NI USRP as network analyzer

[nathanfj]

  I am curious as to whether the USRP hardware can be easily used as a network analyzer to measure transmission (S21).  I think that it would be relatively easy to change the RX and TX frequency, measure magnitude and phase, then change both frequencies and measure magnitude and phase again?  However, I do not know whether it is possible to make it so that the change in the measured phase means something?

  Specifically, I understand that in the USRP-2920 the RX and TX have the same 10MHz clock (so they are phase coherent when set to the same frequency), but that each channel has a different local oscillator so there is a random (but constant in time) phase shift between the two channels and this phase shift changes randomly whenever the LO frequency is changed.  Is there any way around this problem using the settings in the USRP hardware?  (I could use a switch outside the USRP to directly connect TX to RX and measure the constant phase shift after every frequency change before measuring , but this is somewhat undesired...)  Thank you for your help,

Nathan

[James-D]

Hi Nathan,

The USRP hardware is not designed to be a VNA. It is not calibrated in any way and this will make it very difficult to implement an application to measure any S-paramaters with this device. You are correct in your deductions concerning the phase differences between the TX and RX channels. The phase difference will likely change with each change in frequency. There aren't any settings that can be used with this hardware to get around this or compensate for this phase difference shift. You would need to implement a manual calibration procedure in software and this will likely be rather complex, requiring in depth, low-level knowledge of the theoretical and practical concepts of S-parameters. 

Bottom line, it will be very difficult and I have no knowledge of any other users that have implemented this functionality with the USRP. 

Regards,