CNC Probe Working Principle: How Touch Triggering Actually Works
A CNC probe is a measuring head that turns physical contact into a machine-readable signal. This page explains the mechanism, the signal path into the control, what limits accuracy, and when probing is worth the cycle time.
In this article
- 1
- 2
- 3
- 4
- 5
- 6
- 7
What a CNC Probe Is and How It Detects Contact
A CNC probe is a measuring head mounted in the spindle or a tool holder. It carries a stylus with a ruby or silicon nitride ball at the tip. When the ball touches a surface, the head reports a discrete event to the machine control: contact, or no contact. That is the whole output. The control combines the event with the current axis positions to calculate where the surface is.
Most machine shop probes are touch trigger type. Inside the head sits a kinematic seat, typically three balls resting in three cylinders, held closed by spring force. Contact deflects the stylus and breaks the electrical circuit across that seat. The break is the trigger. A strain-gauge head works differently: it measures the bending of the stylus continuously and can report a deflection value instead of a binary event.
Optical and eddy current heads exist too. Optical probes measure without contact and suit soft or delicate surfaces; eddy current probes detect conductive material through a small gap. Both are less common in general milling and turning because they cost more and need a clean environment.
The practical point: a probe does not measure a diameter or a distance by itself. It finds points. The control turns those points into datums, stock conditions, and offsets using the macro programs supplied by the machine builder or probe vendor.
- 1Touch triggerBinary contact signal from a kinematic seat
- 2Strain gaugeContinuous deflection, usable for scanning routines
- 3Optical / eddy currentNon-contact, used on soft or delicate surfaces
From Stylus Deflection to a Machine Coordinate
When the seat opens, the probe transmits a skip signal to the machine control. Older installations use a hard-wired cable or an infrared link; modern heads commonly use a 2.4 GHz radio link, which avoids line-of-sight problems when the head is deep inside a cavity. Transmission latency matters. A radio link that adds 5 ms of delay at a 2,000 mm/min feed rate lets the machine travel roughly 0.17 mm before the control reacts.
The control latches the axis positions the moment it receives the skip signal. This is a hardware latch, not a scan of the position display. The latched coordinates, not the coordinates at the end of the move, are what go into the calculation. If the machine latches on the servo command instead of an encoder feedback signal, expect a systematic error that grows with feed rate.
Probe software then applies a set of corrections. The stylus ball radius is subtracted along the approach vector. The trigger force causes a small pre-travel before the signal fires, so the head is calibrated with a known ring gauge or a calibration sphere to establish that pre-travel value in X, Y and Z. Probe calibration is not a one-time job; it is repeated after a crash, a stylus change, or any spindle maintenance.
Only after these corrections does the control write a work offset or update a tool offset. A probe result that is used directly as a coordinate, without ball radius and pre-travel compensation, will be wrong by a consistent amount that looks like a machine error.
- 1Skip signalLatches encoder position at the moment of contact
- 2Pre-travelStylus deflection before triggering; measured during calibration
- 3Ball radiusSubtracted along the approach direction
- 4OutputWork offset, tool offset, or a reported dimension
What Limits CNC Probe Accuracy
A touch trigger head repeats to roughly 1 μm in a controlled test, but the machine-level result is usually worse. The largest single contributor is the machine itself. Probing measures the position of the machine axis, not the true geometry of the part, so any squareness or straightness error in the machine appears directly in the result. A machine holding ±0.005 mm on a cut part will not necessarily hold that on a probed dimension.
Stylus length is the second factor. Bending of a long stylus under trigger force creates a lobing error that changes with approach direction. A 50 mm stylus is stiffer than a 100 mm stylus, and a 2 mm ball on a long stem deflects more than a 6 mm ball on a short one. The classic countermeasure is to approach the same feature from the same direction every cycle, so the pre-travel value stays consistent.
Temperature and cleanliness also matter. A 5 °C change across a 300 mm steel part moves the dimension by roughly 0.018 mm through thermal expansion. Chips on the surface move the contact point by their own thickness. For a probing routine to mean anything, the surface has to be clean and the part has to be at a stable temperature.
Probing is also a slower measurement than many shops assume. Each point needs an approach, a touch, a retract and a settle. A four-point bore measurement at 300 mm/min with 2 mm clearance typically takes a few seconds per feature. That is cheap for one datum and expensive for a hundred features.
- 1Machine geometrySquareness and straightness errors pass straight into the result
- 2Stylus bendingLonger stylus and smaller ball increase directional error
- 3TemperatureAbout 0.011 mm per 100 mm per 10 °C on steel
- 4Surface conditionChips and coolant film shift the contact point
When Probing Helps and When It Does Not
Probing earns its cycle time when the setup is uncertain. Castings and forgings with variable stock, weldments that move after stress relief, first-off parts on a new fixture, and large parts where a manual edge find is slow all benefit. The routine finds the actual surface, then shifts the work offset so the first cut lands where it should. That prevents a scrapped casting worth more than the probing time.
Probing is a poor substitute for a controlled process. If a batch of parts is drifting because a tool is wearing, probing every part does not fix the tool. It just detects the drift later. Tool wear is better handled with periodic tool offsets or a tool setting probe, not with a work offset update on every cycle.
It also does not replace final inspection. Probing confirms a datum and a few features; it does not produce a dimensional report, and it does not catch every error. A bore probed at four points can still be out of round between them. Complex geometry, thin walls, and features smaller than the stylus ball cannot be probed usefully at all.
For parts we machine to ±0.005 mm, probing is usually a setup tool rather than an in-process gauge. We probe the datum, cut, then measure the critical features on a coordinate measuring machine or with a bore gauge. The probe saves the setup; the metrology proves the part.
- 1Good fitVariable stock, first-off setup, large castings, unstable fixtures
- 2Poor fitTool wear compensation, thin walls, features under the stylus size
- 3Not a replacementFinal dimensional inspection and reporting
Where Probing Changes the Machining Plan
In our own shop, probing shows up most often on 5-axis work where the part is repositioned between operations. After a rotation, the work offset from the first setup is no longer valid. A short bore or face probing routine re-establishes the datum in the new orientation, and the second operation cuts to the same origin. Without it, the second setup depends entirely on fixture repeatability.
On castings and near-net forgings, probing drives the first cut. The routine maps a few reference surfaces, calculates the actual stock distribution, and the control shifts the offset so the finishing allowance is even. This is common on aluminium die cast housings and on engine-related parts where the as-cast position varies by a few tenths.
Probing also supports adaptive routines on thin-walled parts. A light touch confirms the wall is where the model says it is before a finishing pass, which reduces the risk of cutting through a wall that moved during roughing. The probe is not measuring the wall thickness in real time, but it can confirm the setup before the expensive pass.
Across these cases, the same rule applies: probe where the uncertainty is, and only there. Extra probing points on features that are already controlled add cycle time without adding information.
- 1Re-datum after rotationRestores the origin for second-operation cuts
- 2Stock mappingBalances allowance on castings and forgings
- 3Pre-finish checkConfirms wall position before a light finishing pass
A Basic Probing Routine, Step by Step
This is the sequence a typical spindle-mounted touch probe follows in a milling cycle.
- 1Clean the surfaceBlow off chips and dry the area. A coolant film can shift contact by a few micrometres.
- 2Call the calibration valuesLoad the stored stylus ball radius and pre-travel values for the head in use.
- 3Approach with clearanceMove to the start point with 2–5 mm clearance, then feed at 200–500 mm/min toward the surface.
- 4Latch and retractOn the skip signal, the control latches the position and retracts 1–2 mm at the same feed.
- 5Re-approach slowlyRepeat the touch at 50–100 mm/min for the final point to reduce pre-travel scatter.
- 6Repeat from the same directionApproach every point on that feature from the same side so directional error stays constant.
- 7Compute and write the offsetThe macro applies ball radius and pre-travel, then updates the work or tool offset.
- 8Verify with a second pointProbe a known feature once more. A repeat result within a few micrometres confirms the routine.
Probing Method vs What It Tells You
Each method reports a different kind of information. Pick the one that matches the decision you have to make.
| Method | What it reports | Typical use | Main limit |
|---|---|---|---|
| Touch trigger point | A latched coordinate at contact | Datums, stock check, work offsets | Slow per point; directional error |
| Strain gauge scan | Continuous deflection curve | Profile and form checks on simple shapes | Needs slower feed and clean surface |
| Optical probe | Non-contact point or edge position | Soft parts, delicate features | Sensitive to surface finish and light |
| Eddy current probe | Gap to conductive material | Thin walls, in-process gap checks | Conductive materials only |
| Tool setting probe | Tool tip position at the spindle | Tool length and breakage detection | Does not check the workpiece |
| CMM after machining | Full dimensional report | Final acceptance of critical features | Offline; parts wait in a queue |
The Engineering Verdict
Use a CNC probe to find uncertain datums and variable stock, not to replace process control or final inspection. If the setup is already repeatable and the tool is the variable, fix the tool.
Common Questions
Can a CNC probe hold ±0.005 mm on a part?
The probe head itself repeats well, often near 1 μm in a clean test. The machine is the bigger term. Squareness and straightness errors of the machine pass directly into a probed result, so a probe cannot measure better than the machine can position.
On a machine that holds ±0.005 mm on a cut feature, a careful probing routine with calibration and consistent approach direction can be useful in that range. It should still be verified against a known artefact before the values are trusted.
How often does a probe need calibration?
Calibrate after every stylus change, after any crash, and after spindle or head maintenance. In stable production, a monthly check against a ring gauge or calibration sphere is normal.
If the routine writes work offsets that matter to the part, check the calibration before the job, not after a scrapped part.
Does probing add much cycle time?
Each point costs an approach, a touch, a retract and a settle. A four-point bore check at moderate feed typically runs a few seconds per feature.
One or two datums per setup is cheap insurance. Probing dozens of features on every part usually is not worth the time unless the part value is high.
Why does a probed dimension differ from a CMM result?
The two systems touch the part differently. A probe applies trigger force and has directional pre-travel; a CMM uses a controlled low-force head and often scans the surface.
Surface finish, part temperature and fixturing also differ between the two measurements. Expect a small systematic difference and account for it in the tolerance budget.
Can a probe measure a bore diameter directly?
Not directly. The control collects contact points and calculates the diameter from the fitted circle. The result depends on how many points you take and where they sit.
Four points give a usable diameter but little form information. More points on a strain gauge head give better form data at the cost of cycle time.
Is a probe useful on a lathe?
Yes, with a different routine. A tool setting arm or a spindle-mounted head can check a turned diameter or a face position after a changeover.
Lathe probing is most useful for confirming a datum after a bar pull or a jaw change, where the raw stock position varies from bar to bar.
Send Us Your Part and Tolerance Callouts
Upload a drawing and we will review the critical features, suggest where probing or in-process checks help, and return a quote with DFM notes within 12 hours.
12-hour quote100% inspection±0.005 mmNo minimum order quantity
More Machining Notes
We publish setup notes, tooling trials and inspection data from the factory floor.
