Ungrounded PV

2

Comments

  • Cariboocoot
    Cariboocoot Banned Posts: 17,615 ✭✭✭
    Re: Ungrounded PV

    The short version of what Bill just said: when you do ground one side of power you eliminate half the potential problems. :p
  • BB.
    BB. Super Moderators, Administrators Posts: 33,431 admin
    Re: Ungrounded PV

    And if the system can "magically" switch the system earth/safety ground from one side of the power bus to the other at any time--You square the amount of problems/issues.

    There is in the NRTL requirements that a single fault cannot make the rest of the system unsafe. (which is different that a system must be "safe" with two or more failures).

    -Bill
    Near San Francisco California: 3.5kWatt Grid Tied Solar power system+small backup genset
  • jaggedben
    jaggedben Solar Expert Posts: 230 ✭✭
    Re: Ungrounded PV

    First of all, happy Thanksgiving. :-)
    BB. wrote: »
    Because all of the "typical" DC GFI systems I have seen use a ~1-5 amp fuse or breaker between "return" and earth/safety ground--The very act of "detection" removes the ground bond (fuse/breaker opening).

    I guess I'm surprised to realize that the 'typical' DC GFI device used on DC stand-alone systems does this for the battery ground. Hadn't paid much attention, having only installed GT inverters and a few stand-alone inverter systems. Of course for (isolated) GT inverters the fuse method doesn't present the safety issue we are discussing.

    Other than enlightening me on this, most of the rest of what you explained is stuff I think I already understand.
    Again, the rest of a DC power system does not have galvanic isolation with respect to the solar array (this would require an "isolation transformer" in the Charge Controller). The DC return/"ground" is carried through the entire system. So lifting the ground bond for the solar array, also lifts the ground bond for the rest of the system.

    Again, it isn't a requirement of the NEC that the system ground bond be lifted.

    Would it really be that difficult to design a GFP device that opened the + and - conductors from the PV array without lifting the ground bond for the rest of the system?

    Is the 'disaster' the NEC requirements, or is it manufacturers not thinking things through and designing a good GFP device that covers all bases?

    How would you say the NEC should be improved to make things better?
    With regard to GT inverters--The maximum working voltage is limited to 600 VAC... If you run two arrays through a common conduit/junction box, a "ground reversal" can cause the + lead of one array to be at +600 volts and the - lead of the second array to be at -600 VDC -- That is a 1,200 VDC potential between two wires in the same conduit--Appears to be a violation of the 600 Volt maximum differential allowed for "standard" voltage wiring (I asked--I could not get a reply to that violation either).

    You're talking about something that can only happen by mistake or by multiple faults. For mistakes, well, those are mistakes. As for the multiple faults, the whole reason the NEC requires GFDI is to provide an indication of the first fault, so it can be fixed before the second fault occurs in the opposite polarity conductor.

    If you think NEC requirements are a 'disaster', what's your proposed solution to prevent fires or 1200V faults on aging systems?
  • Cariboocoot
    Cariboocoot Banned Posts: 17,615 ✭✭✭
    Re: Ungrounded PV

    Basic problem: ground the (-) on an off-grid system and you can't get a DC GFCI to function because it makes negative and ground essentially the same. As such the GFCI can't measure any current difference between (+) and (-) so ... no GFCI.

    NEC set the requirements and says the existing methods meet those requirements. They just ignore the potential problems this creates for some reason.

    Ideally there would be a "double pole double throw" breaker that would not only lift both (+) and (-) but would then bond both circuitry sides to ground.

    As it is the GFCI circuits have a small resistor to provide negative-ground 'bonding' when tripped. Probably okay for low-current PV, but the potential from a battery bank is enormous.

    My opinion is that they designed the rule for GT systems and applied it to OG without thinking of the consequences.
  • BB.
    BB. Super Moderators, Administrators Posts: 33,431 admin
    Re: Ungrounded PV
    jaggedben wrote: »
    I guess I'm surprised to realize that the 'typical' DC GFI device used on DC stand-alone systems does this for the battery ground. Hadn't paid much attention, having only installed GT inverters and a few stand-alone inverter systems. Of course for (isolated) GT inverters the fuse method doesn't present the safety issue we are discussing.

    It is a suggested way of meeting the "safety requirement" by the NEC. The same circuit/fuse/breaker call outs are in the UL/NRTL requirements documents too.
    Again, it isn't a requirement of the NEC that the system ground bond be lifted.

    However, it is a suggestion on how to meet the requirements to reduce the chance of Arc Faults (which this is very poor at preventing anyway).

    Yes, they say the Mfg. is allowed to meet the requirements using other methods--But they are still "ground fault" detection type circuits that are very poor at doing what the intent was--Reducing the chances of Arc Faults.
    Would it really be that difficult to design a GFP device that opened the + and - conductors from the PV array without lifting the ground bond for the rest of the system?

    You could make a solenoid where you wrap some wire around a mechanical system to trip a contact (or set of contacts)--Basically a remote trip breaker. There must be some out there... However, it would have to trip at ~1 amp, but be capable of handling ~600 amps of fusing current (roughly the fusing current of a #6 wire, or should even be larger fusing current rating when you have 300-400 amp fuses for battery main fuses).
    Is the 'disaster' the NEC requirements, or is it manufacturers not thinking things through and designing a good GFP device that covers all bases?

    This is not the only time that NEC has been a "disaster" when making requirements for DC circuits--I used to design computer systems for telephone applications and the NEC requirements where, at times, in direct conflict with Telephone company requirements and good safety requirements (one that I recall was the phone company required crimp lug terminations on their -48 VDC supply lines to bolt to our input bus bars--The NEC prohibited crimp lugs under the requirement that come electricians may not have the proper crimping tools).
    How would you say the NEC should be improved to make things better?

    First--Like a doctor--Do no harm. Recommending a 1-5 amp fuse between return and safety ground is unsafe and violates even the most basic NEC/NRTL requirements for the last century. Period.

    Next, they have to realize that DC systems do not have isolation between solar panels, battery banks, charge controllers, loads, etc. (no native DC equivalent of AC transformer isolation/current/energy limits). Trying to shoehorn AC fault isolation/limiting schemes in DC systems just does not work.

    Without isolation, that means that any "hot to earth" fault, anywhere in the system (solar panels, charger output, loads, battery bank) will "pop" the fuse and either float the DC bus or even inverter from negative to positive ground--Completely violating the idea of "single pole" fusing/breakers/system grounding/signaling grounding, etc.

    And you don't have fault isolation... With AC circuits, the "branch circuit" where the fault occurs, pulls excessive current and trips the upstream breaker/fuse. Or trips the upstream GFI.

    With a single 1-5 amp fuse in the single return to earth ground bond--A short anywhere in the system (panels, combiner box, wiring to charge controller, charge controller, wiring to battery bus, battery bus, battery bank, each DC load, MSW AC faulted to ground, etc. trips the one "ground fault" detection point in the entire DC power system.

    You could put a dozen two pole breaker with remote trip every where in the system... This would be the equivalent of turning off the AC main power to the entire home if somebody shorts an extension cord in the back bedroom... Sure, it is "safe"--But you lose power to everything (including lights, plunging you into darkness, turning of heating/AC/appliances, etc.) and now you have, some where, a fault (somewhere--who knows where) you now have to isolate and fix.

    If they are going to make such a requirement (like a DC ground fault detection/shut down)--Then they need to use DC to DC isolating equipment (like charge controllers with internal isolation transformers from solar array to battery bank).

    Going back to what they are trying to "fix" is Arc Faults. Midnite has product already that uses Arc Fault detection to shut down power flow (turn off charge controller, maybe new product that are arc fault breakers too???)--Which actually will stop common "series" type arc faults (and not only the hot to earth arc faults--which is still unclear that even an arc fault breaker can stop--very difficult to define/have 100% coverage with distributed DC power generation systems).
    You're talking about something that can only happen by mistake or by multiple faults. For mistakes, well, those are mistakes. As for the multiple faults, the whole reason the NEC requires GFDI is to provide an indication of the first fault, so it can be fixed before the second fault occurs in the opposite polarity conductor.

    Nope--this is the design of the a fuse between return and safety ground. It makes the system unsafe or can actually cause fires because of the inversion of negative to positive ground in the whole system because of a single fault... And it is the detection fuse that actually "stops the hot to ground arc fault", none of the other fancy gear added does anything useful (you can drop the series disconnect breaker, stop the converter from turning off, and the 1 amp fuse opening still stops the short circuit current).
    If you think NEC requirements are a 'disaster', what's your proposed solution to prevent fires or 1200V faults on aging systems?

    1,200 volt faults--Easy, don't put a 1 amp fuse between return and safety ground.

    No over current protection on Return (white wires)--Easy, don't put a 1 amp fuse between return and safety ground.

    There is no good way to stop faults in current limited circuits--specifically arc fault--current limiting is how arc welders work--and fuses/breakers will not... Even arc fault breakers (which are actually useful to stop some types of arc faults) tend to trip when using (for example) brushed motors--and will still only work when the faults are in a "protected" section of wiring (i.e., place the arc fault detection in each series string of the array--more arc fault breakers--and they will still not protect down stream arc faults in wiring from combiner box to charge controller--or open with ground faults).

    Should a MSW inverter with non-isolated outputs even be allowed to be sold? Nope--not if you want Ground Fault/Isolation.

    The problem is that distributed power, the current source nature of solar panels, non-isolating/non-current limiting of native DC equipment is just not near as easy/cheap to bolt on solutions as for AC power systems.

    -Bill
    Near San Francisco California: 3.5kWatt Grid Tied Solar power system+small backup genset
  • jaggedben
    jaggedben Solar Expert Posts: 230 ✭✭
    Re: Ungrounded PV
    Basic problem: ground the (-) on an off-grid system and you can't get a DC GFCI to function because it makes negative and ground essentially the same. As such the GFCI can't measure any current difference between (+) and (-) so ... no GFCI.

    The ammeter method used by ungrounded GT inverters doesn't seem to suffer from this problem. It seems to me the issue is one of product design, and perhaps economics, but not necessarily a 'basic problem.'
    Ideally there would be a "double pole double throw" breaker that would not only lift both (+) and (-) but would then bond both circuitry sides to ground.

    Again, nothing in the NEC prohibits such a method. I'm a little surprised that with the concerns you and bill are expressing that this method is not already being used for the systems we are concerned about. Why do you think that is?
    My opinion is that they designed the rule for GT systems and applied it to OG without thinking of the consequences.

    That's surely an accurate opinion. And it means that the rule as designed works just fine for 99% of the solar systems out there. The question then becomes, how should we word an exception to the rule that applies to that small percentage of systems that it doesn't work well for? (DC utilization only, non-isolated...you guys tell me how the exception should be worded).
  • jaggedben
    jaggedben Solar Expert Posts: 230 ✭✭
    Re: Ungrounded PV
    BB. wrote: »
    Yes, they say the Mfg. is allowed to meet the requirements using other methods--But they are still "ground fault" detection type circuits that are very poor at doing what the intent was--Reducing the chances of Arc Faults.

    GFDI is unable to detect series arc-faults, but it is pretty good at detecting half of a parallel arc-fault before the second half happens. If anyone is paying attention to whether the system is working, then the first ground-fault can be fixed before the ground-fault in the opposite polarity conductor happens.
    If they are going to make such a requirement (like a DC ground fault detection/shut down)--Then they need to use DC to DC isolating equipment (like charge controllers with internal isolation transformers from solar array to battery bank).

    So would you suggest that the NEC require this?
    Going back to what they are trying to "fix" is Arc Faults. Midnite has product already that uses Arc Fault detection to shut down power flow (turn off charge controller, maybe new product that are arc fault breakers too???)--Which actually will stop common "series" type arc faults (and not only the hot to earth arc faults--which is still unclear that even an arc fault breaker can stop--very difficult to define/have 100% coverage with distributed DC power generation systems).

    I think your last parenthetical statement probably explains exactly why the code-making panel decided to require GFDI for PV circuits.
    Nope--this is the design of the a fuse between return and safety ground. It makes the system unsafe or can actually cause fires because of the inversion of negative to positive ground in the whole system because of a single fault...

    Perhaps I should have been really specific. I meant a 1,200V arc fault. That requires multiple faults, one in both a negative and positive PV source circuit conductor. At least it does in systems with GFDI. I did not mean to refer merely to a ground-fault that creates a 1,200V differential between (non-faulted) conductors in a conduit.
    And it is the detection fuse that actually "stops the hot to ground arc fault", none of the other fancy gear added does anything useful (you can drop the series disconnect breaker, stop the converter from turning off, and the 1 amp fuse opening still stops the short circuit current).

    Conversely, without the fuse (or some other GFDI device that opens both PV circuit conductors), the hot-to-ground arc fault will just keep on arc-faulting until it starts a fire. It seems to me you have just described why in most cases the 1amp fuse is a good method for isolated inverter systems that don't have batteries or DC utilization. (It's not a good method for all such systems. See: Bakersfield Fire.)
    There is no good way to stop faults in current limited circuits--specifically arc fault--current limiting is how arc welders work--and fuses/breakers will not... Even arc fault breakers (which are actually useful to stop some types of arc faults) tend to trip when using (for example) brushed motors--and will still only work when the faults are in a "protected" section of wiring (i.e., place the arc fault detection in each series string of the array--more arc fault breakers--and they will still not protect down stream arc faults in wiring from combiner box to charge controller--or open with ground faults).

    So should we all just throw up our hands and wait for parallel arc-faults to burn down people's houses?

    As an installer of (so far almost entirely) GT systems, I think I'm going to be a lot happier when customers in twenty years start to approach me saying "My system is not working, can you fix it?" instead of going to the media saying "my solar system burned up without warning!" To the former, I can say "Your system is reaching its end of life, but for a couple grand we can refurbish it to last another twenty years." To the latter ... well, they won't be talking to any of us.
  • Cariboocoot
    Cariboocoot Banned Posts: 17,615 ✭✭✭
    Re: Ungrounded PV
    jaggedben wrote: »
    The ammeter method used by ungrounded GT inverters doesn't seem to suffer from this problem. It seems to me the issue is one of product design, and perhaps economics, but not necessarily a 'basic problem.'

    With an off-grid system there are two power sources to worry about; the PV and the batteries. It gets a bit complex trying to shut down one or both and keep ESG et cetera. So it's a basic problem with OG but not with GT.
    Again, nothing in the NEC prohibits such a method. I'm a little surprised that with the concerns you and bill are expressing that this method is not already being used for the systems we are concerned about. Why do you think that is?

    I think it's two things; the danger Bill has pointed out hasn't been realized by most and/or the market for such a device is relatively small. Manufacturers don't like to invest in small markets.
    That's surely an accurate opinion. And it means that the rule as designed works just fine for 99% of the solar systems out there. The question then becomes, how should we word an exception to the rule that applies to that small percentage of systems that it doesn't work well for? (DC utilization only, non-isolated...you guys tell me how the exception should be worded).

    I agree. Adjusting the rules for OG would seem the simple answer, but does it not negate the safety issue? Right now we're running with "what we've got" in terms of equipment that satisfies the NEC reg (but presents a new problem that isn't being addressed) but is not the best way of doing it.

    This is another case of "as RE systems become more prevalent the equipment will become more ideally suited to the application". I just remarked on this today in another thread about charge controllers and batteries and how they play off one another and both become better as a result. One day we might see 120 VDC OG inverters as common place. Right now handling that much DC Voltage is tricky and there's not a lot of equipment available to do it. Off grid is still tied to the mobile power applications of the past, but advancing.
  • BB.
    BB. Super Moderators, Administrators Posts: 33,431 admin
    Re: Ungrounded PV
    jaggedben wrote: »
    GFDI is unable to detect series arc-faults, but it is pretty good at detecting half of a parallel arc-fault before the second half happens. If anyone is paying attention to whether the system is working, then the first ground-fault can be fixed before the ground-fault in the opposite polarity conductor happens.

    But in the act of detection, it makes the system unsafe. And it only detects a Hot to Earth fault. It does not detect a return to earth fault or an Hot to Neutral fault.

    If you want ground fault detection, then do what is done every where else and float the hot and return and put a (for example) a set of lights between Hot and earth, Return to earth, and Hot to Neutral (or the resistive/current detection equivalent). Everything OK, Hot to Return is bright light. Hot to Earth and Return to earth is 1/2 brightness (they actually bias the hot and neutral leads to 1/2 voltage).

    If there is a Hot to Earth fault, one ground light glow bright and the other goes out, and conversely for a neutral to earth fault. Has been used on sea going vessel AC power systems for a long time. (Much longer than I have been around).

    This is also used on the 370 VDC circuits used in DC only data centers.
    So would you suggest that the NEC require this? (transformer isolation)

    If you "want safety", yes, the systems would be much safer and probably more rugged against nearby lighting strikes and power surges. Also would limit the maximum surge current to a "reasonable level" instead of the many thousands of amperes a battery bank is capable of delivering into a dead short.

    Would it be "cost effective"--We have isolation transformers in virtually every AC to DC power supply we use today... Many are 60 Hz AC transformers, but a very large number are also DC to DC switching power supplies with high frequency transformers. It is a basic requirement to make the DC side "touch safe" (you have ~1,800 VAC double isolation between AC input and DC output).

    So it can be done, but at a cost of mfg. and probably some efficiency losses.
    I think your last parenthetical statement probably explains exactly why the code-making panel decided to require GFDI for PV circuits.

    But a 1/2 shot at the problem made it worse--Not better.
    Perhaps I should have been really specific. I meant a 1,200V arc fault. That requires multiple faults, one in both a negative and positive PV source circuit conductor. At least it does in systems with GFDI. I did not mean to refer merely to a ground-fault that creates a 1,200V differential between (non-faulted) conductors in a conduit.

    In a GT system, a single fault can create 1,200 VDC (worse case) between the "good neutral" and a parallel "inverted ground" return circuit--That (IMHO) appears to violate the basic 600 VAC maximum differential voltage in standard wiring.

    Also, ground bonded neutral leads are supposed to be White in color--Now, with a grounded Hot and an open earth bond fuse, you have the white driven at full current/energy to the opposite rail (negative rail in a negative grounded system).

    In terms of arc faults--Any DC system with an operating voltage over ~12 VDC is going to be a candidate for arc faults. Any current limited supply (like a solar panel/array) is going to be nearly "circuit breaker" proof (resistant) to fault detection/tripping as the difference between Imp (operating) and Isc (faulted) is way too narrow of spread for traditional fuse/breaker techniques (add variable Imp/Isc due to sun, makes detection a computer/sensor type operation to track diverging power results to highlight possible power system problems).
    Conversely, without the fuse (or some other GFDI device that opens both PV circuit conductors), the hot-to-ground arc fault will just keep on arc-faulting until it starts a fire. It seems to me you have just described why in most cases the 1amp fuse is a good method for isolated inverter systems that don't have batteries or DC utilization. (It's not a good method for all such systems. See: Bakersfield Fire.)

    If they floated the solar array instead of ground bonding (hard ground or fuse ground), then a power lead to short would not have caused any arcing/current to flow. Put in a forced "float" circuit with resistor/active detection circuitry--Any power lead to earth fault would be detected immediately and you can use the resulting signal to notify or shutdown (really does not matter in a single fault unless you want to stop current flow if you are guessing that the fault also reduced wire cross selectional area and would overheat with eventual arc fault).
    So should we all just throw up our hands and wait for parallel arc-faults to burn down people's houses?

    Go with full floating Solar PV power and a float fail detection system--Add DC isolation through "power blocks" (i.e., solar charge controller, AC to DC chargers, Inverters--no non-isolated MSW inverters, use isolated communications networks, etc.).

    Add DC arc fault breakers if you want to stop series arc faults (at least reduce the possibility--although, the results of false tripping/missed trip events may be worse than the problem itself--Arc Fault detection is a gray area).
    As an installer of (so far almost entirely) GT systems, I think I'm going to be a lot happier when customers in twenty years start to approach me saying "My system is not working, can you fix it?" instead of going to the media saying "my solar system burned up without warning!" To the former, I can say "Your system is reaching its end of life, but for a couple grand we can refurbish it to last another twenty years." To the latter ... well, they won't be talking to any of us.

    Many times, electrical systems are not easy to rehabilitate. My old homes have knob and tube wiring--It was like pulling teeth to get my contractor to do any wiring upgrades (home was empty, could knock holes in walls, etc.). A friend an I had to do the last part ourselves when the contractor lost interest towards the end of the project (garage was a nightmare of wiring kludges.

    In-laws have a condo filled with aluminum wiring and a contractor not five years ago who just kept adding loads to existing circuits and other fire code issues (stove hood that vented into ceiling common space, etc.). I at least tried to add circuits to support the new loads and use aluminum to copper splicing products that seem to have some sort of good record (there is not much out there). Waiting to see what happens in another decade with aluminum wiring and galvanized water piping (common areas are copper piping, private runs were done with galvanized).

    -Bill
    Near San Francisco California: 3.5kWatt Grid Tied Solar power system+small backup genset
  • inetdog
    inetdog Solar Expert Posts: 3,123 ✭✭✭✭
    Re: Ungrounded PV
    BB. wrote: »
    ... galvanized water piping (common areas are copper piping, private runs were done with galvanized).

    -Bill

    Did they at least put in dielectric unions where the copper and galvanized meet?
    SMA SB 3000, old BP panels.
  • BB.
    BB. Super Moderators, Administrators Posts: 33,431 admin
    Re: Ungrounded PV

    Yes they did... But the units are now about 40 years old now and shedding rust. We had to put a water filter in the sink cold water supply to filter out the obvious rust.

    -Bill
    Near San Francisco California: 3.5kWatt Grid Tied Solar power system+small backup genset
  • jaggedben
    jaggedben Solar Expert Posts: 230 ✭✭
    Re: Ungrounded PV
    BB. wrote: »
    But in the act of detection, it makes the system unsafe.

    For battery systems, maybe. For GT systems, no, not in and of itself.
    And it only detects a Hot to Earth fault. It does not detect a return to earth fault or an Hot to Neutral fault.

    That is simply incorrect. I think you need to review how these systems work. For a 'return to earth' fault, current divides and part of it goes through the GFDI fuse and trips it.

    There is no 'neutral' except in bi-polar PV systems, and a positive to negative fault is unlikely except through two ground faults, which the GFDI does detect.
    If you want ground fault detection, then do what is done every where else and float the hot and return and put a (for example) a set of lights between Hot and earth, Return to earth, and Hot to Neutral (or the resistive/current detection equivalent). Everything OK, Hot to Return is bright light. Hot to Earth and Return to earth is 1/2 brightness (they actually bias the hot and neutral leads to 1/2 voltage).

    More sophisticated GF detection along these lines is probably a good idea for OG systems and for larger GT systems. Not sure if the NEC needs to mandate it though.
    In a GT system, a single fault can create 1,200 VDC (worse case) between the "good neutral" and a parallel "inverted ground" return circuit--That (IMHO) appears to violate the basic 600 VAC maximum differential voltage in standard wiring.

    In a GT with an NEC compliant GFDI (using any method), any 1,200 V differential is immediately interrupted and no service person will be subjected to 1200V dangers. I don't believe any current will actually flow at 1200V during the interrupting. If you think this is otherwise, I think you need to demonstrate it with a diagram, because I think you are wrong.
    Also, ground bonded neutral leads are supposed to be White in color--Now, with a grounded Hot and an open earth bond fuse, you have the white driven at full current/energy to the opposite rail (negative rail in a negative grounded system).

    Perhaps this constitutes an argument against requiring the grounded conductor of a PV system to be colored white (although I don't agree with it). It's not an argument that has any bearing on whether GFDI should be required for PV systems.

    As all well-trained PV professionals know, a system that is indicating a ground-fault needs to be carefully diagnosed and one must assume that just about anything can be energized. With that as a given, it's probably a good thing for a white conductor to act as a signal to someone doing maintenance, as to whether the system is a grounded system or not.
    If they floated the solar array instead of ground bonding (hard ground or fuse ground), then a power lead to short would not have caused any arcing/current to flow. Put in a forced "float" circuit with resistor/active detection circuitry--Any power lead to earth fault would be detected immediately and you can use the resulting signal to notify or shutdown (really does not matter in a single fault unless you want to stop current flow if you are guessing that the fault also reduced wire cross selectional area and would overheat with eventual arc fault).

    I think you what you are describing is essentially similar to what GT manufacturers are doing with ungrounded inverters (i.e. no isolation transformer). I think we've covered this ground.

    So again, I ask, is the problem with the NEC requirements for ground fault detection? Or is the problem that manufacturers haven't caught up to using better methods for GFDI than a fused N-G bond (where N stands for negative, btw).
  • jaggedben
    jaggedben Solar Expert Posts: 230 ✭✭
    Re: Ungrounded PV
    With an off-grid system there are two power sources to worry about; the PV and the batteries. It gets a bit complex trying to shut down one or both and keep ESG et cetera. So it's a basic problem with OG but not with GT.

    Fair enough.
    I think it's two things; the danger Bill has pointed out hasn't been realized by most and/or the market for such a device is relatively small. Manufacturers don't like to invest in small markets.

    I agree. Adjusting the rules for OG would seem the simple answer, but does it not negate the safety issue? Right now we're running with "what we've got" in terms of equipment that satisfies the NEC reg (but presents a new problem that isn't being addressed) but is not the best way of doing it.

    This is another case of "as RE systems become more prevalent the equipment will become more ideally suited to the application". [etc.]

    Yeah, I agree. I think you've pretty much nailed it as to why we are where we are.
  • BB.
    BB. Super Moderators, Administrators Posts: 33,431 admin
    Re: Ungrounded PV
    jaggedben wrote: »
    For battery systems, maybe. For GT systems, no, not in and of itself.

    Assuming that the act of stopping current flow occurs inside the GT inverter (inverter "inhibits", probably does not open a switch anywhere)... then you have reversed the "ground" from Negative to Positive Ground (or vise versa). That ground reversal can only occur when the fuse opens.

    For example, you can have the Hot to Earth short occur at night--The ground has been faulted, but there is no current flowing... The act of detecting the current flow is caused by the fuse (or breaker) opening. Now you have system ground reversal.

    I agree that GT systems are less likely to be affected by ground fault/reversals. Both sides are usually isolated from any RS 422/etc. communications circuits.

    I don't know enough (anything) about transformerless GT inverters... From one point of view, if the DC input meets the standards for AC power wiring (i.e., 600 VAC maximum working voltage, limited maximum current by design or fusing, etc.)--Then I don't have a specific issue with transformerless designs. May be an issue with lightning strike isolation--But that is probably a "minor" issue (get a direct strike, there is nothing much more that can be done to make things "more rugged".
    That is simply incorrect. I think you need to review how these systems work. For a 'return to earth' fault, current divides and part of it goes through the GFDI fuse and trips it.

    Depends on the resistance of the circuit paths. A 1 amp fuse is going to have much higher resistance (1 amp fuse wire has to get hot enough to fuse--But the rest of the circuit has to be low enough resistance to carry 10x to 100x more current without I2R heating issues). Certainly is possible for the fuse to pop with a neutral to earth fault--but there is nothing in the design to guarantee that will happen (like a calibrated resistance required in the return line just before the GT Inverter / Charge Controller).
    There is no 'neutral' except in bi-polar PV systems, and a positive to negative fault is unlikely except through two ground faults, which the GFDI does detect.

    Sure, there are lots of Hot to Return faults too... Something falls across exposed wiring, a load that goes "low resistance" (H Bridge failure, input capacitor failure, etc.), wire insulation compromised during installation, over heating somewhere in system, etc.

    Anyway, the original requirement for DC GFI was to reduce the chance of Arc Faults--I will suggest that there are many more series Arc Faults (i.e., operating current through a failed connection starting an Arc Fault) than there are Hot or Return to Earth Arc Faults--But in any case, the both can happen and this system does not detect or stop such faults.
    More sophisticated GF detection along these lines is probably a good idea for OG systems and for larger GT systems. Not sure if the NEC needs to mandate it though.

    Originally, all of these "requirements" were based on insurance industry initiatives. UL is Underwriter's Laboratories. It was these insurance requirements that were written into the fire codes of major cities (Like New York City).

    Even today, the NEC is "owned" by the NFPA (National Fire Protection Association):
    The mission of the international nonprofit NFPA, established in 1896, is to reduce the worldwide burden of fire and other hazards on the quality of life by providing and advocating consensus codes and standards, research, training, and education.
    The world's leading advocate of fire prevention and an authoritative source on public safety, NFPA develops, publishes, and disseminates more than 300 consensus codes and standards intended to minimize the possibility and effects of fire and other risks.
    NFPA membership totals more than 70,000 individuals around the world.
    In a GT with an NEC compliant GFDI (using any method), any 1,200 V differential is immediately interrupted and no service person will be subjected to 1200V dangers. I don't believe any current will actually flow at 1200V during the interrupting. If you think this is otherwise, I think you need to demonstrate it with a diagram, because I think you are wrong.

    Not by design... If you have two PV Arrays sharing a single conduit to a pair of GT Inverters, it is very easy to get 1,200 VDC from one Hot to the other "inverted" Return.

    If there was a "code requirement" to keep different arrays that are not connected to a single input bus in separate wiring trays/conduit runs--Then, yes it could be less of an issue (then you get into the whole AC/DC separation issue).
    Perhaps this constitutes an argument against requiring the grounded conductor of a PV system to be colored white (although I don't agree with it). It's not an argument that has any bearing on whether GFDI should be required for PV systems.

    Problem is that the only way to "know" if the white is "Hot" is to pierce the insulation with volt meter probes--A standard AC voltage detector will not work on a DC power system. Granted--all conductors should be treated as hot--without testing--But even during service it is possible for a Hot to Earth fault to reverse the neutral/hot circuits--Do you now get into the requirement to manual ground any "neutral" circuits or disconnect from the source.

    And, then this gets into the issue of no circuit protection devices (breakers/fuses) in an OFF grid system in the "return" leads of a grounded system. That "Hot to Earth" fault could have been done in the battery box and now you have >> 400 amps available in the system neutral (depending on where the fault occurred). So even "grounding" the neutral/return at the array does not ensure a dangerous current can not flow during servicing.
    As all well-trained PV professionals know, a system that is indicating a ground-fault needs to be carefully diagnosed and one must assume that just about anything can be energized. With that as a given, it's probably a good thing for a white conductor to act as a signal to someone doing maintenance, as to whether the system is a grounded system or not.

    If I told you that anything in house hold wiring could become hot with upwards of 10,000 amps feeding the neutral from the pole transformer and I told people to be careful--I would think I was nuts.
    I think you what you are describing is essentially similar to what GT manufacturers are doing with ungrounded inverters (i.e. no isolation transformer). I think we've covered this ground.

    Big difference between a GT non-isolated system and an off grid battery non-isolated AC output (I think--not sure).
    So again, I ask, is the problem with the NEC requirements for ground fault detection?

    It is a suggested solution that is the problem. And it does not do much in the way of fixing the original arc fault problem.
    Or is the problem that manufacturers haven't caught up to using better methods for GFDI than a fused N-G bond (where N stands for negative, btw).

    The problem is not obvious on a "component" level--It is much more obvious at the system level (which is were I did most of my engineering design/manufacturing work).

    The fused safety ground just turns a 100+ years of "safety" on its head... Certainly can be addressed--But it affects almost everything.... So of like changing from DC to AC power system--can be done, some advantages, but everything has to switched at once--Cannot be done piecemeal (at least as done here--Transformer isolated charge controllers, inverter requirements can be phased in--fusing the ground--not so much).

    -Bill
    Near San Francisco California: 3.5kWatt Grid Tied Solar power system+small backup genset
  • jaggedben
    jaggedben Solar Expert Posts: 230 ✭✭
    Re: Ungrounded PV
    BB. wrote: »
    It is a suggested solution that is the problem. And it does not do much in the way of fixing the original arc fault problem.

    I already said I don't think it's the best we can do. Beyond that, I think we'll have to agree to disagree. It's only a real problem for probably less than 5% of PV systems being installed. It does mitigate against one type of arc fault problem. And again, there's nothing that prohibits better solutions.
  • jaggedben
    jaggedben Solar Expert Posts: 230 ✭✭
    Re: Ungrounded PV
    BB. wrote: »
    Depends on the resistance of the circuit paths. A 1 amp fuse is going to have much higher resistance .... [etc].

    That's also true of a hot-to-ground fault. Your original statement - "it only detects a Hot to Earth fault" - is simply not correct.
  • BB.
    BB. Super Moderators, Administrators Posts: 33,431 admin
    Re: Ungrounded PV

    A Return to Earth fault basically bypasses the fuse/breaker (in fact, the other threads suggest a bond wire to by pass the fuse/breaker as a fix for the problem without modifying/removing any of the existing protective circuits).

    A 1 amp fuse has around 0.150 Ohms of resistance. A 6 awg wire has around 0.4 ohms per 1,000 ft...

    So, the ability of the DC GFI to detect Return to Earth faults would need A) current flow and B) enough voltage drop to insure sufficient trip current to pop the fuse (develop a voltage drop on the return wire that is enough to cause 2 amps or more to flow through the 1 amp fuse):
    • V=IR=2 amps * 0.15 Ohms = 0.3 volt bias across fuse

    So--Again, depends on where the fault occurs. And if the fault is near the DC GFI fuse--The fault actually can disable the fuse detection path (the earth ground fault current will flow through a shorted return to earth fault instead of through the fuse)... A single return to earth fault will disable the safety circuit (i.e., a Hot to earth fault can now pass current as there is nothing to stop the current flow--even if the detection fuse pops). Not generally a good design choice.

    To may "what ifs" to make a reliable bonding system.

    Let me as you a question this way--What if the utility pole transformer center tap was not earth bonded and you only had a 1 amp fuse between the White Wire bond block and Earth/water pipe Ground. This would stop any arc faults/short circuit current flow from "Hot" to Earth anywhere in the home or even the three or for neighbor homes connected to the same distribution transformer. And we can hook the sense circuit to turn off the emergency backup generator and somebody will eventually notice that when it does not start the next time the utility power goes out--Should add to overall safety, right?

    You do have the problem that the AC Neutral/White wire can be energized to 120 VAC with respect to ground--And depending on where the fault occurs, there could be from 15 to 10,000 Amperes available on that the neutral which has no fuses/breakers by code requirement (again, anywhere in the home, or even in the neighbors' homes too)... But that is a small issue and it would require a double fault for anything really bad to happen.

    -Bill
    Near San Francisco California: 3.5kWatt Grid Tied Solar power system+small backup genset
  • ggunn
    ggunn Solar Expert Posts: 1,973 ✭✭✭
    Re: Ungrounded PV

    Ideally there would be a "double pole double throw" breaker that would not only lift both (+) and (-) but would then bond both circuitry sides to ground.
    Maybe I am misreading you, but I don't think you'd want to ground both the + and - DC conductors from the array. In a big system that would be a very high current short.
  • BB.
    BB. Super Moderators, Administrators Posts: 33,431 admin
    Re: Ungrounded PV

    The idea (as I understand) of grounding the output of the solar array (at the array itself) was to kill the output voltage of the array so that any power lead to earth short would be at zero volts and prevent current flow (as well as make the output "touch safe" during service/repair). The solar equivalent to turning off a generator.

    In theory, a shorted solar panel will be perfectly OK--The current flow through a shorted array (Isc-array) is not much different than operational current (Imp-array) so there should be no damage.

    In practice boB said that (years ago) panel manufacturers balked at shorting the output of the panel as it may cause damage (should not--But "they" objected).

    The other issue (from my point of view)--If you have a series connection problem in the array which can lead to a series Arc Fault would actually sustain an arc fault with shorted array--Much worse than simply turning the array "off" with a standard breaker/disconnect and stopping current flow.

    This is the problem with distributed/solar/battery power--All of these safety issues that are very difficult to solve. Vs the simple idea of just shutting down the generator and all is then electrically safe. You cannot truly turn off a solar panel or a battery cell.

    -Bill
    Near San Francisco California: 3.5kWatt Grid Tied Solar power system+small backup genset
  • Cariboocoot
    Cariboocoot Banned Posts: 17,615 ✭✭✭
    Re: Ungrounded PV
    ggunn wrote: »
    Maybe I am misreading you, but I don't think you'd want to ground both the + and - DC conductors from the array. In a big system that would be a very high current short.

    "Circuitry sides" ol' bean, not the power source sides. Especially not for the battery connection!
    Shorting the panel output wouldn't cause any trouble anyway as they can feed full current into a short circuit without harm.
  • ggunn
    ggunn Solar Expert Posts: 1,973 ✭✭✭
    Re: Ungrounded PV
    "Circuitry sides" ol' bean, not the power source sides. Especially not for the battery connection!
    Shorting the panel output wouldn't cause any trouble anyway as they can feed full current into a short circuit without harm.
    The inverter inputs, you mean? I'm OK with that, but a large array can have hundreds of amps of Isc. Even though by design the conductors can take it, I'm not sure I would want to short that out.
  • Cariboocoot
    Cariboocoot Banned Posts: 17,615 ✭✭✭
    Re: Ungrounded PV
    ggunn wrote: »
    The inverter inputs, you mean? I'm OK with that, but a large array can have hundreds of amps of Isc. Even though by design the conductors can take it, I'm not sure I would want to short that out.

    Well an off-grid array certainly wouldn't have hundreds of Amps Isc as there are no charge controllers that would take it. My original premise was for enabling a safe DC GFCI for off-grid, as opposed to adapting the GT type which doesn't work so well when applied to OG.

    Come to think of it, 100's of Amps on a GT system would be pretty massive as they usually run quite high Voltage to get the power up.
  • jaggedben
    jaggedben Solar Expert Posts: 230 ✭✭
    Re: Ungrounded PV
    BB. wrote: »
    So--Again, depends on where the fault occurs. ...

    To repeat, again: your original statement - "it only detects a Hot to Earth fault" - is simply not correct.

    Considering, also, that most residential GT systems use #10 wire and that the most likely location of ground-faults is on the exposed wiring of the array, I think there's a good chance the fuse method will catch a majority of grounded-conductor to earth faults on such systems.

    And to repeat, again:
    -Having the 1A fuse system for a GT system is better than not having it.
    -It's not the best we can do.
  • jaggedben
    jaggedben Solar Expert Posts: 230 ✭✭
    Re: Ungrounded PV
    Come to think of it, 100's of Amps on a GT system would be pretty massive as they usually run quite high Voltage to get the power up.

    At typical Isc of ~8.75 it takes 23 combined strings to get 200A of Isc. Depending on the string sizing that could be under 40kW. I suppose that might be 'pretty massive' in many folks' estimation, but it would apply to most any central inverter system, so it's not exceedingly rare.
  • Cariboocoot
    Cariboocoot Banned Posts: 17,615 ✭✭✭
    Re: Ungrounded PV

    jaggedben;

    For purposes of this discussion we really do need to keep the distinction between off-grid and grid-tied. It's like two different worlds!

    Most home GT systems are 10kW or less, so no big currents in the array there.

    The same for the protection discussion; the method being used now is suitable for most GT installs. It's using it off grid that brings up new problems. Or indeed with a GT hybrid system. Throwing batteries into the mix adds a new dimension of excitement when things go wrong.
  • ggunn
    ggunn Solar Expert Posts: 1,973 ✭✭✭
    Re: Ungrounded PV
    jaggedben;

    For purposes of this discussion we really do need to keep the distinction between off-grid and grid-tied. It's like two different worlds!

    Most home GT systems are 10kW or less, so no big currents in the array there.

    The same for the protection discussion; the method being used now is suitable for most GT installs. It's using it off grid that brings up new problems. Or indeed with a GT hybrid system. Throwing batteries into the mix adds a new dimension of excitement when things go wrong.
    The original question was about what the diff is between grounded and ungrounded systems; I didn't see where the discussion veered off into only talking about off-grid systems. I do confess, however, to not reading all of some of the more... lengthy expositions. :D
  • mtdoc
    mtdoc Solar Expert Posts: 600 ✭✭
    Re: Ungrounded PV

    I'm really enjoying the discussion here despite it veering off from the OPs question. For someone like myself trying to learn more about the complicated issue of grounding, bonding and GFPs, it is very educational to read the debate and discourse among the knowledgeable contributors here.

    For my gridtie with battery back up system, I recently decided to bond my battery negative to ground based on what I've learned from the discussion of the issues here and on the Midnite solar forum.
  • jaggedben
    jaggedben Solar Expert Posts: 230 ✭✭
    Re: Ungrounded PV
    jaggedben; For purposes of this discussion we really do need to keep the distinction between off-grid and grid-tied. It's like two different worlds!

    Not sure why you needed to say that, because I've been trying hard to do exactly that. Maybe I've missed a couple instances, but I've been specifically calling out where my statements apply to GT or OG systems. In my last post I was responding to your quote which specifically referenced GT systems.

    But to tie the GT vs. OG discussion back into the ungrounded vs. grounded discussion...

    It seems to me that if charge controllers for OG systems used the same (dual ammeter) GFP method that ungrounded (non-isolated) GT inverters use, this whole issue might be solvable, without the issues Bill keeps harping on. The GFP device would be powered by the PV and would open both sides of the PV circuit if it detected a ground fault. The battery side and the battery grounding connection would be unaffected. As far as the NEC is concerned, it probably needs some language to clarify if the PV circuits in such systems should be treated as grounded or ungrounded (I think it should be the latter, even though I don't think the same for GT systems.) But it's already permitted by the current code.
  • BB.
    BB. Super Moderators, Administrators Posts: 33,431 admin
    Re: Ungrounded PV

    I agree the GT side of the "issue" is less of a problem for the "average" person (there simply is not the access to circuits, and my example of 1,200 volt between two hots in the same conduit is not, in my humble opinion, "death". It just appears to be a violation of the NEC requirements--or a possible slight reduction in safety by reducing safety factors)... It still gets me though that in the NRTL requirement, for inverters, I have had to pass for decades (something like a 30 amp ground bond test current from a "grounded conductor" to Earth Ground -- Then later in that same document, they talk about a 1-5 amp fuse/breaker between return and earth grounds.

    On the DC Off Grid Side--It is still highly dangerous to put any sort of fuse/breaker between a "grounded return/neutral" and Earth Ground--It just breaks the whole single pole fuse/breaker of such a system design (because you still have a large battery bank as an energy source).

    See my analogy (a few posts earlier) to "grounding" a home's neutral with a 1 amp fuse (no other grounding at pole transformer or in other homes on the same transformer)... There is no way that would be safe or legal in the US.

    However, as some folks (in the process) appear to justify--There are other countries that use single pole over-current devices on ungrounded power distribution networks. And those countries do not disappear in a puff of smoke.

    -Bill
    Near San Francisco California: 3.5kWatt Grid Tied Solar power system+small backup genset
  • ggunn
    ggunn Solar Expert Posts: 1,973 ✭✭✭
    Re: Ungrounded PV
    BB. wrote: »
    I agree the GT side of the "issue" is less of a problem for the "average" person (there simply is not the access to circuits, and my example of 1,200 volt between two hots in the same conduit is not, in my humble opinion, "death". It just appears to be a violation of the NEC requirements--or a possible slight reduction in safety by reducing safety factors)...
    Sorry, but I must have missed it. Where is the 1200V differential coming from? That's not an ungrounded characteristic, that's bipolar, which is a different issue. There are certainly bipolar ungrounded systems, but not all ungrounded systems are bipolar, and not all bipolar systems are ungrounded. In any case, the DC conductors from the upper and lower subarrays in a bipolar array (grounded or ungrounded) are never in the same conduit, anyway.