I've just started shopping for a UPS lately, and I've noticed it's tough to find information for run-time at low loads. I think APC is the only manufacturer I've seen that pubishes run-time versus power draw curves, and even those only seem to be available for certain series. My hope is to find something inexpensive that can keep my HTPC, HDHomeRun Prime, and networking gear running for about an hour. At ~40W for the HTPC, 7W for the Prime, 10W for wireless router, and probably 5W each for DSL modem and ethernet switches, that rolls up to under 80W. Maybe I'd get wild and throw my server on there too, which would push me up to about 130W.
Most UPS only seem to have published values for run-time at full load and half load. I could linearly extrapolate run-time based on the % rated load my 80-130W turns out to be, but is that really valid? I have a suspicion the battery discharge curves are nonlinear (maybe APC's curves are applicable to other manufacturers, since sealed lead-acid gel cell batteries seem to be used by pretty much everyone). Or is there is a good de-rating rule-of-thumb to apply to Amp-Hour rating of the battery? E.g. if I need approximately 1A-h @ 120V, that would equate to (10A-h @ 12V) * X (where X is a de-rating factor that accounts for power factor, transformer losses, etc)?
Maybe I'm silly for even targeting an hour of run-time, when most outages I've had fall in 2 categories -- < a few minutes or > several hours (very few in between)