Question PSU Tester Shows Low PG Values for Multiple New PSUs.

[spgfr2017-1]

My PSU tester shows low PG values for two new 1000W units, one from Corsair and one from MSI. The tester was purchased in 2017 and gives an in-range value of 180 for a known good Corsair 750W supply. Both MSI and Corsair tech support recommended returning the units. Should I keep buying new power supplies until I get one that tests good? New tester? Buy from someone other than Amazon? All insights and suggestions welcome.

 

[DSzymborski]

Well, what are the actual data? "Low" is meaningless.

 

[spgfr2017-1]

Corsair, PG = 80, MSI, PG=90

 

[kerberos_20]

PG Values are usually considered normal if it falls within the range of 100ms - 500ms, but can be lower than 100ms at times. 0ms PG values would be considered a failing PSU.

 

[spgfr2017-1]

PG Values are usually considered normal if it falls within the range of 100ms - 500ms, but can be lower than 100ms at times. 0ms PG values would be considered a failing PSU.

What are the implications if PG drops below 100 'at times'? Would you trust a new build to a PSU that fails what appears to be a simple no-load test?

 

[geofelt]

How much did this psu tester cost you?
If it was less than $5000 I would discount the results.
Static testing can confirm a dead psu, but it says nothing about the proper dynamic functioning of the unit.

Are you having a problem with a psu?
Do both cause the issue?
It is very unlikely that two different power supplies of good quality will fail.

 

[spgfr2017-1]

The tester I'm using is a cheap one, but both Corsair and MSI tech support recommended that I return the units based on the out of range low PG values. The power supplies in question are for a new build, and both gave low PG values on first powerup. And yes, I think it unlikely that two supposedly good quality supplies from different manufacturers would fail a simple no-load test.

Does anyone know if the cheapie psu testers actually calculate PG values, or simply report values calculated by the power supply?

 

[geofelt]

I would attribute the low numbers to the tester that you were using, and not to the power supplies themselves.

Did the Corsair and msi reps know about the similar results for the opposite brand?

 

[kerberos_20]

What are the implications if PG drops below 100 'at times'? Would you trust a new build to a PSU that fails what appears to be a simple no-load test?

no clue, corsair page mentions that without any specifics, but PG itself is for powering up, ie youll get power up issues with failing PSU

 

[spgfr2017-1]

I would attribute the low numbers to the tester that you were using, and not to the power supplies themselves.

Did the Corsair and msi reps know about the similar results for the opposite brand?

Did not.

 

[geofelt]

Did not.

They probably assumed that your reported specs were accurate.
Had they known about similar results from the other brand, they likely would be less likely to assume that their own psu had failed.
If you are still in contact, see what they say, possibly they can offer a different explanation.

 

[spgfr2017-1]

No longer in contact. The Corsair psu has already been returned, and I plan to take the msi unit to a local computer shop and see how their testers do.

 

[kerberos_20]

im looking at intels paper regarding PSU timings as intel manages ATX specs, this is for PSU DG rev 1.42

T3 PWR_OK delay[2] 100[2] –500 ms

[2] T3 time is allowed to be lower than 100ms to allow faster computer “turn on” time,
but might break backward compatibility. For Intel testing being less then 100ms will not result in failing timing.

https://www.intel.com/content/dam/w...esktop-platform-form-factors-power-supply.pdf
page 37

 

[spgfr2017-1]

im looking at intels paper regarding PSU timings as intel manages ATX specs, this is for PSU DG rev 1.42

T3 PWR_OK delay[2] 100[2] –500 ms

[2] T3 time is allowed to be lower than 100ms to allow faster computer “turn on” time,
but might break backward compatibility. For Intel testing being less then 100ms will not result in failing timing.

https://www.intel.com/content/dam/w...esktop-platform-form-factors-power-supply.pdf
page 37

Thanks for looking this up - since this is a new build, backward compatibility shouldn't be an issue.

 

[spgfr2017-1]

I've since purchased another psu tester and have received an EVGA 1000GS, which both testers show failed with PG = 0.

I also took the questionable MSI unit to a local shop, and to quote their tech, "My tester couldn't calculate a PG value".

Next week, I'll be receiving an NZXT 1000w psu and yet another tester and will post all the testing data that I've been accumulating.

I'm at a loss to understand what's going on here. I have 2 testers and both testers have indicated failure for three new 1000w supplies from three different manufacturers, yet a spare 750w supply purchased in 2017 tests good on both testers.

 

[kerberos_20]

I've since purchased another psu tester and have received an EVGA 1000GS, which both testers show failed with PG = 0.

I also took the questionable MSI unit to a local shop, and to quote their tech, "My tester couldn't calculate a PG value".

Next week, I'll be receiving an NZXT 1000w psu and yet another tester and will post all the testing data that I've been accumulating.

I'm at a loss to understand what's going on here. I have 2 testers and both testers have indicated failure for three new 1000w supplies from three different manufacturers, yet a spare 750w supply purchased in 2017 tests good on both testers.

on modern systems PG below 100ms isnt an issue, but 0 is questionable, lowest it could prbably go is 20ms for modern standby

 

[hotaru.hino]

The only harm that having a low PG value could do to the system is the computer attempts to power up before the PSU's power rails are stabilized. Powering on before that happens could lead to a lot of scenarios that may not be good for the power supply or the hardware.

In any case, there's the question of how accurate those testers are anyway. If I had one to poke around with, I'd probably check to see if it's 1. Asserting the PS_ON signal during that test and 2. if it's detecting 5V on the PWR_OK line. If so, then I might jerry rig up something with a microcontroller to assert the PWR_OK line at a known value and see what it gets.

Though ideally I would have an o-scope for this test anyway.