What might be the reason, that sometimes (not every time) near 8GHz in all 4 S-parameters occur a peak, which deep is about 0.5dB (S21 and S12). During the measurement calibration process (full 2-port 50MHz-18GHz with 85054D calkit) such a peak is not visible on screen, but performing thru measurement after calibration such a peak occur. We have used different testport cables and calkits, so the problem can't be on that sector. All connectors and calkits components are properly cleaned, gaged and calibrated. We have also measurered the output power from each testports and evertyhing looks fine and meet the specs on that frequency range. Can be the problem source be in the test set, that pin diode dosen't switch excatly the same way every time? All advices are welcome now. Of course we can send our 8722ES for the Agilent repair and service, but the biggest problem is that the peak dosen't occur every time.