AnsweredAssumed Answered

AD9361 Rx CW signal very "dirty"

Question asked by dewilso4 on May 1, 2017
Latest reply on May 2, 2017 by Vinod

I am just starting to learn how to use the AD9361, and I seem to be having some issues getting a nice clean CW signal.  The tuner is installed on a custom board, with signal samples fed to an FPGA, then to a DSP for capture.  I pull out the samples then analyze in MATLAB.  These are the initialization parameters, sampling characteristics, and input signal parameters I am using:

- 26 MHz VCTCXO input

- 30.72 MHz sampling

- LO 900 MHz in one case, and 2147 MHz in another

- CW inject at 901 MHz in one case, and 2148 MHz in another, from -65 dBm down to -95 dBm

- with Rx FIR (filter coefficients unchanged from those specified in the default_init_param object in the ad9361 lib), and without Rx FIR (both scenarios seems about the same)


- I have inspected the DC power inputs with a scope at the chip and see no AC signals which might be mixing in.  

- I have inspected the 26 MHz VCTCXO input with a SpecA at the chip and see no harmonics which might be mixing in.  

- I have inspected the RF signal input with a SpecA at the chip, and it looks clean.

- I have ensured no data corruption in the AD9361/FPGA/DSP signal path by putting the AD9361 in BIST mode, generating a tone, and inspecting the samples in MATLAB.


Below are some MATLAB figures showing the time series and spectrum.  I am trying to understand what might be causing the signal to become so dirty.  Many thanks in advance!

time series CW signal, I/Q and magnitude

Spectrum, CW input, -65 dBm, 901 MHz