151227-2442 EST
FREEBALL:
Some measurements from my home:
Total home load is about 1700 W and I don't know the balance.
At main panel ---
Phase A to neutral 123.7 V,
Phase B to neutral 123.4 V.
Neutral bus bar to ground bus bar 0.8 mV.
Neutral bus bar to copper water pipe as it comes out of the floor 2.0 mV.
It was inconvenient at the time to measure to a point in the backyard, but I expect less than a volt. My water pipe is an extremely good ground rod as it runs about 150 ft to the street water tap.
The purpose of these measurements is to get handle on the quality of your reference point, and to possibly uncover some unexpected unknown problem.
To your specific problem. This was created by you when you opened the EGC path.
Next some measurements on a nearly 250 ft coil of Romex, #14 w grd, to illustrate the affect of the capacitive coupling between the conductors. Note: meters have an input impedance that is mostly resistive plus a small amount shunt capacitance.
A typical digital meter such as a Fluke has a resistive input value of about 11 megohms shunted by a small amount of shunt capacitance. This input impedance remains about constant for all input ranges. A recent, possibly the last 50 years, Simpson 260 has an AC input impedance that varies with the range setting, and is 5000 ohms/volt. On the 250 V range this is 1,250,000 ohms shunted by a small amount of capacitance. This also corresponds to an I load of 0.2 mA at 250 V input. On the 10 V range this is also 0.2 mA at full scale as would be expected.
Experiment on the above mentioned roll of Romex.
Measured capacitance:
White to EGC, black floating --- 5320 pfd = 0.005,32 ufd.
Black to EGC, white floating --- 4800 pfd = 0.004,80 ufd.
This forms a capacitive voltage divider. With 123 V applied between black and white the voltage readings using a Fluke 27 are:
Floating EGC to white --- 55.2 V.
Floating EGC to black --- 65.5 V.
The sum of 55.2 + 65.5 = 120.7 V. You would expect 123 V. The difference may result from the meter input capacitance.
Theoretically the equivalent circuit relative to white is a voltage source of 123 * 4800/ ( 4800 + 5320 ) = 123 * 4800 / 10120 = 58.3 V in series with a capacitance of 0.01012 ufd. At 60 Hz the capacitive reactance is about 230,000 ohms. This is small campared to 11 M-ohm. As an approximation the voltage reading should be about 58.3 V with the Fluke. Meter input capacitance is possibly part of the reason for the difference between theory and measurement.
Your section of isolated cable is likely much shorter than 250 ft, possibly 25 ft. So we can expect a greater difference between an idealized 1/2 of the source voltage and a meter reading. If you used a Simpson 260 on the 250 V range the voltage would be less than your 56 V. Probably much less.
Almost certainly your problem is a so called phantom voltage.
A current measurement using a 1000 ohm resistor would provide an estimate of the length of the isolated EGC.
.
