We recently started using the preselector center function of E4448A as we found that the YTF input filter was off at 15.5 GHz and the measured amplitude was varying excessively over time. While that change improved the repeatability of measurements at 15.5 GHz, the repeatability at 14.5 GHz actually decreased. We were seeing about 1.3 dB of change over a period of about 8 hours. So I did the same measurement manually with a E8257D generator as the input with about -10 dBm input. The maximum amplitude on the PSA was found at about -11.7 MHz offset but the preselector center function set the center frequency to -4.7 MHz offset. With a pulsed input, the preselector center function would usually find the correct offset but sometimes find the lower offset. As that is what we are testing the UUT with, it explains why our test results are no longer stable. I checked a total of 8 frequencies. With some frequencies the preselector center function worked fairly well and others not so well. I checked another E4448A at 14.5 GHz (CW) and the preselector center function had about the amount of error. Is this type of error supposed to be caught during the regular 1 year calibration?