You’ve just run a ballbar test, the report is on screen, and instead of a circle you’re looking at some odd shape. What is it actually telling you? This page is the companion to our main guide on what a ballbar test is and why it matters; here we focus purely on the visual side — the plot itself. The goal is to help you recognize any shape that shows up — an oval, a step, a wavy ring, a sharp spike — at a glance.
We’ve grouped the error types below into two categories. The first group is real machine errors: they come from a physical source — a bearing, a way, a ball screw, a servo setting — and show up on the part. The second group is test-related artifacts: they have nothing to do with the machine itself, but with how the ballbar was set up, its cable, or the test procedure — and anyone who doesn’t know how to read the plot can easily mix the two up and swap out parts that were never the problem. Getting this distinction right is the single most important step in avoiding wasted time and money.
Independence note: the diagrams and explanations below are original educational material prepared by Mentor CNC; they are not a screenshot of any particular brand or software. Renishaw is a trademark of its respective owner.
Part A: Real Machine Errors — faults that originate in the machine itself
1. Backlash — positive: a small outward step above and below the circle
What you see on the plot: A small, sharp step outward from the circle right where an axis reverses direction (typically at 90° and 270°). The size of the step stays constant regardless of feed rate.
Likely cause: Play between the ball screw and nut, worn way-carriage contact, or slack in a worm drive. This is almost always the signature of mechanical play.
Effect on the part: A short flat spot on the machined circular surface, proportional to the square root of the step size — a 10-micron step can produce a visible flat of roughly 1.7 mm on a 300 mm part.
2. Backlash — negative: a jump inward toward the center
What you see on the plot: The mirror image of positive backlash — the step jumps inward toward the center instead of outward.
Likely cause: The backlash compensation value entered into the controller is larger than the actual mechanical play, or a hysteresis (lag) issue in the encoder.
Effect on the part: Sudden inward jumps on circular cuts; dimensional instability. Replacing a bearing or ball screw without first reviewing the compensation value usually doesn’t fix this.
3. Backlash — unequal: steps of different size at the two reversal points
What you see on the plot: The steps at the two reversal points on the same axis are clearly different sizes, or point in different directions (inward vs. outward).
Likely cause: Tension along the ball screw isn’t distributed evenly — the play differs depending on which end of the screw is engaged. A worn or over-tight ball screw, nut, or way can produce the same symptom.
Effect on the part: Accuracy varies by position along the axis — good results in one zone of the machine, poor results in another. On vertical axes, the counterbalance mechanism also becomes a suspect.
4. Cyclic error: regularly spaced waves around the circle
What you see on the plot: A roughly sinusoidal ripple with fairly constant frequency and amplitude, repeating all the way around the circle.
Likely cause: Eccentricity in how the ball screw or encoder is mounted (a screw that turns slightly “off-center”), or a poorly tuned linear position transducer. If the clockwise and counter-clockwise plots are nearly identical, the ball screw is the likely source; if it only shows up in one direction on a vertical axis, suspect the counterbalance mechanism.
Effect on the part: A regularly repeating dimensional deviation on the machined surface — often visible to the naked eye as a “wave” pattern.
5. Lateral play — equal: two nested circles
What you see on the plot: The clockwise and counter-clockwise passes don’t overlap — they appear as two separate, nested circles.
Likely cause: Play in the way-carriage system in the direction perpendicular (tangential) to axis travel. That’s the key difference from backlash: backlash sits along the same axis as the motion (radial), while lateral play is perpendicular to it (tangential).
Effect on the part: Bores don’t come out fully round, and overall positioning accuracy drops. Can be confused with the test artifacts radius change or offset change — rule those out first.
6. Lateral play — unequal: a heart- or teardrop-shaped plot
What you see on the plot: Unlike equal lateral play, the gap between the two nested circles is clearly larger on one side of the axis — the shape looks pulled toward one direction.
Likely cause and effect: Same root cause as equal lateral play (way-carriage looseness), but the amount of play changes with position along the axis — usually a sign that the wear on the way is localized to one region rather than spread evenly.
7. Machine vibration: an irregular, “fuzzy” trace concentrated in one zone
What you see on the plot: Noise that peaks in the region where the ballbar happens to be aligned with the vibration direction; the amplitude changes around the circle, but the underlying frequency stays the same even as feed rate changes.
Likely cause: Vibration originating in the machine itself (damaged ways, an unbalanced counterweight unit, servo loop tuning), or from an external source (a nearby machine, a coolant pump).
Effect on the part: Poor surface finish; vibration amplitude translates directly into surface roughness.
8. Master-slave change-over: symmetric steps at the 45° positions
What you see on the plot: Steps at 45°, 135°, 225° and 315° that grow with feed rate; the pattern mirrors itself when direction reverses.
Likely cause: Some older controllers perform circular interpolation by holding one axis at constant speed while varying the other; to complete a full circle the axes must swap roles at the 45° positions — and that hand-off leaves a mark on the plot.
Effect on the part: Reduced quality on circular cuts. Usually a controller/servo tuning matter, not something that requires replacing a mechanical part.
9. Reversal spikes: a sharp notch right at the direction-reversal points
What you see on the plot: A narrow, sharp notch exactly where an axis reverses direction, growing in size with feed rate.
Likely cause: The motor isn’t delivering enough torque quickly enough to overcome friction at the moment of reversal, so the axis briefly stalls. A backlash compensation value that doesn’t match the servo’s response time produces the same symptom.
Effect on the part: A notch or burr mark on the part at direction-change points. We cover this in more depth from the Fanuc parameter side in our Backlash and Reversal Spike Diagnostic Guide.
10. Scaling error: an oval aligned with the 0°/90° axes
What you see on the plot: An oval elongated exactly along the X or Y axis; the distortion doesn’t change with direction and usually not with feed rate either.
Likely cause: One axis is traveling a slightly different distance than the other for the same commanded move — incorrect linear error-compensation parameters, a loose or over-tight linear encoder, or an overheating ball screw.
Effect on the part: Dimensional error — the part comes out correct on one axis but systematically oversized or undersized on the other.
11. Servo mismatch: an oval at 45°/135° that changes with direction
What you see on the plot: An oval on the 45°/135° diagonal; clockwise and counter-clockwise plots differ, and the ovality grows as feed rate increases.
Likely cause: The servo loop gains of the two axes aren’t matched to each other — the axis with the higher gain leads the other.
Effect on the part: Circles become oval at higher feed rates. The distinguishing test: an equally-shaped squareness error does not change with direction or speed, while servo mismatch does — our Siemens following-error guide covers this distinction from the servo-parameter side.
12. Squareness: a constant oval at 45°/135° that doesn’t change with direction or speed
What you see on the plot: Nearly the same appearance as servo mismatch — a 45°/135° oval — but the distortion stays constant regardless of direction or feed rate.
Likely cause: The X and Y axes aren’t truly at 90° to each other in the test zone — either a localized bend in the axis or a misalignment across the whole machine.
Effect on the part: Machined squares don’t come out square; diagonals measure differently. Repeating the test at different feed rates and in different zones of the travel separates squareness from servo mismatch, and a localized issue from a machine-wide one.
13. Stick-slip: noise that grows at low speed and vanishes at high speed
What you see on the plot: Irregular noise concentrated around one axis; it increases as feed rate drops and can disappear entirely at higher feed rates. This is exactly what separates it from machine vibration.
Likely cause: At low feed rate, the force applied to the axis can’t reliably overcome static friction, so the axis briefly catches and releases. Worn or under-lubricated ways are the typical cause.
Effect on the part: Poor surface finish, especially on low-feed finishing passes — the tool path advances in small catches instead of a smooth arc.
14. Straightness error (tri-lobe): three smoothly-blended lobes
What you see on the plot: Three rounded lobes, but with smooth transitions between them — this is the key detail that separates it from the visually similar test artifact (item 20).
Likely cause: A bow in the machine ways — either localized or along the whole travel. Wear, a past collision, or an issue with the foundation/leveling can all be the source.
Effect on the part: A general accuracy issue across every part made on the machine — not zone-specific, and not something that goes away on its own.
Part B: Test-Related Artifacts — plots caused by setup, not the machine
The shapes in this section usually come from how the ballbar itself was set up or from the test procedure — not from the machine. Mistaking one of these for a real machine fault, and replacing a part that was never at fault, is one of the most common mistakes on the shop floor.
15. Offset change: clockwise and counter-clockwise centers don’t match
What you see on the plot: The center points of the clockwise and counter-clockwise passes are shifted relative to each other.
Likely cause: A long pause between the two directions (during which temperature drifts), a spindle that wasn’t locked, or the test bumping the part attached to the pivot/spindle.
Why it matters: Software can mistakenly interpret this as a large “lateral play” error. Repeat the test without a pause between directions and with the spindle confirmed locked.
16. Plot discontinuity: random spikes and excess noise
What you see on the plot: Spikes or excess noise at random locations that don’t follow any consistent pattern.
Likely cause: A contact issue in the ballbar cable or connector, or a loose connection somewhere in the chain. Note that genuine stick-slip or machine vibration can sometimes look similar — check those sections too.
Why it matters: No reliable diagnosis can be drawn from a plot like this. Check the cable and connections, then repeat the test.
17. Plot rotation: the whole shape appears rotated relative to the axes
What you see on the plot: The entire shape appears rotated around its center by an unexpected angle.
Likely cause: The ballbar software and the machine motion didn’t start in sync — the test program was triggered before the software was ready — or the pivot was positioned poorly relative to the working envelope.
Why it matters: Trying to diagnose anything from a rotated plot is misleading because directionality is scrambled. Check the test program timing and pivot centering, then re-measure.
18. Radius change: clockwise and counter-clockwise passes trace different radii
What you see on the plot: The clockwise and counter-clockwise passes trace circles of noticeably different size.
Likely cause: The ballbar’s physical length changing as it heats or cools during a long pause between directions, or a connection loosening during the move. Sometimes this is actually lateral play present on both axes, appearing this way.
Why it matters: Remove nearby heat sources, let conditions stabilize, and repeat the test without a pause between the two directions.
19. Spiral error: the circle never closes, tracing inward or outward
What you see on the plot: The start and end points don’t line up — the trace spirals inward or outward instead of closing.
Likely cause: A temperature shift during the test changing the ballbar’s length; the cable snagging and nudging the ballbar out of place; the pivot loosening; or a spindle that wasn’t locked and rotated slightly.
Why it matters: This can sometimes be genuinely unequal backlash showing up this way. Rule out heat and connection issues, then repeat the test.
20. Tri-lobe (test artifact): three lobes with sharp transitions
What you see on the plot: The same overall three-lobe shape as the straightness error in item 14, but the transitions between lobes are hard and sharp.
Likely cause: Dirt or wear on the ballbar’s end balls, the part attached to the spindle loosening during the test, or play in one of the ballbar’s connection points.
How to tell them apart: Run the test once, rotate the ball roughly 60°, and run it again. If the lobes shift with it by roughly the same 60°, the cause is dirt or wear on the ball — not a real straightness error; if the shape stays fixed in place, you’re looking at the genuine machine fault described in item 14.
Comparison table: which shape points to what?
| Plot shape | Location | Changes with direction/speed? | Most likely source |
|---|---|---|---|
| Inward/outward step | 90°/270° (axis reversal points) | No | Backlash (positive/negative/unequal) |
| Regular waves | All around the circle | Usually no | Cyclic error (ball screw/encoder eccentricity) |
| Two nested circles | Whole circle | Direction: yes · Speed: no | Lateral play |
| Concentrated noise, constant frequency | Zone aligned with vibration source | Speed: no (frequency stays fixed) | Machine vibration |
| Step at 45°/135° | Diagonal zones | Yes, grows with speed | Master-slave change-over |
| Narrow, sharp notch | Exact reversal point | Yes, grows with speed | Reversal spikes |
| Oval at 0°/90° | Main axes | No | Scaling error |
| Oval at 45°/135° | Diagonal axes | Yes, grows with speed | Servo mismatch |
| Oval at 45°/135° | Diagonal axes | No, constant | Squareness |
| Noise on one axis | Single axis | Speed: yes (grows at low speed) | Stick-slip |
| Three smooth lobes | Whole circle | No | Straightness error (real machine fault) |
| Three sharp lobes | Whole circle | No | Tri-lobe (test artifact — ball dirt/wear) |
| Shifted centers | Whole circle | Direction: yes | Offset change (test artifact) |
| Random spikes/noise | Irregular | Variable | Plot discontinuity (cable/connection) |
| Whole shape rotated | — | — | Plot rotation (synchronization) |
| Two different radii | Whole circle | Direction: yes | Radius change (test artifact) |
| Circle doesn’t close | Whole circle | — | Spiral error (test artifact) |
Frequently Asked Questions
Can more than one error type show up in the same plot?
Yes — in fact this is more common on the shop floor than a single clean signature. Servo mismatch and a mild backlash step, for example, can sit side by side in the same plot. That’s why it’s worth looking at repeated tests at different speeds and directions rather than a single run.
Does this page tell me exactly which error I’m looking at?
No — this page is a starting point, not a definitive diagnosis. The numeric values from the ballbar software, tests at different speeds/directions, and, where needed, a qualified technician’s assessment complete the actual diagnosis.
I can’t tell whether it’s a test artifact or a real machine error — what should I do?
Tighten the connections, let the ambient temperature stabilize, and repeat the suspect test. If the plot changes, it was a test artifact; if it stays the same, you’re looking at a genuine machine error.
Related Guides
- What Is a CNC Ballbar Test? Fast Diagnosis of Sizing Errors and Report Interpretation
- From Part Defects to Fanuc Parameters: Backlash and Reversal Spike Diagnostic Guide
- Siemens SINUMERIK 840D Following Error, Contour Deviation and Axis Service Screen
Disclaimer: this content is for general educational purposes and does not replace official service documentation for any specific machine or controller model. Always consult a qualified service technician before making critical parameter changes.