• TI Thinks Resolved

LMX2571: Spikes in LMX2571 output spectrum

Part Number: LMX2571


I am using LMX2571 in PLL mode using internal VCO for synthesising two VHF frequencies. As per the design I need to switch between these two  frequencies alternatively  in a specified  time interval.

For this I have configured F1 and  F2 registers for these two separate frequencies respectively and for switching between these two signals, both Trctrl mode and Reg control mode was tried.

By observing the direct output from LMX2571 i.e ( output of RFoutTx & RFoutRx) following result was obtained ;

1. When F1 and F2 was alternatively  switched with TrCtrl mode / register select mode - spikes were present in the output spectrum.(snapshot of spectrum attached below)

2. When this F1 and F2  switching was disabled i.e either F1 or F2 is used - there were no spikes in the output spectrum.

Is there any way to remove the spikes from output spectrum for the above mentioned scenario?.

  • Vishnujith,

    This might be the way it is.

    I note the sweep time is 100 ms, which is likley relatively slow in comparison to the switching speed. So if you switch between frequencies A and B and spend equal time on both, I would expect to see spikes at frequencies of : A, B, (A+B)/2.

    Now this spur looks a little less than 80 kHz. If this is where you might expect a fractional spur, another possibility of this is due to settling of the PLL. For instance, you might try inceasing the loop bandwidth (due to higher charge pump current), or also considering using NP0 capacitors in your loop filter as I have seen spurs that jump up right after switching and settle down that are related to capacitor dielectrics.

  • In reply to Dean Banerjee:

    Hi Dean,

    Thank you for your prompt reply. 

    In the above mention issue the spikes are not fixed at certain frequencies rather they come and go at multiple frequencies on both sides of the centre frequency

    *NOTE :This spikes are present only when Rx/Tx switching is enabled ( either by register mode or Trctrl mode).

    I have attached a video of the scenario, i think this will explain the issue in detail.



  • In reply to Vishnujith KJ:

    Hi Vishnujith,

    This is likely a measurement issue. The SpecAn is not synchronized with TxRx toggling. You should connect TrCtrl to the SpecAn's "external trigger". Setup the SpecAn with external trigger and make sure the sweep time is shorter than the TxRx toggling time.
  • In reply to Noel Fung:

    Hi Noel,

    Thank you for your prompt reply. As you suggested I have connected the TrCtrl with Ext trigger of spectrum analyser with the following sweep time,( were Tx/Rx toggling time is 30ms)

    1. sweep time - 30ms. when this sweep time was configured the spectrum was not visible . The attachment below will show the relevant observation.

    2. sweep time - 80ms ( some portion of the spectrum was visible with spikes)

    In addition I have measured Spectrum of another DMR Radio with the same Spectrum analyser settings as of case 2.( sweep time 80ms ). The attachment below shows the relevant details.

    So even with the same Tx/Rx toggling time ( 30ms), with the same Spectrum analyzer settings the output from other DMR radio was fine.

    Hope you will provide alternative solution for this spike issue due to switching of Rx/Tx of LMX2571.

  • In reply to Vishnujith KJ:


    Assuming that this is not a measurement issue (not sure if we can assume this), then this would likely be related to the switching speed of the PLL. If you change the charge pump current, this should have a large impact on this switching speed and therefore your spurs. But if you change this charge pump current and there is no impact, then it is harder to believe that this is due to the switching speed of the PLL.

    Another experiment is to keep the spectrum analyzer and the observe frequency the same, but change the other frequency you change to. This would impact the VCO calibration time (among with many other things).

    Now if one of the above experiments impacts this spur, then you can perhaps impact this spur by increasing the loop bandwidth of the system. On the other hand, if there is no impact, it suggests that maybe this could be measurement related.