In another thread, the topic shifted from my original question about GFCI and AFCI protection devices operating on undervoltage circuits (such as a 120/60 2-pole circuit where a 2-pole breaker would be seeing only 60 volts to ground), to oscillation faults in GFCI (and possibly AFCI) protection device design.
I want to start a new thread so discussion can proceed under a proper title.
The issue is that under some conditions, such as operation at half voltage, or when non-supply sources of fault detection take place (such as an RF field), the devices can go into oscillation, trying to continuously open the a circuit, either unsuccessfully (as in the undervoltage case), or successfully (as in the non-supply case), and repeating that effort continuously with a solenoid probably rated only for momentary and infrequent duty cycles. Continuous operation could overheat the solenoid with the possible risk of causing a short circuit in the device (which is still attached to a source of power) and possibly causing a fire in the building, and injury or death to persons in the building (possibly asleep).
An example of an undervoltage condition that can exist is when a utility transmission or distribution transformer with a three phase delta configuration primary (the most common) loses one of the three hot phase lines, then two of the secondary windings (usually in a wye or star configuation) will be energized with only half voltage due to their primary windings being in series with the remaining two phases (effectively a one phase voltage). I'm told that this condition is supposed to be detected and the entire circuit shut out when it happens to prevent damage. I can't say how many times it actually works correctly because I don't know if this is even the cause in outages I have experienced or known about. But I have experienced cases where the condition occurs and does not result in complete opening of all the phases. In one case, the condition existed for over 3 hours. I have also been told by 2 different people of other cases. One is a case that lasted over 24 hours. Another is a case of involving a 765kV transmission line with one phase down to ground and remaining energized for over 45 minutes (and presumably the other 2 phases energized which would create the single phasing condition at the destination of that line. In my opinion, the cases are sufficiently often that certain protection devices should ideally continue to function correctly, or at a very minimum not contribute further to hazards.
Undervoltage certainly presents many hazards, such as stalled motors that burn up. But for now I want to focus on the GFCI/AFCI risks.
A GFCI/AFCI device designed to operate correctly, or at least lock itself out from operation in certain cases, over a voltage range of
40% to 125%, would eliminate or reduce one risk source. Filtering and shielding of RF energy from falsely tripping the device, or at least locking out the tripping mechanism once tripped, would reduce the other risk source.Under a sufficiently strong RF field, a GFCI/AFCI device could be tripped. But because the device remains energized from the supply side, and the current differential detection circuitry continues to operate after opening the main circuit contacts, it can repeat the tripping action indefinitely. These devices are normally designed under the assumption that once the power source is removed from the faulty load, no further ground fault current will be present and whether the detection circuitry is still active or not is supposedly moot (but not really in the RF field case).
My goal is to have these devices improved so that they function in a non-hazardous way over a supply voltage that between 40% and 125% of the nominal voltage, and/or under an RF field equivalent to 10 watts of power within 1 meter from a half wave radiator (antenna). There are two ways to deal with the RF problem. One is to filter the RF out of the device so it does not falsely trip under these conditions. The other is to design in a mechanism or circuit so that a false trip that cannot be cleared by opening the power contacts will not cause continous tripping or solenoid oscillation.
Possible directions to pursue include working with the NFPA (to add requirements in the NEC that its GFCI/AFCI requirements are only met by devices that meet these improved guidelines) and Underwriters Laboratory (to require devices meet these improved guidelines to be listed). I doubt the pursuit can go through manufacturers since they are unlikely to want to add costly design improvements unless all their competitors are doing under equal requirements. However, some manufacturers may be willing to help push the agenda at the level of NEC and/or U/L requirements.
I know of no injuries or deaths as a result of this hazard. However I am also a believer that we should not wait for an injury or death to proceed with efforts to eliminate or reduce the hazard when we can understand the hazard exists.