Dead Asus P5P800-VM

[tony359]

Hello Momaka,

Thanks for taking the time to give me some advice, appreciated.

Tonight I did some more tests and I noticed that the voltage out of the inductors is not clean at all. Am I mistaken in understanding that that line is the VCore line and it should be very clean as it goes to the processor? See picture below.

I suggest replacing all three MOSFET drivers at once and all of the MOSFETs too. While at it, also check all of the MOSFET Gate resistors and caps, if any.

Very happy to do so. Would you be able to help me sourcing an appropriate replacement? The original components are, obviously, discontinued and I want to be sure I am selecting the right components.

Below is from a previous post:

My Mosfets are 04N03LA and 60T03GH. Those are long discontinues. Would you be able to help me finding a suitable replacement?

For the 60T03GH I found this

https://uk.rs-online.com/web/p/mosfe...gU3849ERaBpI-o

For the 04N03La I found this

https://uk.rs-online.com/web/p/mosfe...Yb6EygMdYB1UdI

They sell in packs of 10 so it would be £26 of components and I'd like to make sure they are suitable - any cheaper alternative is very welcome!

Also, is that discoloration on the MOSFET(s) near the AGP port or just bad lightning? I suspect there may be something going on there as well.

That's a good question. Those tiny resistors are getting very hot but I do not see a short. The resistors read ok. I removed the mosfet and it tested ok. I believe that is the power for the Northbridge - I should check that again.

Last but not least, you may want to replace those Panasonic FL 4V, 680 uF caps, even if they read OK. After a while (may power-on hours), I find that any liquid electrolytic cap on LGA775 mobos with the caps under the heatsink like that will eventually fail or develop high ESR or just get flaky. Using solid polymer caps like Sanyo SEPC or Nichicon LF or similar should fit under there and work great. I typically go with 2.5V, 820 uF.

Sure, why not. I have purchased some Wurth for another motherboard.

https://uk.rs-online.com/web/p/products/8392114/

They are on that other board and they seem to work ok.

No, I do not have a POST card. I was thinking of purchasing one but I have plugged a speaker and there is no error BEEP so I thought I would waste my money. Indeed it would be a good idea to try.

Cheers again for your help.

[dicky96]

WoW! Vcore should be stable, not a sawtooth like that!

Seeing as you have such as high current I doubt it is just because of bad capacitors.

Is the North Bridge getting hot?

Is something else getting hot?

[tony359]

yes the North bridge is getting VERY hot (without heatsink) but to be honest I have another motherboard with the same chipset and the heatsink gets REALLY hot after a while. Just wondering if that's just ok. I mean, even the CPU gets VERY hot on a normal system without the heatsink. But that's fine.

The only "unusual" hot components are those 4 resistors you noticed.

I've measured the VCore at the output of the inductors, is that correct?

[dicky96]

Yeah, the side where they all connect to the Vcore rail and you have all the 2.5V or 3V electrolytic capacitors

To be sure you are checking the right side, what do you see at the other side of the inductors?

[tony359]

The below is what I read on the mosfet side of the inductor, which is what I also read on the drain of the lower mosfet or source of upper mosfet if not mistaken.





I have ordered a new set of mosfets - they're coming from far away so it'll take a while.

Another thing that concerns me is that the Winbond chip doesn't seem to send a "power on" signal out when I attempt to power the board. Now, this system was using a contraption to use a rocker switch at the front but I would still expect the power on signal to come out of the chip somehow. It's concerning.

Still, the board shouldn't burn to death when the PSU is switched on

[dicky96]

So there is a heavy load on the Vcore rail then. Which we knew already from previous evidence. And you say that Vcore to ground does not read very low resistance?

[tony359]

Originally Posted by dicky96

So there is a heavy load on the Vcore rail then. Which we knew already from previous evidence. And you say that Vcore to ground does not read very low resistance?

Uhm... I wonder if I ever checked WITH the CPU installed!

While I read 100-300 Ohm (and climbing) without the CPU, I am reading 17 Ohm steady with the CPU installed.

How does that sound?

[momaka]

Originally Posted by tony359

Very happy to do so. Would you be able to help me sourcing an appropriate replacement? The original components are, obviously, discontinued and I want to be sure I am selecting the right components.

I think Xbox 360 boards use similar MOSFETs and drivers. That's pretty much all I've ever used for replacing MOSFETs, as I used to work in a console repair shop and accrued many dead Xbox 360 PCBs that weren't good for anything else but parts.

I see that you ordered parts already. But if anything happens and they don't arrive, I can check my stash of Xbox 360 PCBs and see if there is anything similar/same.

Originally Posted by tony359

That's a good question. Those tiny resistors are getting very hot but I do not see a short. The resistors read ok. I removed the mosfet and it tested ok. I believe that is the power for the Northbridge - I should check that again.

Yes, that would be a good idea.
By the looks of it, it's either power for the NB or power for the AGP port (1.5V). Probably best to measure the voltages on the input (Drain) and output (Source) of these two MOSFETs. They are used as linear regulators, so it's OK for them to run a little warmer. But from what your thermal camera shows, those appear to be running way too hot. If you can't keep your fingers on them, there may be something wrong with the NB.

Originally Posted by tony359

No, I do not have a POST card. I was thinking of purchasing one but I have plugged a speaker and there is no error BEEP so I thought I would waste my money. Indeed it would be a good idea to try.

Most POST cards are indeed a waste of time/money. But the older PCI-based one will sometimes show useful info for older motherboards. I've had motherboards stuck on a BIOS POST code and not beep to show errors at all (or beep to indicate a normal POST and then hang.) So if you can find one for very cheap (under $5?? I don't know what they go for nowadays anymore), it might not be a bad tool to keep around for working on older motherboards.

Originally Posted by tony359

yes the North bridge is getting VERY hot (without heatsink) but to be honest I have another motherboard with the same chipset and the heatsink gets REALLY hot after a while. Just wondering if that's just ok. I mean, even the CPU gets VERY hot on a normal system without the heatsink. But that's fine.

You shouldn't run anything without a heatsink.
That can only cause damage to the CPU and Northbridge with time, even if you do it only for a brief moment. I don't know where people keep finding this advice, but it's a POOR troubleshooting/testing practice to run anything without a heatsink that's meant to have a heatsink on.

Originally Posted by tony359

Another thing that concerns me is that the Winbond chip doesn't seem to send a "power on" signal out when I attempt to power the board. Now, this system was using a contraption to use a rocker switch at the front but I would still expect the power on signal to come out of the chip somehow. It's concerning.

That is indeed concerning. Makes me wonder if something / the mobo got hit with a lightning strike or a power surge. Might be worthwhile to check the LAN chip too to see if it's getting hot, as that's usually the first one to "bite the dust" in a power surge, besides the LPC / SIO.

Originally Posted by tony359

Still, the board shouldn't burn to death when the PSU is switched on

You're right about that.

Originally Posted by tony359

While I read 100-300 Ohm (and climbing) without the CPU, I am reading 17 Ohm steady with the CPU installed.

How does that sound?

Normal.
CPU's and GPU's tend to have low resistance to ground.

[tony359]

Originally Posted by momaka

I see that you ordered parts already. But if anything happens and they don't arrive, I can check my stash of Xbox 360 PCBs and see if there is anything similar/same.

Hey, thanks for taking the time to reply, appreciated!
Parts will take a while to arrive, I shall keep you posted indeed! Thanks for your offer!

Yes, that would be a good idea.
By the looks of it, it's either power for the NB or power for the AGP port (1.5V). Probably best to measure the voltages on the input (Drain) and output (Source) of these two MOSFETs. They are used as linear regulators, so it's OK for them to run a little warmer. But from what your thermal camera shows, those appear to be running way too hot. If you can't keep your fingers on them, there may be something wrong with the NB.

Well, the Mosfet doesn't get too hot, it's those four resistors.

Q103 - the one close to the hot resistors - reads 3.3 on drain and 2.2 on source. The boardview calls the drain "3V" and the source "1.5V_1/2".

Q104 - closer to the AGP port - reads 2.2 (directly from Q103) on drain and 1.4 on source. Boardview calls the output "1.5V".

This output then goes to the NB, SB and to the AGP slot. AGP slot is not being used. Screenshots attached

Most POST cards are indeed a waste of time/money. But the older PCI-based one will sometimes show useful info for older motherboards. So if you can find one for very cheap (under $5?? I don't know what they go for nowadays anymore), it might not be a bad tool to keep around for working on older motherboards.

Thanks for the advice. The only issue is that I don't know which one is useful and which one is not. I'll do a search.

You shouldn't run anything without a heatsink.
That can only cause damage to the CPU and Northbridge with time, even if you do it only for a brief moment. I don't know where people keep finding this advice, but it's a POOR troubleshooting/testing practice to run anything without a heatsink that's meant to have a heatsink on.

I cannot argue with that.
However my procedure is to power up the board with the back of my fingers on the chip to sense if there is activity at all. I keep the other hand on the PSU switch so I can switch off immediately if it's getting too hot - should be around 60-70 degrees when it burns my skin. I appreciate it's a dangerous method though. Unfortunately to test some of the mosfets I have to run without a heatsink. I have been using the "hand" method or a small metal block I have on top of the CPU to try to remove some heat.

That is indeed concerning. Makes me wonder if something / the mobo got hit with a lightning strike or a power surge. Might be worthwhile to check the LAN chip too to see if it's getting hot, as that's usually the first one to "bite the dust" in a power surge, besides the LPC / SIO.

The MB died on me while it was working. The software crashed, I believe I heard the fans at full speed. I rebooted and it was dead.

The LAN chip gets a little warmer but around 45 degrees - and it's 25 degrees today.

Normal.
CPU's and GPU's tend to have low resistance to ground

I guess this is a good thing as it removes the PCB from the equation - at least for now

Thanks again!

[tony359]

Hi there

Mosfets have arrived! I've removed the old ones but I resisted the urge to just solder the new ones and test and I did some tests without the chips to see if any difference would emerge.

I tested resistance between ground and every pad and I noticed something.

Everything is the same between pads except for the gates of "phase 2" which is the one which got the new driver.

All the other pads read 8.22KOhm - which is basically the value of the resistor located very near the pads and to which they are connected to. Phase 2 gates read 5KOhm instead. I have removed one of those resistors (that's probably the smallest component I have ever removed) and indeed reads 8.2Khom as it should.

I have then noticed that pin 5 (DRVL) of the replacement driver reads 15Kohm while the older drivers read 8Kohm.

Do you think I have to worry about that or is that just a difference impedance given by a different revision of the chip? I could replace the other two if that helps - only thing is that those chips are £6 each! (edit: well, not really. I can have 25 from RS for £3 total )

Thank you!

[tony359]

Hello all

So I went ahead and replaced the two remaining drivers, all the mosfets and all the small 680uF caps. When testing all the Mosfet pads I now get all identical values between phases.

The result is not good unfortunately. Same behaviour - but this time the board draws even more power, 9A, which I guess it's due to the VCore section being more healthy!
The VCore output is still a sawtooth.

The CPU also gets really hot - I used my usual way of "starting the CPU with my bare fingers on the CPU and the other hand on the ON/OFF switch" and after a few seconds I had to switch off immediately as it was getting very hot. Then I installed the heatsink and give the board some more time to boot up but nothing.

In the past the CPU had never become so hot when testing.

I suppose it could be time to give up but I am still willing to try more things if anybody has any ideas.

Thanks for now for all the contributions!

[piernov]

Double check that all the caps are actually properly connected to ground and CPU VCore using the multimeter in resistance mode (or continuity mode).

[ChaosLegionnaire]

yea dont give up just yet. the behaviour seems to have changed which means u might be on the right track. if the mosfets no longer get hot, just the cpu, it means the cpu is dissipating all the power supplied to it which is good but something is still preventing the cpu from starting the system post process.

from your previous posts, it seems the super io chip may be bad if it cant turn the system on from the mobo power header. replacing the super io chip may fix the problem. this board seems to have multiple issues and multiple things broken which prevent it from working correctly. it may be a lot of work but equally rewarding if u can get it going.

[tony359]

Thank you ChaosLegionnaire! Don't worry, I can be quite persistent on things!

I didn't check the Mosfets temps to be honest, I kind of assumed they were still overheating. I'll check that, good point.

On the Super IO chip... this board belongs to an embedded system and it comes with a small daughterboard that fits on the "FPANEL" connector that basically translates a "on/off" switch at the front of the unit in a way which is compatible with the Asus.

I seriously doubt that the manufacturer made some BIOS or HW modifications to the board to change the behaviour of the ON/OFF routine (as they were not selling millions of units) but a custom BIOS is indeed a possibility.

Indeed I understand that the Super IO chip is telling the CPU to power up - on the replacement board I realised that forcing the PSU to give power does NOT start the CPU: the signal MUST come from the Super IO chip. I shall indeed investigate more on that.

I may also wan to replace the PWM chip - probably not relevant but if it's not too expensive I may wanna try that.

Piernov,
I shall check that too, thanks.

[tony359]

Quick update:

Mosfets Still get very hot unfortunately. The heatsink doesn't seem to get hot but I only keep the board on for 20 seconds max. By then the mosfets reach 100+ temps.

Piernov: I am ashamed to say that I had forgotten to solder one leg of one capacitor... Thanks for pointing that out! Still not change though.

I will keep thinking, any input is appreciated!