Solar install turning out really bad - looking for some help/advice please.

[bcroe]

You can actually have up to 25 panels per string, but no more than 5250W DC,
which is why my string is 16 panels.
The way I understand it, if you have shading issues or different orientations, the more panels per string, the better.
Here is a long video on the SolarEdge Optimizers, but the explanation on how they work starts at 21:25.

Thanks for that info. It sounds like my idea was on the right track. IN ADDITION, their inverter
communicates with the power boxes to set the output voltage of each one. That is how they
maintain a 350V string voltage. I didn't hear an explanation of how that communication is done.

This system does solve problems with shading and odd numbers of panels. The claim of 15%
more energy might apply in a difficult situation like that. I don't think they would have any
advantage over a straight forward ground mount string system that had no complex mounting
restrictions. About the only advantage I see over micro inverters, is running a higher DC
voltage over the distance to the inverter/AC connection point. The complexity of the communication
system would worry me, will my HAM transmitter throw it off? And troubleshooting could really
be a problem if something started failing; instead of a few simple parts, you have a community
of computers.

As for your DC wiring currents, they will be the array power output divided by 350 VDC. If that
is conservative, hopefully it will handle any fault situation safely. They keep talking about not
needing fuses with a minimum of long strings. But those long strings will then have a large
current capability outputted from the optimizers, needing larger wire and for me, fuses. I'm
glad to keep it very simple on the ground. Bruce Roe

[sensij]

My permit actually shows #10 ground between the inverter and through the disconnect to the main service panel, though the installer put in #8.
For the hot and natural, he put in #6, though the permit says #4.

There is also a #6 bare ground wire running from the inverter to the existing grounding electrode (but it doesn't pass through any conduit or other meter/disconnect boxes).

My rating is 47.5A for a 12.4kW system going to a 60A breaker.

To what did he connect that bare 6 AWG ground wire inside the inverter? A wire from the inverter straight to the electrode would be required in a transformer based inverter, to ground reference the DC side. If he ground referenced the DC side on this, that should fault the inverter, so hopefully it is just connected to the same ground terminal as the 8 AWG. I don't know how that terminal would accept two wires though.

Also, the way you screwed in the ground in the AC disconnect is not really right. It should be a listed lug, like that shown in this picture. You don't even need any extra wire to do it right, it looks like the looped green wire is long enough to get out to a lug and back with changing anything else.

[sensij]

Not that you need anything else to worry about, but as I was re-reading the SolarEdge document I just posted, there was some text about the combiner box that reminded me to go back and look at your picture in this post.

Your combiner box appears to have two problems:
1) White wire was used for the DC- conductors. In a SolarEdge system, these conductors are floating, not ground referenced. Only ground referenced conductors may be white.

2) The fuses are only on the DC+ wires. Since it is not ground referenced, fuses are required on the DC- wires as well. Those fuses were shown on your permit drawing, so there is a good chance this will fail inspection if the inspector sees it.

Using this combiner is a safety problem and it should be replaced with the permit approved model or an equivalent even if the inspector misses it. The text from the SolarEdge document specifically calls this out:

Note: In rare cases where three or more strings of power optimizers are wired in parallel, fusing will be required to protect the wiring between the power optimizers and the inverter. Because the circuit conductors are not grounded, fuses would be required in both the negative and positive conductors. The recommend fuse size of 20 Amps is calculated using the 15 Amp continuous output current limit of the power optimizer multiplied by 1.25 in accordance with NEC Articles 210.19 and 240.4.

Your FMC routing might run afoul of the codes cited in this paragraph:

Conductor Routing Single conductor cables in exposed outdoor locations within the PV array are permitted by NEC Article 690.31 (B). Article 690 Section IV, requires all DC conductors to be contained in a metallic raceway once they penetrate a building. In addition, beginning with the 2011 code, these raceways must be run along structural members, must be labeled with “Photovoltaic Power Source” every 10 feet, and on every section that is separated by enclosures, walls, partitions, ceilings or floors. The DC raceway must be at least 10 inches below the roof sheathing where it is not immediately below the PV array. Note that a type MC metal clad cable also meets the raceway requirement.

With regard to the questions about ground conductor size, the relevant sections of code are cited. For a 60 A breaker in this case, a minimum of 10 AWG is required. Note that any place the EGC is is not protected (IE, not in conduit), 6 AWG is the minimum.

The equipment grounding conductors should be sized and installed in accordance with the requirements of Articles 690.45(A), 690.46 and 250.122.

I had included a bad link earlier in the thread. The thread discussing the size of the AC conductors is this one.

All code cited is NEC 2011. There are definitely differences in versions earlier and more recent, but this is what I understand to be in effect in the OP's location.

[sensij]

For those who still have unanswered questions about the SolarEdge system architecture, they have a white paper available on their website that explains in detail how their systems should be understood in the context of code. A copy of it is here: SolarEdge System Design and the NEC.pdf.

[Volusiano]

OK everyone, I deleted all my posts that argued about the return wiring and not taking into account the voltage drop on the return as well. You guys are right. Didn't need to read sensij's link though. Just needed to come to my senses.

Guess I had too many beers tonight and argued out of my *ss, LOL!

[sensij]

This thread is an epic fail. Volusiano, pal, you've gone off the deep end. Maybe reading this will help.

[HX_Guy]

I thought the only purpose of the ground wire in a DC system is in case of a lightning strike, otherwise the ground isn't actually used.

The inverter itself is ungrounded per the specs.

[Living Large]

Hm, I don't think so because all you care about is the effect of the voltage drop at the inverter's input, so wherever the current goes after that is irrelevant.

And the current doesn't really make the round trip back (to where?), it just goes out to the grid and ends up wherever it ends up, right? Like maybe to your own loads or your neighbor's circuit panel through their loads to their ground.

Hmmm. Well, let's look at this. From Array+ -----> Inverter+, you lose xV in the wire. From there, the current doesn't disappear or fall into the ground, but goes back to the panel over the same distance and voltage drop: Inverter- ------> Array-, and you drop another xV. So in fact you lose 2xV, not xV.

The suggestion that the current "doesn't really make the round trip back" and "wherever it goes after that is irrelevant" is quite interesting, since without a circuit there is no current through the array. Think of the cost savings in not running the return wire - if the current just flowed out the end.

And remember, the electrons are actually going in the opposite direction of the current. The Bizarro world!

[HX_Guy]

Was it easy to feed the conduits under the fence in your case?

I allowed my installer to run the conduit over the fence because there were already other prior conduits under the fence and there was no more space for more conduits without some major drilling. Plus this is a 1" conduit so it'd be even harder to fit. But it didn't look too bad in the end after I painted everything the same color of the wall (well everything except the labels on the boxes and conduits and the inverters in case of warranty return). In the end it blends in pretty well with the wall after painted.

Yeah, there was plenty of space...the only other conduit already there was the one going to the pool. My installer used 3/4" EMT though, I don't think 1" would have fit. (Permit specified 1"...I'm curious how the inspection is going to go in regards to that.)

[bcroe]

Did I say something wrong or funny? If I said something wrong, I'd like to know what. If I said something funny, it wasn't intended and I didn't mean to mock anybody. But glad I was able to entertain you and make your night, LOL!

I didn't know if that was a joke or not. But if you really think an electrical circuit doesn't
need to have a complete round CIRCUIT back to the beginning, you need to take a break
for basic electricity 101. Bruce

[Volusiano]

Was it easy to feed the conduits under the fence in your case?

I allowed my installer to run the conduit over the fence because there were already other prior conduits under the fence and there was no more space for more conduits without some major drilling. Plus this is a 1" conduit so it'd be even harder to fit. But it didn't look too bad in the end after I painted everything the same color of the wall (well everything except the labels on the boxes and labels on the conduits and the inverters in case of warranty return). In the end it blends in pretty well with the wall after painted anyway.

[HX_Guy]

OK, that is more good info. The inverter has a max 34.5 ADC. Just maybe, your
max is 3 times a panel Impp? What did it read out?

I'd worry that a string of 16 panels would have a Vmpp on the order of 576 VDC.
Do the Optimizers somehow avoid that issue for the 500 VDC max inverter? What
limits do Optimizers placed on number of panels?

You can actually have up to 25 panels per string, but no more than 5250W DC, which is why my string is 16 panels.
The way I understand it, if you have shading issues or different orientations, the more panels per string, the better.

Here is a long video on the SolarEdge Optimizers, but the explanation on how they work starts at 21:25.

http://youtu.be/QsGC3FA5ilk

[HX_Guy]

Worked on tidying up and fixing a couple things today...

First I took care of the ungrounded disconnect box, hopefully I did this properly. Used a 8AWG bare copper wire, scratched off a bit of the paint inside for a good bond, and used the same grounding lug. It looks good to me, but I have no experience with this.







Then I worked on running the CAT5 cable from about a 10' distance away from the inverter, to the inverter.
I got lucky here in the sense that the house already had a cable running to this location from the networking box inside the house, just had to crimp on a connector on this end and plug into the networking switch in the other end.









Cleaned up the rat's nest of conduit as best I could by moving the irrigation control box.





I'm definitely glad I insisted on the installer re-running the metal conduit down and under the fence instead of the crap he originally did, looks a lot cleaner.





Now just have to paint everything the color of the house and it's all done.

[bcroe]

Hm, I don't think so because all you care about is the effect of the voltage drop at the inverter's input, so wherever the current goes after that is irrelevant.

And the current doesn't really make the round trip back (to where?), it just goes out to the grid and ends up wherever it ends up, right? Like maybe to your own loads or your neighbor's circuit panel through their loads to their ground.

HA HA HA HA HA HA HA HA HA HA (stopped to roll on the floor) HA HA HA.
thanks for making my night. Bruce Roe

[sdold]

You have 50 feet so that's 0.3951*50=19.755 milliOhm. Your max input current is 34.5A DC. So the max dc voltage drop over your 50' 6 awg run would be 34.5*0.019755=0.68V dc.

I think it's 1.36V, since the current has to make a round-trip (100').