¶ Check X/Y scale and kerf
Scale is how far the axes actually move. Kerf is the material the beam removes. Check which one is wrong before changing anything. Do not use kerf compensation to hide an axis-scale error.
¶ Make a useful test file
Create:
- one large outside square or rectangle with distinct X and Y lengths;
- one or more inside holes/slots;
- engraved or marked axis labels away from measurement edges;
- generous spacing and lead positions that do not damage the measurement faces.
Use dimensions large enough that measuring-tool uncertainty is small relative to the error. Include a known dimension to catch an inch/mm import problem before cutting.
¶ Test sequence
- Verify the drawing's displayed dimensions in mLaser.
- Use clean, flat, known material and a stable known-good cut process.
- Disable or record contour compensation so its effect is known.
- Calibrate the head, simulate, frame without emission, and cut one coupon.
- Let the part cool; remove dross only in a way that does not change the reference faces.
- Measure outside X, outside Y, and inside features at multiple positions with a suitable calibrated tool.
- Record commanded size, measured size, direction, and repeatability.
¶ Interpret the pattern
| Pattern | Check first |
|---|---|
| Everything is exactly 25.4× wrong | Unit conversion |
| X wrong by a percentage, Y correct | X axis scale/configuration or mechanics |
| Both axes wrong by same percentage | Drawing/import scale before machine calibration |
| Outside too small and holes too large by similar amounts | Kerf/contour side or compensation direction |
| Repeated dimensions scatter | Loose mechanics, backlash, slip, heat distortion, or inconsistent cut quality |
| Error changes with travel direction | Backlash/mechanical or directional cut-quality issue |
¶ Before changing calibration
Repeat the coupon and measure again. Check belts/racks/couplings and axis condition using the approved maintenance procedure, and make sure a rotary configuration was not left active. Export the original parameters first. Only change axis calibration using the documented Gweike method and a traceable measurement. Do not steal a millimetre off a screenshot or recalibrate the machine around one rough coupon.
After axis motion is proven, calculate kerf from measured test features and apply the correct inside/outside compensation in the job workflow. Re-cut and document the result.
This started with an owner asking for a test file on the Gweike M-Series community forum. The method above is ours; you do not need an attachment from that thread to use it.