Key Duplication and Code Cutting Machine Technique

Why this matters

Key duplication is the highest-volume job on the bench and the one most likely to be done carelessly, because it feels trivial. A duplicate cut a few thousandths off the original does not jam on the bench where you can see it; it jams three weeks later when the customer's door has swollen slightly with humidity, and by then the customer is certain the key you cut is bad, not that the machine that cut it had drifted out of registration. The technique that prevents that callback is not in the blank you pick, it is in how the original and the blank are clamped, how worn the cutting wheel actually is, and whether you measured the finished cut against the chart instead of trusting the eye.

Registration is the whole job: both keys have to reference the same point

A duplicator works because the tracer riding the original's bittings and the cutter shaping the blank move in mechanical lockstep, offset by a fixed distance built into the machine. That offset is only correct if both keys are clamped so their shoulder stops, the ledge that seats against the lock face, sit at the identical reference point in their respective jaws. Seat the original fully to its shoulder, seat the blank fully to its shoulder, and confirm neither key can shift under the clamping pressure before the first cut. A key measured tip-to-bow rather than shoulder-to-bow, common on some European and import profiles, needs the tip registered instead, and mixing the two conventions on the same job shifts every single cut by the same fixed amount, uniformly, which is a harder fault to spot than a single bad cut because the key still looks internally consistent with itself.

A single chip of brass caught under either key in its jaw does the same thing at a smaller scale: it lifts that key by the chip's thickness, and every cut position comes out shifted by that exact amount. Clear the jaws by eye and by brush before every clamp, not on a weekly cleaning schedule, because the chip that causes this fault is generated by the job immediately before it.

Cutting wheel condition changes the depth, not just the finish

A machine's programmed offset between tracer and cutter assumes both wheels are the diameter and profile they shipped with. A cutting wheel worn thin, or worn to a rounded rather than a sharp V-profile, removes a slightly different amount of material at the same tracer position than the machine's geometry assumes, and the error runs the same direction on every cut, typically shallow, because a dulled wheel rides slightly higher before it bites. The tell arrives before the depth error is measurable: fresh brass cuts as a clean curled ribbon, and a wheel starting to dull produces fine powder or short chips instead of ribbon. Treat that change in chip character as the signal to replace the wheel, not the manufacturer's stated service interval alone, because two shops running different daily volumes wear a wheel on very different clocks.

Check tracer-to-cutter zero on a schedule, not only when something goes wrong

Keep one known-good key, cut once and measured against the chart, as a standing calibration reference separate from any customer's original. Run it through the duplicator on a fixed cadence, weekly in a moderate-volume shop, and measure the result the same way you would a customer job: calipers against the chart at the end positions. A calibration key that comes back reading true confirms the tracer and cutter are still sitting in the same relationship they were when the machine was last serviced. One that comes back drifted, even by an amount too small to have caused a callback yet, is the earliest available warning that the machine needs adjustment, well before the drift is large enough to jam a customer's lock. Log the result each time: a drift that grows gradually across several weekly checks points to wear, the cutting wheel or a loosening clamp mechanism, while a drift that appears suddenly between two checks points to a knock or a dropped machine, a different repair entirely.

Code cutting is a different machine and a different failure mode

Where there is no original to trace, a code-cutting machine takes the depth-and-space sequence directly from the manufacturer's code sheet and cuts without ever touching a reference key. The first failure point is entering a code against the wrong series: two keyway profiles can look nearly identical at a glance while belonging to different code series with different depth increments, and a code entered against the wrong series produces a key that inserts into the correct keyway but is wrong at every position, because the depths mean something different in each series. Confirm the series against the actual keyway profile in hand, not against the customer's verbal description of the brand, before keying in a single number. The second failure point is spacing rather than depth: the code sets both a depth and a position along the blade for each cut, and a spacing error puts a correct depth in the wrong place on the key, which is a different and often subtler symptom than a depth error, a key that seats and almost turns rather than one that will not seat at all.

Match the machine to the key profile, not just the keyway

A standard edge-cut duplicator, tracer and cutter both riding the blade's edge, is built for the pin tumbler profiles that make up most residential and light-commercial work. Dimple keys, where the cuts sit as points on the flat of the blade rather than along the edge, and most high-security sidebar profiles need a dedicated tracer-and-cutter geometry or a laser-guided milling machine built for that cut pattern. Running a dimple original through a standard edge duplicator either fails to register the cuts at all or, worse, drags the tracer across contours it was never built to follow and damages the tracer tip. Confirm the cut style against the machine's rated capability before clamping anything, not after the tracer has already refused to track cleanly.

Verify the cut against the chart, not against the key's own symmetry

Before handing a duplicate or a code-cut key to a customer, measure at minimum the first and last cut positions with calipers or a key gauge against the depth-and-space chart for that keyway, not just by comparing the new key visually to the original. A drifted duplicator typically drifts uniformly across the whole key, which means the new key can look and feel internally consistent, matching the original at every position relative to itself, while every position is simultaneously off from the chart by the same fixed amount. Checking the two end positions against an independent reference catches that uniform drift; checking only the middle of the key, or only comparing the new key to the old one, does not, because a uniformly shifted key passes that comparison every time.

Machine hazards are mechanical, not incidental

The cutting wheel is a rotating blade close to your hand on every cut, and it throws brass chips as it cuts. Keep the guard in place, feed the key through the vise rather than steadying it with your fingers near the wheel, and wear eye protection for every cut, not only when something looks unusual. If a chip or a burr binds the tracer or cutter mid-cut, switch off and unplug the machine before clearing it; do not rely on the switch alone, because a foot-pedal or momentary-contact control on some duplicators can re-energize the wheel unexpectedly while your hand is inside the guard clearing debris. A cutting wheel coasts for several seconds after power is cut, and it carries the same laceration risk while coasting as it does while driven; wait for it to visibly stop before any hand goes near the cutting area, whatever the switch position shows.

Worked example: the bench log that is the verification, not a separate step

A customer brings in a worn Schlage-profile key with no visible defect but a request for two spares. The technician clamps the original and a matching blank, both seated to their shoulder stops, jaws checked and clear.

Bench log, position 1 through 6, chart values for this keyway: 4, 2, 6, 3, 1, 5. First duplicate cut, measured with calipers at position 1 and position 6 before the key leaves the vise: position 1 reads 4, matching the chart. Position 6 reads shallow, closer to a 4 than the charted 5, a full increment off.

The technician does not hand over the key on the strength of position 1 passing. Because the drift shows only at position 6, the far end of the key from the shoulder stop, the cause is traced to the blank's clamp: a slight rock in the jaw at the tip end, invisible until the cutter reached the position furthest from the fixed shoulder reference, consistent with debris the eye check had missed rather than a global machine offset, which would have shown at position 1 as well. The jaw is cleared a second time, the blank reclamped and confirmed seated with no play at the tip, and the key is recut.

Second bench log entry: position 1 reads 4, position 6 now reads 5, both matching chart. The two spares are cut from this now-verified setup and handed over with the confirmed measurements noted on the ticket rather than "cut and tested by eye."

References

• Manufacturer service and calibration documentation for the specific duplicator and code-cutting machine in use
• 29 CFR 1910.212, general machine guarding, for the guard and feed-point requirements on a rotating cutting wheel
• 29 CFR 1910.133 for eye and face protection against thrown chips during cutting
• See related: Key Blanks and Duplication Reference (blank identification and keyway families), Key Blank and Key Control Inventory SOP (custody of restricted blanks, separate from cutting technique)