Sometimes it works great, sometimes it shows a value way too low.
I have another power meter that measures my Panasonic heatpump (which I read using an esp connected via modbus), which I have a similar statistics sensor for, and that doesn’t seem to fail in the same way.
It doesn’t seem that the DSMR readings have “dropouts”.
Thanks Tom, but is that actually the solution?
I mean; there must be something wrong (most likely that I’m doing something the wrong way), since it bounces between the right amount and the “way off” amount? But then again - it works with my other meter
I think that’s as good as you can do with statistics sensors. I can view my graph over a longer period of time (this is two weeks) and see gaps. There are no gaps on your history graph?
I don’t have any power measuring sensors in my own instance I can test with - all my statistics sensors do regularly go up and down so I can’t replicate the behaviour exactly.
Sorry malbom, I’m all out of suggestions for this one.
If this sensor was defined in YAML, maybe try a duplicate one in the UI and see if that behaves as you expect, but past that, I can’t offer anything else, sorry!
increase your sample size for the statistics sensor. it’s 24 hours or 1440 samples. If you have over 1440 samples, it’s not going to use all 24 hours of samples.
Or use utility meter, which is designed to handle this type of calculation.
No, sample size is required. Calculate out the number of times your sensor updates in a given 24 hour period. Then double it. The max 24 hours will always be honored.