• Please review our updated Terms and Rules here

ECS Elitegroup P5TX-B Pro - Q8 mosfet hotttttt - FP45N03L

J

jc179

Experienced Member
Joined
Apr 21, 2007
Messages
99
Hi Everyone

I am presently using a P5TX-B like the picture below, I picked this board up around 99' when the Q8 in the top right below burned up. At the time I had seen a few of these boards suffer the same failure, so it seems to be some sort of design flaw perhaps with certain revisions. Unfortunately I didn't make much of a note back then as to what Q8 part# wise was on the boards that were actually working without getting Q8 burning hot.

Q8 on my board, I believe was a FP45N03L, 30A Nch mosfet, which seems quite skookum for what its being asked to do here, but on mine it gets too hot imho after running for a while. I've had a larger heatsink attached to it, and a fan, and it then runs cool, and its ran like this for years, but I would like to fix this properly, and find out if anyone out there has one which Q8 does not get roasty hot? (top furthest transistor, which the smaller finned heatsink.

Anyone have any experience they can share with these boards?

thanks,
Jonathan

P5TX-Bpro ECS S7.jpg
 
I agree there is no short, however its hotter than it should be. I had encountered boards that had been running for a long time without this discoloration and parts running quite so hot, so I was hoping to find out what the difference might be and hoped to make the same correction here.
 
As a sanity check, you might check the electrolytics in the circuit for degradation (high ESR, lower than nameplate capacitance).
Other than that, it's just a simple buck regulator.
You are using the correct CPU, aren't you?
 
I would replace the heatsink compound that connects the regulator to the heatsink and make sure there is good contact. Also make sure there is some airflow in that area (no cables in the way etc.).
 
Chuck(G) said:
As a sanity check, you might check the electrolytics in the circuit for degradation (high ESR, lower than nameplate capacitance).
Other than that, it's just a simple buck regulator.
You are using the correct CPU, aren't you?
Click to expand...
That's a good thought, perhaps I should just replace the caps, its old enough to do that.

The CPU is a dual voltage mmx and that is setup correctly, though, I seem to recall from many years ago that using a non-mmx single voltage chip, this didn't get hot at all. Perhaps something is awry with the buck converter for the 2.8v core voltage.

caps sound like something I should have just gone ahead and replace before posting... !
Jonathan
 
Unknown_K said:
I would replace the heatsink compound that connects the regulator to the heatsink and make sure there is good contact. Also make sure there is some airflow in that area (no cables in the way etc.).
Click to expand...
That's all been replaced, carefully, as this mosfet isn't grounded to the board - there was a insulator between the heatsink and the board, and they used that silly plastic clip, I'll see if I can find a nylon bolt and nut that would be a better choice here ...

the location is less than ideal, while no cables are near, this part of the board protrudes out and hides behind the 3.5" drive tray in the case, so I don't think it gets much airflow at all, that's why I added the extra small fan there to help...
 
Just a guess, check that the Mosfet is turning all the way on. The Gate voltage should to be a couple of volts higher than the source to reduce the voltage drop across the Mosfet. If it isn't completely on then it could very be overheating. Could be a design problem since you said a lot of these boards had the issue.
 
I use this exact board w/ a K6-2 475 and Wims patched Bios. Its been extremely stable for me. I havent noticed my mosfet getting too hot, but I can run the system this weekend and double check it.
 
jc179 said:
The CPU is a dual voltage mmx and that is setup correctly, though, I seem to recall from many years ago that using a non-mmx single voltage chip, this didn't get hot at all. Perhaps something is awry with the buck converter for the 2.8v core voltage.
Click to expand...

Lots of boards in this era had crap voltage regulation circuitry that ran smokin hot.

They could be operating the mosfet in the linear region, which would explain the excessive heat production. The mosfet being stone cold when a non-split power plane CPU is used, most likely means that mosfet is used for the vcore.

You could try to reverse engineer the power circuitry to figure out what design faults might be there, but it may be easier to remove the vcore regulation circuitry and use an external buck regulator that doesn't have the same problem and inject it into the board. It'd be easier to do if the board had a slot for a VRM module.
 
GiGaBiTe said:
Lots of boards in this era had crap voltage regulation circuitry that ran smokin hot.

They could be operating the mosfet in the linear region, which would explain the excessive heat production. The mosfet being stone cold when a non-split power plane CPU is used, most likely means that mosfet is used for the vcore.

You could try to reverse engineer the power circuitry to figure out what design faults might be there, but it may be easier to remove the vcore regulation circuitry and use an external buck regulator that doesn't have the same problem and inject it into the board. It'd be easier to do if the board had a slot for a VRM module.
Click to expand...
Got a bug over the weekend and was down for the count!
... The more I look at this, I think your right, I do suspect the mosfet is in a linear mode, I will try to do a bit of probing and confirm that.

All makes sense here, I figure I could just tap in a dc buck converter into the mosfet output pins, and confirm the voltage at the socket is good.

I also have a HOT-541 motherboard, which supports the P55C/mmx chips with an external VRM - is there some sort of a common pinout on these boards that took the external VRM? If that's the case and you have more info please do share it as I would like to put a mmx chip on that board.

thanks,
Jonathan
 
jc179 said:
Got a bug over the weekend and was down for the count!
... The more I look at this, I think your right, I do suspect the mosfet is in a linear mode, I will try to do a bit of probing and confirm that.

All makes sense here, I figure I could just tap in a dc buck converter into the mosfet output pins, and confirm the voltage at the socket is good.

I also have a HOT-541 motherboard, which supports the P55C/mmx chips with an external VRM - is there some sort of a common pinout on these boards that took the external VRM? If that's the case and you have more info please do share it as I would like to put a mmx chip on that board.

thanks,
Jonathan
Click to expand...

The VRM socket on most Pentium boards was an Intel standard.


This is a linear design, but there are some modern switching designs as well.

 
You must log in or register to reply here.
Back
Top