re: Intel D945GTP (PLM?) dies when 4pin Molex connected

I think this might all be getting a bit complicated and muddy! Heres the tale so far, in bullet points!

1. Obtained complete PC
2. Dead on arrival - Pulled PSU and bench tested
3. PSU believed dead, although could be made to run on bench
4. PSU replaced with identical unit from an identical PC - worked for quick test
5. PC died a few minutes after start-up (once it was 35mile from the lab!)
6. Diagnostics showed PSU died when 4-pin connector connected
7. PSU pulled and bench tested - ok
8. Mobo believed faulty - low ohms tests show shorted MOSFETS (in-circuit)
9. Soak tested replacement mobo requested

so, you see the issue - the mobo looks dead, but has it killed the 2nd PSU? I dont want to just attach the new mobo if the PSU will kill it, so will load test the PSU on bench first.

**seanc** · 02-19-2011, 05:02 PM

Re: Intel D945GTP (PLM?) dies when 4pin Molex connected

I'm slightly confused. You say (7) the PSU works when you bench test, but not when you connect it to the board (6).
That still says board short to me.

If you short the PSU on and measure the outputs on the 4 pin P4 connector - does it have 12v?

**Barf** · 02-19-2011, 05:08 PM

Re: Intel D945GTP (PLM?) dies when 4pin Molex connected

Hi, yes you have it right. I bench tested the PSU. With the PS_ON line shorted all the voltages read ok.

I do agree with you, it does seem most likely that its the mobo thats failed. My worry, is that something fails with the PSU when its having to supply a decent level of current. Obviously with only my multimeter on it its not having to do any work! If i can load the PSU up on the bench and prove its ok when supplying a decent amount of current, i'll be happier attaching it to a new board!

In effect, i just want to prove the PSU, so i know i dont have to worry about it!

**seanc** · 02-19-2011, 05:15 PM

Re: Intel D945GTP (PLM?) dies when 4pin Molex connected

Plug a load of old useless hard drives into it or test with an old motherboard you're not worried about.

You could also search the forums, I think Th3_Un1qu3 has made a PSU load tester.

**Scenic** · 02-19-2011, 09:03 PM

Re: Intel D945GTP (PLM?) dies when 4pin Molex connected

sounds like the classic shorted CPU VRM MOSFET problem.
had the same problem with an Asus A8V or something. one of the gazillion of mosfets in the CPU VRM area was shorted. after trying to carefully lift the pins and destroying a couple fets in the process, i thought what the heck, noted down the part numbers and removed all the mosfet pins (in a rather destructive way towards the fets, making sure not to damage any traces).
short on the 4pin P4 plug was gone after that. have to order a bunch of fets and hope that was the only problem of this particular board.

might be the same thing for you, except the 2nd PSU you tested didn't shut down (no short circuit protection?!) and the shorted fet(s) most likely heated up a lot, but didn't smoke..

what make/model are those PSUs?

**Barf** · 02-20-2011, 04:00 AM

Re: Intel D945GTP (PLM?) dies when 4pin Molex connected

Originally posted by seanc

Plug a load of old useless hard drives into it or test with an old motherboard you're not worried about.

You could also search the forums, I think Th3_Un1qu3 has made a PSU load tester.

not possible, i dont have any spare old stuff - computers i only fix when i have to!

**Scenic** · 02-20-2011, 06:02 AM

Re: Intel D945GTP (PLM?) dies when 4pin Molex connected

lightbulbs?
like 12V halogen ones for example

**Barf** · 02-20-2011, 06:05 AM

Re: Intel D945GTP (PLM?) dies when 4pin Molex connected

Originally posted by Scenic

lightbulbs?
like 12V halogen ones for example

thats what im a-thinking. I have spare headlamp bulbs for my car, 12v 55w halogens. Not sure what to load it up to though. From what ive read it seems the 4pin connector was added to provide more power originally for 2x PCIexpress slots at 75w each, so perhaps loading it up to 110 - 165w (2-3 bulbs) would be enough to show any problems under load?

**seanc** · 02-20-2011, 06:27 AM

Re: Intel D945GTP (PLM?) dies when 4pin Molex connected

The 4 pin connector is extra power for the CPU.
The 6 pin connector is for PCI Express.

**Barf** · 02-20-2011, 06:32 AM

Re: Intel D945GTP (PLM?) dies when 4pin Molex connected

Originally posted by seanc

The 4 pin connector is extra power for the CPU.
The 6 pin connector is for PCI Express.

ah, ok then. No 6pin on this. So the question is then, whats the power rating i need to load it to test?

Intel D945GTP (PLM?) dies when 4pin Molex connected

Intel D945GTP (PLM?) dies when 4pin Molex connected

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Related Topics