My experimental solar container is doing exactly what I intended it to do: Teach me how modern solar power systems work. I’m definitely drinking from a firehose, but there are some things that make me stop dead in my tracks that just don’t seem to make sense. RSD (Rapid Shutown Devices) and the requirements of NEC 690.12 and UL 3741 has me flummoxed. The requirements seem bizarre.
Oh, wait, I get it, it’s nonsense. Don’t put panels on your roof, use ground mounts. End of story. That’s the TL;DR explanation. But if you need to know more…
I’m not going to keep typing UL3741 and NEC 690.12, I’m just going to call this stuff RSD, though there are more things in play than just stuff to shut PV panels down. But just to clarify the number soup, UL3741 is the standard that equipment supporting the code requirements must meet, and NEC 690.12 is the code.
So the most obvious part of RSD is that it’s justified as a way to prevent shock hazards to firefighters. Yes, I said justified. It honestly feels like a powerplay for manufacturers of microinverters and some types of rooftop equipment to clear out competitors. When the code was first developed in 2014 there was no evidence that firefighters were endangered by 600V series strings on a roof. There still isn’t a lot of proof, and what experimental and empirical evidence there is still seems sketchy and very much created after the fact to justify the rules.
But I’m not going to engage in conspiracy theories beyond that comment, because I don’t have the kind of deep knowledge I’d need about the history, politics, and technical analysis behind the rules to feel comfortable making accusations. I’m late to the party so I only know what I’ve read since I first wandered into 690.12 and thought “Hey, WTF is all this about!?!”
I’m just going to look at the ramifications and solutions from a DIY perspective. I think professional installers know how to deal with this and have some solutions they are comfortable with. I think many DIYers aren’t even aware it’s an issue (or maybe I’m just the last person to hear about this). In part, DIY folks may not have encountered this issue because, as the TL;DR version says–if you’re doing ground mounts or installations where you don’t really care about code, you don’t need to pay attention to most of this stuff. You are exempt. Mostly. Maybe.
Reading the code leads me to think that my cool giant red pushbutton on the side of my container that triggers the RSD in the EG4 18K PV is not code compliant. It needs to be a switch with “OFF” clearly visible. No it doesn’t, yes it does, no it doesn’t… Well, I’ll figure that out later.
Let’s start with the basic electrical requirement. PV panels mounted on roofs need need rapid shutdown equipment that:
- Reduces the voltage of the conductors within the array boundary (roughly the boundary you might draw one foot away from your array) to less than 80 V within 30 seconds of RSD initiation.
- Reduces the voltage of conductors outside the array boundary to 30V within 30 seconds of RSD initiation.
So the first question is how do you reduce the voltage of a series string of panels to below 80 volts? Simply cutting the connection between the array and the solar controllers actually increases the voltage inside the boundary. Perhaps toss a blanket over the panels? The fact that the requirement is 80V hints at the solution. If the requirement were 0 volts there would be no solution, and microinverters would be in almost as much trouble as series strings. As long as there is sunshine there will be voltage on the input of a microinverter, even if simply opening the line to the grid makes the microinverters turn off their nominal 240VAC output (actual 240/211-264 or 208/183-229). The only practical way other than microinverters to meet the electrical requirement is to have devices between the panels that open the series strings to reduce the voltage below 80 volts. If you have a 400V series string it would take a minimum of 5 devices to chop the string into 80 V segments. At 600V you’d need 8. Yay, more connectors, more electronic devices on the roof baking under the panels or connected to the racks. More wiring to manage. More complexity and elements to fail or fault. The only good news is that it’s fairly easy to meet criteria 2 if you meet criteria 1.
Or you could just abandon DC coupled series strings and AC couple with microinverters. Probably those ground mounts are looking better and better
Unsurprisingly, solutions exist. They come in various flavors, but the most common devices are called module level power electronics (MLPE) that are installed as one MLPE per module for about $50 each–not as spendy as $250 microinverters, but still… A 20KW installation with 500 watt panels would be 40 MLPEs or about $2000–if your inverter can communicate with your chosen MLPE. If not, additional communications devices can be added to wirelessly activate your MLPEs when the connection to the PV panels is cut–for a few more bucks. The largest manufacturer of MLEPs (Tigo Energy) says they’ve delivered over 10 million units. That’s about half a billion dollars worth.
My direct experience with microinverters is not good. Perhaps unsurprisingly, 32 first generation microinverters installed outdoor in line of sight of the ocean (though high enough on a bluff to eliminate salt spray) did not last for the full 25-year warranty period. The manufacturer did not have first generation units to replace them. They started failing at year nine and in less than 15 years all were dead. Replacement constituted an upgrade I needed to pay for, though they “graciously” extended a discount. Given that the original installation company had gone out of business long ago (like 90% of the solar installers from that era), there was no company willing to do the replacement on the ground mount situated on steep and somewhat treacherous ground, I had to do it myself. Not an enjoyable job at 76, though I did it. As a discounted “upgrade” there was no mention of paying for the labor (even if I could find someone to do it), and no willingness to discuss it. You might understand why I’d be skeptical of the 25 year warranty on MLPEs.
The addition of a device under each panel purely for safety reasons would present a significant drag on sales, but MLPEs can also be module-level optimizers, providing MPPT (maximum power point tracking) at the individual module level. MPPT adjusts the voltage and current output of a solar panel (or in straight DC coupled systems, the solar array) to produce the maximum power the panel can achieve given environmental changes like sun angle, shading and temperature. Module-level MPPT also accommodates mismatched panels. Series strings of mismatched panels that are connected in parallel with other series strings (you probably need to read that twice) not only cause current to circulate between the strings, but also defeats array-level MPPT. I’m not certain how well MLPEs play with the MPPTs in all-in-one hybrid inverters like the EG4 18K PV I’m using in my container project. Seems like there might be issues.
Rooftop PV installations are generally more complex in terms of layout and roof orientation, and more prone to shading than ground mount systems. You have to deal with the roof you have, whereas if you have the space (and no HOA to drive you nuts), a ground mount can be optimally oriented in a place with no shading. So MLPEs, like microinverters, can make the best of an inherently limited situation. So salespeople can happly share that MLPEs can increase rooftop energy production by 20%-35% , mitigate shading and reduce losses due to orientation mismatch. They also permit more flexible system designs since they accommodate varying roof orientations and shading. And of course, they can comply with RSD requirements. So hey, all good, right?
Not so fast, bucko. Sticking 40 of any kind of electronics under panels is a potential maintenance nightmare. The folks who make these devices say that maintenance is reduced since the devices are distributed vs. centralized. I guess that depends on your definition of maintenance. If a hybrid inverter goes bad on the side of your house you can replace it or repair it without climbing on a roof. And, given Murphy’s law, the dead or dying MLPE is most likely in the middle of the largest array. You’re likely to need to pull a lot of panels just to get to it. But what they probably mean is you can just let it stay dead. I think that works better with microinverters since it’s exactly what I did when mine started failing. I have no direct experience with MLPEs but it doesn’t seem like that would work. They are interrupting series strings, unless they fail in a manner that closes the connection to the rest of the series string, it seems like one dead MLPE would take out the string. And if they fail closed the system might not meet the 80volt max requirement for RSD. I’ve looked at manufacturers literature and can’t find any support for the notion they reduce maintenance.
Most likely the biggest benefit of MLPEs is that they can meet RSD requirements.
MLPE alternatives
It might seem strange at first, but the racking you chose might allow you to avoid using MLPEs. Some racking systems are now certified as UL3741 compliant, with no other requirement other than installing it properly. This seemingly bizarre exception is mostly about cable management. If the cables are properly tucked away and secured there is less likelihood that a firefighter would be exposed to potentially fatal voltage when fighting a roof fire. And if the wiring is isolated from the racking then the racks wouldn’t become electrified when insulation melts off. I’m not sure how criteria #2 for PV array voltage outside the array boundary gets met.
It should come as no surprise that some solar industry players who make the equipment that directly meets Ul3741, don’t care for these racking certifications. Here’s an article from altenergymag written by an engineer from Solar Edge. As a former advertising executive my bullshit detector is finely honed–mostly from years of lying about other people’s products for money. It’s going off like a fire alarm. Yeah, I see the irony.
While premium racking and careful wire management seems better to me than installing a lot more electronics and a lot more connectors, it does also seem to hint that the codes are not as scientific or logical as one might hope. And from my reading it seems that most ballasted racks for flat roofs are certified, though inherently by design as rows of individual units, those seem even more likely to have exposed wiring. Perhaps it’s simply that ballasted roof racks are usually plastic and the big threat to firemen is electrified racks?
This all seems a bit arbitrary, but maybe it’s all well thought out and experimentally verified. There’s no way for me to tell without becoming far more expert on this segment of the code. Which isn’t going to happen, because for me, this is all moot. The roof of my shop could be a HUGE solar resource if I didn’t mind sticking holes in the roof. But I’m not going to do that–it’s a fifteen year old roof (at least). I have plenty of unshaded space for ground mounts. The roof of my house in Maui could be equally good, but it’s massive, beautiful tile. Replacing a broken tile is a nightmare. I have plenty of room for more groundmounts and my wife would shoot me if I touched the roof. It’s not getting solar racks.
There’s an interesting guy on YouTube called DerekSolarBoi who knows a lot more about all this than I do, and seems very excited about everything about UL3741. There are links in the video description. His videos are really worth a watch if you’re thinking about rooftop solar. He even has a website titled UL3741.com. If you need more info his videos all are worth watching and his flamboyant style is less tedious than my monotone.
Your situation could be quite different. For that matter, I could be reading the code wrong, looking at the wrong explanations, and/or misunderstanding how this all applies. I’d love to hear your comments and corrections. Rants about government overreach is a bit less welcome, but I’ll still read them all. I might not respond to everything, but I really do read every comment. I hope you found this useful if not entertaining. I expect the comments will be more fun than the video.