Ghosts in the Darkness

Phantom power produces dim lights during power outage

During the heat wave that gripped the Northeast this past July, increased air conditioning use strained the electrical system. As a result, one phase wire of a residential pole-mounted transformer melted off at the splice between the distribution wire and the wire to the house service connections. This occurred around 6:30 p.m. and led to the loss of one of the two phases that fed each residence off this transformer, commonly called a “can.” However, the voltage loss wasn’t completely at 0 volts, and it produced some interesting results on this phantom phase.

Initially, incandescent lights on the phantom phase produced a reduced brightness due to the circuits containing only approximately 70V. While 120V AC equipment and appliances on the undisturbed phase were unaffected, some equipment on the disconnected phase turned off on their own or malfunctioned. Later, around 8 p.m., voltage dropped to 4V on the phantom phase, not enough voltage to even make a dim glow in the incandescent bulbs. Periodic swells to 25V occurred, with occasional periods of 40V. As shown in Figure 1, the voltage climbed back to 90V about three hours into the event but fell to below 75V several times before peaking at 95V. Around 11:30 p.m., the local utility lineman arrived, turned off the switch on the can, repaired the melted connection by reconnecting the missing phase wire at the splice and re-energized both phases.

The reduced voltage had different effects on different types of loads: incandescent bulbs were lit dimly; motors strained to spin at rated rpm; fluorescent lights had difficulty lighting or did not light at all, depending on the trip point of the specific light; and some AC-rectified DC loads were unaffected as long as the voltage remained above their regulator’s minimum source rating.

So where did the sometimes nearly 0 volts and other times hazardous-level (phantom) voltage come from that separated the wire from the transformer as shown in the above photo? A survey of the neighborhood found several houses with central 240V HVAC units that were turned on during the fault. The transformers and motors inside the units allowed the voltage of the connected phase to couple into the distribution line of the other phase and provide limited power for that circuit. Because the amount of current that could be supplied in this manner was limited by the impedance of these types, the voltage fluctuated depending on what other single-phase loads were still connected to the phantom phase, and the number and impedance of the 240V loads acted as sources.

At least one cyclic load was present on the phantom phase, as seen in Figure 1, and as it cycled off, voltage increases appeared. As more people came home from work and it got darker, the load on the faulted phase increased, and the phantom voltage sagged to 4V at its lowest point. Later in the night, the load current decreased, and the voltage rose until it was fixed. Due to the inductive and resistive nature of the source(s), the phase angle of the phantom, relative to the connected phase, varied from 4 to 25 degrees, depending on the load and number of sources. Note that this voltage on this phase conductor would normally be 180 degrees out of phase with the other phase conductor. Just prior to being disconnected to allow repairs, the existing phase appeared to change frequency. However, this is more likely explained by the waveform being deformed as shown in the Figure 2.

Observations by residents illustrate how perception and reality can be distinct. One resident, who uses a ground-fault circuit interrupter (GFCI) outlet’s test/reset buttons as an off/on switch for a pool filter, reported the GFCI not operating. However, subsequent tests showed that this should not be the case. While it requires approximately 85V to energize the solenoid to allow the GFCI to be reset, the test button is purely mechanical in nature. It trips regardless of the voltage, and the GFCI trip circuit works down to voltage levels where the shock hazard is basically nil. Another resident indicated that his TV blew up with a pop as the wire burned off from its coupling. When the power returned, this proved incorrect and was unlikely, due to the voltage sagging not surging. Losing the neutral connection and ground reference can result in damaging swells but not likely with an open fault-phase conductor.

These precautions can prevent equipment damage during such events:

  • Turn off or protect 240V loads to prevent them from being damaged when acting as sources for other loads.

  • Disconnect motors to prevent overheating at lower voltages.

  • Keep an eye on electronic loads. Those subjected to the phantom voltage that still run at reduced voltage may malfunction or corrupt data when regulators cannot produce the required voltage levels

  • Disconnect nonessential devices to prevent overloading the existing phase and damaging devices connected to the phantom phase.

Don’t assume that an outlet measuring 0 volts will remain that way, unless proper disconnect and lock-out/tag-out procedures have been followed. Even electricity can sneak in the back door.             EC 

BINGHAM is pursuing a master’s degree in engineering at Cornell University and can be reached at


About the Author

Scott T. Bingham

Freelance Writer
Scott Bingham is pursuing a master’s degree in engineering at Cornell University and can be reached at .

Stay Informed Join our Newsletter

Having trouble finding time to sit down with the latest issue of
ELECTRICAL CONTRACTOR? Don't worry, we'll come to you.