[Nut-upsuser] variables between manufacturers

[Charles Lepple]

On Apr 2, 2019, at 9:42 PM, James wrote:
> 
> I switched my desktop with my raspberry pi because I thought my desktop ups had a larger capacity.
> Now the ups.load of each is not what it was (it should be).

Note that ups.load is a percentage of "full load", which is up to each UPS vendor to define. (Some vendors use different margins, and I don't know that we have done a thorough check to see how well the load percentage tracks Watts or VA ratings when those are available.) If I am reading the model numbers correctly, you have an 850 VA CPS UPS, and a 550 VA APC, so at the very least, you would need to convert to a unit like VA to be able to compare the two directly. Or you can compare the numbers on one UPS with the two different loads.

The APC starts out with a load of 6 and a runtime of 98 minutes. It then changes to a load of 3 and a runtime of 147 minutes (not quite double). If nothing is plugged in, do you get a load of 0? If so, consider that the control electronics will require a certain amount of current from the battery just to keep the output running, and it may be in the same ballpark as a load of 3-6%.

Bear in mind the precision and accuracy of these measurements are not guaranteed by the vendor. Two points may determine a line, but this is an inverse relationship, so a small change in load means a larger change in runtime. You could add a dummy load and see if it tracks linearly up from there.

The CPS UPS seems to be in a similar situation. Starting with a load of 19%, it estimates a runtime of 45 minutes. A load of 9% yields 86 minutes, or nearly double.

In both cases, you may want to check the owner's manual to see if the vendor has typical runtimes. The UPS electronics are probably sized for 20-100% load, given the use case of several minutes on battery to let a generator kick in, or allow an orderly shutdown.