There it is, the pure sinewave power produced by the Belkin PureAV AVU 1500. Many have gotten cynical about the AC power generated by UPS's, having seen so many with horrible looking output. But in this case, the cynicism is not warranted, the Belkin generates pure sinewaves as claimed.
The above picture is driving a simulated audio AC power load. Actually it's a combination of three things totaling 0.7 amps, similar to the idle of my living room system. It's driving a Beringer DCX 2496 digital crossover, the charger for an Apple Macbook Pro (it needed some charging), and the Tektronix 465M itself which is displaying the sinewave. The scope seems to load much like a linear power supply (I'm not sure if it is though) and therefore simulates the load of a Class AB amplifier in idle. All 3 pieces of equipment are plugged into a power strip having no filtering or surge protection, which is then plugged into a single Belkin outlet from the Stage 2 digital filter bank. I have the Belkin itself plugged into another strip, and flip the switch temporarily for these tests. The UPS power kicks in seamlessly, though initially there is a bit of ripple. I also tried the State 1 High Current outlet, and it worked similarly.
Note that the Belkin actually generates balanced AC power, with out-of-polarity sinewaves on neutral and hot lines which add up to 120V RMS. That is typical and expected, sometimes considered "a feature," and similar to what you might get with an isolation transformer, though it could be undesireable with equipment having high neutral leakage (you shouldn't be using such equipment anyway). The above picture is the hot side, the neutral side looked identical on the scope for all tests. I was unable with my safety-oriented test setup to combine phases on the scope.
I couldn't scope the transfer point, but the transition to UPS power and back seemed seamless and didn't cause either the Behringer or a touchy GFCI power strip which drops on brownout to reset. I actually connected the scope input through a Tektronix 10x probe, a 1k resistor, a 1.5 amp fuse on the hot side in a convenient test box I made years ago, a GFCI power strip connected to the power strip the load equipment was tested to, then the Behringer output itself. Safety First!
Running just the digital gear, the Behringer and the Apple, and running the Tektronix scope from a real AC line, gives a slightly less pretty sinewave (because the scope load tends to smooth out the digital glitches), but similar to what you might get on ordinary unfiltered AC power driving digital gear (below).
Driving the most simple linear load at hand, an old fashioned transformer-based Tensor lamp, gave a nicer sinewave, though strangely not as nice as the top one.
Driving no load at all looked even better (unless I got these two pictures reversed). An unloaded test can often be the worst case, if the generator relies on the load to help with the smoothing. But here it does fine, perhaps the best of all.
I've actually been using an AVU 1500 on my living room system for almost a year now, taking on faith that it produces pure sinewaves as claimed. I've had no problem with it at all, even continued listening through Parasound HCA-1500A amplifier right through a blackout. (I was using the Parasound at that time and still now, I will certainly not plug my Krell into the Belkin.) But I bought two other units for Kitchen System and Computer Room, and wanted to run these tests before further deployments.
Further testing shows that under large current draw, the waveform does get worse.
That waveform was produced with 6.3 continuous amp draw, as indicated by the Belkin display. I was running scope, Behringer, and a 2507 DeLonghi heater on "min" (600w) setting. The 6.3 amps (at 118V see below) suggests about 740W continuous nearly resistive load (dominated by the heater).
The above is waveform is a bit disappointing, but not a practical issue as I do not intend to plug in an equivalent Class A amplifier which is about the only thing that would draw that much continuous current. And even the waveform above is fairly low harmonic, probably not harmful to most equipment.
The Belkin specs are a bit confusing. They specify 1500VA and 1000W for input. For output, they don't actually specify a wattage or VA capacity, merely 120V AC and 12.5 maximum amps, non-resistive load. The 12.5 maximum amps must be "peak" amps, not RMS amps. I would then guess the RMS amps maximum would be around (12.5/1.414) 8.8A (which fits nicely with 1000W) but even then probably not on a continuous basis. So with my 740W mostly resistive load, I'm pretty close to maximum output on an actual continuous basis
When I plugged in heater and still with AC power input to the Belkin, indicated voltage already sagged to 114V as read on Belkin display. I am using one of my usual lousy chained household circuits for all this testing, not my dedicated 20A audio power circuit, so quite a bit of sag is expected. I checked Belkin output voltage with heater connected on the following afternoon. When connected to AC power, the Belkin is now providing about 116V as measured by Fluke 8060A true RMS voltmeter (there is some sag in this AC powerline itself as just described). But when I remove AC power, and run from UPS battery, the AC neutral-to-hot voltage actually rises a bit, to 118V. So even with 740W continuous load, the Belkin UPS actually has stiffer power than my lossy AC powerline. I bet it can't continue that much higher, at some point the UPS would shut down when the power line would keep chugging, and you can see from the deteriorating waveform that the UPS is already struggling. It must have some sort of regulation or feedback that is keeping the voltage near optimal within its operating limits, and at 740W continuous load it pushes up against the limits of the inverter power supply, which would explain why the waveform deteriorates. I do not intend to do destructive testing by seeing what happens when I turn heater to medium. At best, it would shut down gracefully. So 740W continuous is as high as I'm going to test.
Here are some real world tests on the AVU 1500 on my living room system. This is what it looks like powering the entire system (Parasound HCA-1500A amp, SVS 1000W subwoofers, everything) with the highest continuous level of pink noise I feel comfortable with (Tact set to -10dB, source is Stereophile CD 2 with -20dB rms pink noise), plus 0.6 amps or so for the scope itself (which is actually more than the amplifiers are using continuously). Total current shown is about 2.2 amps continuous. Still looks very nice, nicer than lots of wall power, though you can see a bit of rounded clipping of the sine waves:
Now this is what it looks like doing the same thing with 42 inch Olevia LCD TV turned on. Current is now shown as 3.3 amps:
Now this looks surprisingly similar to the 740W load above, even though it's not quite 400W. Wall power is usually better than this, but not always, and I've seen worse.
Bottom line, the PureAV backup power is very pure sinewave as advertised up to 250 watts or so, and good enough for most audio uses I have with my current Class AB amplifiers, certainly good enough for keeping everything at idle. It begins to show a bit of strain and loss of sinewave purity when big TV is running also, 400W and above, though still good enough, and even puts out a barely acceptable looking clipped sinewave at 740W with no voltage drop.