Solar Design Thread: 12 Panel Rooftop Array, Boise

This thread is a bit of a “Showing my work out loud” thread in the context of a potential homeowner installed rooftop solar system I was talking with a friend about.

He has a good south facing roof exposure in the Boise area, and based on some ballparking estimates of roof area, the design concept is for 12 panels (2 rows of 6) on an upper story roof. This may adjust slightly based on the actual setback requirements, but shouldn’t meaningfully alter any of the system details.

Boise is, lucky for them, under their own AHJ as far as solar - far as I can tell. Solar Energy | City of Boise contains relevant documents.

This is the submission checklist, which is an awful lot easier than what I had to deal with…

Anyway, various design criteria:

  • As a roof mount system, under NEC 2017, rapid shutdown is required. This means per panel electronics, and there’s no way around it. This means either a microinverter based system, or per panel optimizers. The Enphase IQ8 microinverters might do it except they don’t exist. They’re shipping those on Musk-time, apparently.
  • Backup power is a desired option, if affordable. Generators are cheap, batteries are expensive, and a Secure Power Supply type outlet is also cheap, if one goes that route.
  • Minimum roof penetrations. As a second story install on the rear of the house, ground appearance doesn’t matter (beyond it not looking horrible), so there’s no particular need to match panels and rails. The roof is 15-20 years old, so may be replaced before the project starts, but that’s separate.

So, starting out with the design spitballing, there are two ways to do this - microinverters (with no backup power option), or per panel electronics, a string inverter, and a backup outlet. I’ll spitball prices for both, since it doesn’t really impact anything but the inverter/optimizer cost.

I’m going to spitball this with 320W panels, so 3840W (for $/W calculations).


Panel width varies slightly, but in general, one can assume 1000-1020mm (39.37" to 40.15"). Throw in some UFO mounting widths, and one can assume 40-41" “width per mounted panel.”

For a row of 6 panels, that means 240-246" - or a hair over 20’.

Iron Ridge rail span charts can be found here: XR Rail Family - IronRidge - but for 20psf snow loads, with reasonable winds, the Iron Ridge XR1000 rails (the big boys) can span 8’. And, the Iron Ridge rails support a 40% of allowable span cantilever - so, 3’ 2" or so. On either side.

With an 8’ span and a 20’ panel array length, this can be done with 6 mounts per set of panels - two 8’ spans in the middle, with about 2’ hanging off each side. (2’ - mount - 8’ - mount - 8’ - mount - 2’) However, splice kits can’t go in the center 1/3rd of the spans. So, they have to be offset a bit. A 11’ and 14’ rail gives plenty of wiggle room to mount things without the splice in the dead center.

Iron Ridge flashed feet mount under shingles, the UFOs mount panels to the rails, and some assorted end caps/T bolts/etc are required.

Materials cost, rough estimates:

  • Iron Ridge XR1000 11’ rail x2: $60/ea, $120 total
  • Iron Ridge XR1000 14’ rail x2: $75/ea, $150 total
  • Iron Ridge Flash Foot, x12: $12/ea, $144 total
  • Iron Ridge T-bolt, x12: $2/ea, $24 total
  • Iron Ridge UFOs, x28: $3/ea, $84 total
  • Assorted clamps and end caps: $50
    Total cost for mounting: $572, or $0.15/W


For no reasons beyond “They’re available from a local supplier in large quantities at $220,” I’m specing some Mission Solar 320W panels (60 cell) -

Materials costs, rough estimates:
Mission Solar Energy MSE320SR8T, x12: $220/ea, $2640 total, or $0.69/W.

This can be lowered substantially, if one cares to find cheaper new panels, which can be done. I just can’t get my previous local contacts to respond (at about $0.50/W), and for 12 panels, the freight costs of a pallet don’t really justify the savings unless you get a really good deal. But I can get these all day long at this price.


There’s a bit of handwaving here in pricing, as I’m not calculating wiring costs in great detail yet. There’s a couple hundred dollars of wire either way, and without accurate lengths on the install (which I don’t have), I’ll just ignore those for now.

The first option here is to go with microinverters - which, sadly, means Enphase. I don’t like them, but… they’re about the option.

IQ7+ microinverters (currently backordered) can be found for around $145/ea, but… aren’t actually in stock. Oh well. Maybe they’re tooling up for IQ8, or maybe they’re just suffering supply chain disruptions.

Tigo TS4-O optimizers are around $45/ea.

For a string inverter with backup power, there are two SMA Sunny Boy US options that would work - the 3.0-US, and the 3.8-US. Both will tolerate the specified array, though the 3.0-US has a peak AC output of 3000VA, which means it would likely be clipping a bit more often. The 3.8-US, on split phase, can output 3840VA, which seems about ideal for this system… :wink: The 3.8-US can output a maximum AC current of 16.0A, which actually means the entire system could be wired up with 12 gauge wire - a nice savings over having to go with fatter wires. There are two MPPT inputs, though there’s actually no reason to use both with optimizers. Open circuit voltage of the panels is 489V, which leaves a good buffer up to the 600V max voltage of the inverter, and running voltage is around 410V - squarely in the MPPT operating range of 195-480V of the 3.8-US. You could run two strings of 6 panels, but with optimizers, there’s no good reason to do this, and it would simply cost a bit more in wiring for no practical gain.

For backup outlet use with the optimizers, you also need some way to get around 18V into the inverter for backup use: TECHNICAL NOTE: Sunny Boy US-41: Secure Power Supply Operation with TS4-R-O -

Materials Cost (Microinverters):

  • Enphase IQ7+ x12: $145/ea, $1740
    Total cost for microinverters: $1740, or $0.45/W

Materials Cost (String Inverter):

  • Tigo TS4-O optimizers x12: $45/ea, $540
  • SMA Sunny Boy 3.8-US: $1180
    Total cost for string inverter with optimizers: $1720, or $0.45/W

Notice that the total cost here is the same, either way. Only one gets you a backup outlet.

HOWEVER: I am a bit uncertain as to if the monitoring gateway for the panels is required. I don’t believe it is for main functionality, but it would add some cost to the optimizer based system, if you wanted detailed per panel information.

Total System Cost

I’m adding in $1000 to the system costs here to cover the random other stuff - permits, wiring, conduit, mounting, etc. Plus or minus a bit, it should be about right. There’s no real difference between the two system types in terms of wiring cost, just what runs where. There will be some ground cable and such as well, but, again, no real difference. This is actually a high estimate, especially for the short runs involved in this system, but there’s a lot of “round out” material I’m not calculating out here.

So, for 3840W of solar, installed by a homeowner, the total cost comes to $5932, or $1.54/W. Take the federal 26% off, and you’re at $4389, or $1.15/W, installed.

Commercial installers will be in the $2.50-$3.50/W range, if you find one that’s not out to screw you over, and north of $4/W otherwise.

Wiring Size Calculations

Nothing in here is going to require more than 12 gauge. Nothing in here should be done on 14 gauge. Except ground, that probably requires 6AWG bare copper, though I’m less certain about how to ground roof mount systems than ground mount…

One could do them, but I would be quite surprised if anything came out to other than 12AWG/90C wire. As a bonus, this can be run in 1/2" (or, if you hate tight pulls) 3/4" conduit.

Questions? :slight_smile:

1 Like

[quote=“Syonyk, post:1, topic:426”]The first option here is to go with microinverters - which, sadly, means Enphase. I don’t like them, but… they’re about the [only] option.

Bolded the missing word for you. At least I think you missed a word that you meant to have in there.

EDIT: And question: how much would the cost be to put the rails on? Or are you assuming your friend + you? Assuming someone else, what would you estimate that cost to be, on average, in terms of time and general hourly cost?

I don’t know what labor rates look like, though there’s a decent chance he’d be redoing the roof before putting solar up - or, at least, that section of it. In that case, putting the mounts up would be something the roofers could do and would reduce some of the labor involved later (in addition to getting a better-flashed set of mounts since you can put the shingles around them instead of having to lift them).

So I work at (but I’m a software engineer, and therefore a little divorced from the current technology & incentives side of things). Under $2/w though seems like a really good deal.

I guess there’s a big plus if you have the know-how to source materials from distributors, understand state & local incentives, understand required permitting, and install the system yourself. For a higher $/W is a good place to at-least get quotes from local installers to see how that might compare. Again if you have the technical know-how and can do it yourself you’ll obviously save a lot of money. I think one pro of going with a reputable installer can be support if something breaks (although maybe manufacturer warranties are enough given you know how to install yourself, you can probably debug and fix yourself as well)

Anyways that’s a long way of saying, this looks like a really affordable quote that will have a quick pay-back for your friend. Very cool!

P.S. I don’t have enough experience to say this for-sure but I “feel” like MLPE is the way to go here i.e. go for the IQ7s.

This is what materials cost looks like for installs. My big house system (15.9kW nameplate) was around $1.50/W as well - I didn’t need panel level electronics for ground mount, and I’ve got an awful lot more panel hooked to each of the “big” inverters (east-west strings, so they’re never fully illuminated at the same time, I’ve got 7kW hanging off a 6kW inverter and could have gone with a smaller inverter). If you can get panels cheaper (which is doable, just not entirely reliable), you can get down to about $1.25/W for materials on a roof mount install.

The best I’ve seen out here for installers is $2.50/W, and that’s not a company but a guy I’ve worked with some in the past who is doing them.

I recognize I hold a particularly weird set of biases here, but I would argue that in most cases, the cost delta involved (perhaps not on this system, but on a larger system) is enough that it is worth learning how to do all those things for the savings involved. None of them are particularly hard, though it takes time. Installed cost on my house system was around 25k, and I’m fairly certain that a comparable ground mount by anyone locally would have been north of $75k. I figure spending my weekends in a pandemic year working on it to save $50k (a bit less, after incentives, of course - let’s not discuss rolling financing charges into the bill to get better incentives after you “buy” 0% interest) is a decent use of my time. I have some experience with wiring and such, but if you look at my office system compared to the house, the office system is a tossed together toy (which has been working perfectly for nearly 5 years) - and I’m redoing part of it to make it better after learning what I learned with the house system.

Nothing about a basic solar install is hard, except possibly the plans review process if you live in an area with an adversarial process for that… which I don’t believe Boise is. Unlike most of the rest of Idaho. It requires some care and new skills, but saves an awful lot over the installer prices.

Counterpoint: I know a lot of people who thought they bought from “reliable installers” who are now out of business. They don’t have admin rights on their system to get the per-panel data, and nobody else will touch their system because they didn’t install it. So they basically are stuck with “DIY the repairs on a system they didn’t build,” or “hope they can get someone to come out, reverse engineer the system, figure out the issues, and repair it.” The second sort of thing tends to be billed at a “I don’t want to do this…” rate for an awful lot of hours. Better than the “lease” deals, though, in which case nobody will touch it at all.

I really don’t like that a lot of installers treat the system like Tesla treats their cars - “Sure, it’s on your house, and you paid for it, but it’s still our system and you can’t have the actual data from it.” A lot of installers seem to only allow the customer to access bulk system production data, not per-panel data. And I understand that some of it is to prevent “Why isn’t this panel producing the same as the others?” calls, but… I’ve also seen plenty of photos of installs where panels are shaded by a vent stack for a huge part of the day. “Hiding that you’ve put a panel in a terrible place and it produces almost nothing” isn’t great customer service, IMO. Not as bad as some of the RV installers, though… Handy Bob has some solid rants on them, and I have to agree with him (he’s also one of fairly few people on the internet giving good information about how to charge lead acid).

Good price, yeah, and could come down a bit with some panel price improvements. It’s certainly some non-trivial amount of work, but I’d rather see someone who didn’t work in the space do the work carefully on their house than some of the “professional” installations I’ve seen that don’t even bother to make the conduit routing look good, just slap it across however.

In terms of “quick payback,” maybe? It depends on what the local power company does. Right now, new installs are still on a kWh for kWh net metering, but they reserve the right to change that on demand, and I expect they will come back at some point with better numbers for their “hourly net metering” concept. Their last attempt got smacked down pretty hard by the PUC with something along the lines of, “We said you could come up with numbers, if they were backed by research. You appear to have pulled these number out of your rear ends, and don’t have anything to argue otherwise with, so, no.”

But if you combine reduced grid load and emissions with the ability to run some loads when the power grid is down, it makes sense for a lot of people. Especially now that there are ways to run the string inverters with the optimizers and still get backup power.

So… why? I demonstrated that there’s no meaningful price difference between IQ7+ microinverters (which get you no backup power options) and the string inverter with optimizers (which does get you the backup power option). Even if you want the data gateway for per panel data, it’s still a more capable system for not significantly more money, and Enphase microinverter reliability isn’t amazing. If one if the Tigo optimizers fails, they are designed to “fail open,” and it operates as though there’s no optimizer on the panel.

I “feel” like Enphase buying their way into NEC 2017 and beyond is reason enough to avoid them when possible, because they’ve bought their way to a monopoly on microinverters, and are then playing regulatory capture games to ensure that they look like the best option. Were it me doing a rooftop system, I would pay more to avoid using anything from Enphase for those reasons alone - and you get backup outlets as a bonus.

I’m sure there are some good solar installer companies out there that don’t try to bullshit people and lie about how much “you’re going to save if our numbers are correct and nothing ever changes with net metering from the power companies who have expressed intent to change it, oh, and you can take the full federal tax credit in the first year but we’ve not asked and assume you know all about federal taxes.” I just haven’t met any, and a number of people I know who have gotten quotes after learning about the hidden assumptions baked in feel much the same way. High pressure, “Lie about the equipment the other guy installs,” “sign how for savings!” sales pitches are the norm in the industry, and I simply don’t like that one bit. Throw in that they somehow always manage to come up with the largest system your roof can fit… eh. Not an industry I’m really a huge fan of lately.

However, in any case, it would be interesting for the person involved to get some quotes from recommended companies and compare them.

Totally. You have to weigh that risk with the level of service a local installer might vs. the risk of the company going out of business or not following through with promises of support.

So I have mixed feelings here. I’ve seen dashboards from fiends monitoring software that’s really good and does provide per-panel information. And for someone not tech-savvy a few dollars a month for this service is often worth it. That being said I myself, being a software engineer, really value libre-open-source software, and if given the choice I’d rather self-host and control my software. The idea of “admin” rights (If I understand what you mean by that) I think is a pervasive issue across a lot of industries (cell phone OSs, DRM music/kindle books, IOT, right-to-repair with John Deer tractors (somehow also subject to DMCA laws), ect.)

Fair enough.

FWIW energysage does vet installers to avoid issues like this, and we have installer reviews on our site so you can get an idea about others experiences. That being said I think no one will care about your system more than yourself I guess.

You know more in this department that I do. Naively I assume a per-panel micro-inverter with a build-in optimizer would be more redundant in the case a single panel in an array was misbehaving / under-performing. That being said I don’t know the specifics to optimizer configurations that might completely mitigate these kinds of issues in a string-inverter configuration.

We have a lot of people who have used our platform and are really happy with the installer they decided to go solar with. But in general many solar installers are like car-sales-men, have high pressure tactics, and are dishonest. It’s a very real and big problem (something we work to mitigate on our platform via vetting installers who want to join & finding the best ways to compare “apples-to-apples” quotes from different installers).

But totally agree, if you have the time and motivation and ability to spool up on the technical stuff required to do it yourself, seems like the way to go.

Thanks for the mention of Handy Bob, I’m going to do some reading from him. And probably pass him along to my LDR partner, since she’s got an airstream she’s wanting to put solar on. Of course then you pay the airstream tax to do it properly, but that’s why you bought an airstream, right? :stuck_out_tongue:

EDIT: I presume this is his website? It’s…a mess. sigh

Indeed. If I own it, I should have the ability to get full root to it. Unfortunately, this isn’t global yet. But not giving me the access to production data on my system is a common enough problem.

I’ve found this to be true across the board. Does it take me longer to do stuff? Yeah. Do I either have some confidence it’s done right, or know who to blame when it’s not? Also yes.

I’m sure that’s true now, but how true will it be in 30 years when the promised “savings” haven’t appeared and they realize that it was far more expensive than just buying power? On some of the $4/W+ systems in areas moving away from net metering, that’s going to be true.

Oh, it’s a total mess. And as near as I can tell, just about everything he says is spot on.

What changed? Just more options in the market?

The tech note from SMA talking about how you could run the SPS (Secure Power Supply/backup outlet), with the TS4-O optimizers, by providing ~18V across the secure power supply terminals. That came out last July or so, long after I’d designed my system, so I wasn’t paying close attention to the details.

Prior to that, you could use the SPS only if you didn’t have rapid shutdown hardware in the string. For me, this is how my arrays are set up (if there’s sun, there’s DC to the inverters), so I’ve got my backup outlets. But for roof mount solar, which requires per-panel rapid shutdown (not just array level), SPS wasn’t an option. If AC power cuts out (either from grid failure or disconnection), the inverter commands a rapid shutdown of the roof hardware, dropping the residual voltage well below what the inverter can power up on - so until AC returns, there’s no way for the inverter to ask the optimizers to open up and operate.

However, with this tech note and more recent firmwares, there’s a way to provide just enough power to the inverter (via the SPS switch terminals) to bring the control electronics online enough to cancel the rapid shutdown order - bringing main DC online, and then the inverter can come up and provide power to the SPS.

It’s been a major limitation for the SPS use case, and this seems to have nicely resolved it - adding this an option for people who want rooftop solar with “a bit of backup power when the sun is shining.”

The other option for that is supposedly the Enphase IQ8s, but they’re not shipping and Enphase has refused to answer questions about if/when they’ll actually show up. So, far as I’m concerned, their vaporware until proven available. The Tigo TS4-O and SMA Sunny Boy inverters are mature, available products.

Is it possible to wire up batteries on one of the two MPPT circuits? as a matter of future proofing for batteries? I may be thinking about this entirely wrong, but if there’s a charge controller and the voltage matches up to minimum (100V), can that work?


And… that’s a really good question I have no answer to. You wouldn’t be able to use it with the grid up, as it would just honk on the battery until it hit either the battery limit or the inverter power output, but as a feed for grid down operation, I expect it might. Not sure that there would be any particular priority between solar and battery, though - I don’t know the algorithms used.

I know the Midnite Classic charge controller will eat from a huge range of sources (a guy I talk to on IRC has charged his battery bank off some of the more nonsensical energy sources you can come up with, beating his Classic up all the while - I believe he fed it full bridge rectified 120V for a while and it was confused but worked), but I’m not sure how tolerant the Sunny Boys are to that sort of thing.

… I suppose I do have a spare, I could wire it up and see.