New user here, just had solar switched on yesterday. It appears that the MySolar app is under reporting the amount of PV generation in comparison with what the inverter is reporting. I’m assuming I have in some way installed the CT clamps incorrectly. It appears to be only reporting approximately 88% of what the inverter is stating.
I have a ‘Type 1’ solar installation where PV generation is first used here at the house and only after excess PV is generated does it get exported to the grid. This is another way I can tell that MySolar is getting quite the correct figures because it is understanding how much PV was exported in comparison to my smart meter.
My setup includes the ac-ac adaptor along with the two CT clamps, one for inverter and one for consumption. Any suggestions? Would a photo of my meter box with clamp locations be of benefit?
So did some experimenting - it seemed that when I closed my meter box, the CT clamp was getting squished somewhat and not sitting properly over the wire coming from the inverter. I’ve moved things around a bit and am now getting a much more reliable reading - close to 97% of what the inverter software is saying. Is this within the realms of acceptable accuracy?
What are you asking? If it’s acceptable to you, then obviously the answer is ‘yes’. There’s an article in ‘Learn’ that gives the worst-case accuracy that you can expect, and what you’re seeing is a lot better than that. I recently tried to do a ‘Monte Carlo’ analysis of the input circuitry, reported here: OpenEnergyMonitor 9th of March 2018 meeting summary which seems to indicate that your particular set of parts have conspired against the rules of probability.
If you’re using an emonPi/emonBase, you can tweak the
scales = coefficients in emonHub to adjust the recorded data to agree with your inverter. But bear in mind that c.t.'s are inherently inaccurate at low currents.
Great, thanks Richard. I’ll monitor it and see if I need to adjust the coefficients but it seems pretty reasonable then going by what you said.
Just a query - I note that the emonPi can only accept one optical pulse sensor input. Would that be something that may be changed in the future either hardware or software wise? The smart meter I have has two LED pulse outputs, one for consumption and one for solar production. Could be useful for fine tuning the accuracy of solar measurements.
The pulse input goes to the emon part of the emonPi. The single pulse input is (more or less) inherent in the design of the Atmel 328P architecture - it has two interrupt inputs, one is needed for the radio and the other is available for you to use. You could add an emonTH, which can also accept a pulse sensor (via screw terminals, not an RJ45) which would send the data to the emonPi.
You could add a pulse input directly into the RPi, then you’d need to get the cable out and run some python in the Pi to read the GPIO.
If you look at the STM32 development thread(s), you’ll see that work is being done on that and the minimum spec. that we’re looking at says there will be several pulse inputs available. Anticipating the next question, no I don’t know when.