● Featured Guides
Explore 21 CNC learning topicsLearn condition monitoring from real dataUse the free CNC calculators

CNC Condition Monitoring

How to Read a CNC Vibration Waveform Without Getting Lost

Learn to inspect time-domain CNC vibration for impacts, bursts, drift, repeating cycles, and changes in operating state.

Updated August 15, 20263 minute read
Close view of a 250 millisecond CNC vibration waveform
A short view preserves impacts and waveform shape that may disappear in a whole-cycle summary. Chart created by TWC Industrial from the Bosch Research CNC Machining dataset.

A waveform is simply vibration amplitude plotted against time. It looks busy because machining is busy. Instead of reading every wiggle, look for structure: when the signal grows, when it settles, and whether unusual impacts repeat.

Read at two scales

A one-second view exposes individual bursts and oscillations. A full-cycle view shows entry, steady cutting, tool changes, and exit. Both matter. A brief impact may disappear in a whole-cycle average, while a short view may hide a gradual process shift.

Look for events, not decorations

Useful observations include isolated impacts, periodic bursts, amplitude modulation, clipping, a changing center line, and abrupt state transitions. These are descriptions, not diagnoses. They help decide which supporting data to inspect.

Beware of process boundaries

Tool entry can create a legitimate impact. Rapid motion may look different from cutting. If machine state is available, align it with the vibration trace. Otherwise, compare cycles with consistent timing and note uncertainty.

Make a repeatable review routine

Plot identical time ranges and amplitude scales for comparable cycles. Add RMS and peak beside the plot. When a trace looks unusual, save the raw segment and the production context so another person can reproduce the review.

A repeatable waveform review

Begin with the full machining record and mark obvious states: idle, approach, tool entry, steady cutting, exit and rapid return. Next, inspect the same one-second cutting window across several healthy parts. Use identical axes and plot limits so visual differences are not created by automatic scaling.

If an impact appears, zoom into a short region around it. Check whether it occurs at the same process time on every part. A repeatable tool-entry impact has a different meaning from an impact that begins unexpectedly halfway through a stable cut.

Waveform patternNext check
Single isolated spikeSensor cable, clipping, chip impact and nearby axes
Repeating burstsTool engagement, tooth timing or programmed motion
Gradual amplitude increaseTool wear, warming, load or changing engagement
Abrupt permanent shiftProcess state, fixture, tool or sensor movement

Common mistakes to avoid

  • Letting chart software use a different scale for every file.
  • Looking only at a one-second crop and missing the process state.
  • Removing outliers before checking whether they are real impacts.
  • Comparing cycles that start at different events.

Frequently asked questions

How much waveform should I display?

Use at least two views: the complete relevant cycle and a fixed short window for detail.

Should I smooth the signal?

Only with a documented purpose. Smoothing can hide impacts, so keep the raw trace available.

What does a flat-topped waveform mean?

It may indicate clipping or sensor-range saturation. Check acquisition settings before interpreting the machine.

About the data used in this guide

The charts use selected files from machine M01, operation OP01. The source records tri-axial acceleration at 2 kHz and labels process examples as good or bad. Our initial charts use two files from each label. They are teaching examples, not universal fault thresholds.

Dataset: CNC Machining Data, CC BY 4.0. Recommended citation: Tnani, Mohamed-Ali; Feil, Michael; Diepold, Klaus. Smart Data Collection System for Brownfield CNC Milling Machines: A New Benchmark Dataset for Data-Driven Machine Monitoring. Procedia CIRP 107 (2022), 131–136. Research paper.

Editorial standard

We explain what the selected data supports and avoid naming a mechanical fault when the dataset only provides a good/bad process label. A machine should be inspected by a qualified person before maintenance or safety decisions are made.