As the Expert Amateur name suggests, I’m not a professional electrician, which means no one pays me to do this stuff, but I’ve been working on electrical and electronics projects for a ridiculously long time, and I insist on studying and digging deep enough to know what I’m doing (hence, the Expert part of the name). Still, don’t take anything I say as gospel, I’m perfectly capable of being wrong.
For the TL;DR crowd: Here’s the bottom line. Solar installations are a lightning magnet. Why? Well, metal frames on your roof or in an open area, with metal racking and a bundle of wiring leading directly to the guts of your equipment might have something to do with it. Crack…BOOM…Sizzle. That’s the sound of your investment turning to slag. Okay, that’s dramatic, but really not accurate enough even for TL;DR. We’re not talking about direct strikes–nothing saves your gear if you’re that unlucky. But a strike even miles away can induce high voltages and therefore massive currents in your system. You really, really want to do a good job of grounding your panels and racks to guard against faults, lightning strikes, static discharge, etc. The proper way is to pound at least one 8 foot grounding rod into the ground near the racks, run some bare 6 AWG copper wire directly to the ground rod, and then run more #6 copper wire to any and all earth ground rods in your system, including the one connected to the ground bus of your service entrance. For anyone who cares about why and how, read on. It will be a thrilling ride. Not.
The rest of the story: There are two kinds of grounding for any solar installation that you need to pay attention to:
- Grounding the framework of the panels, mounts, and other hardware.
- internal wiring grounding–how the internal wiring is bonded and grounded.
Both types are safety considerations, but they work differently and have different concerns.
According to US NEC (National Electric Code), Panel frames, racks, and other metal components of the PV array must be (well, the 2023 version says “may be”) earth grounded separately from the general electrical wiring. Separate grounding helps isolate potential faults in the solar system from the main electrical system of the house. This prevents issues in one system from affecting the other. The metal frames of the solar panels and the supporting framework need to be grounded to manage the risk from lightning, or from a wiring fault or damage leaving your PV framework connected to dangerous voltage. Additionally, Transformerless inverters, which are rapidly becoming the standard for solar power, don’t have the inductive isolation that transformer inverters do. The result is that the inverters can backfeed AC to the DC side, DC can make its way to the AC side, and stray voltages can wind up in strange places. So proper grounding is essential. The concern is the possibility of electrocution caused by a fault as well as the possibility of damaging equipment or causing a fire. Even if the shock doesn’t kill you, simply being startled by a shock can be enough to make you react, which could be as bad as electrocuting you if it makes you fall off a roof or a ladder.
Here’s where this all gets weird. Mike Holt, who is considered to be a guru of all things electrical, insists that this separate earth ground is both unnecessary and dangerous. Essentially a nearby lightning strike or even cloud-to-cloud bolt, can induce a voltage across the separated ground rods This high voltage can cause a massive current to flow through the equipment grounds, bonded neutral, and all the internal components and toast the gear. He says the safe and practical way is to use the internal grounding to connect the frame and structure to the main ground bus and from there to the earth ground. He’s not alone in believing this. About 50% of the people who responded to questions about this issue agree with him.
It seems to me that a lightning discharge making it’s way through the ground circuit to the main grounding rod would wreak a lot of havoc on the way, but Mr. Holt is quite insistent about this.
One of the respondents in the Facebook DIY Solar group, Mr. Mayo Tabb, wrote: “Having spent 45 years in the critical power systems for Data Centers I would strongly disagree with the concept of separate isolated grounds. The isolated dedicated computer ground was popular in the late 1970’s. As a previous respondent stated when you have a lightning strike there is a voltage potential between the two grounds that results in a voltage difference within the equipment with catastrophic results. I was called-in many times to examine and explain computer damage when those computers had my companies UPS systems designed to protect the computer. Always found culprit was dedicated isolated computer ground. After many destroyed computers the industry came to understand you need to tie all the grounds together. Many auxiliary grounds are fine as long as they are all firmly connected.”
Following NEC requirements would seem to be the prudent course, connecting the frames and racks to a separate ground rod–which typically means at least one eight foot copper or galvanized steel rod driven into the ground with just enough rod above the surface to connect a ground lug. The ground should be close to the equipment, connected with a suitable wire, typically AWG 6 bare copper wire to make the intention as an external ground clear to anyone that inspects the wiring.
If you consider the points Mr. Holt and Mr. Tabb make to be important you could run an additional ground wire from this second earth ground attached to the solar frames and racking to the service entrance main panel earth ground rod, which would eliminate the possibility of inducing a voltage across the two separated grounding rods. You might wonder “aren’t these two rods already connected together by being stuck in the ground”. But remember that there is a lot of resistance between two spaced grounding rods. The wire provides a low resistance connection, eliminating most of the induced voltage. As another respondent, Kipton Moravec said: “The reason they are all tied together is so they can go up and down TOGETHER with the lightning strike. That minimizes the potential between the systems and prevents the damage.”
One problem with this approach is that most inspectors will wonder what the hell you’re doing. Explaining an “error” in electrical code to an inspector is not something I’d enjoy. A second problem is that if your panels are a few hundred feet from the service entrance, as mine are, all that #6 copper wire will be quite expensive. But external grounding is the most important element in protecting your equipment from lightning, and your fried gear will certainly be more expensive than the wire, though I see the difference in the absolute expense of the added wire vs. the theoretical expense of damage from lightning that may never happen. But lots of DIY installers festoon their equipment with lightning arrestors. Even if you install lightning surge protectors your system will only be as safe as your ground connection. And if the surge is coming from your ground connection then I’m not sure how protective they would be. The massive amount of energy present in lightning means that indirect strikes, and even lightning between nearby clouds, can cause damage. Despite the expense of long wiring runs and the limited electrical code support (connecting grounds is mentioned in the NEC, but it’s a bit buried and not well explained), it seems like the prudent thing to do.
The wiring connecting the frames to the ground rod should be as straight as possible with no sharp bends–lightning takes as straight a path to ground as it can find. In areas with dry soil a single grounding rod is insufficient. Your local electrical inspectors should be able to tell you how many grounding rods you need to be reasonably well protected. Many areas require DC solar wiring (from the panels to the controller) to be run in metal conduit. In my installation this metal conduit was just floating, secured to my container but not connected to the racks. Bonding the conduit to the rack grounding system seems like a wise precaution–so I did that.
For long runs consider trenching and burying the wiring rather than any form of above ground wiring since it will be less affected by lightning. Using conduit is highly recommended for burying solar cables. It’s generally less expansive to bury long wire runs since it can be run in much cheaper plastic schedule 40 PVC conduit rather than metal. Schedule 80 is generally required for above-ground portions. The USA NEC (National Electrical Code) requires 18 inches of cover for PVC conduit (though areas with hard ground freezing require deeper burial of 24 inches or more to prevent frost heaving that might break conduit and wires). Standard solar wire is usually rated for direct burial but it’s unwise to do so, and any long runs should include pull strings for future wire additions or replacement. The conduit leading to your roof or leading to the ends of series strings on ground mounts should terminate in a weatherhead and you should seal conduit ends with electrical putty. Most jurisdictions require that conduit to be metal. If you’re doing long runs you should use at least 1″ conduit even if the conductors can be accommodated with smaller sizes. The number of wires that can be accommodated in each size of conduit is less than you probably think (unless you’re an electrician). There are code requirements and readily available tables that tell you how many. If your wiring takes up more than 50% of the space you are almost certainly already in trouble. The difference in price for larger conduit is trivial compared to the overall costs, and it’s much easier to pull wire–it future proofs the installation. If you need more wires in a few years you have the space.
System wiring bonding and grounding is a different issue with different goals. The aim is to provide a low-impedance path for fault current to ensure any faults trip breakers or blow fuses. In almost all cases there should only be one location in your entire electrical system where your neutral wire is directly connected to the ground bus and from the ground bus to an earth ground. That location is generally in the service entrance box or or first disconnect point. They remain separate in all sub-panels and branch circuits. Neutral-ground bonding is required by NEC at the service entrance and ONLY the service entrance.
In off-grid systems, a single neutral/ground bond in the main electrical distribution panel with connection to an earth ground serves the same function as the service entrance neutral/ground bond.
Bonding neutral and ground is done to provide a low-impedance path for fault currents to return to the source and trip overcurrent devices. The neutral-ground bond provides a very low impedance fault current path. Directly connecting to earth ground (earthing) is done primarily for protection against lightning and static electricity. Earth grounds do not normally carry current, they dissipate static electricity and induced voltage from atmospheric phenomena like lightning. Bonding neutral and ground creates an intentional current path for safety, while earth grounding provides protection against external electrical phenomena. Earth grounds have much higher impedance and are not reliable for fault current
Bonding neutral and ground wires together in locations other than the main service panel or multiple points creates parallel paths for neutral current to flow on grounding conductors. This can permit high currents to flow where you really don’t want it, and multiple neutral-ground bonds can divide fault currents between neutral and ground paths.
There is also an increased shock risk since it can cause ground wires to carry current even in some normal non-fault conditions which means metal cases that would otherwise be safely grounded would actually be electrified by the ground wire. Ironic, and quite dangerous. It also completely screws up ground fault interrupters (GFCIs).
The stray currents can easily make electronic equipment malfunction or cause permanent damage. And multiple neutral-ground bonds make it harder to isolate and troubleshoot electrical faults in a system. In my research I found some YouTube videos of people trying to track down weird currents on their neutral and ground wires. I suspected I knew the source of the problem.
In short, single-point bonding at the main panel ensures fault currents have a clear path to trip protective devices while keeping grounding conductors normally free of current flow. This maintains the integrity of the grounding system for safety purposes.
I have to credit a person I don’t actually know named Mark Bittman who posted a question on the DIY Solar group on Facebook. The question sparked some lively debate and pushed me to dig deeper into the issues. Even though I spent a lot of painful time reading the NEC manual and researching all over the web, I still don’t have complete confidence in my conclusions.
In one of the forums I read through, a member who seemed particularly knowledgeable wrote: “I have had some entertaining evenings with a few bottles of decent beer and wine with some knowledgeable European electrical engineers, and Earthing is always a subject we normally all try to stay away from”.
Added content: A comment on the related video on YouTube caused me to look for a clearer way to explain earth grounding concerns. Here’s my reply: I regret saying “solar arrays attract lightning”. I know better, I was trying to make the point that it’s worthwhile paying some attention to induced current from nearby strikes, but no, they don’t “attract strikes”, they’d need to be a lot bigger and taller to do anything like that, and the video and post are not really about direct strikes, but induced current from nearby strikes. If lightning hits your equipment directly it’s gone, unless you have some very expensive heavy-duty protective equipment. The meat of the article and video is about induced voltage from nearby strikes and general equipment grounding, two distinctly different topics. Induced voltage can be caused by something as remote as a cloud-to-cloud lightning bolt. If a system has two earth grounds it can induce a high voltage between the grounds which causes a huge current to flow because of the difference in potential between the grounds. If there is no ground on the solar array to provide a short path to earth ground a voltage induced or collected on the array framing and structure can cause current to flow through the equipment to the service ground. Not good for your equipment either.
Absent a service ground your entire house has no way to dissipate static discharge, such as lightning, safely into the earth. The service grounds helps protect your home and electrical equipment from damage due to high voltage surges. It also helps maintain the same potential between the earth and any electrical equipment or bonded objects, reducing the risk of electric shock. If you don’t have a ground rod, YOU may provide the path to ground by touching any metal case connected to the system ground, which is pretty much everything. It also provides an alternative path for fault currents back to the power source, especially if a neutral connection is lost. But it’s not a good path, the resistance is variable and may be high.
There is no easy complete solution. Binding (connecting with a ground wire) the service ground to a local ground near the array seems like the best solution to keeping a good path to ground that doesn’t flow though your equipment, while also preventing an induced potential difference between the grounds that can cause a destructive current flow though your equipment. Binding the two grounds also minimizes the effects of ground loops. I’m going to add this summary to the blog post. Thanks for forcing me to clarify, you made me think this through and come up with a better explanation. I hope it helps you.
Another reply I wrote to a comment on the video. Again, I’m not including the question I replied to, so some of the reply will seem a bit disjointed. You can read the full dialogue in the comment section of the video:
Repy: I’m not sure where Will Prowse’s comment appears, it wasn’t directly in the comments here. In the video comments he just said the video and discussion are awesome, which is incredibly generous. But I understand what both he and Mike Holt are getting at. There is good reason to say there is already a path to keep the potential on the frames and mounts close to being the same as the service entrance. The system grounds provide that path, and you can certainly tie a system ground wire to the structure of the array. Code says it’s OK, and many people do that. It’s probably adequate. Probably.
But that’s a somewhat unreliable connection. It relys on a chain of fairly funky screw connections in metal housings and perhaps many short runs of wire (especially in old houses like yours). If your racking is remote from your building I simply don’t see why you’d want to be leading lightning or static induced current into your equipment and then rely on a lightning arrestor to deal with currents that didn’t need to be there in the first place.
The NEC codes go back and fourth on this issue. 2017 says remote equipment (unattached structures) MUST have an earth ground. Later editions say they MAY have. And there’s a vague suggestion that all earth grounds in a system should be bound together. I haven’t found any direct and concise explanation for Mike Holt’s position. I inferr he’s concerned that if the two earth grounds are not at the same potential (which would happen if the connection through the system ground is broken) then the induced voltage of a nearby strike or other static charge will cause a current through the earth grounds to the system ground and bonded neutral that will be hazardous or damage equipment. I agree wholeheartedly with that, but by the same logic, the broken connection in the system ground means there is no earth ground path for your frames and mounts, which is also hazardous. So that leads me to the belt and suspenders conclusion that a local earth ground (proper, under code) bound (connected) to the service entrance earth ground is a good solution. No, it isn’t definitively a code requirement, just (I think) a best practice. As Will Prowse says, it does add cost, but if you’re already trenching to connect your equipment to the service entrance then its a ground rod and some bare #6 wire. Maybe $200 if the run isn’t super long, and if it is, then I’d say doing so raises way above just best practice.
I do love Will Prowse’s videos. I watch them all and read through his forum (even better than the videos). His equipment and knowledge are almost as impressive as his willingness to call bullshit on the myths so many people buy into. He’s done three videos (or maybe two, I’m old, my memory sucks) recently trying to get people to understand how to charge LFP batteries to get the most out of them and keep them balanced. The comments indicate some folks are still insisting he’s wrong. He’s not.
I dimly recall one video where he was trying to track down some mystery currents in his ground and neutral. First of all, it’s remarkable that he was even looking for them, that’s not something many people do. He’s fond of laying panels in his driveway or setting up temporary mounts and most of his inverters are transformerless. Nevada’s dry air is perfect for building static charges. I think I have some notion where his current might be coming from. Of course he does too.
Incidentally, I’ve given up on telling people that DC breakers are inadequate as DC disconnects. They all say “I’ve been using mine for XX years and it’s fine”. Yes, that’s a stupid reply, but I can’t fix that. I don’t use knife switches, but I do use real DC disconnect switches which rely on the same principle–lots of air gap, which is why they are rotary and their connection terminals are in such odd places–like almost 270 degrees apart. There’s no way a snap switch, even one with magnetic quenching, is adequate for more than a few disconnects, and after you’ve seen one burn you wouldn’t trust them either. DC fuses require a lot of care too.
The lightning lottery has about the same odds as Powerball (inversely). I’m really not writing about equipment being hit by lightning. And yes, your house won the lightning lottery, as did millions of others.
There’s an interesting video Here: https://www.youtube.com/watch?v=7DvIjAvMge0 by a guy who did indeed have a nearby strike (which might have been miles away) slag his installation. His solution after a lot of research is to install a number of lightning arrestors (midnight solar SPDs). The workmanship on his solar wiring is impressive, I expect he did his homework. He did indeed come to the conclusion that his grounding was inadequate, that sandy, desert soil means he needed more ground rods for his SPDs to work. But he’s relying on the metal frame of his ground mounts as earth grounds, and I don’t think he’s binding the frame directly to to the earth ground in his solar shed. So I expect those SPDs could get a workout, and they are one shot and then need replacement at 100 bucks each. I’d want to add a few more earth ground electrodes locally to the ground mount and bind them directly with wire to the electrodes at the shed. If nothing else, that might preserve the SPDs, and more likely, will save his equipment from another slagging. In some ways he’s chosen the worst options. He’s got multiple earth grounds (sort of) so Mike Holt would say “that’s bad” but the array earth ground is almost certainly inadequate given the sandy soil. So he might not have to worry about inducing a huge voltage on the system ground and neutral, but he might send a lot of current through the system ground and pop all his SPDs (or toast the gear if it’s a close enough strike).