Uneven light on a light guide column is usually blamed on the LEDs. In most projects we have seen, the lamps were fine and three other numbers were wrong: the mixing distance at the entry end, the tooth pitch on the exit face, and the draft angle on the side walls. Each one controls a different symptom, so measure them one at a time before replacing anything.

The grid of bright dots on the exit face is the exit teeth speaking. Every tooth is a tiny reflector that throws one bright point onto the surface, so the tooth pitch equals the spacing of those points. At arm's length, an indicator light reads perfectly smooth with a 2.0 to 2.5 mm pitch; a display zone viewed close to the face needs 1.0 to 1.2 mm. Going below 1.0 mm is a trap: the tooth root turns needle-sharp, the mould becomes hard to machine and the part scuffs on ejection. Choose the pitch from the viewing distance first, then let the mould tell you if you pushed too far.

The number most often missing from the drawing is the mixing distance: a plain straight section, normally 1.5 to 3 mm long, between the LED entry face and the toothed zone, where the light bounces a few times and blends into one field. When assembly space gets tight this section is the first thing structure cuts. Below 1 mm the square image of the LED projects straight onto the exit face, and no tooth pattern can recover it. Treat it like wall thickness or a snap-fit: write it into the structural input sheet and mark it non-compressible. Along with it, keep a 0.2 mm gap between the entry face and the LED, held by locating posts. Zero gap conducts the LED heat into the PMMA, which yellows over time; a larger gap simply loses light.

The side-wall draft decides where the light leaks out. Total internal reflection in PMMA has a critical angle of about 42 degrees, and any draft tilts the reflection angle. Half a degree to one degree is safe; past two degrees the wall starts glowing while the exit face goes dim. If a deep cavity forces more draft, polish the wall moulding to SPI A-2 or better so the leaked light is mirrored back; a rough wall just scatters it and drops efficiency a notch.

One more symptom worth naming: bright in the middle, dark at both ends, means the tooth density was kept constant along a column where the light fades with distance. Grade it. Start the far end at 1.5 to 2 times the tooth density of the near end, measure the brightness map after the first prototype, then adjust. Prototype money spent there is far cheaper than a mould rework later. And on material: PC works, but it transmits a few points less and runs yellow. In one identical structure, switching from PMMA to PC cost about 8 percent on the light meter, and adding LEDs to compensate was more expensive than just using PMMA.