GF test on transformerless inverters

[jaggedben]

This is more of a theory question....

We have a system with a Power-One transformerless inverter that is frequently throwing a ground fault error on startup in the morning, apparently moisture related. If it doesn't give the error on startup, it doesn't throw it during operation either.

My understanding is that these inverters perform something similar to a Megger test on the conductors prior to startup. (They also monitor for current imbalance between + and - during operation.)

My question concerns how the solar panels affect the pre-start test. Would this test work through the solar cells? If there's a fault on the wiring somewhere between two of the solar modules, would the the test be significantly less likely to catch it? Conversely, is it reasonable to assume that the fault is most likely on the 'home-run' conductors and not between the panels?

IIRC, the inverter was indicating 0.1 Megaohm resistance between some conductor and ground in the prestart test, and this is too little. (3ma at the open circuit voltage, but that's considered dangerous. Arrgh.) I'm curious about your thoughts on how the resistance of the cells compares and how it would affect the test. I don't really understand what the cells would do when a different source attempts to push a voltage through them.

 

[GoldDigger]

It sounds like the test might be "pushing current through the cells" on the way to ground. The results can vary depending on whether the applied voltage is trying to drive current through the cells in the forward or reverse direction and whether the bypass diodes are functioning.
If you apply the same voltage to ground to both + and - ends of the string, one path to the ground fault between panels will always be in the forward direction of the bypass diodes.

 

[jaggedben]

That makes sense.

So if I have a fault on a jumper between two arrays on the same string, the test should catch it in one direction or the other?
Suppose I unplug only one end at a time on the jumper. The test might trip one time and not the other, and that might indicate that the fault is in the jumper?

Practically speaking, I'm trying to figure out if I can use the inverter test to find the fault by strategically unplugging arrays or jumpers. This kind of fault is too small to find with a Fluke.