NFPA 70®-2020, 2023 and 2026 Editions
National Electrical Code®
TIA Log No.: 1924
Reference: 210.8(F) Exception No. 2
Comment Closing Date: September 9, 2026
Submitter: Thomas Deary, Air-Conditioning, Heating, and Refrigeration Institute (AHRI)
Proposed wording for the 2020 Edition:
Revise 210.8(F) to read as follows:
210.8(F) Outdoor Outlets. All outdoor outlets for dwellings, other than those covered in
210.8(A)(3), Exception to (3), that are supplied by single-phase branch circuits rated 150 volts
to ground or less, 50 amperes or less, shall have ground-fault circuit-interrupter protection for
personnel.
…
Exception No. 2: Ground-fault circuit-interrupter protection shall not be required for listed
HVAC equipment. This exception shall expire September 1, 2026 September 1, 2028.
Proposed wording for the 2023 Edition:
Revise 210.8(F)Exception No. 2 to read as follows:
210.8(F) Outdoor Outlets. For dwellings, all outdoor outlets other than those covered in
210.8(A), Exception No. 1, including outdoor outlets installed at the following locations and
supplied by single-phase branch circuits rated 150 volts or less to ground, 50 amperes or less,
shall be provided with GFCI protection:
…
Exception No. 2: Ground-fault circuit-interrupter protection shall not be required for listed
HVAC equipment. This exception shall expire September 1, 2026 September 1, 2028.
Proposed wording for the 2026 Edition:
Revise 210.8(F) Exception No. 2 to read as follows:
210.8(F) Outdoor Outlets. For dwellings, all outdoor outlets other than those covered in
210.8(A), Exception No. 1, including outdoor outlets installed at the following locations and
supplied by single-phase branch circuits rated 150 volts or less to ground, 60 amperes or less,
shall be GFCI protected:
…
Exception No. 2: GFCI protection shall not be required for listed HVAC equipment. This
exception shall expire September 1, 2026 September 1, 2028.
Substantiation: Despite ongoing efforts from the HVAC manufacturers and GFCI
manufacturers, more time is needed to develop products compatible to the GFCI requirements
listed in NFPA 70, Section 210.8(F). Certainly, the HVAC industry and the GFCI industry have
made progress toward this transition. HVAC manufacturers are no longer requesting an
extension to September 1, 2029. Rather, HVAC manufacturers request that Exception No. 2 be
extended to September 1, 2028. This new date represents a compromise and will allow the
HVAC industry to finish its work to ensure that the public is not adversely affected by these
changes. Many HVAC manufacturers have been working with GFCI manufacturers to develop
products that are compatible with each other, and this process will continue over the next several
years.
AHRI is collaborating with Electric Power Research Institute (EPRI) on AHRI Report No. 8029,
which is investigating the incompatibility of GFCI and HVAC systems. Phase II testing of this
project was completed in early 2026.
The following are the preliminary results of this testing
that will be reflected in the final AHRI Report No. 8029 due to be published and released in June
2026:
The preliminary testing results suggest that high-frequency leakage currents associated with
inverter-based operation, especially during startup/transient windows, can interact with GFCI
sensing algorithms in ways not fully addressed by existing standards.
During laboratory testing, a key issue was identified: when the injected frequency was
sufficiently high, the series impedance required to achieve the target high-frequency (HF)
current became low (on the order of tens of ohms), creating a non-negligible parallel path for
60-Hz line voltage. This resulted in measurable unintended 60-Hz differential current appearing
at the GFCI sensing CT during injection. This introduced ambiguity regarding whether trips
were caused by HF current, unintended 60-Hz leakage, or a combination of both.
Phase II analysis identified the fundamental-frequency leakage mechanism as an interaction
between a front-end switching element (contactor) and common-mode filtering components.
Specifically, when the contactor interrupted only one leg feeding the inverter/filter network, the
common-mode choke could behave effectively as a transformer under the partially energized
condition, producing a common-mode voltage on the downstream side and driving current
through line-to-ground filter elements.
This mechanism was associated with large transient current spikes and voltage distortion when
the contactor closed, including cases where neutral-to-ground voltage excursions and significant
current spikes were observed with tripping propensity dependent on the timing of closure
relative to the voltage waveform.
The future will be in GFCI-HF devices which use an internal filtering to look at the frequency of the leakage. Because the human heart is far less sensitive to electrical currents at high frequencies, these devices can safely auto-adjust their milliamp thresholds upward in the kHz range. This allows them to tolerate normal inverter noise while still maintaining strict 4-6 mA life-safety protection at standard 60 Hz.
Yes, in theory.
But a panel mounted breaker is still just as swappable by a future handyman as an old school fuse. It’s not as secure as having the RCD protection built into the HVAC unit itself. The built in RCD can be factory tested. The combination of breaker mounted GFCI and HVAC cannot be factory tested or validated.
Bryce could it be a simple: When HvAC AC kicks on the gfci trips all the time therefore it needs to be more powerful to handle the on and conditions. I spoke to building codes people about that down in Whittier and they simply said the disconnect breakers have to be proper to be gfci for HVAC at disconnects, my question is if they can do it for pool spas why not for HVAC. We know the outlet in attic serving the blower is GFCI. My question is, how come it is taking so long?
Never needed one on my AC unit. Changed the capacitor and several contactors as well as eliminating the timer (don’t need it with a ‘smart’ thermostat). GFCI is not needed for hardwired installations. What’s needed is someone who can work on systems without injuring themselves. This is a solution looking for a problem.
The reason is the advent of the new class of really great speed modulating linear motors, and the tiny insignificant stray currents they product. The GFCI vendors want to make a profit, and don’t want a solution that’s less expensive and more functional, nor do they want to compromise with the European standards which solved this decades ago.
That’s because the European threshold is 30mA, well above the 6mA threshold in U.S. A human body can lose voluntary muscle control between 9-20mA. From a personnel protection standpoint, that is not safe option.
The electrocution rates between 30mA countries and 5mA nominal countries show — the opposite effect of what you might expect from the above. This is not a settled issue internatioanlly.
On the other hand, big legal cases involving very small voltages and currents have become a thing in farm country in the USA:
Small electric currents affecting farm animals and man: A review with special reference to stray voltage
I. Electric properties of the body and the problem of stray voltage
Physiology
Published: July 1990
Volume 14, pages 287–298 (1990)
Of course. Electrocution is the lethal threshold where an electric shock becomes fatal, which is likely going to be well above 30mA. Between 5mA and 30mA, a human may have anywhere from mild shock to loss of muscle control to respiratory paralysis. You would probably die from respiratory paralysis if the exposure isn’t quickly removed, but that is where the European standard comes in. It relies on a lightning-fast trip time to cut the power before respiratory paralysis or heart failure can turn that shock into a fatal electrocution.
The U.S. standard, by cutting power at the nominal 5mA, ensures you retain muscle control and can physically break contact with the source, which would be extremely helpful if you’re in a pool of water. Suggesting 30mA is a viable alternative for a wet, outdoor HVAC environment ignores the incredibly dangerous reality of muscle freeze.
Bringing up multi-million dollar agricultural lawsuits and cow tingles from equipment failures is a dramatic pivot, but has nothing to do with the issue at hand.
If kids were dying right and left that would change my mind. A single odd occurence of malfunctioning or improperly wired equipment hardly creates a need for regulatory intervention.
Vontrail Pargo was the kid’s name. But the regulation that still causes his name to be uttered in the regulatory hallways WOULD NOT HAVE been prevented by the regulations we got. The elephant not in the room is unpermitted work that does not follow established practice.
The only regulation that WOULD have prevented Vontrail Pargo’s death is to require HVAC equipment vendors to have EGC continuity checks integral to the equipment. Integral. Can’t be swapped out. Then operate in degraded mode to force the owner to call for service, without causing heat stroke death risk.
Every rule has consequences and tradeoffs. Among the tradeoffs of GFCI for hardwired critical equipment is the failure of that equipment to operate. Sewage backups, flooded sumps, stranded motorists, heat stroke or cold risks to humans. Vontrail Pargo is not the start and end of the issue.
I guess you could argue any death is a death too many. But death by AC electrocution falls well below lightning strikes and may not justify the regulatory burden.
If regulators were only concerned with fatalities, they would adopt the Brit’a 30mA threshold and call it a day. An estimated 30,000 non-fatal electrical injuries occur every year. How many could be prevented by expanded GFCI protection? Who knows.
One thing the report you posted tells me is that white men in our age group (45-64) are more likely to be electrocuted doing something stupid. Their last words were probably, “this should work”.
