Question about PSU Tester revealing low PG values for several new PSUs.
Question about PSU Tester revealing low PG values for several new PSUs.
They might have thought your reported specifications were correct.
If they had been aware of comparable outcomes from another brand, they would probably be less inclined to believe their PSU had failed.
If you remain in touch, check their statements—there might be another perspective.
reviewing intel's paper about PSU timings when managing ATX specifications, this pertains to PSU DG rev 1.42. T3 PWR_OK delay is 100 microseconds with a range of -500 ms. The document notes that allowing T3 time below 100ms can speed up the computer's startup, though it may affect backward compatibility. Intel testing under such conditions won't lead to timing failures.
I've bought another PSU and received an EVGA 1000GS, which both devices report as failed with PG = 0. I also sent the doubtful MSI unit to a nearby shop, and according to their technician, "My tester couldn't calculate a PG value." Next week I'll get a NZXT 1000w PSU and another tester, and I'll share all the testing results I've collected so far. I'm confused about what's happening. I have two testers, both showing failure for three new 1000w supplies from different makers, yet a spare 750w unit bought in 2017 still works on both.
on contemporary platforms with processing times under 100ms remains unaffected, yet a value of zero seems uncertain, the most likely minimum appears to be around 20ms during active standby mode
The only damage a low PG value could cause is the computer trying to start before the PSU's power rails are fully stabilized. Doing so beforehand might trigger many bad situations that could harm the power supply or hardware.
Regardless, it remains a concern about how reliable those testers really are. If I could experiment freely, I’d verify if the PS_ON signal is present during the test and if the PWR_OK line reads 5V. If both hold, I’d set up a microcontroller to force the PWR_OK line to a known value and observe the result.
Ideally, I would also use an o-scope for this kind of check.