Tolerancing a printed radio cover so it actually drops in

The hardest part of a radio cover is not the shape. It is the half millimeter you cannot see.

A printed part is not the size of the model. PA-CF shrinks roughly 0.8% to 1.2% as it cools, and it does not shrink evenly — a long flat panel pulls in more along its length than through its thickness. If you model a 96 mm window and print it nominal, you get something closer to 95 mm and a head unit that will not seat. So the CAD model is deliberately wrong: we scale the aperture up, then verify against a physical reference chassis.

Second issue: bolt holes are never round. A vertical hole printed by extrusion comes out slightly oval and slightly undersized because of the way the nozzle corners the perimeter. We oversize fastener holes by 0.3 mm and add a 0.4 mm chamfer at the entry, which means the M4 finds the thread instead of the installer wallowing the hole out with a drill.

Third, and the one people never think about: the dash is not flat. Injection-molded UTV dashes have draft angles, sink marks, and a surprising amount of variation between machines. That is why our covers have a compliant flange rather than a rigid perimeter — a couple tenths of built-in give lets the panel conform instead of rocking on a high spot.

We test-fit every batch. One part from every print run goes onto a reference chassis before the rest are packed. It costs us a few minutes per batch and it is the difference between a part you install in five minutes and a part you post an angry review about.