Voltage detected between array negative and ground - dissipates - problem?

[TroySmith80]

I recently wired up a combiner box, where several strings of panels come together, and are combined into a single set of +, -, G wires that run to the inverter. While poking around in the combiner box with a multi-meter, checking for any weirdness, i'm finding that there is some voltage between the array negative output and the ground wire. When i test there, i see 35v or so very briefly, and the voltage rapidly (2-3 seconds) drops to 15v or so, and then continues to drop more slowly, approaching zero. The rate of voltage drop seems to change, perhaps on some kind of exponential curve such that it drops a lot at first and then more and more slowly, and may take a long time to get all the way to zero. Note that this is with the array circuits open, and no current flowing. I'm not sure what to make of this. The way that the voltage seems to dissipate when measured makes me think that there's not really much energy there, but i really don't understand it, so i don't want to dismiss it without further investigation.

Any ideas on what could be causing this, or how to determine if it constitutes a problem? Should I not turn on the inverter before resolving this?

More information if helpful: The array is 40 panels, 65W each. They are arranged in 10 series strings of 4 panels each. The 10 strings are then combined into two series groups of 5 strings each, and the two groups are paralleled and fed to the inverter in a single wire pair. Voc of the final output is 410V. The array ground is connected to a grounding rod and the house distribution panel ground.

[bcroe]

I believe you have a multiple string, higher voltage array similar to mine. These tests were done with
what, a digital meter with extremely high input impedance? The problem is these large systems have
some capacity to ground, and extremely small but finite insulation leakage. So left disconnected in
sun, the system leakages will balance out with fractions of the system voltage on the stray capacities
of the system, mostly between each string and ground. When a very high Z meter is connected, it
will detect these voltages, at least briefly. With so little energy involved, even a good meter will
drain these small stray capacities quite rapidly. An old 1000 ohms per volt analog meter would drain
these charges so fast, you might not see them.

Here 4 @ 70W 120V bulbs are wired in series with test leads. These will tolerate the highest system
voltage, light some on far less. They draw small enough current to light for even minimum sun, but
the impedance is low enough to instantly draw off any static charges. With all array string leads out
but not connected, I first see that each string pair will light the bulbs. I have also added a couple LEDs
with a resistor, to indicate polarity.

Next test, make sure connecting the bulbs between different (isolated) strings causes no current flow.

Next test, one bulb lead is grounded, the other bulb lead is contacted with every individual string lead.
There should be no light, indicating there is no ground of the string circuitry anywhere. Pass these,
you are ready to connect for a power test. Bruce Roe

[TroySmith80]

I have a fluke meter, nothing special, but not junk either. I don't know what the impedence of it is.

Ok, i think i understand your test procedures, that sounds like a good idea. I might actually have a hard time finding some incandescent bulbs, lol.

Thanks for the response, very helpful!

[bcroe]

Here is a picture, good example of something put together from what is on hand. Might be
better using 6W bulbs, that metal top to be changed to blue plastic for safety. The 70W
draw a nice arc at 400VDC.

I have found a lot of uses for incandescent bulbs, beside lighting. Those trying to ban
them don't understand users like me. Bruce Roe
PVblbTstr.JPG