diy solar

diy solar

Feedback on my DIY bifacial panel unistrut ground mount design?

31.65 A Imp, probably 35A Isc.
35A x 1.56 = 54.6A design current.
Assuming 30C ambient and 90C wire, 8 awg for current-carrying conductors and for ground wire. If no more than 3 current-carrying conductors (2 for one string.) 8.7% loss at peak current, good enough for me.

If two strings, 4 current-carrying conductors in one conduit, derating will make 6 awg necessary.


Direct burial?
May not be code.
No do-over.
I'd lean toward conduit. Rather than PVC 18" deep I used steel 12" deep.

Whole rolls generally cheaper per foot than cut to length. Seems like 1000' for 2 conductors plus ground.
 
Not sure about that; when you oversize phase on small wires you generally need to oversize the ground to match as well. The ground would technically need to be #6AWG.
I'm pretty sure in NEC 690 that you don't have to upsize the grounding conductor for PV.
 
Upsize for what, upsized PV +/- wire gauge?
If not required for that, it is useful. Power loss in ground wire isn't a concern. Although, voltage would be.

I would not want my 25% voltage drop on a 600V or 1000V string to show up across ground. Also, don't just limit it to safe low voltage like 48V or less, because you could be standing in a puddle while touching frames. I'd aim for 12V drop in ground wire, tops. (consider a frame not grounded to earth, and human completes connection frame to earth, then current flows through dirt back to SCC.)

Ground needs to be sized for continuous current, can't be downsized as for fuse/breaker protected circuits.
I think it is Isc x 1.56
 
I'm pretty sure in NEC 690 that you don't have to upsize the grounding conductor for PV.
Don't have my code book handy, but the requirement was in 250.122. I know it has changed a few times over the years, but thought it was still a requirement.
 
Don't have my code book handy, but the requirement was in 250.122. I know it has changed a few times over the years, but thought it was still a requirement.

690.45 Size of Equipment Grounding Conductors


Equipment grounding conductors for PV system circuits shall be sized in accordance with 250.122. Where no overcurrent protective device is used in the circuit, an assumed overcurrent device rated in accordance with 690.9(B) shall be used when applying Table 250.122.
Increases in equipment grounding conductor size to address voltage drop considerations shall not be required.
 
"Listed for use in PV systems"
That seems to say DC listed breakers, and diodes, are not allowed as backfeed or overcurrent protection unless explicitly listed for PV.

You quote "assumed overcurrent device rated in accordance with 690.9(B) shall be used when applying Table 250.122"

Does that mean for a 60A circuit, assume 60A breaker (even though there isn't one) and use 10 awg ground wire?!
That can't be correct - do you know what it really means?
 
"Listed for use in PV systems"
That seems to say DC listed breakers, and diodes, are not allowed as backfeed or overcurrent protection unless explicitly listed for PV.

You quote "assumed overcurrent device rated in accordance with 690.9(B) shall be used when applying Table 250.122"

Does that mean for a 60A circuit, assume 60A breaker (even though there isn't one) and use 10 awg ground wire?!
That can't be correct - do you know what it really means?
My post was only meant to address that the EGC does not have to be the same size as the other conductors when those conductors are larger due to voltage drop. This is an exception for PV as the EGC commonly has to be upsized.

I take the first part to mean that if no OCPD in circuit then the 125% rule from 690.9(B) must be applied for the purposes of table 250.122. So if you had a 60A circuit and no OHCP, you would need a 75A OCHP, therefore 75A would be the number you would use in table 250.122, which would be #8 CU. If you had a 48A circuit, no OCPD, #10 CU would work. If a 60A OCPD is used in the circuit then #10 CU would work.

Edit: I just read this over and my head is spinning. LOL
 
If you had a 48A circuit, no OCPD, #10 CU would work. If a 60A OCPD is used in the circuit then #10 CU would work.

Edit: I just read this over and my head is spinning. LOL

If I had 48A circuit and no OCPD, I wouldn't like the idea of #10 ground wire carrying that much current continuously. Which was the point I was getting at.

I thought I read somewhere that both PV wires and ground wire had to be sized for Isc x 1.56, which would be conservative.

Oh, and since shorted wires in a PV system won't blow a fuse, you have to size ground that big all the time. After all, OPCD is required to be at least Isc x 1.56
 
If I had 48A circuit and no OCPD, I wouldn't like the idea of #10 ground wire carrying that much current continuously. Which was the point I was getting at.
I'm not disagreeing with you. I'm just stating what the code is allowing. Personally I like the idea of putting fuses at the source. I alsot think that if a string is fused at the source, is 300 VDC or less, and 30A or less, that we should throw out all the extraneous wiring requirement (like metal conduit) and use "normal" AC wiring methods.
I thought I read somewhere that both PV wires and ground wire had to be sized for Isc x 1.56, which would be conservative.
I've seen that number (1.56) somewhere before. Please let me know of you can find the reference.
 
The math was 1.25x (normal margin for thermal OCP) and 1.25x for cloud edge effects on PV panels.

Here's a circular reference for you:


Somewhere, it came from NEC.

 
I alsot think that if a string is fused at the source, is 300 VDC or less, and 30A or less, that we should throw out all the extraneous wiring requirement (like metal conduit) and use "normal" AC wiring methods.

Just 9kW of arc available. And DC, so it can draw a long flame without extinguishing. What could go wrong? :)

Fusing at the source would be good if the fuse can blow. It can't.
The only way to blow a fuse in a PV system is to have two, maybe three or more strings backfeed into one that is individually fused, and shorted.

Ground fault can find parallel shorts to ground and turn off SCC. Arc fault can do the same for series arcs. But an arc might continue anyway.
If you have RSD and if arc-fault triggers it, have a better chance. Not clear to me that happens.
My TriPower has a "fault" relay which I'm considering wiring to the external RSD keep-alive transmitter, so that might do what I want.
 
I ran a model to estimate 'meaningful' power loss due to wire gauge. If production is greater than loads, and battery SoC is full, then the losses aren't meaningful.

Worst case is winter:
graph1.png
graph2.png
graph3.png

Total loss at 8awg is 2,500Wh, but meaningful loss for 8awg is 1,125Wh.
Total loss at 6awg is 1,600Wh, but meaningful loss for 6awg is 710Wh.
Total loss at 4awg is 1,000Wh, but meaningful loss for 4awg is 450Wh

So the delta is 260Wh power loss for going up to 6awg. (Summer delta is 130Wh).
The delta is 675Wh for going up to 8awg.

One panel produces 2,300Wh in a winter day. 8awg meaningfully loses half of one panel, 6awg loses a third of a panel, and 4awg loses a fifth of a panel.

Of course, this is just a very rough model of demand load. Some use-days could incur greater meaningful power loss for various reasons.

Okay, I'm off to grok the last several comments and get up to speed on the nec language. What a treasure trove for learning the ins and outs of this stuff.
 
Last edited:
Pictures of the completed rack. Thank you everyone for the comments with this!

2023-04-08 10.30.44.jpg2023-04-08 10.31.21.jpg2023-04-08 10.33.32.jpg2023-04-08 10.34.19.jpg
By the way: Yes, the lower edge is close to the ground and snow will accumulate there in winter. In a normal winter, this will occur four times and the accumulated snow will melt by noon. In the worst winter we've had in recent memory (aka this one), it will occur six times and the snow will be melted by the next day, assuming no one is there to push it off. Crushing loads of snowmelt>ice doesn't happen here. I could have designed it higher off the ground, but I judged it not worth the extra effort and expense.

By the way, you might notice a scratch in one of the middle panels. Bobcat prints! The neighborhood bobcat came over and climbed up the thing to check it out shortly after I got all the panels up. Apparently glass is slippery so it left dusty claw marks all up the center. They brushed off easily enough.
 
I have what I think is a dumb question regarding grounding and lightning protection, the SCC GFP system, and my combiner boxes. This is a case of thinking I knew what I was doing, getting the stuff, and going 'wait, how does this fit...'

I have reviewed filterguy's document on grounding and his document on solar panel ground fault protection systems. I believe page 4 of his GFP document most closely matches my system.

This is my system diagram:
20230411_panelwiringdiagram.jpg

This is my combiner box:
combiner.png

My SCCs are Midnight Solar Classic 150s, which I'll run in Follow-Me mode. I intend to use the internal GFP setup. Per the SCC manual, p18:

"If the internal grounding jumper is installed in a Classic [what you do when you are using the unit's GFP system], the battery negative and DC source negative must not be connected to the system grounding conductor anywhere in the system. Grounding of these circuits will defeat the GFP function."

My read of that statement is:
  • At no point outside of my master SCC do I connect PV negative or battery negative to ground.
  • At my master SCC, the PV and battery negative will connect to the internal grounding jumper. I will run ground from that jumper to my main ground at the AC panel (which is a ground wire bound to metal water pipe, I believe).
  • The bare copper ground wire I'm bonding to my PV panel frames and rack will run all the way to that single ground point (or, because of the layout of my equipment room, I'll connect it proximately to the master SCCs internal grounding jumper).
Okay, that seems simple enough. My diagram reflects this. (Although, is it correct that I'm taking the DC ground to the inverter DC ground lug, or should I be wiring the DC ground directly to the earth ground near the AC panel where it's grounded to a pipe? I'm seeing conflicting things on different diagrams, and I don't understand the relationship between the inverter DC ground lug and the AC ground. Are they connected electrically? I think so, but I'd rather not assume and the manual doesn't say.)

Okay we're getting to my question. When I look in that combiner box, there's a ground wire connection coming out of the bottom of the arrester, the blue device on the right. You can see some examples of this in this thread on these style combiner boxes, and here is a screenshot of a pic wattmatter posted there:
wattmattersimage.png

What do I do with that? Isn't that ground wire connected to the PV negative, and thus a no-no per the SCC's GFP system?
 
What do I do with that? Isn't that ground wire connected to the PV negative, and thus a no-no per the SCC's GFP system?

The surge arrestor should be DC open between all wires, except when a surge voltage appears. It would clamp PV+, PV-, Ground all within some voltage limits. That should be fine. Midnight sells their own version like that, with MOV between the nodes.
 
Does Midnight allow battery negative to be grounded? I thought grounding was required by code for higher battery voltages (e.g. 48V charged above 50V), and should be able to clear battery fuse. But I think (not certain) Midnight PV(-) goes through to battery(-).

What Midnight Classic apparently does have is magnetic coupling to PV(+), which avoids the issue of shorted FET causing battery overcharge. But if PV(-) does go through to battery, I'd think that would be incompatible with a second ground bond.

I've bought a Midnight Classic but not put it to use yet.
 
No - per the Midnight classic manual, "the battery negative and DC source negative must not be connected to the system grounding conductor anywhere in the system. Grounding of these circuits will defeat the GFP function."
 
I see. So then electrically things should be wired like this.
View attachment 144310
I hate to say this... but I think you need a separate wire for the array ground. The combiner SPD and SCC grounds can be tapped if you really want to, but preferrably the tap is actually the chassis ground connection point.

All of your enclosures (shunt, inverter, DC disconnect, AC disconnect will need ground connections if they are metal. (This is one of the reasons wireway can be really nice.)
 
Back
Top