MGX7520 onboard graphics problem

[giampitt]

Hello, I have a problem with this motherboard mgx 7520.
This motherboard is from a pinball (revenge from mars); the problem is with the onboard graphics.
After an inactivity period i can see the post screen but only for more or less one minute, than the screen goes grey, it stay powered on but nothing on the screen. If I use a pci vga card the pc works flawless.
If I let the mobo without power for a while (i think something has to discarge) then the onboard graphics work again but again only for one minute and then grey monitor.
I tried changing the power supply, the ram and also the processor, but nothings change.
This motherboard has a known problem with some capacitors (number 1 to 5), 1000uF 10V, but mine looks good, after an inspection i notice that they was already been changed.
I tested with the dmm the 5 capacitors and number 1 to 4 give me a value from 3 to 5 kohm, while number 5 only 25-30 ohm. Thinking it was shorted i decided to change all with panasonic FR caps.
I tested the removed caps with the esr meter and all was good 1000uF and an ESR of 0,14 ohm, testing the pads of the numbered 5 cap on the mobo with the dmm again 25-30 ohm.
What else can i do? that value is normal or there is a short on the board?
I'm not an engenier nor a technician, but i have experience in electronics DIY, but without the schematics of the board i can't go further by myself.



1) 1 to 5 Removed caps
6) AIC1579CS
7) 09N05
8) L1084S
9) PC97317 - IBW
10) CX5520
11) CPU Cyrix GXm-233GP

[giampitt]

Hello, i found another identical mobo, but this one is working.
I check the number 5 caps with the DMM in ohm mode, and in this board the value is 350 ohm, so completely different from the non working one.
I think the problem is in this line, but where?

[Per Hansson]

Now that you have two identical board a nice way could be to use a thermal camera to look for any hotspots that are not the same between the units.

[giampitt]

Originally Posted by Per Hansson

Now that you have two identical board a nice way could be to use a thermal camera to look for any hotspots that are not the same between the units.

I was thinking the same, but i don't have a thermal camera, and i didn't found a cheap one.
I tried to touch with the hand, but in this way i didn't found differences.

[lti]

I don't think 25 ohms would show up in temperature measurements.

Can you see what that cap connects to? The CPU takes 2.9V and 3.3V, and I would guess that the Cx5520 runs off the same 3.3V supply. The 3.3V most likely comes from the L1084S.

Also, check the voltages in both the working and non-working states.