Key Cutting and Code Machine Calibration Standard

Purpose

A code-cut key is only as good as the machine that cut it, and a code machine drifts in ways that feel and look identical to a correct cut. This procedure catches that drift before it reaches a customer's cylinder, especially on a keyed-alike or master batch where the same bitting gets cut a dozen times in a row without ever being compared against a key known to be right.

Without it: a carriage that has crept out of zero cuts an entire day's batch a few thousandths shallow, every key from that batch either binds going in or fails to seat, and the shop does not find out until callbacks stack up across a property where several doors were all cut from the same bad session.

Scope

Covers any bench code-cutting machine, manual or computerized, used to cut a key from a stated bitting or code rather than by duplicating an existing key: the daily zero check, full calibration when that check fails, cutter wheel and carriage condition, and batch verification for any run of three or more keys cut to the identical bitting.

Does not cover blank custody or chain of ownership for restricted stock, which the key blank and key control inventory SOP owns, and does not cover bitting design or the technique of combinating cylinders to a chosen code, which the master key system SOP and the keyed-alike system programming technique own. This procedure assumes a valid bitting has already been selected and asks only whether the machine cut it accurately.

Roles and responsibilities

Procedure

Step 1: Inspect the cutter wheel and vise before the first cut of the day. With the machine powered off, check the wheel by hand for chipped teeth or visible wobble, and confirm the key clamp seats a blank flush against its stop with no gap you can catch a feeler gauge under. Acceptance: wheel shows no chipped teeth and no play when rotated by hand, vise closes flush and holds a test blank without creep when you pull on it. Wrong looks like a stepped or torn cut face on a warm-up cut, which comes from a chipped tooth rather than a bad setting, or the last cut of a run reading shallow while the first read correctly, which comes from vise creep under repeated clamping. Stop rule: any chip, wobble, or vise creep found here, tag the machine out before running the daily check, not after. Hazard: never check wheel play or clear the clamp area with the machine energized; confirm the wheel has coasted to a complete stop and the guard is seated before your hand goes near it, and if the guard will not seat flush after inspection, the machine stays tagged out rather than run with the guard loose, per 29 CFR 1910.212's point-of-operation guarding requirement.

Step 2: Run the daily zero check against a certified check key before any customer code is cut. Cut the check key's known bitting on scrap stock, then measure every cut position with a key gauge against the platform's depth-and-space chart. Acceptance: every position reads within the tolerance stated on your machine's cut sheet for that platform. Wrong looks like one position reading true and an adjacent one reading a few thousandths shallow, which is carriage drift rather than a one-off miscut. Stop rule: any position outside stated tolerance, no customer key gets cut on this machine until step 3 clears it.

Step 3: On a failed check, run full calibration against the machine's factory alignment points. Zero the stylus or cam against the manufacturer's reference, reset the carriage stop, and confirm the depth increment per space matches the chart before cutting anything further. Re-run the check key cut and require two consecutive passing results, not one, before returning the machine to service. Acceptance: two consecutive check-key cuts both within tolerance. Wrong looks like one clean pass after adjustment; a single pass can be luck rather than a fixed correction, and the second pass is what tells the difference. Stop rule: three failed calibration attempts in a row, stop adjusting and pull the machine from service; log it out with the date and reason and call the manufacturer or a qualified service tech rather than continuing to guess at the setting.

Step 4: On any batch of three or more keys cut to an identical bitting, pull one key from the middle of the run for verification, not the first or last. Gauge it against the chart and, where a cylinder pinned to that bitting is on the bench, test it in the cylinder directly. Acceptance: the mid-batch key matches chart depths at every position and turns the test cylinder without binding. Wrong looks like the mid-batch key reading off from the first key cut in the same run, which is the carriage-creep pattern this step exists to catch mid-run rather than after delivery. Stop rule: any mid-batch deviation, stop cutting immediately, return to step 2's zero check before cutting another key on this bitting, and treat every key already cut in that run as unverified rather than assuming only the later ones drifted.

Step 5: Cut, deburr, and stamp each key before it leaves the bench. Deburr both faces of every cut with a fine file or brush; a burr feels fine to the touch but binds a plug with tight tolerances the same way a rushed hand-pinning job does. Test every key in a bench cylinder representative of that keyway before it goes in the batch bag. Acceptance: burr-free cut face, correct batch stamp, turns freely in the bench-test cylinder. Wrong looks like a key that turns with resistance on the bench; resistance on a fresh cut key is the machine or the burr, not the customer's cylinder. Stop rule: any key that does not turn freely gets destroyed and re-cut from a fresh blank; do not file a code-cut key to fit, because filing changes the bitting the record says you cut.

Step 6: Verify unverified keys from a stopped run before releasing any of them. Where step 4's stop rule fired, individually gauge and bench-test every key already cut in that run rather than assuming the first ones cut are safe. Acceptance: each key in the stopped run individually confirmed against the chart and a test cylinder. Wrong looks like releasing the keys cut before the failed mid-batch check on the assumption drift only affects later cuts; carriage creep does not announce when it started. Stop rule: any key in the stopped run that fails individual verification is destroyed and re-cut after step 3's recalibration clears the machine, not filed or released as-is.

Step 7: Log the result of every check, calibration, and batch on the calibration log. Record date, technician, check-key result, any calibration adjustment made, and the batch job number tied to each mid-batch verification. Acceptance: an entry exists for every day the machine was used to cut a code, with no gaps. Wrong looks like a batch job number on file with no matching daily-check entry for that date. Stop rule: a batch found with no corresponding daily check on record is treated as unverified regardless of how the keys tested afterward; the job gets flagged for the customer file and, on a restricted or master system, the affected cylinders get scheduled for a spot recheck.

Step 8: Replace wear components on schedule, not on failure. Track cutter wheel wear against the manufacturer's stated limit and replace before it reaches that limit rather than waiting for a chipped tooth to show up in step 1. Acceptance: wheel replacement logged against the manufacturer's interval, current wheel within its service life. Wrong looks like a wheel run well past its rated life because it "still looked fine." Stop rule: a wheel at or past its stated wear limit does not get one more day's use to finish a batch; replace it first, then run the daily check on the new wheel before cutting.

When the site does not match

No certified check key exists yet for a brand-new keyway the shop just started stocking: cut and independently verify a check key against the manufacturer's factory-sample bitting before treating any key from that platform as production work, and store that verified key as the new reference rather than reusing a customer's key for the purpose. A customer's original key is never an acceptable substitute for a certified check key, even in a pinch; original keys wear, and a worn original will pass a machine that is already drifting in the same direction the wear went. A rush job pressures skipping the mid-batch check on a large run: the check takes minutes, not the hours a stopped run costs once several doors are already installed; it does not get skipped for schedule pressure.

The record this produces

• The daily zero-check log: date, technician, check-key result, pass or fail
• The calibration log: date, what was adjusted, both confirming check-key results, technician
• The batch verification record: job number, bitting or system reference, mid-batch key result, individual verification results for any stopped-and-recovered run
• The out-of-service tag record: date tagged, reason, date cleared, who cleared it
• The wheel and component replacement log: install date, manufacturer interval, replacement date

These are read by the next technician deciding whether a machine is trustworthy for today's batch, by the shop owner reconciling a callback pattern against which machine cut a given job, and by a customer's own audit of a restricted or master system when a key or a door does not perform as expected.

Worked pass: a storage-unit batch that fails mid-run

Monday morning, a six-cylinder keyed-alike batch for a self-storage customer. Step 1's inspection passes: wheel true, vise holds flush. Step 2's daily zero check passes, all positions within tolerance against the check key.

Cutting begins. Keys one through three are cut and bagged in sequence. Step 4 fires at key four, pulled mid-batch: gauged against the chart, position three reads a few thousandths shallow, and it does not seat cleanly in the bench-test cylinder. The stop rule fires immediately; cutting on this bitting stops.

Step 2 is re-run. The zero check now fails on the same position, confirming drift rather than a one-off. Step 3's full calibration follows: the carriage stop had crept during an earlier warm-up run on an unrelated job that morning. Adjusted, re-zeroed, and two consecutive check-key cuts pass clean.

Step 6 applies before anything ships: keys one through three, cut before the failure was caught, are individually gauged and bench-tested rather than assumed good. Keys one and two pass clean. Key three fails the same way key four did, at the same position, which places the actual drift onset between keys two and three rather than at key four. Keys three and four are destroyed and re-cut on the recalibrated machine; keys five and six are then cut fresh and verified.

The batch that leaves the shop that afternoon is keys one and two from the original run plus four freshly cut and verified keys, all six individually confirmed rather than trusted by association with a passing mid-batch check. Cost of the stop: roughly forty minutes of diagnosis and recutting on a job quoted as a same-day turn. Cost of not stopping: several doors on one property cut from a drifting machine, discovered by the customer one lock at a time.

References

• 29 CFR 1910.212, Machinery and Machine Guarding, point-of-operation guarding for the cutter wheel
• Manufacturer's depth-and-space chart and cut-tolerance specification for the platform and code machine in use
• ALOA Security Professionals Association guidance on code-cutting equipment maintenance practice
• See related: Key Blank and Key Control Inventory SOP (blank custody, distinct from cutting accuracy), Commercial Master Key System Setup SOP, Keyed-Alike System Programming Technique howto