Delay in time between AD9371 Rx and observation samples.

Hi,

I am receiving data from both Rx and ObsRx at the same time, same frequency on AD9371. I know AD9371 does not support phase synchronization. 

I calculated the amplitude of the samples for both channels using Amp = (I^2+Q^2). Based on the equations, even if there is a phase delay between the channels it should not affect the amplitude but only the phase. 

I have done two experiments, 

1- Setting both Rx and ObsRx to 122.88Msps

2- Setting Rx to 122Msps and ObsRx to 2*122.88Msps, ignoring every other sample

From the results of 1, I can still see a random delay (varying up to 400ns) between received amplitudes but from 2, the random delay is very small (32ns). 

Would you please let me know if there could be delay between channels when amplitude are captured?

Is there any explanation for the experiment results above? 

Thanks

  • 0
    •  Analog Employees 
    on Mar 23, 2021 4:24 AM

    What is the LO used in ORx path ? I guess the issue here is because of two different LO's (which are not synced).

    Is your phase difference constant on different power ups.

    Can you try giving external LO and make ORx LO Tx Lo. ?

  • Thanks Vinod for your help,

    What is the LO used in ORx path ?

    ORx using sniffer LO (OBSLO_SNIFFER_PLL). Yes, the Rx and ORx are using different LOs.

    When LOs are not synced, does it result in amplitude time shift?

    What about the second experiment I have done? 

    Is your phase difference constant on different power ups.

    For the experiment #1, it is varying on every power up. But #2, it changes a bit (one to four samples that is at most 32ns).

    Can you try giving external LO and make ORx LO Tx Lo. ?

    I do not have access to an external LO now, but I tried Tx LO (OBSLO_TX_PLL) for ORx and the results are still showing random delays between Rx and ORx when both are set to 122.88Msps.

    Could you please try the external LO at your end and let me know the results?

  • 0
    •  Analog Employees 
    on Mar 25, 2021 4:20 PM in reply to mamad2

    What is your application and why do you want phase sync between RX and ORX? Also, can you try by changing the sampling rate and then see the difference?

  • What is your application and why do you want phase sync between RX and ORX?

    Time delay of arrival. We are receiving signals from two channels and calculating the time delay difference between them. 

    Also, can you try by changing the sampling rate and then see the difference?

    I tried and got the similar results. 

    The main question I am asking is that what is the reason experiment #2 I have done is giving me a good results but not #1?  I know the ORx is providing 2 samples per each clock while Rx only one. So, if both channels are set to the same sampling rate, ORx has a clock frequency that is half of Rx in the FPGA design. While, if I set ORx sampling rate two times the Rx one's, they both have the sample clock frequency. Does this could be a reason? 

    Thanks for your help

  • 0
    •  Analog Employees 
    on Apr 1, 2021 11:15 AM in reply to mamad2

    Yes, the delay can vary because of the change in sample rate. The delay through channels depend on the sampling rate and the rate at which the filters are clocked. Also check if the FIFO depth is varying or not for both the cases.