SDR for production test - USRP good enough?

Considering USRP for production RF testing instead of dedicated instruments. Cost is attractive. Is measurement accuracy good enough for production pass/fail?

A USRP (or similar SDR) can be a solid production *stimulus / monitor* if you own the uncertainty budget; it is rarely a drop-in replacement for a calibrated VNA/SA/SG chain when the gate is absolute power, ACLR, or EVM against a customer limit.

What usually works:
- Relative / comparative tests: golden-unit delta, channel power vs a characterized reference, spur presence/absence, simple BER/PER with a fixed gold radio
- Absolute metrology: needs an external cal path (power sensor, calibrated attenuator, periodic SA cross-check) and a written uncertainty stack — “USRP dBFS → dBm” without that is a lab convenience, not a production authority

Practical pattern: SDR for speed + rich I/Q debug, and a thin calibrated instrument (or a sensor + cal) as the pass/fail referee for the lines that appear on the COC. If your limit is loose (smoke / functional RF) and you fixture-cal every shift against a gold unit, USRP-class hardware is often enough; if the limit is a datasheet dB number, budget for one traceable meter.

What limits are on the traveler (power ±XdB, EVM, ACLR, frequency error), and what is the annual volume?

— rfforum 编辑部 / Editorial Team