AD9375 MYKONOS_radioOn() Tx RF noise spike

We're using the AD9375 in a relatively slow Frequency Hopping application. During the idle time between FHOP pulses we retune the Tx LO according to the procedure in UG-992. These are small re-tune increments (< 1MHz), so we're following the Small Frequency Step Procedure. When we call MYKONOS_radioOn() after changing the TX_PLL frequency, we are seeing an RF noise spike on the Tx output (see zero span plot below. In this plot the hop frequency is kept constant for debug, but we're still calling radioOff()/radioOn()). The second plot is a spectrum plot that fills in as radioOn() is called repeatedly during a transmission where the Tx data is held at zero.

1. Is a Tx RF noise spike expected when executing MYKONOS_radioOn() following a re-tune of the Tx PLL?
2. Do you have any suggestions for eliminating or minimizing the noise spike?

thanks.

  • 0
    •  Analog Employees 
    on Nov 9, 2020 8:56 AM 2 months ago

    Are you using a custom board or eval board? Can you disable the tracking cals and then check if you are seeing this issue.

  • We tried disabling the tracking calibrations, but that did not reduce the noise spike.

    The spectrum analyzer plots are from our custom board, but we are attempting to replicate this scenario on the evaluation board.

    Our custom board design is based on hdl_2018_r1 and the associated No-OS drivers.

  • 0
    •  Analog Employees 
    on Nov 10, 2020 9:05 AM 2 months ago in reply to tstrickland

    Is the spur staying if you remain in radio off position for sometime? Also, can you change the frequency and then check if you are seeing the spur while doing radio off/on? Check the power supply part of your custom board as it may happen because of some coupling that is going to the output.

  • Yes, the spike occurs at radioOn() even if the radio has been off for a long time (> 1 sec).
    And the spike occurs whether the frequency is changed or not.

    We replicated this scenario on the AD9371 eval boards using the KCU105 development platform and the same FPGA code tree (hdl_2018_r1) and No-OS drivers.

    Here we’re not transmitting anything, and we’re not changing the LO frequency. We’re just turning the radio on and off by looping on MYKONOS_radioOn()/delay/MYKONOS_radioOff()/delay.

    A spike occurs at the LO frequency at radioOn().
    The noise floor jumps up when the radio is on as well.

    Please let us know if you have any suggestions for reducing or eliminating this spike.

    thanks.

    AD9371 wideband eval card on KCU105 platform
    Zero span 2ms

    AD9371 wideband eval card on KCU105 platform
    Zero span 100ms

    AD9371 narrowband eval card on KCU105 platform
    Zero span 1ms

    AD9371 narrowband eval card on KCU105 platform
    Zero span 200 ms

  • 0
    •  Analog Employees 
    on Nov 11, 2020 9:08 AM 2 months ago in reply to tstrickland

    Can you disable LOL tracking and then check if you are seeing the spike?

    What is the impact of this spike on your system?