night shoal
I wonder if the "negative" untracked consumption is really negative or juste a way to show it differently, maybe since an HA update? The graph scale is not negative and I couldn't find any anomalies in my power-monitoring devices. The green energy is my HVAC system and I know that this kind of load can create negative value but couldn't find anything below 0 in the device history. The "negative" untracked consumption seems to follow the HVAC consumption though.
daring dust
It's not new. Means the sum of your devices is more than your total consumption for that period. Bad configuration or a bad sensor usually
night shoal
okay thanks. i'm using a non-shelly clamp on the green device. though i configured it as a 80A clamp it seems it's causing issues. i bought a shelly clamp, we'll see
btw Shelly EM gen3 is -40% https://www.shelly.com/fr/products/shelly-em-gen3
dusky dagger
Mmmh I really wish gen4s were that cheap too
Anyway, I had the same with the Innr plugs and I think the summatieve power has a minimum increase of 0.2kwh or something so it just went through the roof. I'm using the Zeinmann summation integral (or whatever it's called) with an accuracy of 4 decimals and I think it's better than the plugs themselves. Not sure if it's underreporting though.
pale python
I am having this problem too, but with suspiciously large untracked consumption near it
so I figure it could be a timing issue somehow? some consumption reporting got delayed or early?
Those steady columns are ~0.052 kWh
then suddenly it's lower by 0.046, then by 0.282 (pushing it into the negatives), but then it's higher by 0.198, then by 0.036
I'll check when the day is done whether it averages out to the steady figure
or whenever it stops being crazy
pale python
One complication is that I have solar panels, so I wonder if these negative consumptions could be from solar production that my net meter measured but weren't properly matched up to output from the solar production web portal's API
That also would explain why it's steady at night
steady sail
@pale python You're not showing the PV so hard to tell. But 0.052kWh = 5.2Wh is not much. It's equivalent to running a (European) kettle for 9 seconds...
pale python
<@86575941819588608> You're not showing the PV so hard to tell. But 0.052kWh = ...
Well it shouldn't be much, because the only things left that I am not measuring, are like... my lights, a ventilation system, and a hob. at the time it was also my computer screens but I fixed that
maybe some devices like routers, idle TV etc.
Here is my untracked consumption for april 11
And here is my solar production for the same day
I think there is something very suspicious about the way the production starts that day, right?
I think this fits perfectly with the theory that the solar production of the hour before the first bar is simply being reported late
that explains the extreme negative at the start, and some further, less obvious noise would be plenty to explain the rest
Anyway, that gives me something to look into improving, I'll do that next, and I'll make my own thread if I have questions about that since I don't even know if OP has solar
steady sail
Okay, but two lightbulbs is already more than 5Wh 😄
But yeah, that solar start looks suspisous. As that lines up with the dip at 6:00 and the peak at 7:00.
This should also look weird in the energy usage graph. As you should see energy return but no PV at 6. And if you look at the PV energy entity itself it probably makes a jump as well.
Then the question is, what integration do you use for PV?
thick pumice
One complication is that I have solar panels, so I wonder if these negative cons...
weren’t properly matched up to output from the solar production web portal’s API
Does that API report energy or does it report power? And what is the latency?
Unless you’re using one of the few integrations that handles importing historical data, you could have an hour shift on your solar energy.
Perhaps try lining up the solar graph in HA with the data provided by the web portal and see if they match up
pale python
> weren’t properly matched up to output from the solar production web portal’s A...
The error isn't consistent enough for this explanation, it just has a few spikes. Plus I already confirmed the entity enables itself late explaining the first spike
It's the Growatt integration, it reports total energy but I don't know what the API does on the inside
Here's what that looks like. I hadn't posted it here because at this point it seems like a different problem from OP's, narrowed it down to Growatt integration being like this
100% explains the FIRST incidence, don't know what explains a later spike but it easily be some random API latency or something
I suppose it could still be some timing issue where there is a built-in anti-throttling measure that avoids polling when the sun is down and it keeps going and resumes an hour too late
Growatt is really annoying, I wish I had a physical device metering instead