Converting 50kW system from micro inverters to 8x SMA inverters

[bcroe]

I have a similar problem walking on my metal house roof, new in
2021. I would not try to go up there unless the sun has completely
dried it out. Then I wear these roofing boots, cougar paws made
for the situatuion. There are no panels on the roof, but I have wires
for lightning mitigation and communication antennas to rework.

I am in the process of making a series of clamps that attach to the
ridges on the metal panels, 6061 aluminum and 18-8 stainless bolts,
just like my solar array. A piece of conducting cloth tape goes down
to avoid direct metal to metal, clamp to roof contact. I plan a safety
rail up from a designated accesspoint and along the ridge, to attach
my safety cable.

Ready to use clamps can be bought, they are quite expensive, they
are not custom to my needs, and the retainer screw probably scratches
up the metal covering. There will also be a bit of rubber path coating
attached, going up to the ridge.

These clamps will not be subject to any continuous stress for things
like holding down an antenna in a storm. Pictures another day.
Bruce Roe

[pclausen]

I got 2 panels dismounted and on the pallet last night before sundown. Started up again this morning at 6am and the metal roof was quite slippery from the dew on it, but I managed.

After loading the first row of 5 panels onto the pallet, they started getting real heavy, especially since I had to pull them over the ridge to load them on the pallet, and with 5 panels on there, I had to start raising them higher and higher.

Fortunately I figured out that once the 1st row was gone, I could just shift the next 2 rows up one and sit loosely on the mounts after removing the inverters and associated cabling:



That worked well for the first section of 20 panels. The middle section will be trickier because I can't enter each row I'm working on from the side. So I might have to carefully rest the first panel in each center section on the section I completed.

I'm measuring DC voltage on each string in each section as I go along. I DON'T want to take it all apart again later if I'm not seeing DC at the inverter!

But right now its unbearable hot and humid being mid day. The green roof panels burns you, as do the blank frames, so I'm taking a break inside with the AC on.

[pclausen]

Thanks for that link nerdralph Placed and order for a couple of those along with some 3 hole ones for some of my other arrays where the PV wires just exit 1/2" conduit now with electrical tape to "seal". Didn't even know they made these. They should work perfect for what I'm doing.

[nerdralph]

Agree that drilling holes in a junction box and using grommets is certainly not ideal. I searched quite a bit back in 2019 when I did that particular ground array and never did find anything I could use to transition PV wires from strings into a junction box and or directly to conduit.

These should work for you. The M8437GBT should fit 5 RPVU #10 wires.

https://www.heyco.com/Solar_Power_Components/product.cfm?product=Solar-Liquid-Tight-Cordgrips

[pclausen]

I would take those 2 cracked panels out while the weather is nice, and swap them with a
convenient pair in the string that already has 2 cracked. So when trouble comes, it will only affect one string and be easily serviceable.

Good idea. I'll will definitely put that on my list.

Your 20 July readout shows line voltage of 129.18 plus 130.19, near 260VAC. Some inverter voltage
monitors are programmed for shutdown at 264, you are very close and subject to any line variations.

Yeah, check out this graph of my line voltage at the shop from today:



The main culprit, I think, is that I have 200 ft of MCM350 Al from the 50 KVA transformer on the street to my house. And then another 200 ft of MCM500 Al from the house to the shop building.

Those micro inverters do cut out quite a bit around mid day, even through I worked with Ubiquiti to update the firmware to be less sensitive, but they never could fully fix it. And that was with line voltage less than 264V. Fortunately I have a grid guard code for my SMAs, so I can then them to not cut out above 264 (currently have then set to 270 which seems sufficient).

What is your snow experience?

Snow is generally not too bad here in Central VA. When we do get it, it typically melts in a day or two. I do try to get out there and keep it off the ground mounted panels. The shop roof I leave alone. Although it does sometimes take several days to melt there since the pitch is very gentle.

[pclausen]

I measured my PV cables, and the OD is right around 1/4". So I went ahead and ordered those cable glands I linked to earlier. They are for cables 4-8mm (0.16-0.31inch) so they should be just about right for my wires. I'll be sure to drill the holes for the glands on the downward facing (South) side. I suppose I could look at NEMA 4X enclosures like this one:

https://www.amazon.com/VEVOR-Electri...NrPXRydWU&th=1

But that might be overkill for what I'm doing. Once the install is done, it will never be opened again, so a fancy lockable door is not really needed.

I did screw on some support 2x4's to the pallet:



If I was really paranoid, I could also strap it to a couple of a solar panel roof anchors on the other side of the ridge. That would for sure keep it in place I would think, unless some crazy storm comes through while I have panels on the pallet. I plan to start with the first section early Wednesday morning (assuming my PV cable gets here Tuesday as the UPS tracker is currently showing) and maybe get all 42 panels from all 3 sections done if I hustle that day. Might spill over into Thursday, which looks ok for the first part of the day. The rest of the week, not so much.



Looking back from when I initially installed the shop roof array, it looks like I had all the prep done on 10/31/2016:



And the first few panel mounted around 5pm:



And was about 40% done on 11/4/2016:



And down to the last few panels on 11/6/2016:


So getting it all done in one day by myself might be a stretch. But when I did it originally, I had to load one panel at a time on the tractor after installing the micro inverter and landscape cable, drive it into position, climb the latter, pull the panel off the tractor, and then mount it, then climb back down the latter and repeat the process for the next panel.

This go round, I'll be on the roof the whole time, so I anticipate it going a LOT faster. But we'll see. Some of the fasteners might not want to come off easily after 6 years.

[J.P.M.]

All good points. The pallet looks worse in the pic than it actually is. Other than those 2 board that were not nailed on straight, the wood is sound. It is a couple of months old and it what supported 15 panels I picked up recently to add another row to my pole barn. I will screw a couple of 2x4 to the side of the 6x6 blocks to tie them to the pallet to keep them in place.

Agree that drilling holes in a junction box and using grommets is certainly not ideal. I searched quite a bit back in 2019 when I did that particular ground array and never did find anything I could use to transition PV wires from strings into a junction box and or directly to conduit.

I had the same issue in 2020 when I did the pole barn extension. This is what I ended up doing for that install:



Looks like something like this would be a better choice:

https://www.amazon.com/dp/B08TCF3S13...HJ1ZQ&th=1

On the pallet: it looks to me like about the only thing keeping it from sliding down the roof is the sort of notch in the beams at the peak. I appreciate that it's a temporary thing but so is wind.

I've put penetration's on the bottom side of things that needed penetrations, but never a junction box. NEMA 4 or 4X which is what I usually specified and was spec'd when I was engineering process equipment and their control systems.

[bcroe]

I once had a broken ceramic insulator on a huge diode, but it still tested
good. Months later, reverse leakage had become unacceptible. I
concluded semi conductors must be sealed to last. I would take those
2 cracked panels out while the weather is nice, and swap them with a
convenient pair in the string that already has 2 cracked. So when trouble
comes, it will only affect one string and be easily serviceable.

I must note just how much more wiring is need by a micro inverter array,
than my string array. Higher losses too. Your 20 July readout shows
line voltage of 129.18 plus 130.19, near 260VAC. Some inverter voltage
monitors are programmed for shutdown at 264, you are very close and
subject to any line variations. That happened here, short term fix was
re program inverters, eventually got the utility to lower voltage. 5 years
later I added some much bigger wire to lower voltage and increase efficiency.

What is your snow experience? Bruce Roe

Sn4iE2.JPG

[pclausen]

Good suggestion. I'll definitely do that with the new J box on the roof.

Played around with my dashboard a bit more:



Sill need to work on my weekly and monthly graphs.

[Ampster]

As far as the grommets are concerned, let gravity be your friend, Put the holes in the bottom of the J box and use drip loops. The weather head you posted in another picture is a good example,

[pclausen]

All good points. The pallet looks worse in the pic than it actually is. Other than those 2 board that were not nailed on straight, the wood is sound. It is a couple of months old and it what supported 15 panels I picked up recently to add another row to my pole barn. I will screw a couple of 2x4 to the side of the 6x6 blocks to tie them to the pallet to keep them in place.

Agree that drilling holes in a junction box and using grommets is certainly not ideal. I searched quite a bit back in 2019 when I did that particular ground array and never did find anything I could use to transition PV wires from strings into a junction box and or directly to conduit.

I had the same issue in 2020 when I did the pole barn extension. This is what I ended up doing for that install:



Looks like something like this would be a better choice:

https://www.amazon.com/dp/B08TCF3S13/ref=sspa_dk_detail_4?pd_rd_i=B08TCDW489&pd_rd_w=gD2xx&content-id=amzn1.sym.999c0877-3704-4f0f-9726-eebf80846a35&pf_rd_p=999c0877-3704-4f0f-9726-eebf80846a35&pf_rd_r=B07F3RQPH780PTS4333V&pd_rd_wg=QM0Ui&pd_rd_r=45febf78-2a67-4134-84ed-01b194bc9160&s=hi&sp_csd=d2lkZ2V0TmFtZT1zcF9kZXRhaWw&spLa=ZW5jcnlwdGVkUXVhbGlmaWVyPUExNldRTTJUQlhXRUJCJmVuY3J5cHRlZElkPUEwODMzNTEzMzRLSzQ2U0ZMR0k3TSZlbmNyeXB0ZWRBZElkPUEwMjQwODYxRjNCTVFRQjg3NUMxJndpZGdldE5hbWU9c3BfZGV0YWlsJmFjdGlvbj1jbGlja1JlZGlyZWN0JmRvTm90TG9nQ2xpY2s9dHJ1ZQ&th=1

[J.P.M.]

I ran out of wire when making the 52 3.5ft extensions I need when converting the shop roof from micro inverters to strings. Another roll should finally get here by next week.

When I installed the 56 panels on the shop roof, I divided them into 3 sections (4x5, 4x4 and 4x5) so that I wouldn't have any issues with expansion and contraction with temperature changes.

The mounting system Ubiquiti used was pretty unique:



And here's one of the separations:



My plan is to work on one section at a time, and remove the top 3 rows of panels from each section, and stack them on this pallet:



So I'll have 12 or 15 panels on the pallet, depending on what section I'm working on. I'll put each panel on the pallet face down so that I can remove the micro inverters as I go. With the top 3 rows of panels gone, I'll be able to get to the micro inverters on the bottom row of panels and remove them without having to remove or lift those panels at all.

The 4 AC stings currently come down through the roof like this:



And the individual strings are joined in this box:



I'll probably scrap that junction box and get another one just like it and drill 8 small holes and then use robber grommets for each PV wire, much like I did here for one of my ground arrays:



Fun times ahead!

On the 3d photo:
1.) That pallet looks a bit worn and battered to be able to take maybe a 500+ lbm. load.
2.) What is/will keep the wood blocks at the left side in place when the pallet is loaded ?

On then 6th photo:
1.) I don't know the answer to this question and maybe you've already addressed the situation, but when, if at all does code allow holes to be drilled in a junction box ? Sec. 314.15 seems to address it, but those are not drainage openings. In any case, it still looks like the AHJ must sign off on it field modifications. Just sayin'.
2.) As shown, what will happen to liquid that drips onto/around the grommeted holes, especially as the grommet material dries out/contracts/cracks ? Just my opinion, but that looks like a lot of trust to place in rubber/neoprene/whatever grommet.
Condensate will drip off the underside of panels to whatever is below. I'd suggest that makes the underside of panels a damp/wet location.

[pclausen]

I ran out of wire when making the 52 3.5ft extensions I need when converting the shop roof from micro inverters to strings. Another roll should finally get here by next week.

When I installed the 56 panels on the shop roof, I divided them into 3 sections (4x5, 4x4 and 4x5) so that I wouldn't have any issues with expansion and contraction with temperature changes.

The mounting system Ubiquiti used was pretty unique:



And here's one of the separations:



My plan is to work on one section at a time, and remove the top 3 rows of panels from each section, and stack them on this pallet:



So I'll have 12 or 15 panels on the pallet, depending on what section I'm working on. I'll put each panel on the pallet face down so that I can remove the micro inverters as I go. With the top 3 rows of panels gone, I'll be able to get to the micro inverters on the bottom row of panels and remove them without having to remove or lift those panels at all.

The 4 AC stings currently come down through the roof like this:



And the individual strings are joined in this box:



I'll probably scrap that junction box and get another one just like it and drill 8 small holes and then use robber grommets for each PV wire, much like I did here for one of my ground arrays:



Fun times ahead!

[pclausen]

Had a chance to play a little with InfluxDB and Grafana today in Home Assistant. Got a rough dashboard going that looks as follows:



The large graph is probably the most interesting one, especially the red line (Hillside #1). That's the array that gets partial shade from the Hillside #2 array in the am.

Needs a lot more work, but its a start. I can get a lot more out of this already than what the Sunny Portal offers. I completely gave up on it when I discovered it doesn't support more than 4 inverters per site.

Ultimately, I want to create a dashboard similar to this one:

[pclausen]

Good point. I'll just leave them in place then since they are already run. I got 2 more inverters online this morning. These are the arrays on the hillside where some panels have cracked glass. For now, I just wired each array as 2x12 strings.

This is the configuration:



So that array has one cracked panel in each string.

The other array is like this:



So string A has 2 cracked panels and string B has none.

Looking at the output of each of those strings:



It looks like I am taking a small hit, maybe. But given other variables, like dirt on the panels, there may be no performance hit at all. One string with a cracked panel is only 8 watts lower than the one with no cracked panels. So until I start seeing the strings with cracked panels perform worse that those without, I'm tempted to just leave well enough alone. All 4 panels that cracked happened at the same time about 2 years ago when I was using a stump grinder to do some trenching in front of the shop building. (I know, I know)