On Jan 26, 9:14, dvcorbin(a)optonline.net wrote: can't Yeah, I thought about explaining why the paragraph around the idea that .5mA kills is drivel, and gave up. common. I am currently (preofessionally) developing a product which normally operates off of AC power (via a wallwart), but needs to remort low AC conditions and fall back to battery. over 850mA depending on what it is doing. The voltage fluctuations out of the wall wart (which is rated as 11.8V @ 1A) will rise as high as 17V when under a minimal load. really cut some regulation requirements on the board as well, since the components WILL tolerate this range of voltages. load is significantly greater than the voltage variation based on fluctuations in the AC (eg during a brown out). This has required the development of software that is constantly monitoriing the "active" state of many of the devices to "calculate" the current load, and then going through a transform to estimate the RAW AC that is providing power to the wart. Well, there might be a way to deal with the problem. Most of the poor regulation is down to the following. Most small DC wall warts consist of a transformer feeding a bridge rectifier with a moderate electrolytic capacitor across it. Under low or zero load, the capacitor charges to the peak voltage. As you apply more and more load, you get more and more ripple, and the average voltage goes down (in fact, because there is resistance between the transformer and capacitor, so does the peak). However, you could analyse the ripple and work out what the input was doing. At moderate loads, the capacitor will charge rapidly to something near the peak voltage on each half-cycle, and discharge relatively slowly; at high loads, you'll get a more symmetric ripple. If you want to know in detail, you could sample it, but my guess is if you're just looking for brownouts all you'd need to do is compare the amount of ripple to the average voltage and apply some rule of thumb. Or compare the slope of the rise to the slope of the fall. Of course when the ripple goes away completely you've got no mains input at all :-) -- Pete Peter Turnbull Network Manager University of York