Voltage measurement and Uninteruptable Power Supply


One of my uses for the IoTaWatt is to measure main power interruptions. The primary power brick (5V) is plugged into na Uninterruptable Power Supply (UPS) and the voltage reference power brick is plugged into the regular power main. I ship off the sampled voltage reference data to external storage for analysis and alerting. My goal is to automate long recording of how many power interruptions and excess voltage events (arbitrary definition) we have, and how long they last. Sort of a poor man’s Dranetz (for us crusty old IT people who had to prove power DIDN’T cause a crash)…

I have only had the IoTaWatt running for a couple of weeks, no power events in that time to correlate. Just putting this out there to see if there are any flaws in my expectations or connections.


Maybe just unplug the VT to simulate a power failure and see what happens?


Did that, ‘obviously’ everything drops to zero at the status page - no reference=no mains power. I still need to tweak the event detection part in the receiving side. IoTaWatt captured the transient no problem.


One thing to keep in mind is that IoTaWatt records everything as 5 second averages, so a power interruption that is shorter than that will show up as a dip rather than zero, and depending on how it straddles a measurement period, maybe not much of a dip for a real short failure.


Here is a long term view of the recent wind storm and subsequent power failure, time on generator, restoration.


Here is the return of prime power


And here is the period of the failure and move to generator. I don’t have it on a UPS.


@frogmore thanks for sharing real-world data. To me, the ‘dip’ is when the IoTaWatt was off because there was no main power, during what appears to be a switchover to generator. No data=diip in the graph.

I use a small UPS for my low power devices (Rasp. Pi) to avoid ‘pulling the plug’ when there is a power disruption. Looks like the IoTaWatt SD card can survive such events without issue. Since there is a gap in data (none for power off w/o UPS or nil for UPS-protected logic power), the net result is the same. So, with @overeasy statement of the 5 second average, I have enough on the external receiving side to define a crude rule for ‘power disruption’ that will satify my curiosity.


The 2nd low point is when the generator turned on after the power was out for about 30s plus time for it to come up to speed. Here is a close up.


Here is an earlier event before the power really went out. Probably a big tree hitting the power line somewhere.