Ground Fault Protection

[keithwhare]

I intended Positive Connection as meaning bolted up / solid connection (on purpose) as opposed to faulted connections later in the discussions...

I fairly positive that my brother-the-electrical-contractor uses the terminology "permanently bonded".

Keith

[Halfcrazy]

Hi, it is me, Bill (BB.) from the Wind Sun forum.  

7. If you have a car radio (am/fm or CB, Ham, etc.) that uses the case as a ground/safety ground reference... And have an antenna cable with braid to case and the braid tied to earth ground for lightning protection, you have a hard return path through the braid from - Varray to safety ground, through short to + array and can be carrying 10's of amps (or more) of Isc from the array.

So, in the above, your - bus which as always been assumed to be at or very near earth ground / safety ground potential is now at Varray--which is both a voltage hazard and a current hazard.

Bill in this example the GFP would not trip because the bond in the radio would bypass the GFP sensing circuit.

[Kent0]

Bill,

On further review, I agree with you about the battery negative bus being at a hazardous voltage under ground fault conditions. The diagram below explains how this occurs. In terms of safety, it is a compromise. The immediate hazard of the ground fault arc is removed and a potential fire is hopefully averted. I say hopefully because, as mentioned before, its not perfect. For example, if someone installs a piece of automotive radio gear that has the chassis connected to battery negative and the antenna connected to a properly installed ground, the GFP system will be rendered ineffective and PV fault currents may flow on the antenna cable shield.

It seems like the Classic that has the responsibility for the dc system bonding jumper should check on startup, and daily thereafter, to confirm that there is only one dc system bond. When multiple dc system bonds exist, the class should remain inactive (which provides no protection, but at least requires attention) and display a warning that the GFP system has been compromised.

This diagram shows the internal GFP in the Classic charge controller. The hazardous voltage possible when a ground fault exists is the same for the dc GFP used by other manufacturers.




[attachment deleted by admin]

[boB]

That's why the OLD way of shorting out the whole array was better.  Just way more expensive

[BB.]

Thank you Kent0 for the drawing... Yes, that is one of the faults conditions I am concerned about.

-Bill

[BB.]

After a PM to/back from boB--I will post this statement. This is my personal opinion and has not be blessed by Midnite Solar, this forum, or any others. This is just my opinion/interpretation of the various regulatory codes as they exist today (as I understand them). (boB is not making add this--It is my honest opinion ).

The NEC DC GFP system has little to no safety improvement and in fact violates several NRTL Recognition/Listing requirements for multiple countries.

And because of the fundamental flaws in the NEC DC GFP requirements/system, it should immediately be disabled and permanent DC Return/Earth Grounds be used in place of the sensing fuse in DC / Off Grid Systems as a safety/fire hazard.

This includes Hybrid inverter systems that use DC Solar Charge Controllers as part of their DC power design.

I am not too keen on having NEC DC GFP on Grid Tied (Utility Interactive) Inverter systems (I don't think it adds very much in the way of safety) and it does create a nightmare in terms of a ambiguously grounded solar array. And it creates potential confusion of what is or is not grounded in the DC PV Array side of the circuitry--which is always a potentially dangerous situation. Also, I do not know how the NEC DC GFP system affects the new non-galvanically isolated GT inverters. I don't like NEC DC GFP and believe it makes things less safe for a service person--but I don't know any method of bypassing the NEC DC GFP circuitry--and in any case, it may be different by vendor/model number anyway.

Now the details why:

Per Kent0's nice drawing--Yes, that is one of the two (as far as I can tell) failure modes, that DC GFP, is designed to detect/protect against, it actually makes the system very unsafe and an actual safety/fire hazard if it operates as designed (per NEC requirements--Again, not picking on Midnite--They are giving their customers the option of NEC DC GFP grounding or the traditional hard ground of the return bus/earth ground that does not "meet" current NEC).

If there is a fault between +Vpanel (assuming negative ground system) and frame/earth ground, this causes the whole DC battery wiring to no longer meet ELV ratings (more or less, no longer "touch safe") because of the opening of the Ground Fault Sense Fuse/Breaker. Now all down stream 12-48 VDC components/wiring/devices/etc. have to rated for 150+ VDC and Highpot tested to a minimum of 750 VDC. (It has been decades since I have been at this level of design and test for safety/code issues--so I have probably messed some of the details up--but the basics, I believe, are still correct).

And this causes a whole set of nightmares for device manufacturers. For example, the fuse/breaker/PTC needs now to be rated to 150 VDC (or higher for high voltage array product). And if there are any capacitors from -DC to Frame ground, those now need >150 VDC rating. Also, need to check barrier requirements between -DC (and +DC) and Frame ground now that you can have >150 VDC between DC power and frame ground.

Now this means that the DC side of a PV system cannot be considered to be SELV (Safety Extra Low Voltage).

Assuming Wiki is more current than my knowledge:

http://en.wikipedia.org/wiki/Extra-low_voltage

Separated or safety extra-low voltage (SELV)

IEC defines a SELV system as "an electrical system in which the voltage cannot exceed ELV under normal conditions, and under single-fault conditions, including earth faults in other circuits".

This outright affects and violates UL/NRTL registration/listings...

The only way that a NEC DC GFP system would not violate the SELV clause is if Voc-cold is guaranteed to never exceed 120 VDC. I am not sure how solar panels are rated to ensure that this requirement could be satisfied, especially now that Listed Solar panels are rated to 600-1,000 volt (600 VAC / 1,000 VDC?).

And even if the the SELV rating of 120 VDC is not exceeded, I still believe the inversion of the safety ground reference is unsafe because the normal single pole DC breaker/fuses are now on the return lines when the ground reference is reversed (and the Ground Fault detection fuse/breaker is opened). This is also a violation of multiple safety requirements and create both a serious shock hazard and fire hazard.

Personally, I would recommend that people never install (or configure) a NEC type DC GFP system. It is not safe.

-Bill

[Kent0]

The NEC DC GFP system has little to no safety improvement and in fact violates several NRTL Recognition/Listing requirements for multiple countries.

I have to vigorously disagree. The dc GFP does provide significant reduction in fire risk from PV arrays. At present the dc GFP is the only device that provides any protection from arcing faults between positive and ground. Regardless of your opinion of it, the NEC is the letter of the law in the US, Canada, and many other countries and no NRTL listing is possible unless it complies with the NEC. That's not to say I think the dc GFP scheme is perfect, there is certainly room to improve it.

Some ways to do this would be:
1) At system startup, and daily thereafter, check the uniqueness of the dc system ground connection. Any inverter or charge controller responsible for establishing a dc system bond should temporarily disconnect it and confirm that there are no other system bonds. If necessary, the system should remain deactivated until corrections are made. This is especially important for charge controllers because flaws in premises wiring can effect the GFP as seriously as system installation errors.

2) Charge controllers should be required to open the PV array dc grounded conductor as allowed in NEC 690.5(A); not just open the dc system bond. This would remove the high voltage risk on all the user equipment. There would still be a high voltage risk on normally grounded wires in the PV array but there is no reason that risk should carry onto any dc wiring on the premises. Unfortunately, this requires the addition of a switch in the charge controller between PV- and Batt- and it requires the PV array negative to be wired directly to the charge controller. I think this was actually the intention of 690.5(A) and the NRTL that has listed the system used by all charge controller manufacturers has been negligent in allowing the dc system bond to be opened rather than the grounded conductor. The system bond is not a grounded conductor. NEC 690.5(A) only refers to opening grounded conductors not the system bond.

[BB.]

Good Afternoon Kent,

I have to vigorously disagree. The dc GFP does provide significant reduction in fire risk from PV arrays. At present the dc GFP is the only device that provides any protection from arcing faults between positive and ground. Regardless of your opinion of it, the NEC is the letter of the law in the US, Canada, and many other countries and no NRTL listing is possible unless it complies with the NEC. That's not to say I think the dc GFP scheme is perfect, there is certainly room to improve it.

I respectfully disagree.

The ONLY point at which the NEC DC GFP supplies arc fault protection is for (assuming positive ground) is for +Vpanel to Earth Ground faults that are over ~0.5 to 5 amps. It does not protect against + to - arc faults (which I would guess are the majority of the arc fault failures). And it is not really arc fault sensitive--it is just earth fault current sensitive). Also, to stop arc faults to earth, the Fuse/PTC/Circuit Breaker needs to be rated for the maximum interrupt current of the system... That means > 150 VDC and ~80+ amps (for the larger DC controllers) of the solar panels. And if you go >250 VDC, the typical breakers/fuses/PTC are not rated for near those high of DC voltages.

And if there is a +Vbattery to Earth Fault, the interrupting device (ground fault detector/fuse/breaker/ptc) must be rated to the Isc of the battery bank--which for a 80 amp charge controller and a 5% rate of charge battery bank at 12-48 volts is probably well over >> 5,000 amps of DC current--Also an almost impossible job for a small/cheap ground fault detector fuse/breaker/PTC.

Then we get into the part of of the UL/IEC documents which specifically state:

IEC defines a SELV system as "an electrical system in which the voltage cannot exceed ELV under normal conditions, and under single-fault conditions, including earth faults in other circuits".

There have been lots of times my design job would have been much easier if I did not have to consider the wide range of power systems/installations I generically had to address, both to meet code/safety requirements/and marketing requirements.

I do not see any DC power device that goes through same safety requirements that a simple 120/240 VAC device has to go through (including double insulation, or earth grounding, or leakage current tests).

And, given now UL/NRTL requires you to evaluate the "earth faults in other cirucits" requirement--and NEC DC GFP is currently a requirement--I don't see any easy way of passing this short of a fully double insulated DC to DC conversion section (meeting the same requirements as a typical 120-240 VAC power supply).

Some ways to do this would be:
1) At system startup, and daily thereafter, check the uniqueness of the dc system ground connection. Any inverter or charge controller responsible for establishing a dc system bond should temporarily disconnect it and confirm that there are no other system bonds. If necessary, the system should remain deactivated until corrections are made. This is especially important for charge controllers because flaws in premises wiring can effect the GFP as seriously as system installation errors.

Obviously more expensive and, again appears to violate the SELV requirements unless there is a hard return to earth bond in the system--exactly what you are trying to detect and fault the system over.

2) Charge controllers should be required to open the PV array dc grounded conductor as allowed in NEC 690.5(A); not just open the dc system bond. This would remove the high voltage risk on all the user equipment. There would still be a high voltage risk on normally grounded wires in the PV array but there is no reason that risk should carry onto any dc wiring on the premises. Unfortunately, this requires the addition of a switch in the charge controller between PV- and Batt- and it requires the PV array negative to be wired directly to the charge controller. I think this was actually the intention of 690.5(A) and the NRTL that has listed the system used by all charge controller manufacturers has been negligent in allowing the dc system bond to be opened rather than the grounded conductor. The system bond is not a grounded conductor. NEC 690.5(A) only refers to opening grounded conductors not the system bond.

I do not have a copy of NEC (too many years since I had a need for one).

But, I would find it hard to believe that NEC would allow a switchable Return / Earth Ground bond--given all the trouble I had to go into designing a "fool proof" Return to Earth Bond in my systems.

If there was a return path switch that opened the "negative" current path in the controller and there was still a "negative" to earth bond on the battery side of the system--I could see that requirement working. Although, it would:

A) require some sort of active circuitry to figure out the difference between Earth Current Flow and Return Current flow... This is easily and cheaply done with 120 VAC Ground Fault Detectors (current transformer). DC, not so easy. And in any case, GFI's are only required for branch circuits to prevent shock where water is expected (sinks, pools, out doors, etc.). It is not required on equipment that uses a permanent "safety to earth" bond.

B) or require a double insulated, isolated, DC to DC switcher for solar charge controllers--a complete different architecture than is normally done today in Solar (although very common in my industry).

And this does not even address the fact we have a fault, for which NEC DC GFP is designed to "protect against" causing a reversal of "DC Neutral" and "DC Hot"...

This is the equivalent of placing a rinky-dink fuse in the Earth to Neutral bond of a split phase 120/240 VAC home main panel.

Try looking at a Line A to Earth fault in that case... You will now see that the neutral has no over current protective devices inline... And the wire may run from #4 to #14--any random neutral/earth fault can cause excessive current flow in the neutral if the fault is at a 14 awg branch circuit (where it is, statistically, more likely to happen).

And to protect against ARC Faults, that fuse would need to be rated at 10,000 AIC (amps interrupt current) like all AC Mains Breakers/Fuses are required (maximum current supplied by a pole mounted transformer)--A simple ACG fuse or equivalent does not meet this limit.

Very Respectfully,
-Bill

And to be clear, I intend this to be a friendly discussion... I enjoy debating (and learning) technical interchanges. Doing this purely by written communications prevents the normal voice inflections and body language that comes through with face to face talks. 

If anything here appears to be overbearing on my part--It is simply my failure to communicate my intentions clearly...

Although, I still think (from day one) what NEC did with DC GFP is dangerous. 

[boB]

Just use a DC GFP using circuit breakers that has the 2 higher current disconnects ganged with the 1/2 Amp fault breaker Like the  OBDC-GFP/2 and use one high current breaker for positive and the other for PV negative.

That's not going to easily be put into a charge controller though.

boB

[BB.]

Just an FYI to everyone...

With the help from the folks at Midnite solar, I have just submitted a 26 page document to John Wiles asking for him (and NEC) to review the entire NEC DC Ground Fault Protection requirement in face of the serious issues of safety that affect Solar PV Power Systems when installed per NEC Code Requirements.

If anything further happens, I will let you guys know.

Thank you all for the ongoing discussions that we have had about safety. I have documented many of those in the submission.

Cross posted to:

http://www.wind-sun.com/ForumVB/showthread.php?t=10650

Sincerely,
-Bill B.

[Halfcrazy]

Here is a link to Bills paper http://www.midnitesolar.com/pdfs/DC-GFP-Draft3-5.pdf

[keyturbocars]

Good job Bill (and the rest of you guys that helped with this)!

Looks good.

Edward

[laszlo]

Here is a link to Bills paper http://www.midnitesolar.com/pdfs/DC-GFP-Draft3-5.pdf

I don't know what kind of system BB is thinking of -- in my system, the negative bus is locked behind a battery box and runs in conduit and wiring boxes -- there is no way to come in contact with it. The way I understand it from Kent's drawing, the equipment ground would still not have a high voltage under ground fault condition, even when using the DC-GFP required by NEC.

So what is all this fuss about? Is BB using an RV, where it *would* matter a great deal if the frame was energized?

[BB.]

laszlo,

I am sorry, I have not be monitoring this thread/forum, hence the reason for the late answer.

A very simple reason that the 1 amp fuse/breaker in the "neutral" to Earth Ground connection is dangerous (and explained in the too long paper) is the fact that a "hard" earth grounded neutral insures that "neutral" (negative battery bus in the case of negative ground system) never exceeds zero volts with respect to ground.

Since the "neutral" never exceeds ground potential, there is never any reason to place a fuse/breaker on the neutral/return wiring.

I can have a 4/0 battery ground cable go to the negative (neutral) bus and run anything from 4/0 cable to 18 awg wiring anywhere in the system without any fuse/breaker on the neutral wiring.

If, as has been shown here, you get a + or "hot" cable fault to earth (or some other battery/solar/load + or "hot" to earth fault), this simply pops the 1 amp neutral to earth fault detection fuse/breaker. And, depending on exact location of fault and protective devices installed) can now earth reference the 4/0 wiring behind a 300 amp fuse (if anyone installed the fuse/breaker on the battery bus) driven by a 12-48 volt battery bank.

And, now the neutral (formerly earth referenced negative battery/solar bus) is energized around -12 to -48 volts and 300 amps (just an example) of battery +bus current.

And that un-fused distribution circuit with a 14 awg neutral/-bus wire going to your sump pump, radio, fan, led lighting has 300 amps of battery bus voltage/current available.

At that point, the 14 awg wiring is your fuse (rated ~166 amps) and will fail... How it will fail and what it will take down with it--all depends on your configuration and the specifics of the fault.

There are lots of other safety/regulatory issues too (UL/NRTL requirements for grounding of conductors, isolation between "low voltage" and "high voltage" circuits, that floating/both hot circuits need multiple pole/gang trip protective devices, issues with ground references of communications circuits like RS-232 & RS-485 etc....).

Imagine if your your very nicely negative grounded car battery system all of a sudden could flip to positive ground with a simple + wiring short... What are all of the issues that you can think of (car/ham radio antenna grounding, unfused negative wiring, cigarette lighter grounded outlet now hot, what happens to all of the various computers, sensors, servos in your car, you now have to disconnect the positive side of the battery first to avoid grounding your metal wrench instead of the negative side first).

From a systems point of fuse, placing a 1 amp fuse between "neutral/-battery" and "earth/frame" ground is ugly with its multitude of safety/functionality implications.

-Bill

PS July 2, 2018... The link to the my white paper is available through:

https://us.v-cdn.net/6024911/uploads/attachments/512/1965.pdf

-Blll

[laszlo]

Hi Bill!

I have been busy with the job so I didn't read this until now but  I took the time to read and digest what you are saying, and  I 'get it'.

I thought the only issue was a human coming in contact with the DC negative bus in the case of a ground fault, but as you meticulously describe, in the case of a GF the neutral/negative now  has -negative potential compared to the ground and there is no overcurrent protection so it will fault uncheched across the negative/neutral in the same way or worse than  it would if you touched the positive/hot.  Yikes!

Actually, I did have this big question mark in my head when I took the big fat  bonding jumper off in the Magnum in order to enable the GFP in the Classic -- is the #6 that I have running to the Classic ground terminal going to as good as that 2/0 or higher bonding jumper? Luckily I saved the jumper so I can put it back.

So what is Wiles saying? Are they going to revise the standard to address this concern?