Rectifying Circles Not Round: Laser Cutting Calibration

The direct answer: A circle that cuts as an oval indicates a mismatch between commanded motion and actual motion in one axis. The primary causes are incorrect steps-per-millimeter calibration for the stepper motors (typically 3–8% error) and uneven belt tension that causes axis lag (1–5% differential). To rectify, measure the cut circle’s major and minor axes, compute the axis-specific scaling factor, adjust the steps per mm in the controller, and then balance belt tension to within 10% of the target frequency (80–120 Hz for GT2 belts).

You programmed a 50 mm diameter circle. The laser cut a shape measuring 50.2 mm in X and 48.5 mm in Y. That is an oval. The error is not random. The X axis moved farther than commanded; the Y axis moved less. You checked the belts. They seem tight. You checked the grub screws. They are secure. The problem persists.

This article quantifies the relationship between motor calibration, belt tension, and circle roundness, and provides a field-proven procedure to restore accuracy to ±0.1 mm over 500 mm.

The Geometry of an Oval: What the Numbers Tell You

When a laser cutter produces an oval instead of a circle, the error is almost always axis-specific. A perfect circle requires that the distance traveled in X equals the distance traveled in Y for every diagonal move. If the X and Y axes have different scaling factors, a programmed circle becomes an ellipse.

Measure the cut part:

Measured dimension	Calculated error	Interpretation
X diameter = 50.2 mm, Y diameter = 48.5 mm	X = +0.4%, Y = –3.0%	X overshoots, Y undershoots. Scaling mismatch.
X = 49.8 mm, Y = 49.8 mm (but circle is skewed)	No scaling error	Axis squareness problem (non-90° gantry).
X = 50.1 mm, Y = 50.1 mm but with flat spots	No scaling error	Belt slack or backlash in one direction only.
Diameters are correct, but the circle is wavy	No scaling error	Belt tooth engagement or bearing issues.

The most common case is a scaling mismatch: one axis moves a different distance than commanded for the same number of steps. The cause is either incorrect steps-per-mm calibration or belt tension differential that changes the effective pulley diameter under load.

Stepper Motor Calibration: Steps Per Millimeter

1 The Formula

The distance a stepper motor moves per step is determined by: $$\text{Steps per mm}=\frac{\text{Motor steps per revolution}\times \text{Microsteps}\times \text{Pulley reduction}}{\text{Pulley circumference (mm)}}$$Steps per mm=Pulley circumference (mm)Motor steps per revolution×Microsteps×Pulley reduction

Where:

• Motor steps per revolution = 200 for standard NEMA 17/23 steppers
• Microsteps = 4, 8, 16, 32, 64, 128 (set by driver switches)
• Pulley reduction = 1 (direct drive) or gear ratio (e.g., 3:1 for belt reduction)
• Pulley circumference = π × pitch diameter (mm)

Example, standard configuration:

• 200-step motor, 16 microsteps, direct drive, GT2 pulley with 20 teeth, belt pitch 2 mm
• Pulley circumference = 20 teeth × 2 mm = 40 mm
• Steps per revolution = 200 × 16 = 3200 steps/rev
• Steps per mm = 3200 / 40 = 80 steps/mm

If the controller is set to 80 steps/mm but the actual movement is 79.2 steps/mm (1% low), a commanded 100 mm move will travel only 99 mm. The circle will be undersized in that axis.

2 Field Calibration Procedure

Equipment needed: Digital calipers (0.01 mm resolution), straightedge, tape measure (for large travels), test pen.

Step 1 – Home the machine.
Home both axes. Jog to a known position near the center of the work area.

Step 2 – Mark a reference point.
Lower the laser (or attach a pen) and mark a dot at the current position. Record the machine coordinates.

Step 3 – Command a long move.
Using the control software (LightBurn, LaserGRBL, Ruida, or Trocen), command a 300 mm move in X (or Y). For belt-driven machines, use the longest practical move — at least 200 mm — to minimize measurement error.

Step 4 – Measure actual travel.
Measure the distance between the start dot and the endpoint. Use calipers for moves under 200 mm, a steel rule for longer moves.

Step 5 – Compute correction factor. $$\text{Correction factor}=\frac{\text{Actual distance}}{\text{Commanded distance}}$$Correction factor=Commanded distance- Actual distance

If actual = 298 mm, commanded = 300 mm, correction factor = 0.9933 (0.67% short).

Step 6 – Adjust steps per mm.$$\text{New steps/mm}=\text{Current steps/mm}\times \text{Correction factor}$$New steps/mm=Current steps/mm×Correction factor

If current = 80 steps/mm, new = 80 × 0.9933 = 79.46 steps/mm.

Step 7 – Save and retest.
Update the controller configuration. Repeat the 300 mm move. The error should be ≤0.1 mm over 300 mm (0.03%).

3 Typical Steps per mm Values for Common Drives

Motor	Microsteps	Pulley (teeth)	Belt pitch	Steps per mm
NEMA 17	16	20	GT2 (2mm)	80.00
NEMA 17	16	16	GT2 (2mm)	100.00
NEMA 23	16	20	GT2 (2mm)	80.00
NEMA 23	16	25	GT2 (2mm)	64.00
NEMA 17	32	20	GT2 (2mm)	160.00
NEMA 17	16	12	T5 (5mm)	53.33

Note: The calculated value is theoretical. Always verify with a physical cut or pen draw. Machining tolerances in pulleys (typically ±0.1 mm pitch diameter) and belt stretch under load can shift the effective steps per mm by 0.5–1.5%.

The Hidden Variable That Changes Scaling Under Load

1 How Belt Tension Affects Positioning Accuracy

A timing belt is not perfectly rigid. Under acceleration and deceleration, the belt stretches. The amount of stretch depends on belt material, width, length, and tension.

The effective positioning error from belt stretch during constant velocity is negligible. The error occurs during acceleration and deceleration. When the motor starts moving, the belt stretches before the load (gantry) moves. When the motor stops, the belt overshoots or oscillates.

For a 1 kg gantry, a 6 mm wide GT2 belt with 100 N tension (approximately 10 kg-force) has a stiffness of about 20,000 N/m. Under a 0.5 G acceleration (5 N force), the belt stretches 0.25 mm. That 0.25 mm error is not recovered immediately — it results in a positioning lag that persists until the next deceleration.

If X and Y belts have different tensions, the lag is different per axis. The result: circles become ovals, and rectangles become parallelograms.

2 Measuring Belt Tension Without a Gauge

A belt tension gauge (e.g., Gates Sonic Tension Meter) is a professional tool. In the field, you can use the frequency method:

1. Pluck the belt like a guitar string.
2. Measure the frequency using a smartphone app (Spectroid, FFT Analyzer, or Guitar Tuner).
3. Compute tension from frequency.

Formula for open belt spans: $$T=4\times m\times L^2\times f^2$$T=4×m×L2×f2

Where:

• $T$T = belt tension (N)
• $m$m = belt mass per unit length (kg/m) — GT2 6mm: 0.022 kg/m
• $L$L = span length between pulleys (m)
• $f$f = fundamental frequency (Hz)

Example: GT2 belt, span = 0.8 m, measured frequency = 100 Hz.$$T=4\times 0.022\times (0.8{)}^2\times (100{)}^2=4\times 0.022\times 0.64\times 10000=563\text{ N}$$T=4×0.022×(0.8)2×(100)2=4×0.022×0.64×10000=563 N

That is extremely high. Normal tension for a GT2 belt on a laser cutter is 30–80 N.

Target frequencies for common spans (GT2 6mm belt, 30–80 N):

Span length (mm)	Low tension (30 N) frequency (Hz)	High tension (80 N) frequency (Hz)
400	47	77
500	38	62
600	32	52
700	27	44
800	24	39

For a typical 500 mm span, aim for 45–55 Hz.

3 Balancing X and Y Tension

The most important parameter is not absolute tension, but tension balance between the X and Y axes. Measure both belts using the same span length (or adjust the formula for different spans). The frequencies should be within 10% of each other.

Frequency difference	Effect on circle roundness	Action
<5%	Negligible	Acceptable
5–10%	Slight oval (0.1–0.3 mm over 100 mm)	Acceptable for general work
10–20%	Visible oval (0.5–1 mm)	Adjust
>20%	Severe oval (>1 mm)	Immediate correction

4 Adjusting Tension

Most laser cutters use one of three tensioning methods:

• Screw-adjustable motor mount: Loosen mounting screws, slide the motor away from the pulley, and retighten. Adjust incrementally.
• Eccentric pulley: Rotate an off-center pulley to increase or decrease tension.
• Spring-loaded tensioner: Add or remove shims.

Procedure:

1. Loosen the motor mount bolts (but not completely).
2. Insert a hex key or pry bar to apply gentle tension.
3. Pluck the belt and measure frequency.
4. Adjust until frequency reaches the target.
5. Tighten bolts. Remeasure — tension often drops 5–10% when bolts are tightened. Overshoot by 10% initially.
6. Repeat for the other axis.

Warning: Overtensioning destroys bearings, increases motor load, and causes skipped steps. A belt that sounds like a low bass note (60–80 Hz) is acceptable. A belt that sounds like a guitar string (100+ Hz) is too tight.

 Diagnostic Decision Tree for Non-Round Circles

Use this sequence to isolate the cause. Do not skip steps.

Step	Test	Observation	Action
1	Measure X and Y diameters of cut circle.	X ≠ Y by >0.3 mm over 100 mm	Go to Step 2.
2	Perform steps-per-mm calibration on X and Y.	One axis has >0.5% error	Adjust steps/mm. Retest.
3	Measure belt tension frequency on X and Y.	Difference >10%	Balance tension. Retest.
4	Measure X and Y diameters of the cut circle.	Difference >0.5 mm over 200 mm	Square gantry. Retest.
5	Examine circle quadrant points for flats.	Flats present at reversal points	Cut a large square, measure diagonals.
6	Reduce acceleration by 50% and retest.	Circle improves	Keep lower acceleration.
7	Reduce microstepping to 16× and retest.	Circle improves	Use lower microstepping.

---

Preventive Maintenance Schedule

Interval	Action	Target
Daily	Visual belt inspection	No cracks, missing teeth, fraying
Weekly	Pluck test, record frequency	Within 10% of baseline
Monthly	Steps-per-mm verification (100 mm move)	Error <0.1 mm
Quarterly	Full calibration (300 mm move both axes)	Error <0.2 mm
Annually	Replace belts (GT2 belts degrade after 2000–3000 operating hours)	New belts

---

Conclusion

A non-round circle is a quantitative diagnostic. The difference between X and Y diameters tells you the axis scaling mismatch. That mismatch comes from incorrect steps-per-mm calibration or unbalanced belt tension. Steps-per-mm calibration is a one-time setup, verified by measuring a long-axis move with calipers. Belt tension balance requires periodic adjustment, measured by pluck frequency.

The sequence is: calibrate steps per mm first, then balance belt tension, then verify squareness and backlash. Changing tension after steps calibration will alter the effective scaling — always recalibrate after adjusting belts.

A laser cutter that cuts circles to ±0.1 mm over 200 mm is achievable with proper calibration and maintenance. The tools required are a set of calipers, a smartphone frequency analyzer, and 30 minutes of systematic testing.

Read More: How to Laser Cutting Gears: Precision Efficiency?

 Frequently Asked Questions

Q: My circles are round, but the size is wrong. Is that a belt tension issue?

A: No. The wrong size with correct roundness indicates steps-per-mm scaling error, but is equal in both axes. Calibrate steps per mm. Do not adjust belts.

Q: Can belt tension change over time?

A: Yes. New belts stretch 0.5–1.0% during the first 100 operating hours. Re-tension after the break-in period. Belts also loosen if mounting bolts vibrate loose — use threadlocker.

Q: What is the acceptable runout for a laser cut circle on a 100 mm diameter?

A: For general fabrication, ±0.2 mm. For precision work (interlocking parts), ±0.1 mm. For engraving (visual only), ±0.5 mm is acceptable.

Q: My machine uses lead screws instead of belts. Does this article apply?

A: Partially. Steps-per-mm calibration still applies. Belt tension does not. For lead screws, backlash (nut wear) is the primary cause of non-round circles. Adjust or replace the anti-backlash nut.

Q: How do I calibrate steps per mm if my controller does not support changing the value?

A: Some low-cost controllers (e.g., older K40 stock boards) have fixed steps per mm. In this case, you must change the pulley or belt pitch to achieve the correct scaling, or upgrade the controller. Consider a GRBL-based board (e.g., LaserBoard, Cohesion3D) that allows full calibration.

References

1. XC3000S Series Laser Cutting System User Manual. Raytools AG. Section 2.1.5: Transform and scale functions for graphic correction.
2. HypCut Laser Cutting Control Software User Manual, Version 1.4. Section 5.1: Graphics editing and geometric transformation.
3. ISO 1101:2017. Geometrical product specifications — Geometrical tolerancing — Tolerances of form, orientation, location and run-out. (Reference for roundness tolerance definitions)
4. Gates Corporation. (2018). PowerGrip GT2 Belt Drive Design Manual. Gates Technical Manual 17195.
5. LightBurn Software Documentation. (2024). Machine Calibration: Steps per mm and Axis Scaling. LightBurn Help Center.
6. Adams, S. M. (2005). Sheet Metal Bend Allowance. KETIV Virtual Academy. (Reference for precision measurement principles)