GFCI protection with VFD

winnie

Senior Member
Residual current detection (such as used by a GFCI) will be defeated by any sort of galvanically isolated power transfer.

If you have a GFCI feeding a normal isolation transformer, then a ground fault on the secondary of the transformer will not be detected by the GFCI. Similarly if the GFCI feeds a motor and the motor runs a generator, etc.

Most VFD outputs are _not_ isolated from the input; a ground fault on the output will be detected as residual current at the input. Additionally the _switching_ in the output of a VFD will often cause large high frequency capacitively coupled current to flow, and will likely trip a GFCI as ptonsparky says. In my lab we need to have our residual current detection set at several amps to avoid tripping on this capacitive current.

-Jon
 

wireday

Senior Member
Personal protection, I have seen this work with line reactor to VFD, Wanted to know if was still GFCI protected.
 

Jraef

Moderator
Staff member
Location
San Francisco Bay Area, CA, USA
Occupation
Electrical Engineer
Most VFDs will trip GFCI breakers, it's a known issue based on the way a VFD functions and how GFCIs sense what it thinks is (or might be) a "ground fault" based on current flowing that does not return. If you have seen a VFD work on a GFCI breaker, it was luck...
But what is the application that needs a GFCI and a VFD? Pool or spa pump? If so, is it a single phase into a 3 phase VFD going to a hard wired pump?
 

synchro

Senior Member
Location
Chicago, IL
Occupation
EE
As has been said above a GFCI will very likely trip on the output side of a VFD, and even the input side if there isn't sufficient filtering from reactors, etc.

A different question (which I think the OP is getting at) is whether a GFCI on the input of the VFD can protect against low level ground faults (~5 mA) on the output of the VFD. I don't think the GFCI would be effective in this case because a ground fault on one of the output phases would appear as a small increase in the pulsed currents on the IGBTs on that phase and through the filter capacitor that shunts the DC bus. Granted a small fraction of the of this low level ground fault current may bypass the filter capacitor. But that means the sensitivity of the GFCI would be severely degraded for detecting a ground fault on the VFD output.
 

kwired

Electron manager
Location
NE Nebraska
Most VFDs will trip GFCI breakers, it's a known issue based on the way a VFD functions and how GFCIs sense what it thinks is (or might be) a "ground fault" based on current flowing that does not return. If you have seen a VFD work on a GFCI breaker, it was luck...
But what is the application that needs a GFCI and a VFD? Pool or spa pump? If so, is it a single phase into a 3 phase VFD going to a hard wired pump?
Expanded GFCI requirements of 2017 and 2020 NEC might make this a bigger issue than it used to be. For most part will only effect cord and plug supplied equipment though.
 

Jraef

Moderator
Staff member
Location
San Francisco Bay Area, CA, USA
Occupation
Electrical Engineer
Expanded GFCI requirements of 2017 and 2020 NEC might make this a bigger issue than it used to be. For most part will only effect cord and plug supplied equipment though.
That's why I asked. There was a thread in here some time ago where we all went off discussing this, turned out the OP was misinterpreting the term "short circuit and ground fault protection" used in the NEC as meaning GFCI (which is does not). But the more recent changes now DO require GFCI in places where it once did not, and pool and spa equipment is one of those.

I too doubt that a GFCI breaker feeding the line side of a VFD is going to detect and trip if there is a ground fault on the LOAD side of the VFD, because the AC-DC-AC conversion process is basically isolating the load side from the line side. The VFD may trip on an output ground fault, most do, but not at Class A GFCI levels.
 

kwired

Electron manager
Location
NE Nebraska
That's why I asked. There was a thread in here some time ago where we all went off discussing this, turned out the OP was misinterpreting the term "short circuit and ground fault protection" used in the NEC as meaning GFCI (which is does not). But the more recent changes now DO require GFCI in places where it once did not, and pool and spa equipment is one of those.

I too doubt that a GFCI breaker feeding the line side of a VFD is going to detect and trip if there is a ground fault on the LOAD side of the VFD, because the AC-DC-AC conversion process is basically isolating the load side from the line side.The VFD may trip on an output ground fault, most do, but not at Class A GFCI levels.
I won't claim to know exactly why VFD's don't necessarily play well with GFCI's, but do agree that they won't detect a ground fault on the VFD output, too much isolation with AC-DC and back to AC between input and output for that to happen. Also agree that VFD's are pretty sensitive and can pick up a ground fault at pretty low levels on the output, but they do this to protect the drive from excessive current more so than anything else. Not guaranteed to give you class A GFCI level protection though.
 

winnie

Senior Member
We probably disagree on terms, not physics.

The output of a VFD is _not_ isolated from the input. Any current flowing out of the VFD to ground must be present as residual current on the input. The rectification, filtering, and modulation don't change this. The electrons have to come from somewhere, and without something like a transformer, they come from the supply to the VFD.

What the rectification, filtering, and modulation do change is the frequency characteristics of the residual current, perhaps taking this current out of the frequency range detected by the GFCI (do normal GFCIs detect DC faults? what about 360 Hz faults?)

Also the PWM switching causes lots of capacitive leakage current which must be dealt with, and can trip the GFCI. I wouldn't call this a nuisance trip since such capacitive current can zap you.

-Jon
 

synchro

Senior Member
Location
Chicago, IL
Occupation
EE
The output of a VFD is _not_ isolated from the input. Any current flowing out of the VFD to ground must be present as residual current on the input. The rectification, filtering, and modulation don't change this. The electrons have to come from somewhere, and without something like a transformer, they come from the supply to the VFD.

What the rectification, filtering, and modulation do change is the frequency characteristics of the residual current, perhaps taking this current out of the frequency range detected by the GFCI (do normal GFCIs detect DC faults? what about 360 Hz faults?)

Also the PWM switching causes lots of capacitive leakage current which must be dealt with, and can trip the GFCI.
-Jon
With the following integrated circuit used in GFCIs, the detection of unbalanced (common-mode) current requires that the current be above some +I[SUB]th[/SUB] threshold for a fixed duration T on positive half-cycles, or below -It[SUB]h[/SUB] on negative half-cycles in order to trip the interrupter.

https://www.onsemi.com/pub/Collateral/RV4141A-D.pdf

So GFCIs using this IC would have the external component parameters set for the duration of a 60 Hz (or 50 Hz) half-cycle. I think you could shorten the time T somewhat with the external components, but the detection method itself does not seem to be very suitable for short duration isolated current pulses such as those from a rectifier like that in a VFD.

There are DC-type GFCIs, and a least some (if not most) of them have a toroidal core but they apply an internally generated AC waveform to a winding in order to drive the core somewhat into saturation. Then any unbalanced DC currents through the "sense" windings on the core would shift the B-H magnetic curves and circuitry would detect the the resulting asymmetry of the nonlinear I/V characteristic and trigger the interrupter.

It seems to me that splitting the DC bus in half with an electrolytic cap on each half, and then connecting the two halves with a DC-type GFCI could be a good configuration. This would help filter out a lot of the unwanted differential mode "noise" coming from the output drivers and the input rectifiers. The DC bus would have to be completely split on all phases, which I don't think all VFDs have for connecting any link chokes.

The only thing is that there's a lot of common-mode noise voltage inherent on the 2-level PWM used in VFDs. And then because of the capacitance to ground on the VFD output cables, it results in a lot of common-mode output current from the VFD. So the threshold for ground fault detection might have to be quite high for it not to be tripping on these currents.
By the way, I came across an IEEE paper about methods to reduce common-mode noise on VFD outputs, but even with an IEEE membership the price was more than I wanted to shell out just for curiosity.
 
Last edited:
Top