How to Make a CNC Tracer

• 1
  Key takeaways
• 2
  What a CNC tracer actually measures
• 3
  Choosing the probe, scale and frame
• 4
  Frame, stylus and interface details that matter
• 5
  Calibration routine and drift control
• 6
  Common errors and how to avoid them
• 7
  How to make a CNC tracer, step by step
• 8
  Build versus buy versus send out

A CNC tracer follows a surface with a probe and outputs the difference between where the surface is and where the drawing says it should be. The classic version is a bench gage: a part sits in a fixture, a stylus rides along a profile, and a linear scale or dial indicator shows the deviation. On-machine versions mount in a tool holder and sweep the same profile after machining.

The output is one or two numbers per point. That is the whole point. You are not building a coordinate measuring machine, and you should not try. A tracer answers questions like: did the radius blend cleanly, did the bore drift, is the profile within 0.02 mm of the master.

This makes the tool useful for first-article checks, for tool wear trending, and for incoming inspection of a part you will copy. Aerospace brackets, medical housings and engine components are typical candidates. Flat plates with one critical bore usually do not need one.

The limit is access. A tracer needs a clear path for the stylus. Undercuts, deep pockets and internal channels are out of reach unless you build a special stylus. If your part has those features, plan for a different method.

Start with the probe. An LVDT or linear-variable probe gives a continuous analog signal, which is what you want for profile tracing. A touch-trigger probe gives discrete points and is better for hole location. A laser triangulation head avoids contact but struggles with shiny aluminium and with coolant mist.

For the scale, a glass linear encoder with 1 μm resolution is enough for a ±0.005 mm class build. Magnetic tape is cheaper and more tolerant of dust, but its resolution is coarser. Do not mix a fine probe with a coarse scale; the scale sets your floor.

The frame decides whether the numbers mean anything. Granite surface plate, cast iron column, or a heavy steel weldment all work. Aluminium extrusion bolted to a wooden bench does not. Thermal drift of 2 °C across a 300 mm steel column moves the reading by about 6 μm.

Plan the travel before you buy rails. A 400 mm × 150 mm working envelope covers most bracket and housing work. Longer travel needs a stiffer column, and stiffness grows with the cube of the section height, so a tall thin column is a trap.

Control electronics can be as simple as a digital readout with a data port. If you want logging, a USB encoder interface plus a laptop is enough. Keep the electronics away from the coolant line.

Mount the scale on the moving member, not on the base. Any flex between the stylus and the scale shows up as error. Bolt the scale rail to a machined surface, not to a painted one, and shim it flat to within 0.05 mm over its length.

Stylus geometry sets your effective resolution. A 2 mm ruby ball on a 40 mm stem is a good general choice. Longer stems increase bending and reduce contact force repeatability. Keep the stylus as short as the part allows.

Contact force should be low, around 0.1 N to 0.5 N for a spring-loaded probe, so you do not mark soft materials like aluminium or brass. Too low and the stylus skips on a rough surface. Test on a scrap part before you trust a reading.

Cable routing is a common failure point. Route the probe cable with a service loop and clamp it so it cannot pull on the stylus. A dragging cable can add 5 μm to 20 μm of false reading.

For the interface, a simple comparator circuit with a go/no-go light is enough for shop floor use. Add a digital output only if you need to log data. Extra features add noise and cost.

Calibrate in the same room where you will use the tracer. Bring the master and the machine to room temperature for at least 4 hours. A tracer calibrated in a cold morning shop will read wrong by afternoon.

Zero on the master, then move to a second artifact of a different size. The difference between the two readings is your scale error. Adjust the scale factor by that ratio and repeat until both agree within 1 μm.

Then check repeatability. Touch the same point 20 times and record the spread. Anything over 2 μm means something is loose or dragging. Fix the mechanical issue first; do not compensate in software.

Recheck every 8 hours of use, after any move of the machine, and after a temperature swing of more than 3 °C. Log the check so you can see drift trends over weeks.

For parts with tight tolerance, hold the part in the same fixture used for calibration. Fixture-induced distortion of 10 μm is easy to create with a single over-tightened clamp.

The most common error is a frame that flexes under hand pressure. Push the stylus with your finger and watch the readout with the probe disconnected from the part. If the number moves, the frame is moving. Add gussets or go to a heavier base.

Second is thermal drift. A tracer near a door or under a skylight will drift through the day. Move it to an internal wall and shade it. A 3 °C change over a 300 mm steel column is roughly 9 μm of false reading.

Third is stylus damage. A bent stylus reads a different effective length and shifts every number. Keep a spare and check runout with a dial indicator monthly.

Fourth is electrical noise. If the reading jumps when a nearby machine starts, your cable is picking up interference. Reroute, shield and ground at one end only.

Finally, do not trust a tracer for features it cannot reach. Undercuts, deep bores and free-form surfaces need a different tool. Using the wrong gage gives you confident wrong numbers.