Is there any way to test the 'cleanliness' of the power your are getting at your PC from the UPS and also from the wall?
I have a powerful workstation (Dell T7400) that is now on its 3rd PSU in 30 months (all under warranty thankfully), but 3PSU's seems like a lot to go thru in 2.5 years, especially given that I have owned over 20+ machines in the last 15 years and never had a PSU fail on any other machine.
I have a 10 year old APC SmartUPS 2200NET, (now on its 2nd set of batteries), but I wonder if over time they start to give off 'dirty' power that maybe contributing to the premature failure of my PSU's? I don't know enough about elextricity to quantify what I mean by 'dirty' power, but the dell tech that swapped out my PSU's suggested I may want to look into it.
I just bought a newer, (expensive) workstation I to be honest I am afraid to plug it into the UPS, but at the same time don't want to drop another $500-$1000 on a UPS without a need.
Best Answer
Clean power is determined with an oscilloscope and frequency meter. A voltmeter isn't enough, it only shows you RMS voltage to let you know the system is supplying power and if it's capable of supplying voltage at the load you have on it. For the SmartUPS 2200, nominal output is 120V with a less than 5% variance at full load.
The frequency meter tells you how well the UPS is calibrated to output 60Hz power. It can go off tolerance over the years. The SmartUPS is supposed to sync with the grid as long as it's got power and hold the frequency to 57 - 63 Hz when delivering offline power.
The oscilloscope is necessary to make sure you're getting a clean sine wave without power spikes, noise, etc. which is what "clean power" is all about. Some of the original UPS systems output a square wave or stepped square wave attempt to simulate the normal sine wave delivered by the power company. APC systems like your SmartUPS 2200 are spec'd to deliver sinewave output, but if the output switching transistors are having problems, this may have distortion, stepping or noise disrupting the delivery of a clean sine wave output.
Also, at 10 years old, I would also suspect any MOV protection the system has to offer has degraded pretty badly. Metal Oxide Varistors can only take so many power outages, spikes, etc before they lose their ability to suppress the hash spikes that are thrown at them.