Post Go back to editing

FMCOMMS5+ZYNQ3 unstable received baseband spectrum in qpsk Simulink example

Category: Software
Product Number: FMCOMMS5

Hello All,

I executed QPSK example provided by Mathwork in Simulink, the received baseband spectrum of the signal is changing amplitude a lot! I looked at the I and Q time domain received data, the envelope of both I and Q varying unexpected and there are many spikes on the I and Q data.

I just loop backed the transmitter interface to receiver interface with a short coax cable, the cable is tested!

Here the retrieved data:
L` X@A0 X@A0, L` X@A0 X@A0, L` X@A0 X@A0, L` X@A0 X@A0, L` X@A0 X@A0, L` X@A0 X@A0, c<!;8UY#A<!9 `ello world 006 Hello world 007 Hello world 008 Hello world 009 Hello world 010 Hello world 011 Hello world 012 Hello world 013 Hello world 014 Hello world 015 Hello world 016 Hello world 017 Hello world 018 Hello world 019 Hello world 020 Hello world 021 Hello world 022 Hello world 023 Hello world 024 Hello world 025 Hello world 026 Hello world 027 Hello world 028 Hello world 029 Hello world 030 Hello world 031 Hello world 032 Hello world 033 Hello world 034 Hello world 035 Hello world 036 Hello world 037 Hello world 038 Hello world 039 Hello world 040 Hello world 041 Hello world 042 Hello world 043 Hello world 044 Hello world 045 Hello world 046 Hello world 047 Hello world 048 Hello world 049 Hello world 050 Hello world 051 Hello world 052 Hello world 053 Hello world 054 R1L% $b)J&I :1L% $b)J&I b1L% $b)J&I b1L% $b)J&I b1L% $b)J&I b1L% $b)J&I b1L% $b)J&I b1L% $b)J&I b1L% $b)J&I b1L% $b)J&I b1L% $b)J&I b1L% $b)J&I b1L% $b)J&I b1L% $b)J&I b1L% $b)J&I b1L% $b)J&I b1L% $b)J&I b1L% $b)J&I b1L% $b)J&I b1L% $b)J&I b1L% $b)J&I b1L% $b)J&I b1L% $b)J&I b1L% $b)J&I b1L% $b)J&I b1L% $b)J&I b1L% $b)J&I b1L% $b)J&I b1L% $b)J&I b1L% $b)J&I b1L% $b)J&I b1L% $b)J&I b1L% $b)J&I b1L% $b)J&I b1L% $b)J&I b1L% $b)J&I b1L% $b)J&I b1L% $b)J&I b1L% $b)J&I b1L% $b)J&
 
As can be seen part of transmitted data retrieved correctly!!

Does anybody know what are the causes of spikes in the received IQ data?

what are the causes of I/Q envelope variations over time? Could it be discontinuity on the ADC and DAC?

Please find attached received I/Q data.

zoomed version:



Kind Regards

Arash



info added!
[edited by: Arashjafari at 3:30 PM (GMT -4) on 28 Jun 2022]
  • Can you please check the IQ data by feeding the CW tone, also can you please provide more details like how setup and connectivity to understand more about the problem.

  • Here I generated a CW tone in the Simulink, send it to FMCOMMS5 transmitter block, received it at FMCOMMS5 receiver block. please find the Simulink setup an result below:

    I checked the generated TX CW on the spectrum analyser, the CW signal is stable in Clear/write trace mode and low sweep time. When I stopped the simulation there is still a signal centered at LO frequency(e.g 2.4 GHz) with -60 dBm power level. It means stopping BB data does not disconnect the LO from analog mixer which is bad. Is there a way to stop internal LO signal generation ?

    I performed another test by feeding the RX interface with a CW tone generated from external signal generator(e.g R&S). The received data seems to be much better:

    The offset from 0 Hz is coming from unsynchronized LO(external generator and on-board LO).


    I'm getting this error at the boot

    [ 437.434037] ad9361 spi32766.1: ad9361_dig_tune_delay: Tuning RX FAILED!

    Maybe this can cause some problems.

    All in all, I executed the QPSK example with default configuration generated by MATHWORK, I didnt change anything, I suppose the BER should be almost 0 with a loopback cable.

    Kind Regards

    Arash

  • It means stopping BB data does not disconnect the LO from analog mixer which is bad. Is there a way to stop internal LO signal generation ?

    Yes , this transceiver is of direct conversion type and hence there will be LO leakage into the output. Internal LO leakage tracking calibrations are run inside the chip, which causes better rejection of the LO leakage. As per datasheet, LOL  should be around -50dBFs at the TX output, so your observation is correct.

    he offset from 0 Hz is coming from unsynchronized LO(external generator and on-board LO).

    In the plot at the RX output, the image power is same as that of the tone power,, which is not correct. What is the LO frequency for RX and TX? What is the tone offset from LO that you are feeding from generator?Are you running RX quadrature tracking cal? Send us a RX capture in frequency domain with markers on tone ,Lo and image.

    I'm getting this error at the boot

    [ 437.434037] ad9361 spi32766.1: ad9361_dig_tune_delay: Tuning RX FAILED!

    Are you using the latest drivers for configuring the chip? Are you using no-OS drivers?

  • Dear Srimoyi,

    Thank you for the reply!

    Are you using the latest drivers for configuring the chip? Are you using no-OS drivers?

    No, I'm not using Linux kernel drivers and libiio from the ADI , I'm using MATLAB authorised drivers, I think MATLAB uses Linux Kernel as well.



    In the plot at the RX output, the image power is same as that of the tone power,, which is not correct. What is the LO frequency for RX and TX? What is the tone offset from LO that you are feeding from generator?Are you running RX quadrature tracking cal? Send us a RX capture in frequency domain with markers on tone ,Lo and image.

    The LO frequency is set to 2.4 GHz, tone offset is around 16.28 kHz. Yes the quadrature error correction is enabled. 

    Kind Regards

    Arash

  • If you keep the RX input tone offset at say 2MHz from RXLO, are you seeing the same output? Can you share the snippet of your FFT plot? Are you feeding the input into 1st channel RX1 or 2nd channel RX2?

  • Hello,

    No changes observed, the image is still there.



    What do you mean with snippet of mz FFT plot, I|m using real data time domain input, the span is automatically set according to the BB sample rate of mz receiver block.

    I'm using channel number 1, RX1.

    Kind Regards

    Arash

  • Dear Sirmozi,

    The more important issue I do have is the I/Q envelope variations while the signal from transmitter loopbacked to the receiver.

    Kind Regards

    Arash

  • I|m using real data time domain input, the span is automatically set according to the BB sample rate of mz receiver block.

    Are you feeding real data to the input?In that case, the output spectrum is correct and you will see both image and tone at same power levels as on either side of the LO. You need to feed complex signal in order to see the image rejection at the output.

    The more important issue I do have is the I/Q envelope variations while the signal from transmitter loopbacked to the receiver.

    Can you try loopback with external LO source for RX and TX LO and then check if you are seeing the same amplitude variation?

    Refer to the below link:

     RE: QPSK Modulation Issues in ADRV9361_z7035  

  • I|m using real data time domain input, the span is automatically set according to the BB sample rate of mz receiver block.

    Are you feeding real data to the input?In that case, the output spectrum is correct and you will see both image and tone at same power levels as on either side of the LO. You need to feed complex signal in order to see the image rejection at the output.

    The more important issue I do have is the I/Q envelope variations while the signal from transmitter loopbacked to the receiver.

    Can you try loopback with external LO source for RX and TX LO and then check if you are seeing the same amplitude variation?

    Refer to the below link:

     RE: QPSK Modulation Issues in ADRV9361_z7035  

  • I|m using real data time domain input, the span is automatically set according to the BB sample rate of mz receiver block.

    Are you feeding real data to the input?In that case, the output spectrum is correct and you will see both image and tone at same power levels as on either side of the LO. You need to feed complex signal in order to see the image rejection at the output.

    The more important issue I do have is the I/Q envelope variations while the signal from transmitter loopbacked to the receiver.

    Can you try loopback with external LO source for RX and TX LO and then check if you are seeing the same amplitude variation?

    Refer to the below link:

     RE: QPSK Modulation Issues in ADRV9361_z7035