Electrical
Photovoltaic (PV) system wiring field guide: NEC Article 690
Wire the array into a code-legal system: DC strings sized for the cold, rapid shutdown, the disconnects, grounding, and the interconnection that decides whether the breaker even fits.
Direct answer
PV system wiring is the electrical side of a solar array: the DC strings, the inverter, the disconnects, the grounding, and the interconnection to the service, all governed by NEC Article 690. Cold-temperature string voltage, rapid shutdown, and the 120 percent interconnection rule control the design, while the adopted code edition and the AHJ govern.
Key takeaways
- NEC 690.7 requires PV string voltage calculated at the site's lowest expected temperature, since cold raises Voc past the inverter and system voltage limit.
- System voltage caps: 600 V residential one and two-family, 1000 V most commercial and multifamily, up to 1500 V ground-mount utility-scale.
- NEC 705.12 120 percent rule: main breaker rating plus 125 percent of inverter output cannot exceed 120 percent of busbar ampacity, PV breaker opposite the main.
- NEC 690.12 rapid shutdown drops conductors outside the array boundary to 30 V or less within 30 seconds of initiation.
- Never mix MC4 connector brands across a mated pair; cross-mated contacts overheat in full sun and start rooftop fires.
What PV system wiring is, and why the electrical side decides the callback
Photovoltaic system wiring is the electrical side of a solar array: the work that takes the DC the modules make, runs it through the inverter, and ties the AC output back to the building's service, with the disconnects, grounding, and rapid shutdown that keep it safe. NEC Article 690 is the part of the code written specifically for PV, and it governs almost every decision on this list.
There are two physical jobs on a solar install, and they are usually done by two different trades. One is getting the array onto the roof without leaking, which is the racking, the flashing, and the structural load. The other is wiring the array into a working, code-legal electrical system. This guide is the second job. The roof-mounting side is its own discipline, covered in the rooftop solar mounting and racking guide, and EV charging, which often shares a service with PV, has its own guide too.
The thing that separates a clean PV job from a callback is rarely the panels. It is the string voltage that climbed past the inverter rating on the first cold morning, the rapid shutdown that was never tested, the load-side breaker that broke the busbar rule, or the MC4 connector from one brand crimped onto a pigtail from another. Those are the four that fail inspection, and the four this guide spends the most time on.
Grid-tied, grid-tied with storage, and off-grid
A PV system comes in one of three architectures, and the architecture decides what wiring you are dealing with before you size a single conductor.
Grid-tied, also called net-metered, is the common one. The array feeds a grid-following inverter, the inverter ties to the service, and any power the building does not use runs backward through the meter to the utility. There is no battery. When the grid goes down the inverter shuts off, because a grid-following inverter cannot make its own reference and is not allowed to backfeed a dead utility line where a lineman might be working.
Grid-tied with storage adds a battery and usually a hybrid inverter. Now the system can ride through an outage, which means part of the load gets separated onto a backed-up subpanel and the inverter has to form its own grid when the utility is gone. The battery pulls NEC Article 706 and its own disconnect and listing requirements into the job.
Off-grid has no utility connection at all. The array charges a battery bank through a charge controller, an inverter makes AC from the battery, and the whole thing has to carry the load on its worst day with no grid to lean on. It is the least common of the three on commercial work and the most demanding to size.
Most of what follows applies to all three. The DC side, the disconnects, the grounding, and the rapid shutdown do not care whether there is a battery. The interconnection and storage rules are where the architectures part ways.
The module, the string, and how voltage adds in series
A PV module is a DC source, and the two numbers that drive the wiring are on its nameplate. Voc is the open-circuit voltage, the voltage the module produces with nothing connected. Imp and Isc are the operating and short-circuit currents. The module also lists temperature coefficients, which say how those numbers move as the cell temperature changes, and the coefficient for Voc is the one that sizes your strings.
Wire modules in series and you have a string. In series the voltages add and the current stays the same, so ten modules at 40 V each make a 400 V string that still carries one module's worth of current. That is how a PV array reaches the few hundred volts a string inverter wants to see. Wire strings in parallel and the currents add while the voltage stays put, which is how you build power without pushing the voltage past the equipment rating.
The string is the unit you size everything else around. Its voltage sets whether you are inside the inverter's window and under the system voltage limit. Its current sets the conductor and the fuse. Get the string count wrong and nothing downstream is right.
Why does PV string voltage matter in cold weather?
Cold raises PV voltage, and a string sized at room temperature can climb past the inverter's maximum and the system voltage limit on the coldest morning of the year. That is the single most common DC design error on a solar job, and it does not show up until the weather does.
The physics is in the temperature coefficient. A crystalline silicon module's Voc rises as the cell gets colder, by a fraction of a percent per degree below the 25 degree C rating temperature. On a clear, cold winter morning with the sun just hitting a cold array, the string sees its highest voltage of the year, well above the nameplate Voc. NEC 690.7 requires the maximum system voltage to be calculated at the lowest expected ambient temperature for the site, using the module's temperature coefficient or the correction table in the code, and the informational note points to ASHRAE extreme-minimum design temperature data as the source for that low.
The number you are protecting is the equipment rating. Residential one and two-family systems are capped at 600 V, most commercial and multifamily at 1000 V, and ground-mounted utility-scale up to 1500 V under the conditions in the code. The cold-corrected Voc of one module, times the number of modules in the string, has to stay under that ceiling and under the inverter's listed maximum input. Size the string for the cold, not the spec-sheet Voc. The max-modules-per-string number is the inverter's maximum input voltage divided by the cold-corrected module Voc, rounded down.
Confirm the section numbers and the voltage limits against the adopted NEC edition. The coefficients are on the module datasheet, and the cold design temperature is specific to the site.
Voc,cold = Voc × [1 + β × (Tmin − 25°C)]Nmax = Vmax,inverter / Voc,cold- Voc
- Module open-circuit voltage from the nameplate, measured at the 25 degree C rating temperature
- Temperature coefficient (beta)
- How much Voc changes per degree of cell temperature, from the module datasheet
- Maximum system voltage
- The highest voltage the array can reach at the coldest expected temperature, capped by the equipment rating
Inverter types: string, microinverter, optimizer, and hybrid
The inverter turns the array's DC into grid-synchronized AC, and which kind you use changes the whole wiring topology.
A string inverter takes one or more series strings on its DC input and makes AC at one central box. It is the workhorse of commercial PV and a lot of residential. The DC runs from the roof down to the inverter, so there are energized DC conductors in the building, which is what drives the rapid shutdown question.
A microinverter sits under each module and makes AC right there. There is no high-voltage DC anywhere; the conductors leaving the module are already AC. That solves the DC rapid shutdown problem at the module by design and handles per-module mismatch, at a higher per-watt cost and with more devices on the roof that can fail.
A power optimizer is a module-level DC-to-DC converter paired with a string inverter. The optimizer conditions each module's output and provides the module-level shutdown, while the string inverter still does the DC-to-AC conversion. Optimizers and microinverters are both module-level power electronics, MLPE, and both are common ways to meet rapid shutdown.
A hybrid inverter adds battery management to a string or central inverter, handling the array, the battery, and the grid in one unit. It is what most grid-tied-with-storage systems are built on. Pick the topology for the site. Shading and module-level shutdown push toward MLPE, and a clean, unshaded commercial roof can run straight strings to a central inverter for less money and fewer roof-level devices.
String inverter or microinverters: which belongs on this roof?
A string inverter is cheaper per watt and simpler to service, but the string is only as strong as its weakest module, so shade on one panel drags the whole string down. Microinverters and optimizers isolate each module, so shade or a mismatched panel only costs you that panel, and they provide module-level rapid shutdown without a separate device.
The honest version is that the choice comes down to shading and roof complexity against cost. A large, unshaded, single-plane commercial roof is where straight strings into a central or string inverter win: fewer devices on the roof, lower cost, and rapid shutdown handled at the array boundary by a listed system instead of per module. Break that roof up with dormers, vents, and shading, or split it across many small planes facing different ways, and MLPE earns its keep, because every module operates at its own best point and the per-module shutdown comes built in.
The maintenance tradeoff cuts the other way. A string inverter is one box at eye level you can swap in an afternoon. A failed microinverter is on the roof under a module, which means racking work to reach it. Neither is wrong. Match it to the roof and the budget, and put the reasoning in the record.
| Factor | String inverter | Microinverter or optimizer |
|---|---|---|
| Shading and mismatch | String drops to the worst module | Isolated per module |
| DC in the building | Yes, energized DC conductors | Microinverter none; optimizer yes |
| Rapid shutdown | Needs a PVRSS or UL 3741 system | Module-level, built in |
| Cost per watt | Lower | Higher |
| Service access | One box at ground level | On the roof under the modules |
The DC side: PV wire, rooftop heat, and the home runs
The DC side of a string-inverter system runs from the modules to the combiner or the inverter, and the conductors have to survive a rooftop, which is a brutal place for wire. Exposed single-conductor runs use PV wire or USE-2, both rated for wet, sunlight, and the temperatures a rooftop reaches. Standard building wire is not rated for that exposure and does not belong in free air on a roof.
The sizing starts with the module short-circuit current, and NEC 690.8 compounds two factors that people miss. The maximum circuit current is 125 percent of rated Isc under 690.8(A), and the conductor is then sized at 125 percent of that maximum under 690.8(B), so a DC source-circuit conductor lands near 156 percent of Isc before any conditions of use. Rooftop temperature is the next part people underestimate. Conductors in conduit on a hot roof run far hotter than the air around them. Through the 2014 NEC a graduated rooftop adder applied; the 2017 and later code keeps a single sizable adder, commonly cited around 60 degrees F or 33 C, for raceways less than about 7/8 inch above the roof, on top of the normal ambient correction. Those corrections lower the conductor's ampacity, so the wire that looked fine at the table value can be undersized once you derate the 156 percent base current for the roof it lives on. Confirm the factors and the adder against the adopted edition.
Modules connect to each other and to the home runs with MC4-style connectors, and where more than a couple of strings combine, the strings land in a combiner with overcurrent protection. Both of those have their own rules, below.
MC4 connectors: one brand, matched and listed
MC4-style connectors are the snap-together DC connectors on every module lead, and the rule that gets violated most is the simplest one: do not mix brands. A connector is listed as a matched pair from one manufacturer. Mating a connector from one brand to a connector from another, even when they click together and look identical, is not a listed connection, and the code requires intermating connectors to be of the same type and from the same manufacturer, or specifically listed as compatible.
The reason is not bureaucratic. Cross-mated connectors can look fully seated while the metal contacts barely touch, and a high-resistance DC connection in full sun makes heat, melts the housing, and starts the kind of rooftop fire that put rapid shutdown in the code to begin with. Crimp the contacts with the tool made for that connector, seat them until they lock, and keep one brand from module lead to home run. If a panel ships with one brand and your home-run pigtails are another, you change the pigtails, not the listing.
This is the inspector's easy catch and the field's easy shortcut. Do not give them the catch.
Combiners and string fusing
When more than two strings land in parallel, each string usually needs its own overcurrent device, and that is what the combiner box is for. With one or two strings, a fault in one string cannot push more current backward than the other strings and the modules can safely carry, so fusing is often not required. Add a third string and a fault in one string can be backfed by the other two, which can exceed a single string's rated current, so each string gets a fuse sized to the module's series fuse rating.
The combiner is also where the string home runs become one larger DC feeder to the inverter, and where DC arc-fault and ground-fault detection often live. Use fuses and holders listed for PV DC service. A standard AC fuse is not rated to interrupt DC, where there is no zero-crossing to help quench the arc, and that mismatch is a real hazard, not a paperwork issue.
Confirm the string count that triggers fusing and the fuse sizing against the module datasheet and the adopted code edition.
What is NEC 690 rapid shutdown?
Rapid shutdown is a code requirement that lets a firefighter de-energize the conductors of a rooftop PV array from a single switch, so the roof is safe to cut and walk during a fire. It lives in NEC 690.12, and for buildings it is one of the most consequential requirements in Article 690, because a live array cannot be turned off the way a normal circuit can. The sun is still shining.
The mechanism is a shutdown initiator, often combined with the service or PV disconnect, that on operation brings the array conductors down to a safe voltage within a set time. The widely adopted limits are that conductors inside the array boundary, commonly one foot from the array, come down to a controlled level, and conductors outside that boundary come down to 30 V or less, within 30 seconds of initiation. The common ways to meet the inside-the-boundary limit are module-level power electronics, microinverters or optimizers that shut down at each module, or a PV hazard control system listed to UL 3741 that treats the whole array as an evaluated unit and can meet the requirement without per-module electronics.
Two cautions. First, the voltage limits, the timing, and the array-boundary distance have moved between code cycles, so verify the exact numbers against the adopted NEC edition rather than carrying an old value in your head. Second, a UL 3741 system is only compliant as the listed combination. Swap in a different module, racking product, or inverter than the listing covers and you have broken the compliance, even if every part is fine on its own. This is the requirement an AHJ will test at inspection, so it is the one to commission for real, not assume.
The DC, AC, and utility-accessible disconnects
A PV system needs a way to disconnect every source and every part you might work on, and on a service-connected system the utility usually wants a disconnect it can reach and operate. The DC disconnect isolates the array side, the AC disconnect isolates the inverter output, and the utility-accessible AC disconnect, where the utility requires one, is an exterior, lockable means the lineman can open without entering the building.
A DC disconnect has to be rated to break DC, the same way a DC fuse does, because DC has no zero-crossing to quench the arc. Many string inverters integrate a listed DC disconnect, which can satisfy the requirement at the inverter. The utility disconnect requirement is set by the local utility's interconnection rules as much as by the NEC, so confirm what that specific utility wants early, because it changes where the equipment lands on the wall.
The AC side: inverter output, OCPD, and listing
On the AC side the inverter output is a continuous source, so it is wired and protected like one. The inverter's AC output circuit conductors and its overcurrent device are sized at 125 percent of the inverter's rated continuous output current, the same continuous-load logic the code uses elsewhere. From the inverter the AC runs to its disconnect and then to the point of interconnection, which is the next section and the one with the rule that fails the most plan reviews.
The inverter itself is listed to UL 1741, the standard for inverters and interconnection equipment, and for grid interaction it has to be listed for that purpose, which in recent practice means UL 1741 with the supplements that cover grid-support functions. Keep the AC conductor sizing, the OCPD, and the inverter listing consistent with the equipment label, because the AHJ reads the label.
What is the 120% rule for solar interconnection?
The 120 percent rule limits how much PV you can backfeed into a panelboard on a load-side connection. When the PV breaker is at the opposite end of the busbar from the main breaker, the sum of the main breaker rating and 125 percent of the inverter output current cannot exceed 120 percent of the busbar's rated ampacity. It comes from NEC 705.12, and it is the calculation that decides whether your solar breaker fits in the existing panel or whether the service has to change.
The reason is heat. When the utility feeds one end of a busbar and PV feeds the other, current from both sources can stack in the middle of the bar without either breaker seeing the total, so the bar can overheat while both breakers sit happy. The 120 percent allowance is the margin that keeps that worst case under the bar's rating. Put the PV breaker at the opposite end from the main, because the rule depends on that physical position.
A worked case. A 200 A busbar with a 200 A main: 120 percent of 200 is 240 A. Subtract the 200 A main and you have 40 A of room for the PV breaker, so a 40 A backfed breaker, carrying up to a 32 A inverter at 125 percent, just fits. Want more PV than that and you are looking at a supply-side connection, a busbar or main-breaker change, or a derated main. Run this number before you promise a customer a system size, because it sets the ceiling on a load-side tie. Verify the percentages and the section against the adopted edition; the values have been stable, but the section organization has shifted between cycles.
| Input | Value |
|---|---|
| Busbar rating | 200 A |
| Main breaker | 200 A |
| 120 percent of busbar | 240 A |
| Room left for PV breaker | 40 A (240 minus 200) |
| Inverter at 125 percent into 40 A breaker | Up to about 32 A continuous |
Supply-side tap vs load-side breaker
There are two places to tie PV into a building, and the 120 percent rule only applies to one of them. A load-side connection lands on a breaker in a panelboard downstream of the main, and it lives under the busbar math above. A supply-side connection, also called a line-side tap, taps the service conductors between the utility meter and the main service disconnect, ahead of the main breaker.
Because a supply-side tap is ahead of the main, it is not on a busbar with the loads, so the 120 percent busbar calculation does not apply. That is exactly why it is the move when the load-side panel cannot take the breaker. The tradeoff is that you are working on the line side of the service disconnect, where the connection has to be made with listed tap or splice hardware suitable for line-side use, the tap conductors and their overcurrent protection have to be sized for the service, and the utility almost always has to approve it because the meter enclosure may be involved.
The supply-side rules sit in NEC 705 as well; confirm the section against the adopted edition. Reach for supply-side when the busbar math runs out, and price the added work and the utility coordination into the job from the start.
Grounding and bonding the array
A PV array has two grounding jobs: bonding all the metal so a fault has a path back, and connecting that system to earth. The module frames and the racking are bonded together and back to the equipment grounding conductor, often with washer-style bonding devices, WEEBs, that bite through the anodized aluminum to make metal-to-metal contact, or with listed lay-in lugs on a conductor. Anodizing is an insulator, so a bolt through a frame is not a bond by itself. The bonding device is what makes it one.
The equipment grounding conductor, the EGC, carries fault current back to trip the device, and it runs with the circuit conductors from the array through to the inverter and the service. The grounding electrode conductor, the GEC, ties the system to the building's grounding electrode system. Most modern PV is functionally ungrounded, with neither DC conductor bonded, and relies on the inverter's ground-fault detection rather than a solidly grounded DC conductor, but the equipment grounding and bonding still have to be continuous and correct.
The failure mode here is quiet. A frame that was never properly bonded sits at the array's potential during a fault, and the detection may not see it, so the array stays energized when it should not. Bond every frame, size the EGC for the circuit, and where the array grounding ties to the building system, coordinate it with the building's grounding rather than driving a separate, isolated rod. The grounding article, NEC 250, works alongside Article 690 here.
Placards and labeling
PV systems carry more required labeling than almost anything else an electrician installs, because the people who need the warnings, firefighters and the next electrician, did not see it go in. The placards mark the rapid shutdown switch and tell a responder the building has PV and how to shut it down, mark the DC and AC disconnects, identify the system at the service, and warn about the dual power sources at the interconnection. The rapid shutdown label in particular has a specified format and wording so a firefighter recognizes it on any building.
The exact labels, their wording, and where they go are spelled out across NEC 690 and 705, and the requirements have changed between editions, so pull the current list from the adopted code rather than reusing an old label set. This is the cheapest part of the job to get right and a common reason for a failed final, because the wiring can be perfect and the inspector still red-tags it for a missing or wrong placard.
DC arc-fault and ground-fault protection
DC PV circuits at 80 V or more between conductors have to be protected against arc faults, under NEC 690.11, with a listed PV arc-fault circuit interrupter. A DC series arc, from a loose connector or a damaged conductor, does not self-extinguish the way an AC arc does, and it is a leading ignition source in rooftop fires, so the detector has to sense the arc's signature and shut the inverter down. In most systems this function is built into the inverter's listing under UL 1741 rather than a separate box.
Ground-fault protection is the companion. A ground-fault detector senses current leaking from a current-carrying conductor to ground and interrupts the circuit before it becomes a shock or fire hazard, and it is also usually integrated into the listed inverter. Both functions are part of why the inverter listing matters: when you stay inside the listed equipment, you get the arc-fault and ground-fault protection the code requires. Confirm the 80 V threshold and the section numbers against the adopted edition.
Battery storage and NEC 706
Battery storage adds NEC Article 706 on top of everything in 690 and 705. The battery is its own source with its own disconnect, its own overcurrent protection, and listing requirements for the cells, the enclosure, and the energy storage system as a whole. On a grid-tied-with-storage job, the hybrid inverter and the battery have to coordinate so that during an outage the system can form its own grid and feed only the backed-up loads, never the utility.
The interconnection of storage has its own version of the source-connection rules, and lithium battery systems bring listing and sometimes location and ventilation requirements that vary by jurisdiction and product. The point for the PV wiring side is that adding a battery changes the disconnect count, the interconnection path, and the listing requirements, and it pulls Article 706 into the design. Confirm the storage rules and any local amendments with the AHJ and the equipment listing.
Commissioning and testing the array
Commissioning a PV system is where you prove the array you wired matches the array you designed, and it is the step that separates an energized system from a working one. Before the inverter ever closes, you check the DC side. Measure each string's open-circuit voltage and confirm it matches the calculated Voc for the conditions, confirm polarity on every string so nothing is backward, and verify the strings that should match each other do. A string reading half the expected voltage has a module wired wrong or a bad connection, and you find it now, not at the next maintenance visit.
Insulation resistance testing with a megohmmeter checks that the DC conductors are not leaking to ground through damaged insulation or a wet connector, which is a fault the open-circuit voltage check will not reveal. Then you test rapid shutdown for real: trip the initiator and confirm the array conductors come down to the controlled voltage in the required time, because that is the function a firefighter will rely on and the one the AHJ will ask you to demonstrate.
Record the string voltages, the polarity check, the insulation resistance results, and the rapid shutdown test. An energized array with a green light is not a commissioned system.
AHJ inspection and utility PTO
Two separate approvals stand between a finished array and a system that can run. The AHJ inspection confirms the installation meets the adopted code: the rapid shutdown function and its labeling, the disconnects, the grounding and bonding, the interconnection and the busbar calculation, and the conductor and overcurrent sizing. The utility's permission to operate, PTO, is a separate sign-off that the interconnection meets the utility's own requirements, and a system is not allowed to export until the utility says so, regardless of what the AHJ approved.
The order matters and the timelines are not yours to control. The AHJ inspection comes first, then the utility processes the interconnection and grants PTO, and the gap between them can be weeks. Tell the customer that up front, because an array that passed inspection but is sitting dark waiting on the utility is a support call you can avoid by setting the expectation early.
Commercial-rooftop and data-center PV
On large commercial and data-center rooftops, PV stops being a residential-scale exercise and the wiring decisions scale with it. Systems run at 1000 V DC to cut conductor size and string count across a big array, the strings combine through multiple combiners into large DC feeders, and the rapid shutdown question gets answered at array scale, often with a UL 3741 hazard control approach over per-module electronics on a clean, unshaded roof.
The interconnection on these systems is rarely a simple load-side breaker. The PV ties in at a switchboard or through a dedicated service, the supply-side and feeder-tap rules in NEC 705 come into play, and the coordination with the building's electrical design and the utility is a project in itself. The roof loading, the conduit routing across a membrane, and the weight of combiners and conduit also land back on the roofing side, which is why the racking and the electrical have to be coordinated from the start instead of stacked after the fact.
What to document
A PV system that nobody documented is a system the next electrician has to reverse-engineer off the roof. The record is what proves the string voltage was checked against the cold, that the busbar math was run, and that rapid shutdown was tested, not assumed.
| Side | Requirement | Note |
|---|---|---|
| DC | Cold-corrected string voltage under the equipment rating | Use the site low temp and the module Voc coefficient |
| DC | String fusing where more than two strings parallel | PV-listed DC fuses at the module series fuse rating |
| DC | PV wire or USE-2, rooftop-derated ampacity | Apply the 310.15 rooftop adder plus the 125 percent factor |
| Both | Rapid shutdown tested and labeled | Verify voltage and timing at the AHJ test |
| AC | Inverter output OCPD at 125 percent | Match it to the inverter label |
| AC | Interconnection method and busbar calc | Load-side 120 percent or a supply-side tap |
| Ground | Frame bonding and EGC/GEC | WEEB or lay-in lug, EGC sized to the circuit |
| Labels | Placards per NEC 690 and 705 | Pull the list from the adopted edition |
Common mistakes
- String voltage exceeding the inverter maximum or the system voltage limit at the coldest expected temperature.
- No rapid shutdown, or a system that was installed but never tested at the initiator.
- Breaking the 120 percent busbar rule on a load-side interconnection.
- Mixing MC4 connector brands across a mated pair.
- Pulling undersized or non-PV-rated DC wire, or skipping the rooftop temperature derate.
- Bolting module frames to the rack without a listed bonding device and calling it grounded.
- Missing or wrong placards and rapid shutdown labels.
- Using AC-rated fuses or disconnects on the DC side.
Field checklist
Want this checklist to run itself on every job — with photo proof and a signed record crews can hand the customer? That's FieldOS.
Standards and references
NEC Article 690 is the PV-specific part of NFPA 70, and it governs the array wiring, the maximum system voltage calculation at 690.7, the DC arc-fault protection at 690.11, and rapid shutdown at 690.12. Interconnection to the premises and the utility lives in NEC 705, including the load-side busbar rule commonly cited as the 120 percent rule under 705.12 and the supply-side connection rules in the same article. Battery storage brings in NEC 706, and grounding and bonding run through NEC 250 alongside 690.
On the listing side, inverters and interconnection equipment are listed to UL 1741, which is where the grid-support, arc-fault, and ground-fault functions are evaluated. PV hazard control systems for rapid shutdown are listed to UL 3741, and a system is only compliant as the evaluated combination. Conductor ampacity, the rooftop temperature adder, and the continuous-load factors come from NEC 310 and the general rules of the code.
Two hedges worth stating plainly. The section numbers, the voltage limits, the timing, the array-boundary distance, and the busbar percentages have all moved between code cycles, so the adopted edition and any local amendments control, not the numbers in any guide. And the utility's interconnection rules and the AHJ sit on top of the NEC, so confirm both early. Stress three things on every job: size the string for the cold, prove rapid shutdown works, and run the 120 percent calculation before you promise a system size.
Units, terms, and synonyms
PV wiring carries its own vocabulary, and the same idea shows up under different names across a datasheet, a plan set, and a spec.
Voc is open-circuit voltage and Isc is short-circuit current, both from the module nameplate. A string is modules in series; an array is the strings together. MLPE, module-level power electronics, covers microinverters and optimizers. PVRSS is a PV rapid shutdown system, and PVHCS is a PV hazard control system listed to UL 3741. PTO is permission to operate, the utility's go-ahead. The DC side is everything before the inverter; the AC side is everything after.
- Voc / Voc(cold)
- Open-circuit module voltage, and that voltage corrected up for the coldest expected temperature
- String / array
- Modules in series make a string; strings together make the array
- MLPE
- Module-level power electronics, meaning microinverters or power optimizers
- Rapid shutdown
- De-energizing array conductors to a safe voltage for responders, under NEC 690.12
- EGC / GEC
- Equipment grounding conductor and grounding electrode conductor
- PTO
- Permission to operate, the utility's authorization to energize and export
- PVHCS / UL 3741
- PV hazard control system listed to meet rapid shutdown as a whole evaluated array
FAQ
What is NEC 690 rapid shutdown?
NEC 690.12 rapid shutdown lets a firefighter de-energize a rooftop array's conductors from one switch. On initiation, conductors inside the array boundary drop to a controlled level and those outside fall to about 30 V within 30 seconds. Module-level electronics or a UL 3741 system meet it. Verify the limits against the adopted edition.
What is the difference between a string inverter and microinverters?
A string inverter converts the DC from series strings at one central box, so shade on one module drags the whole string. Microinverters convert at each module, isolating shade and mismatch and providing module-level rapid shutdown, at higher cost and with roof-level units to service. Match the choice to shading and roof complexity.
What is the 120% rule for solar interconnection?
The 120 percent rule limits load-side solar backfeed under NEC 705.12. With the PV breaker opposite the main, the main rating plus 125 percent of inverter output cannot exceed 120 percent of the busbar ampacity. A 200 A bar with a 200 A main leaves about 40 A for PV. Verify against the adopted edition.
Why does PV string voltage matter in cold weather?
Cold raises a module's open-circuit voltage, so a string sized at room temperature can exceed the inverter maximum and the system voltage limit on the coldest morning. NEC 690.7 makes you calculate maximum voltage at the lowest expected temperature using the module's coefficient. Size the string for the cold Voc, not the spec sheet.
Can I mix MC4 connectors from different brands?
No. Mate connectors of the same brand, or ones specifically listed as compatible. Two MC4-style connectors from different makers can click together and look seated while the contacts barely touch, and a high-resistance DC connection in full sun makes heat and starts fires. If pigtails and module leads differ, change the pigtails.
Do I need fuses for my PV strings?
With one or two parallel strings, a fault cannot push more than the modules safely carry, so fusing is often not required. Add a third string and the others can backfeed a fault beyond one string's rating, so each string needs a PV-listed DC fuse sized to the module's series fuse rating. Verify the count against the code.
Supply-side or load-side: which solar interconnection do I use?
A load-side connection lands on a breaker downstream of the main and must pass the 120 percent busbar calculation. A supply-side, or line-side, tap connects ahead of the main on the service conductors, so the busbar rule does not apply. Use supply-side when the busbar math runs out, and expect utility approval.
What wire is used for the DC side of a rooftop solar array?
Exposed single-conductor DC on a rooftop uses PV wire or USE-2, both rated for sun, wet, and rooftop heat. Standard building wire is not rated for that exposure. Size the conductor for the NEC rooftop-temperature adder plus the normal ambient correction, and apply the 125 percent factor for continuous PV source-circuit current.
What is permission to operate (PTO) and why is my solar system off?
Permission to operate is the utility's separate sign-off that your interconnection meets its requirements. A system can pass the AHJ inspection and still sit dark, because it cannot export until PTO is granted, which can take weeks after inspection. The inspection and the PTO are two different approvals on two timelines.
Does a PV string inverter system have energized DC in the building?
Yes. A string inverter runs energized DC conductors from the roof down to the inverter, which is why rapid shutdown exists. Microinverters convert at each module, so the conductors leaving the array are already AC and there is no high-voltage DC in the building. Optimizers still carry DC to a string inverter.
People also ask
Codes cited in this guide
This guide is written and reviewed against the published standards below. Always confirm the current adopted edition with the authority having jurisdiction.