Exit Device and Panic Hardware Installation Technique

Why this matters

An exit device that looks right from across the room can still fail the one thing it exists to do: release the latch in a single motion, every time, under load. The egress compliance check measures that outcome with a force gauge after the fact. This article is the technique that gets you there in the first place, the hand skills the compliance check assumes you already have. Most panic hardware callbacks do not trace to a wrong device or a code violation; they trace to a rod cut a little short, a strike set a hair off the latch throw, or a coordinator that was never actually timed. Get the technique right and the compliance check is a formality. Get it wrong and you are back on that door in a month, on your own time, explaining to a manager why the exit they are legally required to keep operable now sticks.

Read the opening before you touch the device

A device mounts to a template, not to the door as it actually sits. Before you drill anything, check the opening itself: run a straightedge along the latch stile top to bottom and look for daylight, work the door on its hinges and feel for play or sag, and measure the frame gap at the head and at the latch side with the door closed. A door that has dropped a sixteenth of an inch on its hinges will take a perfectly installed device and turn it into a misaligned one, because the device is square to the door and the door is no longer square to the frame. Fix the door first: tighten or replace hinge screws into solid wood or a reinforced jamb, shim a sagging frame, or flag the opening as needing carpentry before hardware. Mounting a device to a door that will not hold still is the single most common reason a "good install" reads as a bad one two weeks later, once the door has settled again under daily traffic.

Mounting technique

Dry-fit the device against its template before drilling, and confirm the template's reference line runs parallel to the door edge, not just "close." On hollow metal doors, through-bolt into the reinforcement channel rather than the door skin; the skin alone will not hold a device against repeated impact, and a fastener biting only sheet steel backs out within a season. On wood-core and most aluminum storefront doors, use the manufacturer's specified screw and depth, not whatever happens to fit the hole in the truck bin.

Snug every fastener finger-tight first, all of them, before you fully torque any one. Then torque in a cross pattern working diagonally across the mounting points, the same sequence you would use on a wheel. Torquing sequentially end to end pulls the chassis into a slight twist as the far end draws down last, and that twist is invisible on the bench but shows up as uneven latch engagement once the device is loaded. If the touchpad or crossbar does not sit dead parallel to the door edge after final torque, back off and re-sequence rather than shimming the low corner; a shim compensates for the twist without curing it, and the twist comes back the first time someone leans hard on the bar.

Rod technique on vertical-rod devices

Cut vertical rods with the door in the closed position and the device fully engaged, measuring from the rod guide down to the point the coupling needs to reach, never by eye against the old rod. Cut square with a fine-tooth blade, then deburr the cut end on both the inside and outside edge before threading it into the coupling; a burred thread cross-threads on the first turn and gives you a rod that looks engaged but is only caught on the first pass of thread. Check full thread engagement against the manufacturer's cut sheet for that device, and never trust the last thread alone to hold; if the sheet is not on hand, engage the coupling until it bottoms, then back off only the amount needed for free rod travel.

Rod length is a moving target across a single door: too long and the top latch drags or holds the door fractionally off the frame, which reads later as a door that "sometimes doesn't latch"; too short and the top latch never reaches its strike at all. Cut a fraction long on the first pass, test, and trim rather than cutting to a calculated length and hoping. A rod you trim twice costs a couple of minutes; a rod cut short the first time costs a new rod and a second trip to the supplier.

Setting the latch-to-strike relationship

Before adjusting anything, read the keeper for witness marks. A keeper that shows a bright, centered burnish line is catching the latch cleanly; a mark biased to one edge tells you which way the strike needs to move before you touch a screwdriver, and a mark that runs across the full depth of the keeper rather than a clean line tells you the latch is dragging on entry, not just landing off-center.

Adjust horizontal position in the strike's mounting slots first, in small increments, testing the door close after each pass rather than chasing it in one large move. Once horizontal position is correct, set keeper depth so the latch seats with a small amount of float, enough that the door does not bind on temperature swings but not so much that the latch rattles under wind load. A keeper set bone-tight to the latch on a summer install will bind by the first cold snap, because the door contracts and the frame does not contract at the same rate.

Timing a pair of doors

On a pair, close the inactive leaf first and confirm it seats fully before the active leaf's latch reaches the strike; the coordinator bar exists to enforce that sequence mechanically so a fast-closing active leaf cannot beat the inactive leaf to the frame. With both leaves open, trip the coordinator by hand and watch where the astragal lands relative to the active leaf's latch path. If the astragal blocks the active leaf even slightly, the pair will read as "won't latch" on every fast close, and the fix is re-timing the coordinator's trip point, not forcing the active leaf harder. Test the sequence at a normal closing speed, not a slow hand-guided close; a coordinator that looks fine walked through by hand can still lose the timing race under a closer running at full speed.

Dogging and electrified functions

Before you set any dogging mechanism, check the door edge for a fire label, because fire exit hardware never carries a functioning dogging feature and a mechanically dogged fire door has effectively lost its latching function while it sits dogged. If the label is painted over, worn illegible, or there is no door schedule to confirm the rating, treat the leaf as rated and skip dogging entirely; that call belongs to the egress compliance check, and this technique note exists only so you never set a hex-key dog on a door before that check has happened. On a device that is confirmed clear for mechanical dogging, verify the dog engages and holds under the door's own closer tension, not just under hand pressure, and back the cam off fully before you leave unless the customer has specifically requested the door held open.

Before opening a device rail or a power supply enclosure to wire an electric latch retraction or an alarm switch, kill the circuit at its breaker or disconnect and verify dead at the terminals with a meter; a latch retraction motor still bites at low voltage, and a shorted rail arcs across a screwdriver tip. Complete the wiring, close the enclosure fully, and confirm no tool or lead is left resting against the rail before you restore power at the breaker. Test the function at the device itself once power is back, not just by watching the breaker reset without tripping; a breaker that holds tells you the circuit did not short, not that the device works.

Final function test

Test from both sides: push from the egress side with the closer running at full speed, not held slow by hand, and confirm the latch releases and the door swings clear. Then confirm the door closes and latches unaided, top latch included on a vertical-rod device, from a full open position. Where the door still fights you, isolate before you adjust: hold the closer arm and swing the door by hand to feel whether the drag is the closer, the hinges, or the device itself, because adjusting the device to compensate for a heavy closer just relocates the problem to the next service call. If isolating the closer means loosening or repositioning its arm, the arm is under spring tension the moment it is unpinned; ease it off slowly with a hand on the arm the whole time rather than backing off the fastener and letting it swing free, because a closer arm releasing suddenly under its own spring load will strike whatever is in its path.

Worked example

A small restaurant's back-of-house exit gets a new surface vertical rod device on a hollow metal door, replacing an old rim device the owner wants upgraded after a failed inspection on an unrelated finding. Reading the opening first: the door hangs true, no daylight along the stile, frame gap even top to bottom. No carpentry needed.

Mounting: dry-fit confirms the template line, through-bolts land in the reinforcement channel, all six fasteners snugged finger-tight, then torqued in a cross pattern. Crossbar sits parallel to the door edge on the first check.

Rod: cut long on the first pass at just under the measured length, tested, trimmed, deburred, threaded into the coupling until it bottoms then backed off for free travel. Full thread engagement confirmed against the cut sheet.

Latch-to-strike: keeper shows a bright centered burnish mark on the bottom latch, no adjustment needed there. The top latch, tested from full open, does not reach its strike; this is the one step that fails on the first attempt. The rod was measured with the door propped rather than fully closed, which put the strike contact point off by the amount the door still travels in its last few degrees of swing. It is re-measured correctly with the door held closed, a second rod is cut to the corrected length, deburred and threaded, and the top latch then seats clean from full open on the retest.

Dogging: fire label present and legible on the door edge, so the dog is left disengaged and the cam backed off fully.

Final test: pushed at full closer speed from inside, latch releases clean, door swings, both latches seat unaided from full open on the retest, checked three times. The closer arm was never disturbed during this job, so no isolation step was needed.

Verification

Before calling the job done: confirm the crossbar or pad sits parallel to the door edge under final torque, confirm every rod shows full thread engagement at its coupling, confirm the keeper shows a clean centered burnish mark rather than an edge-biased one, and run the door at full closer speed from both directions at least three times. Confirm the top latch seats unaided from a full open position on a vertical-rod device. On an electrified device, confirm the function at the device itself with power restored, not just a breaker that held. Photograph the finished install the same way you would document a compliance finding, because the next tech who opens this door in five years is working from your rod cut and your torque pattern whether you left notes or not.

References

• BHMA / ANSI A156.3 (Exit Devices), device listing and performance classification
• 29 CFR 1910.333(b)(2), de-energize and verify dead before opening an energized device rail or power supply enclosure
• Manufacturer installation templates and cut sheets (rod length, thread engagement and torque values are device-specific; confirm against the sheet for the model on the door)
• See related: Commercial Panic Bar / Exit Device Installation howto (device types, code overview, brand comparison), Panic Hardware and Egress Compliance Check SOP (the force and geometry measurements this technique is built to pass), Mortise vs Rim vs Surface Exit Device Retrofit Decision Matrix (device selection), Door Closer Adjustment and Service SOP