R
Member
Low-loss devices look easy to measure until the numbers stop repeating. A filter or cable with only a fraction of a dB of loss sits close to the noise floor, and that is where small setup choices start to dominate the result.
The first thing I check is dynamic range, not the headline specification. Reduce the IF bandwidth and the trace cleans up, but the sweep slows down and drift creeps in; widen it and the measurement is fast but noisy. I usually start narrow, then widen only until the trace noise is about a tenth of the loss I am trying to resolve. Second, averaging helps only if the drift is under control. Averaging a drifting trace gives a smooth but wrong answer, so I re-check the calibration and the connector torque before I trust an averaged number. Third, cable movement matters more than people expect: flex a test cable during a low-loss measurement and watch the trace move. If it moves, the fixture, not the device, is what you are measuring.
A useful sanity check is to measure a known through standard right after the device, under the same settings. If the through does not read back cleanly, fix the setup before arguing about the device. What is the smallest loss you have measured confidently, and what limited you first, noise or drift?
The first thing I check is dynamic range, not the headline specification. Reduce the IF bandwidth and the trace cleans up, but the sweep slows down and drift creeps in; widen it and the measurement is fast but noisy. I usually start narrow, then widen only until the trace noise is about a tenth of the loss I am trying to resolve. Second, averaging helps only if the drift is under control. Averaging a drifting trace gives a smooth but wrong answer, so I re-check the calibration and the connector torque before I trust an averaged number. Third, cable movement matters more than people expect: flex a test cable during a low-loss measurement and watch the trace move. If it moves, the fixture, not the device, is what you are measuring.
A useful sanity check is to measure a known through standard right after the device, under the same settings. If the through does not read back cleanly, fix the setup before arguing about the device. What is the smallest loss you have measured confidently, and what limited you first, noise or drift?
