1
\$\begingroup\$

I have a 2-pronged fan plugged into a GFCI in my basement. Twice now, as I was turning it off (out of maybe 30 times I have turned the fan off), the GFCI tripped.

Now, I understand that GFCIs trip when they detect a current imbalance, however, I'm at a loss to figure out what could have caused that current imbalance. Here are the relevant facts:

  • The fan has a plastic case and buttons (except for the metal fan cage). I was not touching any metal part of the fan when turning it off.
  • The fan sits on a wooden pallet and has rubber feet.
  • The fan's cord is in good condition (it was running very close to some metal components, including the fan cage, but there are no cracks). In any case, why would a cord fault cause a GFCI trip only when the fan turns off?
  • The GFCI is working properly (I have a drill press plugged into the same GFCI, no problems).

Where could the current imbalance be coming from?

As requested, pictures: of socket of fan

\$\endgroup\$
  • \$\begingroup\$ Is the drill-press (whatever that is?) unearthed too? \$\endgroup\$ – Andy aka Feb 9 '18 at 15:55
  • 1
    \$\begingroup\$ @LorenzoDonati I think he might mean one of these or these \$\endgroup\$ – Andy aka Feb 9 '18 at 16:11
  • 1
    \$\begingroup\$ @Andyaka Ouch! linguistic barrier in action here, I took it for a joke. Silly me! Rereading the phrase with the explanation of what "drill press" means clears it up! doh! \$\endgroup\$ – Lorenzo Donati supports Monica Feb 9 '18 at 16:40
  • 1
    \$\begingroup\$ @dpdt I think it is the drill press having an asymmetrical leakage to ground close to the trip limit. Along comes a surge or disturbance (that could also be as a result of the fan being activated) and the GFCI trips. You could (with utmost care and no children in the room) disconnect the drill-press earth lead and wire a sensitive ammeter between the disconnected earth wire and proper earth on the GFCI and take a measurement. EXTREME CARE NEEDED OF COURSE. \$\endgroup\$ – Andy aka Feb 9 '18 at 16:46
  • 1
    \$\begingroup\$ Well, if only those two times happened, it could be even something that you wasn't aware when it happened. For example, wet hands or ambient moisture together with poor insulation between the plastic buttons and the inner metal chassis. It's in the basement: if it is humid, some condensation inside the fan might explain the issue. \$\endgroup\$ – Lorenzo Donati supports Monica Feb 9 '18 at 16:49
0
\$\begingroup\$

Assumptions

There does not have to be any leakage to ground for a GFCI to trip. There just has to be an unbalanced current detected.

A GFCI is like a balanced CM choke that reverse the polarity of one winding to use the differential current for activate a latching relay.

If the energy of the difference is enough to trip the breaker, it will. But normally the bandwidth of any transformer and CM chokes included cannot span much more than a few decades of frequency and the transient arc of a switched inductive load (fan) may span many more decades. So I think the mathematical problem is how much energy is imbalanced in the windings over the whole bandwidth not just at the line frequency. THe technical problem is the sensor is design for line frequency and the input here is an impulse.

The solution may be a 1 nF plastic cap ( rated for Vac) across the switch. Then the problem may be during switch on. So try an RC snubber across the switch resistor to limit the bandwidth to 1KHz for fBW~ 0.5 L/R or ~ 0.5 RC using 1nF.

To test for the problem , use the largest inductive load. ( transformer, solenoid, motor etc) THe probability to switch off during peak current is increased with the number of trials. 20 ought to be enough if marginal, 5 or less if significant. It must be stored energy not resistive.

Normally it is done by creating an inbalance of 100uA one line or the other of the GFCI CM Choke at line frequency which translates into 12mW * x ms? which will be less energy than the 1/2LI^2 of the inductive load being switched off. It will take much more energy to cause some of the spectrum to be imbalanced from mismatched windings as a function of geometry , stray capacitance and frequency.

\$\endgroup\$
  • \$\begingroup\$ Where else will the current imbalance come from? Kirchoff's Current Law states that current into hot prong = current out of neutral prong if there are no other current flow routes. Is the current being capacitively coupled to ground? \$\endgroup\$ – dpdt Feb 10 '18 at 0:15
  • \$\begingroup\$ Its the flux that must be balanced not just the current and my theory is that there enough imbalance on some kV switch impulse arc from switching LdI/dt at peak current that the balance is poor for some reason for this spectrum that does not show up at the fundamental due to CMRR vs f \$\endgroup\$ – Tony Stewart Sunnyskyguy EE75 Feb 10 '18 at 2:03

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.