for accurate measurements, calibration over frequency is needed. picking a more flat frequency range improves accuracy.
Thanks for your reply. To solve the problem mentioned above, the current solution is as follows: since the measured power signal is several kilowatts, it will be attenuated by a weak directional coupler before connecting to AD8362. In the attenuation of the coupler, the output signal power corresponding to the same power signal has a certain relationship with the frequency. In the use of AD8362, the output signal corresponding to the same power signal has a certain relationship with the frequency. If the two complementation can complete accurate measurement. Does this method correspond to the flatness of the adjustment frequency in your answer?