How are the caps looking? They're the prime candidates for something that degraded over time, and that board's from the bad old days for cap plague.

I have that board, working. Capacitors TC9, 15, 17, 28 were swollen and leaky, and the spill from TC9 had corroded some traces near the AGP slot. Wasn't very bad so I chose not to do any repair other than change those caps, but I'll have to keep an eye on this over time. Can you check the BIOS isn't corrupted? Here is mine:

dionb wrote:

How are the caps looking? They're the prime candidates for something that degraded over time, and that board's from the bad old days for cap plague.

All recently replaced with Rubycons I yanked out from a discarded Gigabyte NF2 board (dead SB)

quicknick - I'll try flashing your BIOS and see if anything changes. What release is it though? The 10/15/1999 one? (I have downloaded a 2000 and a 2003 release but none of these 2 worked)

Is the Bios battery OK? Some board refuse to work with a dead battery installed 😵

Brand new, and it did POST with a dead battery that it came with. (measured about 0.16v)

Flashed the BIOS,no change.

Replaced some of the little 100uF 16V buggers near the MVP3 chipset and still nothing.

What should I try next?

Check the board visually. Are there any scratches?
And inspect chips in QFP packages (the south bridge and others): there could be short circuits between pins or cold joints.
Also if you have a multimeter it's better to mesure CPU voltage (a choke near a cpu socket).

Roman555 wrote:

Check the board visually. Are there any scratches?
And inspect chips in QFP packages (the south bridge and others): there could be short circuits between pins or cold joints.
Also if you have a multimeter it's better to mesure CPU voltage (a choke near a cpu socket).

No scratches, no bridges.

I'll measure the choke near the socket and report back.

LE: Measured, got 0.10V on the choke.

PcBytes wrote:

LE: Measured, got 0.10V on the choke.

CPU voltage depends on jumper settings for particular CPU model but 0.10V is obviously too low.
Could you provide photos of the whole mainboard and a part around CPU socket?

Roman555 wrote:

PcBytes wrote:

LE: Measured, got 0.10V on the choke.

CPU voltage depends on jumper settings for particular CPU model but 0.10V is obviously too low.
Could you provide photos of the whole mainboard and a part around CPU socket?

Sure. It's set to 2.8v by the way. ( although I use a ceramic Pentium, not a MMX)

I think voltage settings (2.8v) for non-MMX Pentium are wrong.
Do you have the manual? If not it's here
Check the CPU setting for your CPUs:

I tried 3.3v as well and it does the same.

So VRM doesn't work... Try to clean the soldering from flux.
Sometimes a PWM controller refuses to work because of leaks.
BTW, did you mesure CPU voltage with CPU installed or without CPU?

Roman555 wrote:

So VRM doesn't work... Try to clean the soldering from flux.
Sometimes a PWM controller refuses to work because of leaks.
BTW, did you mesure CPU voltage with CPU installed or without CPU?

With CPU.

Good. Sometimes, no CPU - no CPU voltage.
Now disconnect all power wires from the board, so you would be sure that the board is completly turned off.
Measure resistance between Vcore rail (VRM output - big choke near socket where you've measured CPU voltage) and ground rail (mount hole near CPU-FAN connector). If there's short circuit the VRM doesn't start.

Roman555 wrote:

Good. Sometimes, no CPU - no CPU voltage.
Now disconnect all power wires from the board, so you would be sure that the board is completly turned off.
Measure resistance between Vcore rail (VRM output - big choke near socket where you've measured CPU voltage) and ground rail (mount hole near CPU-FAN connector). If there's short circuit the VRM doesn't start.

Okay:

Red probe on choke pin near VRM - infinite
Red probe on choke pin near CPU - infinte
Black probe on either choke pins - 154 Ohms both ways

It's OK.
Could you clarify why so much soldering was done around MOSFETs ? Were they all replaced by working ones?
I'll try to figure out what's wrong but it involves a lot of measures. Something like this. And result is not guaranteed.
In the red circle there's something like filth. Am I wrong?
Arrows show where probes are applied to measure voltage when the mainboard is turned on. So be very careful. A black probe is always on the ground rail.

Roman555 wrote:

It's OK. Could you clarify why so much soldering was done around MOSFETs ? Were they all replaced by working ones? I'll try to f […]
Show full quote

It's OK.
Could you clarify why so much soldering was done around MOSFETs ? Were they all replaced by working ones?
I'll try to figure out what's wrong but it involves a lot of measures. Something like this. And result is not guaranteed.
In the red circle there's something like filth. Am I wrong?
Arrows show where probes are applied to measure voltage when the mainboard is turned on. So be very careful. A black probe is always on the ground rail.

Soldering was done because this board had missing MOSFETs and voltage regulator (although it did POST with new MOSFETs, they were later killed by a K6-2, legit) and I replaced them again now with a new set from a known working Socket 462 KT133 board. While the voltage regulator holds up, I have had the MOSFETs blow with two K6-2s set accordingly. Anything else (like K5 PR100, non-MMX pentium chips and MMX as well) would work fine (with the exception of the MMX which I suspect it's dead as it never POST'd)

The red circle is actually a Zener diode in SOT-323 package which I replaced. Doesn't look pretty but it makes good connection.

Okay:

Diode under fan header: 0.04 Ohms on the left, infinite on the other
Linfinity chip : infinite reading on both pins
Choke between ISA slots: infinite both ways
Voltage regulator: tab gives infinite, right tab gives 4.6 Ohms, left gives infinite
MOSFET in middle: tab gives infinite, right tab gives infinite, left gives 0 Ohms reading
MOSFET in the left: tab gives infinite, right tab gives infinite and left gives infinite as well.

Attached below is a picture I took when I first got it to work.