I used two AD9371（USRP N310）, one as tx and one as rx.
tx sends QPSK data, rx receives data and calculates EVM.
The calculation results show that the EVM of different sub-carriers differ greatly. What may be the cause of this phenomenon？
One reason could be difference in reference clocks. Are you providing same external reference clock to both the evaluation boards ?
If you test the boards separately , Tx monitor is spectrum analyzer and…
Can you share the constellation plot?
If you test the boards separately , Tx monitor is spectrum analyzer and Rx signal send from signal generator are the boards working fine?
The two AD9371s use different external reference clock. Maybe this is the reason why the EVM of different sub-carriers are quite different.
When I use one AD9371 to conduct similar test, the maximum difference of EVM of different subcarriers is 3.5dB. Is this a normal phenomenon?
This is a constellation plot of two different subcarriers. It is obvious that the EVM of the first subcarrier is smaller than that of the second subcarrier
Looks Good. What RF frequency are you measuring this ?
Lower frequency will have better EVM as phase noise is better.