AD9361 Reference Clock Harmonics

Dear ADI Team

My customer is using AD9361. They have done measurements with AD-FMCOMMS3-EBZ_Evalboard with AD9361 (on-board 80 MHz reference used). They have carried out measurements for all channels for GSM/ DCS/ PCS.  However they see a lot of spurs in the spectrum.

From the frequency contour plots they conclude that the majority of the spurs is caused by harmonics of the reference clock (i.e. n x 80MHz or n x 40 MHz). This was confirmed by using an external signal generator  and using other values of the reference clock frequency (e.g. reference clock frequency of 65 MHz). Then they still see the spurs on the multiples of 65 MHz.

Can you please advise how to improve on the spurs observed in our measurements?

Thanks

IK

  • 0
    •  Analog Employees 
    on Jan 25, 2021 3:53 PM 1 month ago

    Yes these spurs at multiples of the ref_clk frequency, will be there and are seen on eval board. You can use an external filter to eliminate these spurs or you can use frequency pan accordingly so that the spurs donot fall in your band of interest,

  • 0
    •  Analog Employees 
    on Jan 26, 2021 3:53 PM 1 month ago in reply to srimoyi

    Please see the feedback from the customer:

    1.Where shall we put the filters at the clock input of AD9361? Or should we suppress them at the RF output. ? In the latter case, this will be problematic because on certain channels the spurs are very close to the carrier.

     

    2. The carrier channel frequenices are given by the standard, regarding the frequency plan, do you suggest to dynamically use different reference clock frequencies? This would imply having to use two VCXOs and switch the reference frequency during operation which looks to be difficult.

    Could you please elaborate both options you provided in a bit more detail?

    Thanks

    IK

  • 0
    •  Analog Employees 
    on Jan 29, 2021 6:25 AM 28 days ago in reply to IrfanKhan

    You can frequency plan and select the ref_clk such that its harmonics fall out of your band of interest and then filter the Rf output to remove the harmonics.

    Yes. I was talking about band pass filtering at the RF output.