My PC randomly powers on for about one second and then immediately shuts down. It worked normally for roughly a month, suddenly stopped, then came back after sitting unused for two or three months. It ran for five or six days before failing again.
I suspected the motherboard's RT3607BCP PWM controller, so it was replaced, but that made no lasting difference. A technician also heated and reworked the board, after which the computer worked for one day before failing again. The PSU has been replaced, and I have already checked the GPU, RAM, and CPU.
Sometimes pressing the power button starts the system for a second, sometimes repeated attempts do nothing, and occasionally it starts by itself for a second before shutting down. It never reaches BIOS or Windows. I also tested the power-switch pins directly with a screwdriver, so the case button seems unlikely to be the cause.
Could this be a cracked solder joint, damaged trace, short circuit, VRM fault, or another intermittent motherboard problem? What should I check next?
5 Answers
If you have board-level tools, monitor the startup sequence instead of heating or replacing more components at random. Check whether 5VSB is stable while the system is off, then observe PS_ON# and PWR_OK during the one-second attempt. If PS_ON# is released again while the PSU rails remain stable, the motherboard is requesting shutdown, pointing toward the Super I/O, PWRBTN input, standby logic, or power-good circuitry. If PS_ON# remains active but PWR_OK or a voltage rail collapses, investigate a board short, VRM, MOSFET, or driver fault.
Before doing more board rework, disconnect the case power-switch connector completely and start the system by briefly bridging the two power-switch pins with a screwdriver. A stuck switch, damaged cable, or short on the PWRBTN line can cause random starts and shutdowns. Since you already tried bridging the pins, the switch is less likely, but leave it disconnected during testing to be certain.
I already tried starting it by shorting the motherboard pins with a screwdriver, and it behaved the same way.
Because the PSU and major components have already been tested, the motherboard is a strong suspect, but replacing the RT3607 controller does not eliminate every VRM-related fault. A failed MOSFET, gate driver, enable signal, power-good circuit, cracked trace, socket contact, or another voltage rail could still stop startup. Inspect and reseat the CPU, check the EPS connector, look for bent or corroded socket pins, and try testing with slightly less cooler pressure in case the board is flexing.
Also verify the replacement PSU was tested with its own modular cables. Modular cables are not universally interchangeable, even between units from the same manufacturer. If the cabling is correct and a minimal outside-the-case test behaves the same way, replacing the out-of-warranty motherboard may be more practical than continuing repairs without proper board diagnostics.
Test the motherboard outside the case with only the CPU and cooler, one RAM stick, the PSU, and a graphics card only if the CPU has no integrated graphics. Remove drives, front USB connections, RGB devices, expansion cards, and anything else. This can rule out a case standoff or peripheral short.
The fact that heating or reworking the board temporarily restored operation does make a cracked solder joint, trace, or contact issue plausible, but it is not conclusive. Repeatedly heating random areas can cause more damage, so the fault should be isolated with measurements instead.