Method · how it's made
How cadgent dials in a fit
The loop behind the HS-02A insert: cheap physical sweeps, a few sharp rules, and a refusal to call anything "fits" without holding it.
The fit-gauge loop
One continuous clearance against a physical object: measure with calipers, print a thin-ring gauge that sweeps the diameter, find the snug ring by hand, bake that number into the real part, reprint once. A gauge is a few grams; reprinting the whole part per guess is not.
- Unambiguous.Present only the test surface. Any feature that reads differently by how you hold the part corrupts the result. (A +1.2 mm clearance hole once got felt instead of the seat.)
- Match the final print.Same printer, filament, layer height, and wall count as the shipping part, so the tolerance transfers. Infill and speed are free to crank.
- The print is ground truth.A measured print beats a datasheet. When the spec and the part disagree, the part wins.
The tool stack
No single CAD kernel. Whatever measures or cuts the geometry most honestly for the job.
The verification discipline
Claims are tiered by evidence, and the tier is never upgraded for free. A render proves a render. Only a printed, measured, hand-checked fit proves a fit.
| rendered | it exists on screen |
| virtually checked | geometry measured, booleans verified, bounds proven |
| physically fit-tested | printed and confirmed against the real hardware |
An adversarial print-readiness gate runs before any "ready." Its job is to disprove the claim, not bless it. If it can't, the verdict stays UNPROVEN, and so does the part.
See it run end-to-end: the HS-02A insert →