GFDI issue

[sunnybai]

Do NOT put a megger on the panels. It can/will create a fault where none existed before! If you need to use a megger (megohmmeter) to check the wiring, you will need to disconnect the panels and the inverter first.

To check the panels, one good way is to take each one in turn out of the series string and replace it with a jumper. When the GF goes away, you have found the panel.
Also, if the voltage to ground that you see is a solid one (carefully put a 100K ohm 1 watt resistor in parallel with your meter leads) then the value of the voltage will tell you where along the string the fault is. Do all of this with the GT inverter disconnected from the panels. If the voltage goes away when you disconnect the inverter, then look for an internal problem there.

The voltage goes away when I disconnect the leads from the inverter, and test to ground. So you are saying there could be an internal problem with the inverter?

Thanks.

[Sunking]

The voltage goes away when I disconnect the leads from the inverter, and test to ground. So you are saying there could be an internal problem with the inverter?

No could be about it. The fault is in your inverter.

[inetdog]

Could you explain in further detail what is in red and how this is connected and how it determines where in the string it goes away?

Sure: Let us say that we have six panels in series which each have a Voc of 40 volts. The total voltage across the string when disconnected from the inverter is therefore 240 volts. But when measuring from either + or - to ground with no fault present, we see zero volts from each of them individually to ground.
The 100kohm resistor is to force some current to be drawn so that we will not read a voltage which results from a very high resistance, not relevant, leakage.

Now, if the ground fault is at the + of the top panel, the meter will still read 0 volts to ground from + and will read -240 volts to ground from the - lead of the bottom panel.

If the ground fault is anywhere inside the top panel, we will read between 0 and +40 on the + lead and between -240 and -200 on the - lead. We have set up a simple voltage divider, where the earth ground is connected in the middle of the previously isolated panel array.

If the ground fault is somewhere in the fourth panel from the bottom, the - lead will show between -160 and -120 (the voltage from the fault point to the - end of the string.) And the voltage from the + lead to ground will be between 80 and 120.

But any of this assumes a strong, not intermittent or-heat related ground fault. It is possible from the description that if the fault is in a panel it is dependent on the temperature of the panel in that case the test will still show 0 volts from both leads to ground until the panels heat up. The panel temp should be about the same whether they are delivering current or not, so the test may have to be done during full sun.

This is not related at all to the more direct and conclusive test of removing one panel at a time looking for the fault to go away, but it does not require diving as far into the wiring harness right at the start.

[sunnybai]

No could be about it. The fault is in your inverter.

Confirmed, Kaco technical support agrees that it is the inverter. New inverter gets shipped out tomorrow.

A bit of history; this particular string has been operating trouble free for approx. just over a year.

No mature trees or squirrels around and I was meticulous in making sure wires are routed properly. Although I still performed a thorough inspection of all conductors.

Sunkings approach is based on FACTS. I appreciate that.

I will confirm operation when new inverter is installed.

Thank you all for your insight and factual information. This forum is excellent.