This is one of those "should have figured that out myself" problems, but I figure I will warn anyone who might be in my shoes.... Check the specifications carefully before investing in an APC UPS. Not that I'm going to suggest going with another manufacturer, just that I'm going to suggest paying close attention to the exact model you purchase.
When we moved, I had to bid farewell to to my huge, honking 30A 240V Matrix 5000 5kva UPS with extended run battery packs. Yep, that was one sweet puppy back in the day when I had the home data center and poor power. Just before we left, the downsizing of the data center had it running at about 30% capacity and it could keep everything in the computer room and my office happy for about 6 hours during a power outage.
Moving to the new place, I was downsizing substantially and needed to get a smaller UPS. I was going to miss the readouts on the front of the system, so I decided to get a BackUPS XS 1500LCD from APC. I had measured my power needs and it seemed that it would meet my needs.
In comparison to the nearly-twice-as-expensive SmartUPS 1500, it seemed like a good deal: 90% of the max load, about half the run time, most of the same features, a cute little LED readout, and a button on the front to disable the alarm. What could be nicer?
Well, I tried to run the BX1500LCD at about 350W (supposedly it does 16min at 432W) and I was seeing nowhere near 16 minutes. As a matter of fact, with half that load, it didn't indicate 16 minutes of run time. I contacted APC and they were somewhat useful, but doing a lightbulb test was an impossibility without taking the equipment out of commission first. After a little digging, I realized that all my positive UPS experiences with APC had been on the SmartUPS and that this wasn't the first (runtime or power conditioning) issue I'd had with the BackUPS family. To be fair, my parents run an ancient power- strip-style BackUPS and it seems to work OK for them... but that's not OK for me.
Enter the SmartUPS 1500. A bit uglier (but now available in black, so that at least goes with my decor), but my first test out of the box was brilliant. At about 350W (2009 Mac Pro doing "normal stuff", Mac Mini running compiles, laptop, 30" display, Gigabit Ethernet switch, portable phone, and VOIP router), it started its no-power run at 32 minutes, dropping immediately to 25 when it went on battery. However, I was determined to watch it down to 4 minutes (my breaking point, because that's about where I can guarantee that everything will automatically shut down). Tracking every minute, it asymptotically approached running for 32 minutes... color me happy!
There will be a follow-up as I run the lightbulb test on the other unit. Frankly, if it's defective, it's going back.