Currently the NEC does not require a roof top disconnect for solar panels. Should it?
There is a story out of California of a fire that could not be put out until someone with the proper PPE could cut the wires on the roof.
This does not seem to be addressed in the accepted changes to the 2011 either.

I really think thats not gonna work out on a 4/12 pitch house roof.

Why not?

I suppose you could build a Unistrut rack up there or mount it on the side of the Chimmey, Then since its the ist disconnect should be readily accessable so put a latter going to it.
Most have conduit on outside of house down to an accesible one at ground level.
Sometimes common sense has to enter into the picture.

Since these things are producing pretty modest power I would think a DC rated snap switch in a bell box would do the trick, particularly if you had one per array.
Maybe a contactor with a switch at ground level would work better tho.

I seem to have heard somewhere that turnout gear and a fire ax worked well.

Another was to cover the solar panels with a tarp!

Greg,


I have read several stories about fireman getting zapped from a PV system while they are fighting a fire. I have also seen a new PV system where the converter is right on the roof. It sends down AC power and it has overcurrent protection. This way you can use regular NM wire in the walls which helps to protect the cable from sunlight and damage. Also it helps when you lose AC from the grid, the DC stops on the roof and doesn't continue down like the normal PV systems do.

When you put the inverter on the roof you invoke 690.14(D)


italics indicates handbook commentary

The disconnect referenced in 690.14(C) is the building disconnect and follows the basic rules of a service disconnect. Obviously the folks who wrote this do not think the DC side is worth worrying about and they are not particularly interested in anything outside the building beyond the serviceability of the inverter.

I think that ignores the danger of high voltage, high current DC but nobody asked me. All I can say is be careful up there.
If nothing else, take off your jewelry

Moeller Germany("an EATON brand" now) is just promoting a "firemen disconnect" for pv installations with DC capabilities up to 1kV in a variety of forms.

If actuated it shortens the dc part, so no dangerous voltage shall be present. I think they are not yet available in numbers but might be sold with you quickly if there is a real demand in the market.

Didn't find an English description. Maybe to be sold in the US not as "Moeller" but directly as "Eaton". Btw, also here the standards are behind reality.

Here a link to a German descrpition on www.moeller.net with some photos:
http://www.moeller.net/binary/presse/publications/de_special_09_09.pdf

Greg.,

IIRC the NEC code commite did talk about the PV DC disconnection but it seems they never get to the moot point for safety issue and what I am aware with roof top photocell array they should have DC disconnect switch and I did see at least two did not have proper disconnect switch on the roof top set up the other issue is voltage class that something I know it will cross the line here.

I am not suprised with solar panels can crank pretty serious voltage I know the last one I did couple quick repairs the DC solar panels there were running about 370 or so Volt DC so that one reason why I feel it allready cross the line for safety on resdential area.

Wolfgang I did see the link about the solar disconnect switch device that look good.

I will dig up and find our verison { France } and compared to the USA verison as well.

Merci,Marc

I think if the roof is flat & readliy accessible Yes there should be one there , But in a case like that its a different set up then on a standard house set up. I think in the house case it should be like a service . Right away as you enter the dwelling or outside ground level.

I am not a big fan of running DC PV wiring inside of the house even if it is in pipe. There isn't any over current protection or disco until the end of the pipe.

It's not often that I get to use the same reply for two threads ... this one and the current one about working space

Let me start at the very beginning ...

The NEC is not, as we are clearly told in its' introduction, either a design manual or an instruction manual. As much as inspectors love to have things spelled out -in minute detail, they have the same duty as everyone else apply their judgement. Ours is a system of 'common law,' where we rely upon past practices as well as our neighbors to guide us in our interpretations of the law.

The locations of disconnects is one such area, where design judgement comes into play. For example, the disconnect for rooftop equipment may very well be best located - in terms of access by service personell - in a place that can only be accessed from a ladder. This would naturally place the disconnect above the 6-1/2 ft above grade that is referred to in various parts of the NEC.

Lest I belabor the obvious, it's also quite possible that 'good design' may call for more than just a single code-mandated disconnect.

We saw this issue present a different face recently, when an article in the IAEI magazine advocated the use of GFCI breakers, rather than devices. While that may make your monthly testing easier, it's of little comfort to the rooftop worker when one needs to be reset.

SOLAR, and other generation methods, have concerned me from the start, simply because the feeders are 'live' even when the switch at the panel is "off." It's more of a concern with solar, since multiple panels are often used together for a single feed - and there's no way to turn a solar panel 'off.' A fault in one unit can damage every other unit, unless there is some means to isolate each panel. Likewise, a short in a unit can make everything 'live,' for as long as the sun shines.

I would not, at this point, worry too much over stories of 'firemen getting shocked.' Not that I doubt the accounts; rather, I doubt the competence and completeness of the reporters. I also visit a 'self reliance' forum, and the creativity of the members who have 'gone solar' is exceeded only by their ignorance. I suspect that most of the problems will come from installations that have myriad code issues- the location of the disco being the least of them.

I & anyone dealing a lot with the NEC knows its Not a design Manual But after saying that there sure is a Lot of Do"s & Can nots in it that affect each design.
As far as Disconnect switchs its kinda a rule of thumb . The ist Disconnect what ever the power source is Readily accessible (there are Exceptions such as behind locked doors). After that on a case by case basis they can be
not readly accessible.
Yoopersup

I don't really see a problem with the disconnect being on the roof.
Reason I say that, is because if there was a fire that required the disconnection of the panels, most fire appliances that I am aware of, carry a pike pole, these are inherently made from fibre-glass,with a steel tip, all that would be required, is an actuator (on the disconnect) in the shape of a ring to put the hook of the pike pole though and give the actuator a quick pull.
Having said that, I'm not entirely sure that down is off (isolated) over in the US, with a disconnect?
It wouldn't take much to isolate a solar installation at all, if that were the case.

Just my $0.02 worth.

Trumpy:

Yes, down is "OFF" on this side of the world.

If this was really an emergency disconnect to be operated with a hook, why not a pullout with a ring on it?

Not for nothing but some of the roofs around here are very high. You need extension ladders to get to them. I don't think the fireman could reach the disco unless they were in a bucket truck.

Also the story I read about the fireman getting shocked came from a NFPA magazine.

I also like the new PV systems where they change the AC into DC right on the roof. Then the lines from the roof down can be inside or outside or wherever you want to run them. If the AC shuts down, the Inverter shuts off at the roof and the lines down are not energized. That is much safer in my eyes.

Harold:
I have an issue with unprotected conductors inside of a structure, AC or DC.

690.14(C) covers that

Greg:
Yes, that's the Article.

I was going toward the following scenario:
Panels on roof (resi, SFD); DC thru roof into attic (accessable via pull down stairs); combiner box in attic; EMT thru second & first floor to basement; DC disco in basement w/inverter, net meter; POCO tie-in thru backfed CB in panel.

The DC conductors in the above scenario are unprotected (no OCP or disco) until they get to basement.

You still need the above defined disconnect.

They seem to assume PV sources are by nature current limiting.

I still believe the real answer is a gated output on each collector. They could just add a big FET power transistor on each output that would need to see the "gate" voltage to turn it on. Then all you need to do to disable each array is open the gate voltage circuit.

Greg,

Yeah, the disconnect is readily accessible in the attic with the pull down stairs, while the wire is sparking down in the basement. I guess that makes me feel safe.

Is it any less safe than having the utility service disconnect in the basement and the wire sparking in the attic?

Greg,

At least there is over current protection in the basement panel.

NFPA has said, unlike the utility, solar collectors are somewhat self protecting. If the conductors are sized to more than the maximum output of the collector the conductors are capable of handling anything the collector can throw at them.
Because this power is so expensive on a per watt basis they usually oversize for voltage drop to the point that overcurrent is seldom a problem.
I was always curious what happens when you have a bolted fault in a solar array.
If I have a 1kw array feeding a 1kw toaster wire heater it produces 3400 BTU of heat. (give or take)
If I have a 1kw array feeding a bolted fault where does the 3400 BTU go? Do the collectors burn up or does the imbalance of impedance cause the voltage to drop to close to zero, cutting off the current?

I thought that the DC was ungrounded so that a double fault would be required.

I thought that PV physics has current saturation issues such that bolted shorts are NOT the same as we see with batteries or infinite AC buses.

I didn't say a ground fault, only that it was bolted aka a dead short.

I suppose I was asking about the saturation issue.