I got called to a member of this forum to assist with a F23 error on a Sunsynk.
Normally F23 means that the inverter is sensing a leakage on something like the PV the PV wires. Previously we found wires that got caught between the panels and the rails, or even wires that got damaged where they go through the roof, but at this client we could not find anything wrong with the wires.
He is experiencing alarms early mornings, late afternoon and then up to 22H00 in the evenings. Once the panels starts producing, he wont experience a problem again.
I had all three wires from the panels to the combiner box disconnected and did a few tests. Meggered the wires and the lowest recorded reading was 110 M Ohms on a 1000 volt setting. So no problem there. Then I tested the voltage per panel, adding 1 by one till all strings were connected and saw that each panel gave 45.5 volt although the VOC rating was 43.2V. This was done late afternoon and I would not expect a VOC that high. The last test was interesting. I measured between earth and Positive and got a reading of 42volt. Earth to negative was 0V. I then decided to measure that “Leakage voltage” again starting from a single panel and adding one by one after recording the reading. Each panel i added contributed just over 5 volt to the reading. I have never seen something like this before and this morning I decided i would try it in my own installation. 2 x 400Volt strings and 4 x 200 volt strings. I could not find any DC component between Pos-Earth or Neg-Earth.
At the client, I thought maybe if i connect the panel earth to the council earth I would get rid of the leakage, but when i did, the reading went from 42 volt to 20 volt.
Anyone here that can maybe explain this as I have never seen this in all the years i have been playing in this field. All i know is that he has some unknown panels that i have never heard of before and can only suspect that the panels might be the cause here but don’t know how to prove that.
Can you swap out or remove the suspect panel? If the fault goes away, you can prove that the problem is with the panel. I don’t know how that would be possible, because there should not be any path from the panel connections to the frame?
And then once you have the panel on the ground, you can do more thorough tests.
The problem is, its not one panel. Each of the 8 panels add about 5v to the total of 42 volts. I added one by one to the series string and the value increased with everyone i added.
That indicates something else is amiss. Usually this sort of thing happens when the backing cracks and the panels are wet, then there is leakage to the rails. What happens when you disconnect all the panels (measuring DC and megging to rails).
Thinking about it, it could also be some sort of capacitive coupling.
Yeah, it’s probably not really worth trying to fix it an any other way than replacing with a recognizable brand. I don’t know why anybody still buys random brands? It would still be interesting to find the actual cause.
The client suggested we replace the panels, but I was not comfortable selling him new panels only to sit with the same issue after replacement. I have 8 x 450 JA solar in my warehouse and will repeat the test later today using them. If i can be sure that the panels are causing the issue, I can test the panel to better understand the issue.
Of course there is the risk of DC breakdown on solar cells (if you push too high a voltage across them in reverse), but if it wasn’t for that… or if you didn’t care too much about the panel, I wonder if one can just megger them from each power cord to the frame. That will tell you if it leaks… I am just concerned that it might also damage the panel? Something to research a bit. Maybe at 250V you can get away with it?
Breakdown voltage, in reverse, is usually at least 10V per cell, so depending on where the short is and how many cells are between the short and the cable you are testing. Of course it also depends on current, the small amount of current a megger puts through there, for a short period, may not be damaging at all.
