I have installed PNA in atteched configuration. Sometimes value of CH B/A is 5dB apprx. higher than expected , if I preset NA every thing become OK. As it is observed purely randomly I don’t know what to do. However, I have fixed it using preset every time before measurement. This problem is fixed not resolved. Any suggestion regarding this is greatly appreciated.
When the problem occurs, can you check to see if you have changed the values of the source or receiver attenuators. In normal mode, they are changed to compensate for offset so the test port power remains correct. Sometimes this leads to issues when the test port is bypassed. There are some preferences that can be set to turn off the mathematical compensation. Normally, I would not think this is an issue with your setup, but something to check.
From your PM to me, you say nothing changed. So the only thing to do next is see what is different between good and bad. Take your good state, single sweep and leave in hold, save the state as a .csa file; if you see the problem again, hold the sweep, save the state as a .csa file. Post the two files and we can see why they are different.
You say everything is fine after preset, but then you don't say how it becomes bad. does it drift (you PM said no drift was seen); does it just jump, does someone set a magnitude offset? How does it go from good to bad, suddenly and unexpectedy while no one is watching? Usually, something must be done to trigger a state change. Maybe the step attenuator is getting flogged by accident without you knowing it? Is it a 5 dB step attenuator you have in that box? It would be an odd hardware problem, never heard of it before, but there's always a first time.
In looking at your instrument states, we noticed that the power levels were different. In one case it was set to -10 dBm and in the other it was -2 dBm. At - 2 dBm, you are probably compressing the 83020A amplifier, which would cause a drop in power in the B receiver relative to the A receiver. This makes the B/A trace lower than that at -10 dBm, which is what you observed.