DAC38RF82: Spur generated from DAC

Part Number: DAC38RF82
Other Parts Discussed in Thread: LMX2592, LMX2592EVM,

Hello,

I am using DAC38RF82 interfaced with Kintex ultrascale FPGA.

I am finding a spur generated from DAC when sampling rate is 6.4 GSPS.

 This spur starting from 1.5GHz with power level 0dBm and spur power level is around -33dBm at 

Where as with 5 GSPS spur is present but power level is down.

I have attached 2 images where you can see spur.

We want to understand from where the spur is getting generated and how to arrest that. Kindly help us in this.

  • Naveen,

    Are you using the internal DAC PLL? Are you using any interpolation? What is the SYSREF frequency? Is SYSREF continuously running? Is the DACCLK from a clean source? Have you tried filtering this source? Are you using switching power supplies for the DAC or anywhere else on your board? If so, what frequency are they running at? 

    Regards,

    Jim 

  • Are you using the internal DAC PLL?

    Ans: No, it is generated external.

    Are you using any interpolation?

    Ans: We are not using any interpolation, it is operated in 8411 mode (8 bit)

    What is the SYSREF frequency

    Ans: Sysref frequency = 25MHz.

    Is SYSREF continuously running

    Ans: it is continuous sysref.

    Is the DACCLK from a clean source

    Ans: Yes, we are using LMX2592 to generate 6.4GHz. Clock is clean 

    Have you tried filtering this source?

    Ans: No, currently no provision to filter on board, But clock seems ok.

    Are you using switching power supplies for the DAC or anywhere else on your board?

    Ans: We are using power supplies similar to DAC EVM. Switching Frequency is around 2 MHz.

    Regards,

    Naveen

  • Naveen,

    What is the 5GHz spur you are talking about? I do not see this in your spectrum. What is the 1.5GHz spur at 0dB? makes no sense. With your settings, the worst spur I am seeing is at 1.594GHGz at -51dBm. Is your input tone 2.40250GHz? How did you come up with a SYSREF of 25MHz?

    I would suggest running the DAC in NCO only mode and see if any spurs are still present. In this mode with the NCO set to 2.4GHz, I see a spur at 86.68MHz at -59dBm and one at 1.60GHz @ -60dBm (2nd harmonic).

    Regards,

    Jim 

  • Hi Jim,

    With sampling rate of 6.4 GSPS spur is getting in the order (Fs/2) - Fin. The spur power considerably high from Fin = 1.7GHz

    With sampling rate 5 GSPS spur is getting generated but power level is around -50dBm as you can see in the attached picture.

    Sysref Frequency= LR/(10*1*32) where LR = 8 Gbps and  F=1 and K=32.

    We will check using NCO.

  • Naveen,

    This spur is a clock mixing spur. I think you need to filter your clock source. Are you using the single-ended clock input or differential clock input? What is the amplitude of this signal? Do you have an option to bring the clock in from an external source?

    Regards,

    Jim

  • Are you using the single-ended clock input or differential clock input?

    Ans: I am using differential clock input

    What is the amplitude of this signal?

    Ans: 1 dBm

    Do you have an option to bring the clock in from an external source?

    Ans: We don't have option to bring clock in from external source.

    Is there any way to reduce spur in DAC?

  • Naveen,

    You mention Fin is 1.7GHz but your plots have a marker at 2.4GHz. What Fin are you using? When sampling at 6.4GHz, does the spur amplitude change as you change the Fin? Does the spur reduce when SYSREF is turned off? Are there spurs when doing the NCO only test?

    Regards,

    Jim

    1004.DAC38RF82_NCO_Only_Test.pptx

  • Hi Jim,

    I have shared plot for only 2.4GHz

    My Fin is from 100MHz to 2.4GHz.

    When sampling at 6.4GHz, does the spur amplitude change as you change the Fin?

    Ans: Yes.

    Does the spur reduce when SYSREF is turned off?

    Ans: I didn't try by turning off SYSREF.

    Are there spurs when doing the NCO only test?

    Ans: Didn't test NCO till now.

  • So you are sweeping the Fin from 100Mhz to 2.4GHz with the sample rate at 6.4GHz. When you IF is at 1.5GHz, you start to see a spur, correct? If so what is the frequency of the spur. What happens when the IF goes from 1.5GHz to 2.4GHz? Is the spur gone? If not, does the frequency track with the IF? Does the amplitude change between 1.5GHz and 2.4GHz? Can you send more plots?

    If sampling at 5GHz, and you repeat the sweep test, you see the same spur but at a reduced amplitude. Have you tried sampling at other frequencies like 7GHz or 6.5GHz? Trying to see if the spurs are somehow related to the sample rate. 

    What is the  reference frequency used by the LMX device? Have you tried changing the VTUNE external filter components?  You may want to show your schematic on the High Speed Clocks forum to verify the LMX circuit is properly set for 6.4GHz operation. 

  •  Yes, I am sweeping Fin from 100Mhz to 2.4GHz. Spur is getting in the order (Fs/2) - Fin

    What happens when the IF goes from 1.5GHz to 2.4GHz? Is the spur gone?

    Ans: No.

    Below are the more plots

    Have you tried sampling at other frequencies like 7GHz or 6.5GHz? 

    Ans: No

    What is the  reference frequency used by the LMX device?

    Ans: 100MHz

    Have you tried changing the VTUNE external filter components

    Ans: No.

    Below is LMX Interface

  • Naveen,

    I do not see any of these spurs you are reporting. I am using a LMX2592EVM to provide a 6.4GHz clock to a DAC38RF82EVM. The data from this test is attached along with the configuration file used by both boards. Is there a chance you have some clock signals crossing the DAC output which might be causing these spurs? Please send your board layout if you would like for us to review it.

    Regards,

    Jim

    DAC38RF82_Fs_6.4G_811_int_1x.cfgLMX2592.tcsDAC38RF82_6400M_Int_1x_real_811.pptx

  • Hi Jim,

    Thanks for your inputs.

    Kindly share your ti mail id so that we can share layout with you as we cannot share the layout in open forum.

    Regards,

    Naveen.A