Post Go back to editing

AD6688 has spurs at the input

Category: Hardware
Product Number: AD6688

Hello,

We have a custom designed PCB for AD6688 connected to an Kintex UltraScale+ FPGA. We have JESD Receiver core running on the FPGA to receive the ADC data.

The ADC is setup with the following configuration:

fs = 3GHz,  Real input, only enable ddc0, NCO=1/4fs, Decimation Rate=2, enable HB1 filter and 6dB gain, Complex output, I/Q data rate =1500MSPS

L=8, M=2, F=1, S=2, K=16, N'=16,HD=1

The Receiver IP is setup for a lane rate of 7.5 Gbps.

The input sampling clock of AD6688 is multiplied by 2594, and the reference input clock of 2594 is 100Mhz. The 204B link reference clock of FPGA and ADC is provided by LMK04828, and the frequency is 187.5MHz. They are refer to that same 100MHz clock.

When the normal configuration of this adc is completed and the link between ADC and FPGA is completed, the input port of ADC is tested by spectrometer and it is found that there is 750MHz input spur.

Could you please help me figure out why this spur is observed and how it can be minimised.

Regards.

  • Hi  

    Thanks for using AD6688.

    Could you please provide a screenshot of the FFT?

  • Hi JAlipio

    Our application is to do RTSA function on the spectrometer, and this is the fft spectrum corresponding to the span size.As shown in the following figure

    I tested at the input port of ADC with a spectrometer and found that there was a spur of 750MHz with an amplitude of about -117dbm.As shown in the following figure

  • Hello,

    Can you explain why your FFT display shows a -88.5 dBm spur at 1 GHz when the AD6688 is configured for 3 GSPS operation with complex NCO set to 750 MHz along with dec-by-2.   If you are concerned about a spur that measures to be -117 dbm at the ADC's differential input(?) than one would expect the narrow span FFT display to be centered around DC....not 1 GHz?


    The AD6688 has a 1.7 Vpk-pk input with 200 ohm input impedance hence would require a 2.6 dBm CW tone at its input (assume perfect matching and lossless balun) to provide a 0 dBFS level on its FFT.   The FFT display shows marker levels referenced to dBm.  Does this imply that the actual spur level at 1 GHz is  -91 dBFS level?

    With regard to origin of 750 MHz spur, it is worth noting that the internal DSP of AD6688 operates at a fabric rate of 750 MHz so it is possible that this may produce a low level spur to appear due to digital coupling back to analog input?  Another possible cause could be the 187.5 MHz reference signal of which its 4th harmonic would appear at 750 MHz.................albeit if the signal is a square wave with near 50% duty cycle, the odd order harmonics would be of the greatest magnitude thus more likely to an issue than the 4th harmonic.