RF test in production - OTA vs conducted for IoT?

Conducted test is fast and repeatable but misses antenna issues; OTA catches them but slows the line. For IoT volumes, where did you draw the line between conducted and OTA testing?

For IoT volume, most lines we hear about use conducted as the workhorse and OTA as a gate, not the other way around.

A pattern that scales:
1. Conducted: 100% or high sample for TX power, freq error, ACLR / mask, RX sensitivity / PER on a golden fixture. Fast, repeatable, catches most silicon / soldering faults.
2. OTA / chamber: qualification + periodic audit (first article, weekly / lot sample, after antenna / enclosure change). Catches keep-out, detune, cable/antenna assembly, and "works on fixture, dies in housing".
3. Field / golden sample: a small set of units that travel with the antenna + enclosure revision so you can correlate conducted vs radiated when yields drift.

Draw the line by failure cost: if a bad antenna only shows up as returns, under-sampling OTA is expensive; if your antenna is a known-good module with a locked keep-out, heavy conducted + light OTA often wins. Share band, whether the antenna is on-board vs cable, and your scrap/return cost - that usually sets the sample rate.

— rfforum 编辑部 / Editorial Team