CNC Machining Center Step: 6 Steps to Set Tool Length
Setting the step of the knife on a CNC machining center means finding where each tool tip sits in machine coordinates and storing that number so the control can correct for it. This page walks through the full procedure: zero reference, touch-off, offset entry, verification, and the checks that stop a 0.2 mm error from scrapping a 4,000 mm part. Written for machinists and process engineers who already run the machine and just want the setup to repeat.
Key takeaways
What the CNC machining center step of knife controls
On a vertical machining center the spindle nose is a fixed point in machine coordinates. The tool tip hangs below it by a distance that changes with every tool you load. That distance is the step of the knife. The control adds it to the programmed Z value, so the same G-code block cuts the same depth whether the tool is a 50 mm face mill or a 3 mm end mill.
Get it wrong and the error shows up as a uniform depth shift across the whole part. A 0.10 mm mistake in the offset moves every floor, pocket, and shoulder by 0.10 mm in the same direction. That is different from tool wear, which drifts during the cut, and different from thermal growth, which changes slowly over hours.
The offset lives in the tool offset table, usually under the wear or geometry column. Geometry holds the nominal length you measured at setup. Wear holds the small correction you add as the edge dulls. Keep the two separate. Mixing them makes the next setup unpredictable because you no longer know which part of the number came from measurement.
On a horizontal machining center with a rotary table, the same logic applies through the B axis, but the reference surface has to be re-established after any pallet change. On a mill-turn center the offset usually includes both the tool length and the turret station position, so read the machine manual before you assume the number transfers.
Checks to run before you touch off any tool
Clean the taper and the tool holder before the tool goes in the spindle. A chip 0.02 mm thick between the taper and the spindle face becomes a 0.02 mm Z error that no amount of careful touch-off will remove. Wipe the taper with a lint-free cloth and inspect the holder for dents.
Confirm the tool number in the control matches the physical pocket. On a 24-station magazine it is easy to load tool 12 into pocket 14 after a changeover. If the control calls T12 but pocket 14 spins into position, the offset you just measured belongs to the wrong tool.
Set the work zero first. Touch off X and Y, then Z on a known surface such as the top of the vise jaw or a gauge block. If the work zero is set after the tool offsets, every offset needs to be re-measured because the reference moved. The order matters more than the method.
Let the spindle run at cutting speed for 10–15 minutes before the final measurement on any job with a tolerance tighter than ±0.02 mm. The spindle and housing reach a steady temperature and the nose position stabilizes. Measuring cold and cutting warm is a common source of a slow Z drift in the first hour.
Which touch-off method fits your part
A 0.1 mm shim and a slow jog is accurate enough for most 3-axis work at Ra 1.6–3.2 μm and tolerances around ±0.05 mm. You feel the drag, note the machine Z, and subtract the shim thickness. It costs nothing and it works on any surface, including rough stock.
A dial indicator on a magnetic base gives better repeatability, roughly ±0.005 mm on a flat reference. Use it when the part has a tight floor depth or several tools that must match each other. The indicator tip must sit on a clean, flat surface, not on a scale or a rust patch.
A tool presetter moves the measurement off the machine. You get a printed length for each holder, and the operator types it in. This is the right choice for a shop running 127 machines where setup time is the bottleneck. The trade-off is that the presetter reading only matches the machine if the spindle taper is clean and the holder seats the same way.
An automatic tool length sensor inside the machine is the most repeatable option, typically within a few microns. It also catches a broken tool before the next block runs. The catch is that the sensor stylus wears and drifts. Check it against a gauge block every few weeks.
On 5-axis work with a tilting head, none of these methods is enough on its own. The tool tip position depends on the rotary axes, so you need to verify the offset at two or three tilt angles before trusting it. A single Z touch-off on a 5-axis machine can hide a 0.05 mm error that only appears at 45°.
Step by step: setting the tool length offset
Run these in order. Skipping step 4 is the most common cause of a scrapped first part.
- 1Step 1 – Establish the work zero in ZTouch the reference surface with a known tool or a probe. Set G54 Z to that surface. Write the value on the setup sheet. Do not change it after this point.
- 2Step 2 – Load and identify the toolInsert the holder, confirm the pocket number matches the tool number in the control, and call the tool in MDI. Watch the magazine index to the right pocket.
- 3Step 3 – Touch off the tool tipJog to within 1 mm of the reference, then switch to 0.01 mm or 0.001 mm increments. With a 0.1 mm shim, stop when the shim drags; note the machine Z and subtract 0.1 mm.
- 4Step 4 – Enter the offset and verify the signType the value into the geometry column for that tool. Check the sign: on most controls a longer tool needs a more negative Z offset. Run the tool to Z0 in air and confirm the tip sits on the reference surface.
- 5Step 5 – Repeat for every tool in the programTouch each tool to the same surface. Keep the reference identical. If a tool cannot reach the surface, use a gauge block of known height and subtract it from every reading.
- 6Step 6 – Cut a test feature and measureFace a small area or cut a 2 mm deep pocket, then measure the depth with a micrometer or height gauge. Adjust the geometry offset by the difference if it exceeds 0.01 mm.
- 7Step 7 – Record and lock the setupLog tool number, offset value, reference surface, and time. Save the offset table to the machine if it supports it, so the next run starts from a known state.
Touch-off method comparison
Pick the method that matches your tolerance and batch size.
| Method | Repeatability | Best for | Watch out for |
|---|---|---|---|
| 0.1 mm shim | ±0.03 mm | 3-axis rough and semi-finish work | Feeling the drag varies by operator |
| Dial indicator | ±0.005 mm | Tight floor depths, matched tools | Tip must sit on a clean flat surface |
| Offline presetter | ±0.01 mm | High-mix shops, fast changeover | Only matches if the taper is clean |
| In-machine sensor | ±0.003 mm | Lights-out and long runs | Stylus wear and drift over weeks |
| Probe on spindle | ±0.005 mm | Parts with complex datums | Probe length offset must be current |
The setup is only as good as its verification
Touch off every tool to one reference, verify in air before the first cut, and log the numbers. That single discipline removes most first-part scrap on a CNC machining center step of knife setup.
Common questions on the step of the knife
Why does my first part come out 0.2 mm too deep every time?
The most common cause is a work zero that was set after the tool offsets. If G54 Z moves, every offset is wrong by the same amount in the same direction.
Check the order: set work zero first, then touch off tools. Also confirm the shim thickness is subtracted. Forgetting the 0.1 mm shim on five tools adds up to a consistent 0.1 mm error.
How often should tool length offsets be re-checked?
Re-check after any tool change, after a crash, and after the machine has been idle for more than a shift. On a running job, check the tool that does the critical depth once per shift.
For a 5-axis job, re-verify at the tilt angles you actually use. A single vertical check does not cover a tilted head.
Can I copy an offset from one machine to another?
No. The offset is relative to that machine's spindle nose and work zero. Two machines with the same model still differ by the taper and the zero setting.
If you have an offline presetter, use it to generate a nominal length, then verify it on the machine before the first cut.
What causes a slow Z drift during a long run?
Thermal growth in the spindle and the bed is the usual cause, typically 0.01–0.03 mm over the first hour. Let the machine warm up at cutting speed before the final measurement.
Tool wear adds a second, smaller drift. Track it in the wear column rather than the geometry column so the two effects stay separate.
Does the shim method work on a 5-axis machine?
It works for the vertical reference, but it does not capture the error that appears when the head tilts. The tool tip moves along a different vector once the rotary axes turn.
For 5-axis work, verify the offset at two or three tilt angles against a gauge block or a probe. If the readings disagree by more than 0.02 mm, check the rotary axis center and the tool holder runout.
What tolerance can I hold after a careful setup?
On a well-maintained machine with a clean taper and a verified offset, ±0.005 mm is achievable on critical features, with surface finish in the Ra 0.8–1.6 μm range.
That assumes the material is stable and the cut is not dominated by tool deflection. A long 3 mm end mill in aluminium will deflect far more than the offset error.
Send your drawing, get a setup-ready quote
Upload the part and we return a quotation with free DFM analysis within 12 hours. Tool offsets and fixtures are planned before the first chip.
12-hour quoteNo minimum order quantity100% inspection before shipment
More setup notes from the shop floor
We publish setup notes, tooling trials and inspection data from the factory floor.
