Something I’ve noted as my detections have progressed, Sense doesn’t seem to take phase into account at all. For 120 V devices this could speed things up a bit because most people don’t relocate their appliances frequently. If they aren’t moving the device then it is unlikely that they will plug it into a different phase. It could help provide a nice little error check for whether or not a device is really the same one that has been detected or something else. Houses with multiple refrigerators/freezers for example would be much easier to correctly detect as there is a decent chance they could be on different phases. Certainly there are items that get moved a lot (my wife’s heating pad) but I don’t know many people that move major appliances around too often.
For 240V this would help eliminate misidentifying large resistors. My oven was recently detected and I was thrilled. I was less thrilled when Sense tried to tell me my oven was on when it was really the Keurig. As the heating elements in the oven are 240V and the heating element in the Keurig is 120V this should have been a no-brainer. The 240V load causes both legs to spike, the 120V load will only cause an increase in one. I spent most of that day playing whack-a-mole as Sense detected multiple items in the house as Oven and I had to repeatedly report that the oven was not actually on.
Perhaps I am incorrect and this data is being considered. This issue was mostly present during the recent “degraded performance” period.
From what I understand Sense detects legs individually. Each sensor acts independently of the other for detection purposes. Which is why so many people have problems detecting vacuums and other “portable” devices, laptops, chargers, etc…
For example Sense might detect your vacuum accurately when plugged into your living room outlet but when plugged into your bedroom it might not detect it as that circuit might be on another leg or recognize it as a completely separate device. If it’s on the same leg it should recognize it.
I have no idea how it handles devices using both legs aka a 240v device like a stove. But I would imagine Sense would have from type of algorithm for that.
Did you plug anything different into the one leg that would cause noise or something that might confuse Sense?
Phase data on each leg is used by Sense for identification. This video gives you an idea on how Sense categorizes clusters of on/off transitions using phase and power of a single leg, plus many other measured parameters/features.
Thank you both for your thoughtful replies. I’m presently working with support on the issue and they have some concerns about the quality of the data that is being collected. Specifically they believe that my 240 V loads are not showing up evenly across the two phases. I’ve taken some physical actions to ensure that my CTs are fully closed and remain that way. I also verified voltages and current draw matched the app. All seemed well.
One thing somewhat unique is my voltages fluctuate quite a bit. Specifically phase to neutral voltages. They are still within the utility’s parameters so there isn’t much to be done. The root of that issue is 7 houses sharing one ancient transformer.
I’m guessing that some of this issue will be related to calculating wattage for each leg separately. Since my voltages can vary quite a bit from one another the resulting wattages will vary too. If the algorithm is comparing first by amperage the problem shouldn’t exist because phase to neutral voltages will not be a factor. Since I believe it’s converting to wattage first there will be a substantial error.
I’ll update as we get to an answer. Perhaps my charts from the lab section will finally convince the utility to upgrade the transformer.
