| The existing design does not allow oscillation | under a fault condition for which the GFCI is | designed.
This is the kind of statement that can be used to say that anything is just fine the way it is. The whole issue is that the conditions for which is was designed is incomplete.
| Installing the device backward, as | you asked about, would allow oscillation under | a fault condition for which the GFCI is designed. | | The point is that installing a GFCI backwards would | not resolve the problem you want resolved, and would | introduce a problem in addition to whatever you see | as a problem with the current design. Thus, to meet | your requirements, a re-design would be needed. I | believe that is what you have in mind with what | you wrote below.
Yes, a re-design is what I meant. The backwards installation idea was meant as an illustration to understand the nature of the problem.
| How much of a problem exists today with GFCI's tripping or | failing due to strong RF fields? How much of a problem | exists that would be resolved by designing GFCIs to operate | at 60 volts? If it is extensive enough, perhaps some | manufacturers will make what you want available. As far as | making such a device a code requirement, you can write a | proposal to have the requirement included in the NEC.
Since encountering the RF field problem and posting about it a few times, I've heard from 3 other people who have experienced the same thing. In all cases the problem was resolved quickly because someone was there to notice. But it doesn't take much imagination to realize that the behaviour of a "cheap" solenoid continuously activating in such a small device could very well result in that solenoid burning out. The results can vary from a device that no longer provides protection to a device that is the ignition source.
As for the operation at 60 volts, I was only curious because I looked into using a 120/60 volt system, where each current carrying conductor would have a 60 volt potential to ground. While I would not expect the power to the solenoid to be referenced to ground, I did wonder if the sensing would be affected in some way. Given that a GFCI device is considered a suitable replacement for a 2-prong receptacle in a circuit without a grounding wire, it would seem it has no ground reference dependency at all, and should be able to operate on the 120 volts that is present, although at the risk of one conductor still being live after the contacts open, if the contacts are only on the supposed hot side (120/60 has both current carrying wires as hots at 60 volts to ground).
That simply got me thinking about the 60 volt issue even though it was not relevant for why I first considered it. In the scope of that issue I do realize that sometimes utility power can fall to half voltage due to loss of a phase leading into a transformer delta primary. And sometimes when that happens the system does not shut off the affected circuits, and the result is 2/3 of the single phase customers having half voltage. Under such a condition, some people who don't fully understand the problem may try to work with equipment in improper ways to see what is going on. They may end up with ground fault conditions. The voltage may be half as much, but it still poses some danger. Given the realistic possibility of a half voltage condition, I think it needs to be considered in the design. I have heard that some GFCI devices will detect being miswired and trip open on such cases. That's a good idea. So why not this one, too.
Making the solenoid so it can operate within expected ranges of half voltage (nominal 60 volts, but could fall to 50 volts), might be the more expensive part of the design. Making a solenoid or control circuit not oscillate is, I think, a less expective design. Of course this is based on what I do know about the design internals, which due to manufacturer secrecy are limited to just the basic theoretical illustrations, which didn't even show which side the solenoid was even powered from (so they are certainly not anywhere near complete).