Grounding and Bonding Technique for a Residential Array

Why this matters

A module frame that is mechanically clamped to a rail and a module frame that is electrically bonded to the grounding system are not automatically the same thing, and the failure is invisible from the ground and often invisible from the roof too. An integrated grounding clamp that seats crooked rides on top of the module's anodized coating instead of piercing it, holds the module exactly as securely as a correctly seated one, and bonds nothing. This article is the field technique for making every bond in a residential array a verified bond rather than an assumed one, from the module frame to the panel, including the equipment grounding conductor connections a new EV charger circuit adds to that same panel.

What "bonded" actually requires at a module frame

Anodizing is an electrically insulating oxide layer applied to aluminum for corrosion resistance, and it is not conductive. A grounding path between a module frame and the racking depends on something physically breaking through that layer to reach bare aluminum underneath, not on the two metal parts simply touching. This library's grounding electrode conductor reference covers the code path the bond ultimately connects to; this article covers what has to happen at the clamp itself for that connection to exist at all.

Installing an integrated grounding clamp correctly

Seat the clamp square across both module frames it bridges, not skewed, and torque it to the range the clamp manufacturer specifies, say a torque range on the order of 90 to 120 inch-pounds for illustration since the actual figure varies by clamp model and must come from that clamp's own instructions. A clamp under-torqued rides its teeth against the anodizing without fully piercing it; a clamp installed skewed can seat its teeth against one frame correctly and miss full penetration on the other, since the piercing action depends on the teeth meeting the frame surface squarely. Confirm the clamp manufacturer's listing covers the specific module frame profile you are working with; a clamp listed and tested against one frame manufacturer's anodizing thickness and alloy is not automatically listed for another's, even when the clamp physically fits.

Continuity: the one gate, and reading it in the field

There is one test that resolves whether a given clamp bonded or not, and it is the same test regardless of how the clamp looks: continuity between the module frame and a known-good ground reference, most practically the array's equipment grounding conductor at the nearest junction box, read with a continuity-rated multimeter or a low-resistance ohmmeter. Run this check clamp by clamp as the array goes up, not as a single pass at the end. A tech who waits until the last module is set and then finds one frame in the middle of the array reading open has to retrace the whole string of connections to find which of a dozen clamps is the failure, instead of catching it in the thirty seconds after that one clamp was installed.

Case A: a bond that passes

A clamp is seated square across two module frames, torqued within the manufacturer's range, and the continuity check between the frame and the array's equipment grounding conductor reads a closed circuit at a resistance consistent with a solid metal-to-metal path. Visually this clamp looks identical to every other clamp on the roof: same brand, same torque mark, same position. The only thing that distinguishes it as correct is the meter reading taken at install.

Case B: a bond that fails, and why it looks fine to the eye

A second clamp, installed on the same roof an hour later by the same tech using the same tool, is seated very slightly skewed, close enough to square that nothing about it looks wrong on inspection. The teeth on one side pierce the anodizing cleanly; the teeth on the other side ride on top of it, held in place mechanically by the clamp's clamping force but making no electrical contact with bare aluminum. Torque reads correctly on the wrench because torque measures the fastener's mechanical load, not whether the teeth broke through a coating underneath it. The continuity check between this frame and the equipment grounding conductor reads open. Nothing about a visual inspection, a torque check, or a tug on the module would have caught this; only the continuity reading does, which is why the check runs on every clamp rather than on a sample.

The fix is to back the clamp off, reseat it square, and re-torque, then re-run the continuity check rather than assuming a re-seat fixed it. And if the second check still does not hold: inspect the frame surface itself for enough tooth damage, from a clamp seated and reseated more than once at the same spot, that a fresh contact point on the frame is needed rather than another attempt at the same worn one.

Bonding the rail splices themselves

Where two rail sections join end to end, do not assume the mechanical splice bar that holds them in line also bonds them electrically. A splice bar seated against anodized rail faces has the identical problem a mis-seated module clamp has: it holds the joint together mechanically while the anodizing keeps it from being a reliable current path. Install the listed bonding hardware the rail manufacturer specifies for that splice, a bonding clip or jumper rated for the rail profile, and run the same continuity check across the joint that you run at every module clamp, rail to rail, before you trust the splice to carry a bond past it. A rail run with ten module clamps that all pass their individual continuity checks can still present an open bond overall if one splice in the middle of the run was never bonded, because the clamps only prove each module is tied to its own rail section, not that the rail sections are tied to each other.

Routing and terminating the equipment grounding conductor

Where the shop's practice is to run a standalone equipment grounding conductor alongside the source-circuit conductors, a common choice because it gives an inspector something visible to trace even though the rail system itself may be listed to serve as the bonding path, route that conductor along the rail, secured at the same intervals the source-circuit whips are secured, and land it at each rail section through a listed lug rather than looping it under a clamp screw that was not designed as a termination point. At the junction box or disconnect, size the lug to the conductor's actual gauge; forcing an oversized conductor into an undersized lug by half-mooning the strands to fit reduces the contact area to less than the conductor's full cross-section and gives you a joint that reads fine on a quick continuity check today and loosens under thermal cycling over the following seasons. Torque every lug to the value printed on the equipment it is landing in, not to a hand-tight-plus-a-turn habit; a torque value on a lug label exists because that specific terminal was tested at that specific clamping force.

On a service call rather than a new install, inspect an existing equipment grounding conductor's lug connections for corrosion or a green-tinged oxide film before you reuse them, particularly on a coastal or high-humidity site. A lug that visually still grips the conductor can carry a heavily oxidized contact surface underneath, and oxidation reads as measurable added resistance on a continuity check even when the mechanical connection looks intact; where a check on an existing termination reads noticeably different from a freshly landed one nearby, reterminate it rather than accepting a visual pass.

The EV charger tie-in: one panel, one bond point

Adding an EV charger circuit to a panel that already carries the PV interconnection breaker raises one grounding question worth checking explicitly: does the new circuit's equipment grounding conductor land on the same bonded ground bus the PV interconnection's conductor already lands on, or did the retrofit add a second ground bar that is not solidly bonded to the first. Two ground buses in one panel that are not tied together present a fault current with two paths of different impedance instead of one, the same class of problem Case B demonstrates at a module frame, just relocated to the panel. Verify it with the same gate: a continuity check between the new circuit's grounding conductor and the existing PV grounding conductor at a shared reference point, not a visual check that both terminals are labeled ground.

Where the EV charger circuit instead feeds a detached structure, such as a garage that is not part of the same building the array sits on, that structure needs its own grounding electrode system under NEC 250.32 in the edition your AHJ has adopted. This is a different rule than the one this library's grounding electrode conductor reference states for the array itself, that a PV system mounted on a building with an existing service electrode does not need a separate electrode of its own. Do not carry that logic over to a fed detached structure; a structure supplied by a feeder is its own case under its own code section, and treating it as exempt because the array on the main building was exempt is exactly the kind of same-looking-rule-applied-past-its-condition error that shows up as a failed inspection months later.

Verifying the finished bonding path

Before the array is closed up, confirm continuity end to end: every module frame to the array's equipment grounding conductor, that conductor through every rail splice and lug to the junction box, and from there through to the panel's grounding bus, including any new EV charger circuit sharing that bus. Where the design relies partly on the rail system's own listing to carry the bond, verify that listing against the installed rail model and the module frame it touches, since a listing covers a specific combination and not aluminum racking in general. A finished array with every module producing normally and every connector properly mated can still carry an open bond on one frame that nobody will notice until a fault current has nowhere else to go; the continuity check is what stands in for the fault this bonding path exists to handle safely, run once in advance rather than found out once in an emergency.

References

• NEC 690.43, PV Equipment Grounding, and NEC 250.122, equipment grounding conductor sizing, in the edition your AHJ has adopted
• NEC 250.32, grounding and bonding of a separately fed structure, for a detached-structure EV charger circuit
• Grounding clamp manufacturer's listing and torque specification for the specific module frame in use
• See related: the PV System Grounding Electrode Conductor per NEC 690.47 reference, for the code path this technique's bonds connect to
• See related: the Solar Racking and Roof Mounting reference, for rail hardware and mounting-system context