Corsair HX1000i PSU Efficency calculation bug

[ronzino]

I have read that there are some undocumented behaviour about corsair iCue or Corsair Link internal calculations, but I don't think that the problem I am reporting is related to this.

According to HWiNFO64 description PSU Efficiency is the ratio "PSU Power Sum"/"PSU Power".

Well this ratio seems to be not correct for two reasons

1) quite often the numerator is bigger than denominator. (ok, I know, this can be due to by incorrect measuring of input and output power BUT IN ANY CASE THIS MUST lead to a > 100 % efficency, but this never happen

2) the ratio calculated is wrong even if the numerator is smaller than denominator

Some examples of recorded values

PSU Power Sum 125,472W
PSU Power 124,000W
PSU Efficiency 89,9%

Instead the efficiency should be 101,18%

PSU Power Sum 119,089W
PSU Power 124,000W
PSU Efficiency 89,3%

Instead the efficiency should be 96,039%

 

[Martin]

No, it's the ratio between Output and Input power (not "sum" and measured).

 

[ronzino]

mmm I am sorry Maritin but there is no Input power in HwInfo and furthermore if I disable the measurement of all 3.3 5 12 volts (or their currents) suddendly the sum goes to zero (and of course it is correct) BUT also th e efficiency goes to zero. This means that PSU Efficiency is calculated using PSU power sum: so zero at numerator means zero for effiency.

pictures attached

 

[Martin]

The input power is not displayed in this case because it's calculated (not measured) based on power curves (models).

 

[ronzino]

The input power is not displayed in this case because it's calculated (not measured) based on power curves (models).

mmm I am starting to see, some more questions

1) So do you mean that in HWiNFO you have coded the power curve of this PSU like the following? So you know the output power (DC Load X axis in the graph) so you know the efficiency by best fit curve ?

2) what is it the sensor called "PSU Power" ? it is said to be the measured PSU power, but is it the measured input or the measured output ?

3) it seems to me that you use DC volt and ameper readings from 3.3 5 12v line to calculate the DC load, and so with the curve you can have the Efficiency, is it correct ?

4) BUT as you have two curves (one for AC 240 V AC ad another one for 115 V AC) why if I disable the monitoring of Input voltage the Efficiency is still showed ? HWiNFO should change the curve according to AC input

 

[Martin]

1) Yes
2) "PSU Power" should be the total measured output power read as a single value from the controller. However this value is "unofficial", we assume this when we apply the PMBus specification to the Corsair protocol.
3) Yes, we used the same method as the Corsair software does - it calculates the total output power as a sum of all particular rails rather than relying on the "unofficial" total "PSU Power".
4) Because the last valid Input voltage is used (it's sufficient to sample this once as it's unlikely to change)

 

[ronzino]

Thank you Martin now everything is clear to me.
Many thanks for you time.

 

[ronzino]

Sorry for resuming this, but i have just seen that the ATX 24 pin specification provides on pin 14 blue wire a negative -12v tension.
I don't know what is used for by motherboard but for sure its power is not taken into account when calculating PSU power (sum). As I guess that not so much current is driven out by -12v line I wonder if the missing little piece of the puzzle "PSU Power" = "PSU Power (sum)" is the missing measure on -12v line.
I guess that -12v line measures are not exposed by Corsair, so that is why they are not reported in HwInfo as well :-(

 

[SpeedyIV]

Hmm... You're right. Pin 14 carries -12VDC. I wonder what this is used for and how much current is typically required on this rail.

 

[Kiriakos-GR]

Hmm... You're right. Pin 14 carries -12VDC. I wonder what this is used for and how much current is typically required on this rail.

Why wondering when PSU attached label this has that info in writing.

 

[Kiriakos-GR]

As I guess that not so much current is driven out by -12v line I wonder if the missing little piece of the puzzle "PSU Power" = "PSU Power (sum)" i

With 30.000 Euro worth measuring equipment - an power analyzer, you can come close to the truth.
All that you have, its a 1000W PSU with a cheap power meter on-board.

APC Smart-UPS C Series 1000~1500VA inaccurate LCD Status Power meter - Details

APC Smart-UPS C Series 1000~1500VA inaccurate LCD Status Power meter - Details

www.ittsb.eu