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

[pclausen]

I have finally reached the breaking point in replacing failed micro inverters and want to convert to all string inverters. I have a couple of SMA 7.7s deployed already, and they are great and rock solid.

Here's a summary of my installation:



And here's how the arrays are currently setup:



So mostly Micro Inverters, but a couple of string based arrays as well.

Here's a visual of the layout:



And here's my proposed solution for going all string based:



And what that would look like visually:



Except for the shop building, I can easily remove all the inverters and rewire for DC strings. The shop roof is going to be a pain, and is of course also the array that is a huge pain to replace dead micro inverters on.

From what I understand, there is no issue with me reusing the THHN 10 gauge wire in the conduits for the DC runs. That sure will save me a lot of time and money. Per the chart above, all strings will be well below the 600V rating of THHN.

I was going to redo that 3x15 array in the foreground with 2x SMA 7.0s, but it will be a lot less work to just leave the 7.7 already there and then add a 3.8 to handle the last 15 panels.

I'm getting the failed micro inverters replaced under warranty. I got about 25 failed ones right now, and just don't want to go through this painful process over and over again. I figured I should be able to cover most of the cost of getting the string inverters from selling the new replacement and good used micro inverters.

Anything you guys would do different?

[bcroe]

I am wondering, are you going to run DC over to the building and
put all the string inverters there? That would reduce wire losses a
point or 2, and better protect the inverters.

From my experience, there is an advantage of making all the
inverters the same size. Then a spare kept on hand, could
immediately be used to replace any failed string inverter.

If snow is an issue, there is a lot you could do to reduce the
problem. Mounting panels landscape is one, leaving a half foot
gap between a row and the next higher is another. Landscape
mounting here meant reversing every other panel, half had easy
reach for the string connections, and the other half needed a
short MC4 extension cord. good luck, Bruce Roe

[Ampster]

You should be able to sell the micros to recover some of your costs. I have sold old Enphase M215s on EBay, The SMA inverter are reportedly rock solid and last a long time.

[pclausen]

bcroe Yes, I definitely plan to run DC to the back of the shop building where all the string inverters will be located. Here's an overall view from my house looking down at the shop:



And a closeup of the 400A meter base and solar subpanel as they are currently: (there is another SMA 7.7 inside the shop right behind the solar subpanel)



My plan is to move the 200A solar subpanel all the way over right next to the meter base, and then fill up the wall with SMAs as follows:



I was tinkering with the design this morning and was wondering if there would be an issue with only using 2 of the 3 MPP inputs on some of the inverters and still get full output? If so, I was thinking something like this would make more sense and maximize the voltage on each string for the least amount of loss:



Since the cost difference between the 6.0, 7.0 and 7.7 is minimal, I figured I might as well add more 7.7s since I already have a couple of them. This would also give me vacant MPP inputs for adding more strings later if I wanted to. Looking at the efficiency curve of these inverters, there does not seem to be a loss at all from running at less than 80% capacity.

Here's a closer look at the 400A meter base:



It is actually a sub-panel of sorts and does not belong to the POCO. The main 400A sub-panel is up at the house and this one on the shop runs back up to the house via that 500 MCM Al cable you see. So losses are minimal even when pushing 200A.

Ampster Yes, I plan to sell the inverters. The sell new on Amazon @125 and Streakwave has 30 in stock they are selling for $96 each. I figure I'll offer then for $75/obo to start to see how it goes. About 30 of them will be brand new warranty replacements from Ubiquiti and the rest will be used but functional. I will also have a bunch of Y-cables and such to go with them.

This is all the crap I got initially back in October of 2016:



I'm of course keeping the panels and mounting hardware and the non-solar related stuff, but everything else will be put up for sale, along with 48 more of those 250W inverters and Y cables I got later.

[solardreamer]

Just curious, why did you go with sunMax microinverters in the first place?

[pclausen]

At the time I was looking (summer/fall 2016), these seems to be very competitively priced ($203 for a 260 watt panel and $92 per inverter). I also really liked their roof mounting system and the whole concept of individual panel monitoring. But probably the most determining factor what that I was already a heavy user of Ubiquiti network gear and cameras and like the concept of a "unified" experience of all my gear and had visions of sunMax integrating into that stack for a true single pane view of all my stuff.

Had it not been for the extremally high failure rate of the microinverters, which are otherwise very stout with a cast aluminum enclosure and low power Bluetooth communication with Gateways connected via Ethernet directly to my local network.

Here's how many failed inverters I have in my system right now: (red - I see the inverter but power output is 1 watt or so. orange - no comms)



I get about 10-20 failed inverters every 12 months. So I decided to end it this year and go 100% string inverters and the 2 SMA 7.7s I have are rock solid.

I did have to register and get a grid guard code so that I could bump the allowable AC output to be 272 volts, in order to overcome the fact that my local utility really likes to delivery your power "hot". In my case that is at least 250 volts as observed in the morning and evening when I know the solar plant isn't pulling up the numbers.

Anyway, after upping the max AC voltage, they have been solid as a rock going on 2 years now. The first one was a 40 model and the second one is a 41. Does anyone know if there is any difference between them? My guess is that maybe the only thing is some additional mandated safety features on the 41?

[solarix]

What is the age of the microinverters?

[oregon_phil]

Mounting system/hole pattern is slightly different between -40 and -41. -41 has integrated sunspec transmitter for communication with rapid shutdown devices via DC lines. Firmware is not interchangeable.

[pclausen]

The original 80 were purchased 10/2016 and deployed 12/2016 so about 5.5 years. I raised enough stink with them in 2017 that they sent me 56 replacements around 6/2017. I deployed 48 of them on the hill side arrays in fall 2017. Since then I loose about 10-20 per year, both originals and replacements. Its a known manufacturing process flaw that causes some capacitors to be installed too tightly during the wave soldering process which cause premature failure.

The other issue I have with these inverters is that during the middle of the day on sunny days, I get these over-frequency errors. I worked extensively with their developers in the spring of 2017 on this issue and they went so far as to ship me some very expensive data logging equipment that I connected to my mains and we setup VPN to allow them access to said equipment as well as my gateways to tail logs on the inverters. We went through several different firmware versions and while the issue got better, it was never completely resolved.

They stopped development and abandoned the whole SunMax line in 2019 and halting software development sometime in 2018. Unfortunately Ubiquiti has a bad habit of dropping product lines at the drop of a hat, and SunMax was no exception.

I'll definitely give the guys I worked with an A for effort in trying to sort out my issues and sending me 56 free inverters and continuing to honor warranty replacement for all the 130 or so inverters I have deployed. But I just don't want to put up with the labor and return shipping cost until I reach the 10 year warranty expiration. Hence my decision to cut my losses and move to string inverters. And I think I should be able to recoup most of that cost from selling all those 130 inverters and y-cables and such.

[slinthicum]

Quite a story that to some degree makes me thankful my system does not have microinverters. Yes, the old-style string inverters fail (mine did before the completion of its first operational year), but it is a single device and not multiple devices providing the opportunity to have to deal as you've had with the "labor and return shipping costs until you reach the `10 year warranty expiration."

[Mike90250]

On an occupied building ( HOUSE ) you may have to use a Rapid Shut down System to meet local code, using optimizers at each panel as a shutdown device, to make the array "fire department" safe.

[solardreamer]

Sounds like Ubiquiti was in over their heads and sunMax was/is hardly representative of microinverters in the market. There are much more reliable microinverters available.

[peakbagger]

My comments for what they are worth on the proposed Inverter Installs. Good that you are on the north side of the building for cooling. I cannot see how deep the roof overhang is but by the looks of the dirt on the siding, it may not be deep enough to keep the inverters out of the rain. If I was going through the hassle to install inverters I would consider installing an overhang below the existing overhang that sticks out past the front of the inverters with some extra. Anytime I can keep electronics out of drip line I do so. The other thing is considering installing one may be two horizontal runs of electrical gutter (wireway) with a internal divider to keep the DC separate from the AC. Sure conduit will work but that is lot of "spaghetti", gutter will look a lot neater. If you just put a gutter under the bottom set you can run verticals to the top row or just run an upper and lower gutter.

Makes me glad I only have 4 microinverters and 3 string inverters in my collection.

[solarix]

I rest my case.

[pclausen]

I definitely agree that the SunMax inverters are not representative of micro-inverters in general. Like I said, there was a flaw in the manufacturing process that cause a lot of failures. So I have nothing against microinverters in general, but in my use case, with no shade to speak off for 98% of my panels, they don't offer an advantage over string inverters.

I do have a couple of strings that will be slightly impacted by shade in the am, and one of them only in the winter months. I know SMA has this thing they call ShadeFix, that supposedly compensates for a few panels seeing shade. So what does the MPPT actually do when a couple of panels in a string see shade?

peakbagger Great suggestions! I think I'll incorporate both. Really like the idea of a wireway. Much cleaner than the "spaghetti of conduit I already have.

So one question I have that might have been overlooked, is if I can use only 2 of the 3 MPPT inputs and still reach the full potential of the SMA SB inverters. Looking at the specs for the 3 larges models, I see this:



So it looks to me like I can easily reach the 7,700 watt rating of the 7.7 model by using inputs A and B only with like 15 panels in each string and prove about 4,000 watts per input (470 volts @ 8.63A) which is within the specs.

Ignoring shade concerns, are there any downsides to only using 2 inputs on a 3 MPPT capable inverter? Seems to me I can reach higher voltages that way by having more panels in series across 2 inputs vs. fever panels per string in series when using all 3 inputs, and thus reduce loss, not to mention simplified wiring as I need to run fewer wires back to the inverter location, which for some arrays, is almost 200' away.