Roof Penetration and Flashing Technique for Solar Mounts

Why this matters

Every array on a sloped roof needs a way through the shingles to structural framing, and every one of those holes is a leak waiting to happen if the flashing is wrong. A failed standoff flashing does not announce itself at handover. It shows up one to five winters later as a stain on a bedroom ceiling, usually blamed on the roof rather than the array sitting on top of it, and the fix is a tear-off under twenty-plus modules that were never designed to come off again. The other failure mode is faster and worse: a lag that never reached the rafter holds the mount through a calm spring and lets go in the first real windstorm, taking part of the array with it. Both failures trace to the same handful of moves done wrong, and both are fully preventable at the drill.

Work the roof under its own hazard, not the array's

Before the first pilot hole, this is roof work under the job's fall protection plan, tied to activity classification, typically the six-foot trigger under 29 CFR 1926 Subpart M for construction-classified roof work. A roof surface in full sun also runs hot enough to burn skin left against it for more than a few seconds, so knee pads and gloves are not comfort items on this job. Where the penetration being cut sits within reach of an already-energized string, a repair or an add-on to an existing array, treat every nearby conductor as live under the energized DC array safety standard before a drill touches the deck; a rapid shutdown switch reduces conductor voltage outside the array boundary, it does not make the roof safe to drill into blind.

Confirm the rafter before you commit a hole

Drill a pilot hole at the marked location and pull the bit back out with the shavings still on the flutes. Rafter wood comes out as long, fibrous shavings; sheathing alone comes out as fine, powdery dust. This is the same test the crew uses when resetting mounts after a reroof, and it holds here for the same reason: a lag with full thread engagement in dimensional lumber and one seated only in three-eighths of an inch of OSB look identical from the roof surface, and only one of them holds a sustained wind-uplift load.

If the shaving test comes back dust only, the hole is not driven. Cap it and move to the nearest point that still falls inside the racking manufacturer's engineering letter for this site, its stated maximum span between attachments, not a distance you eyeball. A standoff moved past that span to chase a rafter has traded one failure mode for another; now it holds, but the rail between it and its neighbor is unsupported past its rated length. If no point within the allowable span lands on structural framing, the fix is blocking between rafters to an engineer-approved detail, not a longer lag driven deeper into sheathing hoping to find something solid.

Set the penetration perpendicular to the roof plane

Drill square to the roof surface, not plumb to the ground. On any pitch steeper than a few degrees the two are different angles, and a bit walked in at the wrong one bores an oval rather than a round hole. An oval hole does three things wrong at once: it gives the flashing boot a gap to bridge instead of a clean seat, it changes the lag's effective embedment depth from what the pilot test just confirmed, and it concentrates bending stress on the shaft where it exits the deck. Check the angle against the roof surface with a rafter square or angle finder held to the shingles, not against a level, before the bit touches the deck.

Drive the standoff to the letter's numbers, not by feel

Set the lag diameter and minimum thread embedment from the racking manufacturer's engineering letter for this specific rail and roof, typically a stainless lag of five-sixteenths or three-eighths inch driven at least two and a half inches into the rafter on a standard residential mount, more where the letter's wind and snow inputs call for it. Drive to hand-tight with an impact driver, then finish to the torque value in that same manual using a calibrated torque wrench, and paint-mark the fastener head once it is there.

Do not chase a loose feel by cranking harder. A rafter that strips under the wrench has already lost thread engagement, and more torque past that point does not add holding power, it just spins in a hole that is now oversized. Stop, back the lag out, and either move to an adjacent rafter within the letter's allowable span or step up to the next fastener size the manual permits for a stripped hole. A lag that never reaches its stated torque is logged as a miss and corrected the same visit, not left for the annual inspection to find.

Weave the flashing into the shingle courses, not onto them

Lift the shingle tab above the penetration and slide the flashing's upper edge underneath it, so the flashing sits over the course below and under the course above, the same lapped order the shingles above it already follow. This is the primary seal. Water runs downhill over each layer the way it runs off a fish's scales, and a flashing installed backward, with the course above lapped over the flashing instead of under it, catches water at the top edge and channels it straight at the penetration it was meant to protect.

Sealant is the secondary line, not the primary one, and it goes above the flashing, along the top edge and over the exposed edge of the base plate, never underneath it. A bead of caulk run along the flashing's bottom edge to seal it down traps whatever water gets past the shingle lap against the deck instead of letting it drain, which is the single most common flashing defect on this shelf and the reason a leaking mount so often gets blamed on the roofing rather than the install. Extend the flashing at least six inches upslope of the standoff, more where the flashing kit for that base plate specifies a wider footprint, so the lapped area under the course above has margin on every side.

If the lifted tab cracks or its seal strip tears while you work under it, that tab does not go back down as-is. A cracked tab over a flashing is a leak path that stays invisible until it fails, because a rack of modules covers it for the next two decades. Replace the tab or set a compatible dab of roofing cement at the crack before closing up, and note it on the mount's photo log.

Cap every exposed thread

Any lag thread left exposed above the flashing or the base plate corrodes and wicks water down its own shaft. Seal exposed threads with the compatible sealant for the fastener and flashing metals in use, stainless with stainless-rated caulking on the assembly, and check that the standoff cap or boot fully covers the shaft before moving to the next point. In coastal or high-chloride environments this step is not cosmetic; dissimilar-metal corrosion at an exposed thread runs faster there than the shingle wearing out around it, and a thread left bare is the fastest path a mount has to losing its rated capacity years ahead of the roof covering itself.

Verify before the row goes under a rail

Before rails or modules cover the work, lift the course above each flashing one more time and confirm it sits flat with no gap, no daylight along either side of the flashing, and no fastener visible. Photograph every mount from an angle that shows the flashing lap, the same practice the shop already uses for warranty documentation, because this is the last point in the install where anyone will ever see this joint again. Where a roof section is new to the crew, a garden-hose test from downslope for several minutes with someone watching the underside from the attic is worth the extra twenty minutes before the row is loaded with modules; it is not worth it after.

A worked pass, including the miss

Architectural asphalt shingle roof, 2x8 rafters at 24 inches on center, engineering letter calling for a five-sixteenths inch stainless lag at a minimum two and a half inch embedment and a maximum four-foot span.

First standoff of the row: pilot hole at the marked point returns fine dust only. No lag driven. The mark sits four inches off the nearest rafter centerline, more than the manual's foot-slot adjustment can absorb in place, so the whole standoff shifts one and a half inches along the rail, still inside the four-foot span to its neighbor. The new pilot hole returns long fibrous shavings. The tab above is lifted, the flashing's upper edge slid under the course above and over the course below, penetration drilled perpendicular to the roof plane and checked with an angle finder, lag driven to hand-tight then to the manual's torque spec and paint-marked. Sealant runs along the flashing's top edge and the base plate's exposed lip only. The lifted tab reseats flat with its seal strip intact.

Second standoff, four feet upslope: pilot returns shavings on the first try, torque confirmed, flashing woven and sealed the same way. Before the rail goes on, both flashings are lifted and checked, flat, no daylight, no exposed thread, both photographed.

References

• See related: the racking and mounting reference for roof-type flashing systems and manufacturer standoff types.
• See related: the roof condition assessment SOP for the framing survey this attachment layout is built from.
• See related: the energized DC array safety standard for work near an already-energized string during a repair or add-on.
• IRC Section R907 for reroofing and roof-covering modifications.
• Racking manufacturer engineering letter and flashing kit installation instructions for the specific rail, base plate and roof covering in use.