Hello!
Since December I am observing a strange peak in solar production in the evening. It started in Dec. sporadically. I did not really take notice of it because it was not there daily. In Jan. there were no occurrences but since mid of Feb it happens nearly daily.
Here is an example:
The peak does not represent the real solar production. As new user I am not allowed to send a second image. I can send the pic later.
Does anybody have an idea what this could be?
Looks like it’s after sunset!
Have you looked at the histories of the entities involved?
Yes exactly. Also it is a small solar system with just 800Wp.
The Inverter entities deliver plausible data:
How are you converting W into kWh for the energy dashboard?
Edit: I notice that your power sensor reads unavailable at about the time the spike appears.
How are you converting W into kWh for the energy dashboard?
Just wanted to point out that an energy production above 800Wh is not even possible with my small solar system.
I notice that your power sensor reads unavailable at about the time the spike appears.
Yes the AP Systems EZ-1 inverter (https://emea.apsystems.com/de/) turns off when there is no production. Do you think this is worth to dive into?
Strange though that this was not a problem in last summer.
I think that’s likely to be your problem. The energy dashboard can cope with zero, but not unavailable. I don’t have any generated power, but if the system has been down, when it comes back I get a spike covering all the energy used since the last reading.
Not the same thing, I know, but it shows how unavailable sensors can produce odd results.
Edit again:
That’s 800 W, isn’t it, not Wh? Power, not energy?
I don’t think that’s generally true. It may be a specific integration is not handling something correctly, but generally a sensor going unavailable and then back to its original value shouldn’t mess up the statistics.
That depends on how the sensor behave.
If the sensor gets zeroed during the off period and HA is not knowledgeable about it, then HA might interpret the value wrong, like the sensor have reached its max value and have started from zero again.
How are you converting the power value of your AP Systems E-Z1 sensor (W) into an energy value (kWh) for the dashboard? If you’re not using a Utility Meter helper, that might resolve the issue.
Are you using modbus to get your energy readings? If so it looks like the latest version isn’t playing nicely with a lot of integrations and causes spikes when devices have either network issues/become unavailable.
Thanks alot for all the responses!
First @jackjourneyman: I have a “today production” entity from my inverter that gives me the kWh. Indeed the state changes to unavailable at sunset:
@WallyR gave a good hint: The today production is value that comes from my inverter what means that I do not integrate the production over the day in home assistant. Assume during sunset there is a period where the inverter goes off and comes back later it will send todays total production. Eventually the energy dashboard will interpret it differently compared to the previous values during the day?
@Neil_Brownlee Nope, the inverter has an own integration. I can not say if the integration is based on modbus under the hood.
I’m using the EZ1-M too.
The integration normally gives 6 “total_increasing” - sensors.
A: total lifetime
B: total daily
C: total P1
D: daily P1
E: total P2
F: daily P2
(all in kWh)
For the energy dashboard you should only use A.
The EZ1-M is connected via w-lan and has a local api.
No modbus used and no utility meter helper needed.
Thanks alot that did the trick!
Obviously the issue really was that the state changes to unavailable caused the issue somehow. Anyway with the lifetime production even the “wrong” production bars got updated and fixed!
