How linear is the power sweep? Can we do a source power calibration as a function of frequency and assume that the power sweep is accurate and linear?
It depends on the power linearity specification of the network analyzer. For the N52300-020 PNA-L network analyzers, it is 1.5 to 2 dB. So theoretically, you could be up to 2 dB off, which is significant. Most analyzer, in a narrow frequency range, and narrow power range, will have better linearity. For example, a quick test of one of the analyzers at the factory showed that for the above frequency and power setting, the linearity error was less than 0.1 dB. You can perform an experiment (described below) to determine the linearity in your setup. Set the trace to S21, frequency sweep, with a power level in your power sweep range, let’s say -15 dBm, 1800 to 2117 MHz. Perform a source power cal at -15 dBm. Then copy channel 1 to channel 2. On channel 1, change the sweep type to power sweep, with a CW frequency of 1800 MHz, and the desired power sweep range (lets say 30 to 0 dBm). On channel 2, change the sweep type to power sweep, set the sweep range to the same as channel 1 (-30 to 0 dB
