AC Units tripping GFI protection

[kwired]

I think the MFG's are spewing information to the CMP's that they want them to hear so they will add more GFCI requirements even if not all that justified.

 

[ActionDave]

The manufactures have benefited greatly from changes in the NEC and I am sure there is a large amount of lobbying and schmoozing that goes on but the blame for this GFCI foolishness and other crap code changes lies at the feet of the CMPs. They are the ones that vote on the proposals.

 

[mbrooke]

So what are we as an industry going to do about this? Why is no one taking action as a group?

 

[ActionDave]

Because we are all busy working.

 

[ActionDave]

I think the NEC is loosing some credibility given that there are more local amendments removing sections than I remember in the past.

 

[synchro]

From the technical side, I think the best way to address the tripping problem on AC units with VFD's is to apply appropriate filtering to both the equipment and within the GFCI breaker. This would reduce the high frequency common mode noise reaching the breaker and also reduce the breaker's response to that noise. I think such a balanced approach is better than applying more filtering on just the line input of the AC unit, because that will be more costly, larger, and potentially result in more power loss than if the filtering in the detection circuitry within the breaker is also enhanced.

I believe most if not all GFCI breakers have a trip time vs. applied leakage current that's significantly faster than the UL 943 requirement. This may be (at least in part) to provide comfortable margins for component tolerances, operating conditions, etc. if there are no other significant constraints. And so there should be some room to reduce the bandwidth of the filtering within the breaker, and possibly use additional filter poles while keeping an adequate dynamic response. So I think there may be some opportunity for a new version of a breaker targeted specifically for this application if it can be justified in this market. Or better yet, maybe they can make it viable for general usage.

A reference to such filtering is made page 10 of the ON Semiconductor application note for their NCS37010 IC made for GFCI devices (available at the link below):

"Filtering
The analog signal capture portion of the IC includes a single pole filter that can be set externally with Cidf. This provides an additional layer of protection against false tripping under steady state noise conditions. High frequency steady state noise is common with pumps, motors or other cyclic noise generators. Cidf 220 nF 1 kHz low pass.
For additional filtering suggestions please contact ON Semiconductor."

https://www.onsemi.com/pdf/datasheet/ncs37010-d.pdf

 

[mwm1752]

EPD breakers work - they are just not for personal protection -

 

[ActionDave]

From the technical side, I think the best way to address the tripping problem on AC units with VFD's is to apply appropriate filtering to both the equipment and within the GFCI breaker.....

What about

EPD breakers work - they are just not for personal protection -

I would have no objection to EPD protection in some of the recent areas where GFCI protection has been added.

 

[mbrooke]

EPD breakers work - they are just not for personal protection -

As do ground proving devices.

If the CMPs reasoning is based on missing EGCs, and requiring GFCIs instead of other approved means, this in of itself is bias in the code making processes. The NEC can not endorse only one method.

 

[don_resqcapt19]

Two more proposed TIAs have been submitted on the issue of 210.8(F), both are attempting to set the effective date of this requirement as applied to AC units to 1/1/23. The proposed TIA numbers are 1589 and 1593. The four previous ones have all been rejected and I do not expect that either of these will pass.
Part of the substantiation for 1589 says:

Substantiation: GFCI protection was expanded in the 2020 NEC without HVAC component
and equipment safety standards being harmonized with GFCI amperage limits. Currently the UL
standard that HVAC equipment is listed to (UL 1995) has no requirements for leakage current if
the unit is hard wired, as most residential air conditioners/heat pumps are. In the future, HVAC
equipment will be listed to UL 60335-2-40, which sets a limit of 10 milliamps of leakage current.
However, this new standard is not mandatory until 1/1/2024. UL 943 is the standard to which
GFCI breakers are listed and are required to trip at 5 milliamps of current. Even if HVAC
equipment is listed to the UL 60335-2-40 standard, there is no guarantee it will be compatible
with UL listed GFCI breakers This lack of coordination is what is leading to the nuisance
tripping that customers are dealing with.
Until both equipment and component standards are updated, designers, installers, AHJs, and
consumers are forced to choose between an NEC 2020 compliant installation or an operational
installation. In jurisdictions that have adopted 2020 NEC with 210.8(F) intact, there have been
numerous instances of field tripping of the GFCI breaker on ductless mini splits, units containing
power conversion equipment, and on many single-stage units. ...

 

[mbrooke]

60335-2-40 takes me to this? >>

IEC 60335-2-40:2018 RLV | IEC Webstore | water management, smart city, environmental protection, environment

IEC 60335-2-40:2018 RLV Standard | water management, smart city, environmental protection, environment | Household and similar electrical appliances - Safety - Part 2-40: Particular requirements for electrical heat pumps, air-conditioners and dehumidifiers

webstore.iec.ch

 

[kwired]

What about

I would have no objection to EPD protection in some of the recent areas where GFCI protection has been added.

Unless the "noise" that causes nuisance trips still effects those, then we haven't gained much.

 

[Ravenvalor]

I install the Panasonic DC bathroom exhaust fans. I noticed that they have a noise filter inside the electrical enclosure. Is this perhaps there to prevent the GFCI that may be feeding it from tripping?

Thanks.

 

[don_resqcapt19]

...
I believe most if not all GFCI breakers have a trip time vs. applied leakage current that's significantly faster than the UL 943 requirement. This may be (at least in part) to provide comfortable margins for component tolerances, operating conditions, etc. if there are no other significant constraints. And so there should be some room to reduce the bandwidth of the filtering within the breaker, and possibly use additional filter poles while keeping an adequate dynamic response. So I think there may be some opportunity for a new version of a breaker targeted specifically for this application if it can be justified in this market. Or better yet, maybe they can make it viable for general usage.
...

If they took as long to trip as UL 943 permits, most would think that they are not functional when the try the test button. The resistor for the test button typically around 15k ohms and that would mean you would have to hold the button in for 3.7 seconds if the device was set to maximum permitted time to trip.

 

[synchro]

If they took as long to trip as UL 943 permits, most would think that they are not functional when the try the test button. The resistor for the test button typically around 15k ohms and that would mean you would have to hold the button in for 3.7 seconds if the device was set to maximum permitted time to trip.

If a GFCI is redesigned to utilize the longer time allowed by the UL 943 trip curve, then I believe the "supervisory circuit" UL 943 specifies to allow manual testing could be designed to hold the test current imbalance at a constant level for the required duration after the button is initially pushed. So then the user could release the button quickly and the proper test would still occur. Of course they'd have to wait several seconds for the test to be completed.

 

[don_resqcapt19]

If a GFCI is redesigned to utilize the longer time allowed by the UL 943 trip curve, then I believe the "supervisory circuit" UL 943 specifies to allow manual testing could be designed to hold the test current imbalance at a constant level for the required duration after the button is initially pushed. So then the user could release the button quickly and the proper test would still occur. Of course they'd have to wait several seconds for the test to be completed.

It is very unlikely such a change would make it through the STP for UL 943.

 

[synchro]

If they took as long to trip as UL 943 permits, most would think that they are not functional when the try the test button. The resistor for the test button typically around 15k ohms and that would mean you would have to hold the button in for 3.7 seconds if the device was set to maximum permitted time to trip.

Existing GFCI breakers respond to the test button by tripping the mechanical switch in what appears to be a small fraction of a second. That delay includes the detection circuitry and its filtering, the SCR, solenoid, and trip mechanism. The ON Semiconductor IC at the link I provided above specifies a response time to 5 to 20mA ground faults of 150 milliseconds typical. The single pole noise filter suggested on pg. 10 of this document will settle to within 1% of its final value from a step input in 0.73 milliseconds. So I suspect that there may be significant room to reduce the bandwidth of such filtering to provide additional noise rejection without getting anywhere near the maximum delay allowed by UL 943. But it would take some digging to know how much opportunity there is without noticeably affecting the response to a test button.

 

[don_resqcapt19]

Existing GFCI breakers respond to the test button by tripping the mechanical switch in what appears to be a small fraction of a second. That delay includes the detection circuitry and its filtering, the SCR, solenoid, and trip mechanism. The ON Semiconductor IC at the link I provided above specifies a response time to 5 to 20mA ground faults of 150 milliseconds typical. The single pole noise filter suggested on pg. 10 of this document will settle to within 1% of its final value from a step input in 0.73 milliseconds. So I suspect that there may be significant room to reduce the bandwidth of such filtering to provide additional noise rejection without getting anywhere near the maximum delay allowed by UL 943. But it would take some digging to know how much opportunity there is without noticeably affecting the response to a test button.

The requirement in UL 943 is to simulate a ground fault. This is done by the test button connecting a resistor between the hot conductor on the line side of the CT to the neutral on the load side of the CT.
This is the test button requirement from UL 943.

5.15.3 The supervisory circuit shall operate to create a current imbalance of the sensing toroid, but shall
not rely on the equipment grounding conductor to operate.

No idea of what the IC link says, as the governing document is UL 943. That specifies the maximum time to trip as the quantity of 20/current in mA raised to the 1.43 power.

 

[SSDriver]

I have now fixed multiple HVAC and pool pumps(with drives) tripping GFCI's by simply installing Siemens brand GFCI breakers. Unfortunately this usually means I'm installing a Siemens Spa panel inline somewhere for the HVAC units.

 

[kwired]

I have now fixed multiple HVAC and pool pumps(with drives) tripping GFCI's by simply installing Siemens brand GFCI breakers. Unfortunately this usually means I'm installing a Siemens Spa panel inline somewhere for the HVAC units.

So the question is what is different about the Siemens GFCI's?

If the manufacturers want to push the CMP's to keep expanding GFCI requirements sure would be nice if they fixed some those bugs that apparently Siemens has found a way to fix.