Seven Ways the User for CNC Machining Sets Tool Offsets
Tool setting decides whether a program runs to ±0.005 mm or scraps the first part. This guide walks through seven setting methods used on the shop floor, the tolerance each one can hold, and when a given method is the wrong choice. Written for machinists and process engineers who have to pick a method before the spindle starts.

In this article
- 1
- 2
- 3
- 4
- 5
- 6
- 7
Key takeaways
Bare Touch-Off: The Baseline Every User for CNC Machining Learns First
Bare touch-off means jogging the tool down until it touches the workpiece or a gauge block, then zeroing the Z axis. It costs nothing and needs no extra hardware. The spindle stops, the operator brings the tool down in 0.01 mm increments, and the moment a chip or a curl of aluminium appears under the flute, the axis is zeroed.
Repeatability is the problem. Feel, spindle speed, and how clean the tool tip is all change the number. On a 12 mm flat end mill touching a milled surface, two operators will land about 0.05 mm apart. On a long 3 mm drill, the spread can reach 0.1 mm because the tip deflects before it cuts.
Use bare touch-off for roughing, for stock that will be removed anyway, and for the first setup on a casting where the datum is still 2 mm of skin. Never use it to set the final Z on a 0.05 mm tolerance bore. The error goes into the control and stays there for the whole run.
One habit keeps it honest. Always touch on the same feature, always with the spindle stopped, and always after wiping the surface with a rag. If the surface has a burr from the previous pass, take a light 0.2 mm facing cut first. You are setting to the surface you actually cut, not the one you planned.
Paper Shim and Dial Indicator Methods
The paper shim method inserts a 0.05 mm feeler or a strip of cigarette paper between the tool and the workpiece, then jogs down until the paper drags. Subtract the paper thickness and you have your offset. It works well on small tools that would shatter if touched to metal, and on parts where a witness mark on the face would fail inspection.
Accuracy sits around ±0.03 mm. The paper compresses, the drag feel varies between operators, and a damp workshop changes the paper. It is a good compromise for second operations on a part that is already finished on one face and must not be scratched.
A dial indicator is the better tool when you need a real number. Mount a 0.01 mm indicator on a magnetic base, sweep the tool tip or the spindle taper, and read the deflection. Sweeping a 10 mm end mill body gives you concentricity as well as position. Combined with a gauge block stack, this method holds ±0.01 mm without an electronic edge finder, which matters if you run older machines.
Both methods need a reference you trust. Check the indicator against a known gauge block before the setup, not after the part is scrapped. A 0.01 mm indicator that reads 0.02 mm is a common failure in a busy shop, and nobody catches it until dimensions drift.
Electronic Edge Finders for X and Y Datum
An electronic edge finder is the workhorse for X and Y. The body is a shank with a spring-loaded tip. When the tip contacts the workpiece, a circuit closes and an LED lights. Spindle speed stays at 500–1,000 rpm, and the operator feeds in 0.01 mm steps until the light flickers.
On a clean, deburred edge, an edge finder with a 10 mm tip holds about ±0.01 mm. The tip has a known radius, so you add or subtract that value in the offset. For a Ø10 mm tip, that is 5 mm. Write it on the setup sheet so the next operator does not guess.
Accuracy falls apart on three things. A burr on the edge lifts the tip early. A chamfer makes the contact point unpredictable because the tip rides the 45° face. Magnetic chips stuck to the workpiece close the circuit before real contact. Deburr, brush, and wipe before you sweep the edge.
Use the edge finder on the same two faces you use to define the program datum. If the part is programmed from a corner, find both faces and subtract the tip radius from each. Do not find one face and derive the other from the drawing unless the drawing dimensions match your setup.
Gauge Blocks, Laser Systems, and Probe Cycles
A gauge block stack plus a 0.001 mm dial indicator is the low-tech way to get high accuracy. Build the stack to the nominal dimension, sweep the tool to the block, and read the deviation. With Grade 0 blocks and a temperature-stable room, this holds 0.002–0.005 mm on Z. It is slow, and it only works where you can physically fit the stack.
Laser tool setters mount inside the enclosure and measure tool length and diameter by breaking a beam. They shine on small drills and engraving tools that would break on contact. Repeatability is typically 0.002–0.005 mm, and the cycle runs in a few seconds. The trade-off is price and the need to keep the lens clean.
Probe cycles are the method that scales. A spindle probe touches the workpiece, the control calculates the offset, and the value is written into the tool table without operator arithmetic. On a 40-tool job, that removes 40 chances for a transcription error. The probe also measures stock condition before the first cut, which lets the program shift the work offset.
On our 16 simultaneous 5-axis machining centers, probe cycles run at the start of every job so the work offset reflects the actual casting, not the nominal one. That is how a 4,000 mm part stays inside ±0.005 mm across a run. Reach for probes when the batch is larger than a handful, when tool count is high, or when the raw stock varies.
How to Set Offsets Step by Step
Follow this order on any mill or turn job.
- 1Clean the machine and the partWipe the table, the vise jaws, and the workpiece. Chips under a jaw move the datum by 0.02 mm or more. Deburr the edges you will touch.
- 2Load the tool and measure its lengthPull the tool out of the holder by hand, then seat it with the correct torque. Set length on the tool presetter or the machine setter. Log the number on the setup sheet.
- 3Establish the work offsetFind X and Y with an edge finder or probe. Find Z on the same surface the program uses. Add the tip radius where the method requires it.
- 4Verify with a single test cutCut a 0.2 mm facing pass or a 5 mm slot in scrap stock. Measure it. If the number is off by more than 0.02 mm, re-check before running the part.
- 5Record everything before the runWrite down tool numbers, offsets, method used, and the temperature. The next operator needs the method, not just the final value.
- 6Re-check after the first partMeasure the first part in the fixture. Thermal drift in the first 30 minutes can move Z by 0.01–0.03 mm on a warm spindle.
Seven Tool-Setting Methods Compared
Accuracy figures are typical shop-floor values, not best-case lab numbers.
| Method | Typical accuracy | Best use | Main risk |
|---|---|---|---|
| Bare touch-off | ±0.05 mm | Roughing, castings | Operator feel varies |
| Paper shim | ±0.03 mm | Small or fragile tools | Paper compresses |
| Dial indicator | ±0.01 mm | Older machines, no probe | Indicator out of calibration |
| Electronic edge finder | ±0.01 mm | X and Y datum | Burrs and chips |
| Gauge blocks | ±0.002–0.005 mm | Z on high-tolerance work | Slow, limited space |
| Laser tool setter | ±0.002–0.005 mm | Small drills, engravers | Dirty lens |
| Probe cycle | ±0.002–0.005 mm | High tool counts, batches | Probe stylus damage |
Pick the method before the spindle turns
Use touch-off for roughing, an edge finder for X and Y datum, and a probe or laser setter when tool count or batch size makes arithmetic risky. The method you choose sets the ceiling on the whole job.
Tool-Setting Questions Engineers Ask
How often should tool offsets be re-checked?
Check Z on every new tool and after any crash, however small. On long runs, re-check the first tool every 4 to 8 hours because thermal growth moves the spindle.
If the shop temperature swings more than 5 °C between shifts, re-check at the start of each shift. A part that measured 0.005 mm over tolerance at 4 pm may be right at 8 am.
Does coolant affect the reading?
Yes. Coolant on a surface changes the electrical contact of an edge finder and adds a film that a tool touches early. Dry the face before you sweep it.
Flood coolant also cools the part during a long cut, so the offset that was right on a cold part may be wrong once the part warms. Let the part stabilise before the final measurement.
What tolerance can a probe realistically hold?
A calibrated probe with a clean stylus holds 0.002–0.005 mm on a stable machine. The stylus ball must be clean and undamaged, and the probe must be calibrated with a known ring gauge.
On a machine with 0.001 mm feedback and a warm spindle, that accuracy is achievable in production. On a machine with worn ballscrews, the probe is accurate but the machine cannot move to the number.
When is bare touch-off still the right choice?
When the feature will be removed in the next pass, when the stock is a rough casting with 2 mm of skin, or when the only alternative is stopping the machine for an hour.
Never use it for a finishing pass on a tolerance under 0.05 mm. The error goes into the control and repeats on every part.
How do you set a tool on a lathe?
Touch the tool tip to a faced and turned diameter, then record X. For Z, face the end of the bar and record Z. Use a 0.01 mm indicator on the turret to confirm the tool is on centre.
A tool set 0.05 mm off centre on a 20 mm diameter will cut a taper and wear the insert edge unevenly. Set centre height first, then the offsets.
What belongs on a setup sheet?
Tool number, holder, length offset, diameter offset, the method used, the datum faces, and the temperature at setup. The method matters because another operator repeating the job needs to reproduce the same number.
Add the test-cut result and the first-part measurement. When a dimension drifts three days later, the sheet tells you where the setup started.
Send the Drawing, We Will Check the Setup
Upload your model and we will return a quotation with free DFM analysis within 12 hours, then start production within 24 hours.
12-hour quote100% inspection±0.005 mmNo minimum order