voltage at ground wire

BB  and anybody interested in this issue.

Fluke has addressed the high impedance DVM problem with a new line of "electricians"   DVM's that have 10k ohm's /volt input impedance.  These will load down a circuit enough that stray voltages are drained off. May I recommend the Fluke 117 electricians DVM . It is specially designed for the commercial electrician.  I bought one and it is very handy for this situation.  It will not be useful in very low power electronics , digital circuits, etc. use due to the fact that it will load down a circuit and give false readings due to its loading. Fluke standard is 14 megohm input impedance as seen on the 77,87,177,377 etc.

I use a 7 watt incandescent lamp with alligators  as a loading device. You don't need anything larger than that.

david

BB. said:

This meter also has a "low impedance" mode of ~3,000 Ohms (as I recall):

https://www.amazon.com/gp/product/B019CY4FB4

I like it because is also has an AC/DC current clamp too... Has been working well for me.

-Bill

Ditto what Bill says. That's the meter I've been using, and I'm very happy with it.

Does it do in rush for current peaks like motor starts? 

@solarhills,

you  are seeing some leakage current from somewhere in the system......if you are disconnecting BOTH positive and negative from the remote array AND negative is not grounded at the array, then this is some leakage. The voltage can be quite high and can present a shock hazard. To mitigate this hazard I would suggest cleaning the array combiner with an electronic cleaner, degreaser to remove built up dust, etc.  I use CRC QD Electronics Cleaner. This product is safe to use on electronic circuit boards, breakers, switches, etc. as it does not leave a residue. Do not in any case use a product such as WD-40 as it leaves a petrochemical residue which will break down during an arc and produce hazardous gas and will destroy contacts as in breakers, switches, etc.  Recheck  voltages.  One or more panels may have a leakage problem from the wiring to the frame, either internally or at the MC connectors.  CRC QD can be used with plastics and insulators as well.   A good multimeter should show this. Check the current with a sensitive meter, if it's only a milliamperes or so it will not not be a hazard but currents of 30 milliamperes can stop the heart!  Normal GFCI outlets trip at 3 milliamperes , the special GFCI for machinery trip at 30 ma. (This is for your concrete floor mounted hydronic zone heater, furnace, etc.). NEC guidance on the issue is to tie the grounds togather,  bonding is the word used in the NEC. If you are in a lightening prone area I would suggest a bare copper 6ga. minimum wire to be connected between the ground rods. This will prevent the lightening from entering the house and will bond the entire system.  On checking the panels themselves disconnect each one individually and check for voltage from wires to frame or cover panel to exclude light and use the ohm's function to check resistance from wiring to frame....it should be 20 megohm or so. 

@ MrM1,

this is meter does not do peak inrush. There are two kinds of amp clamp meters. This one is a Hall effect sensor as it is AC/DC. Inductive sensors sense inrush current better bur cannot sense DC.   I Have to be very selective in my test instruments as I do live 480 volts including at the primary feed and absolutely require IEC1010 cat 4 instruments as a lightening strike on the power grid miles away from you can induce a 8000 volt spike on the power line. This will explode a cheap multimeter in your hands, with possibly fatal results. Cheap meters are fine on a branch circuit as the capacitance of the wiring will dampen the spike, not to be used in the main box or feeder lines.