PcChips Athlon Socket A M825G Motherboard FSB 100, 133, 166
EC7, EC5, EC1: OST 1500uf 16v all in parallel
EC2, EC6, EC11, EC15: OST 3300uf 6.3v all in parallel
D3, D23, D24, D2: JS B55QS03 TO-262 common cathode Double Diode
Q2, Q3: Nikos Niko-Sem P50N03LDG N-MOSFET
U3: Fairchild KA7500C dual PWM controller
Q50, Q11: W2F
Q49, Q10: W1P
Phantom shorts: all dual diodes measure 7.0 ohms.
EC7, EC5, EC1, EC2, EC6 all bulged. New ones of those and the board would not start. The power supply shuts down as if there's a short. Board starts if Vcore power source CPUPW1 is disconnected. All pins in CPUPW1 measure as shorted so 12v is shorted to ground. I removed the CPU and tested the semiconductors. Almost all shorts among the 4 double diodes and MOSFETS except for Q3. Q3-gate was not shorted. All pins of Q2 were shorted.
Many MOSFETS have phantom shorts, shorts from the design or failure of surrounding circuitry and not itself. The differences between MOSFETS in the same position in different phases often give hints to which ones have failed without having to pull them to test for phantom shorts. Both Q2 and Q3 for this two phase FET-DIODE buck converter are in the same position and Q3 measures better than Q2. MOSFETS where all pins appear to be shorted also give away a bad one because no design allows all 3 pins to appear shorted. Q2 has all pins shorted.
Two strikes against Q2 so I pull it. It is shorted on all 3 pins. CPUPW1 is no longer shorted and other shorts have disappeared. I put in another one and the board shuts down after about 5 seconds. A voltage test shows that Q2-gate and Q3-gate are receiving the signal from the KA7500C through Q50, Q11, Q49, Q10. A voltage test on the source and drain show that the board is receiving 12v and producing Vcore properly.
Turns out I had subbed in a test CPU that didn't work because I hadn't changed the FSB to match. I removed the test CPU and put the original CPU back in and the board works. The original CPU and memory were undamaged.
All fixed with new caps and 1 MOSFET.
EC7, EC5, EC1: OST 1500uf 16v all in parallel
EC2, EC6, EC11, EC15: OST 3300uf 6.3v all in parallel
D3, D23, D24, D2: JS B55QS03 TO-262 common cathode Double Diode
Q2, Q3: Nikos Niko-Sem P50N03LDG N-MOSFET
U3: Fairchild KA7500C dual PWM controller
Q50, Q11: W2F
Q49, Q10: W1P
Phantom shorts: all dual diodes measure 7.0 ohms.
EC7, EC5, EC1, EC2, EC6 all bulged. New ones of those and the board would not start. The power supply shuts down as if there's a short. Board starts if Vcore power source CPUPW1 is disconnected. All pins in CPUPW1 measure as shorted so 12v is shorted to ground. I removed the CPU and tested the semiconductors. Almost all shorts among the 4 double diodes and MOSFETS except for Q3. Q3-gate was not shorted. All pins of Q2 were shorted.
Many MOSFETS have phantom shorts, shorts from the design or failure of surrounding circuitry and not itself. The differences between MOSFETS in the same position in different phases often give hints to which ones have failed without having to pull them to test for phantom shorts. Both Q2 and Q3 for this two phase FET-DIODE buck converter are in the same position and Q3 measures better than Q2. MOSFETS where all pins appear to be shorted also give away a bad one because no design allows all 3 pins to appear shorted. Q2 has all pins shorted.
Two strikes against Q2 so I pull it. It is shorted on all 3 pins. CPUPW1 is no longer shorted and other shorts have disappeared. I put in another one and the board shuts down after about 5 seconds. A voltage test shows that Q2-gate and Q3-gate are receiving the signal from the KA7500C through Q50, Q11, Q49, Q10. A voltage test on the source and drain show that the board is receiving 12v and producing Vcore properly.
Turns out I had subbed in a test CPU that didn't work because I hadn't changed the FSB to match. I removed the test CPU and put the original CPU back in and the board works. The original CPU and memory were undamaged.
All fixed with new caps and 1 MOSFET.
Attached Files