Fortron ATX 300GTF fails due to sustained excessive +5v load
Today's "patient" is a 2-year old Fortron ATX 300-GTF, a very cheap, but usually decent PSU rated for 300W with passive PFC. Background information:
For the first 2 months of its life, it was used to provide power for one of the kids' "gaming" PCs, before it was replaced with an Enermax device. At this point, the Fortron was 'modded' in such a way that its fan would run directly from the +12v rail, PSON was shorted to ground, and the PSU was used to provide the EMF for an electrolysis demonstration, which ran occasionally, for a few days at a time.
Two 'dummy' loads were always used as follows to reduce crossloading issues:
3.3v = 1R resistive 'dummy load', 11 watts.
12v = Cooling fan, and 10R load, total about 20 watts.
Main load was on +5v line, usually about 27A, hence about 135 watts.
Yesterday I decide to set up the 'kit' differently, but don't bother measuring the 5v load, only the voltage, which was (on hindsight, abnormally low), 4.70V. The setup runs for about an hour, after which the PSU shuts itself down, and never starts up again.
While the fan was running until the moment the unit shut down, the PSU was much hotter than usual, so with another PSU and the Fluke, I measured the current (simulating voltage of 4.70V) on the +5v: it was 37.1A. The Fortron's rated maximum was 30A.
Observations:
1. 5Vsb still works fine
2. When PS_ON is grounded, 12.1 volts appear for a moment on the +12V line, before plummeting to zero as the unit shuts down - with 5Vsb still working. However, no voltage ever appears on 3.3v or 5v.
3. Measuring with a ohmmeter at the ATX connector, to ground:
3.3v:8R, unchanging
5v:15R, unchanging
12v:150R, increasing to over 2K.
Any suggestions as to where the problem may lie? Otherwise I'll continue this thread with a component-by-component autopsy, complete with gory pictures
Note: While a non-IT-related load was used in this case, many people still use older motherboards and GPUs that draw almost entirely from the +5V line. Often they will heavily overclock, possibly leading to a similar scenario.
For the first 2 months of its life, it was used to provide power for one of the kids' "gaming" PCs, before it was replaced with an Enermax device. At this point, the Fortron was 'modded' in such a way that its fan would run directly from the +12v rail, PSON was shorted to ground, and the PSU was used to provide the EMF for an electrolysis demonstration, which ran occasionally, for a few days at a time.
Two 'dummy' loads were always used as follows to reduce crossloading issues:
3.3v = 1R resistive 'dummy load', 11 watts.
12v = Cooling fan, and 10R load, total about 20 watts.
Main load was on +5v line, usually about 27A, hence about 135 watts.
Yesterday I decide to set up the 'kit' differently, but don't bother measuring the 5v load, only the voltage, which was (on hindsight, abnormally low), 4.70V. The setup runs for about an hour, after which the PSU shuts itself down, and never starts up again.
While the fan was running until the moment the unit shut down, the PSU was much hotter than usual, so with another PSU and the Fluke, I measured the current (simulating voltage of 4.70V) on the +5v: it was 37.1A. The Fortron's rated maximum was 30A.
Observations:
1. 5Vsb still works fine
2. When PS_ON is grounded, 12.1 volts appear for a moment on the +12V line, before plummeting to zero as the unit shuts down - with 5Vsb still working. However, no voltage ever appears on 3.3v or 5v.
3. Measuring with a ohmmeter at the ATX connector, to ground:
3.3v:8R, unchanging
5v:15R, unchanging
12v:150R, increasing to over 2K.
Any suggestions as to where the problem may lie? Otherwise I'll continue this thread with a component-by-component autopsy, complete with gory pictures
Note: While a non-IT-related load was used in this case, many people still use older motherboards and GPUs that draw almost entirely from the +5V line. Often they will heavily overclock, possibly leading to a similar scenario.
Comment