Post Go back to editing

AD9082 has wrong points at 5G sampling rate

Category: Software

Test environment parameters:
1. ZU19MPSOC + petalinux2018.2 + adilinux API
2. Input parameter clock 1280M, ADC sampling clock 5120M
3. CDDC mixing to 1700M, FDDC not mixing
4. IQ mode, LMFS=8222
At 5120 sampling rate, there are the following 3 problems
1. Under the 2x decimation of CDDC and 2x decimation of FDDC, there are errors after data collection
As shown

2. CDDC is extracted by 4 times, and FDDC is extracted by 1 times. There is a bulge after data collection, but there is no wrong point
As shown

3. Under the 2x decimation of CDDC and 1x decimation of FDDC. After the positive wave is amplified, the waveform is not smooth.
As shown

4. Why is there a wrong point in the data when the 2x decimation of CDDC?

  • Hello,

    Comments to your observations are as follows:
    1)   With an ADC rate of 5120 MSPS, CDDC=2x decimation and FDDC=2x decimation of FDDC is not a valid mode of operation since the interface data rate between CDDC output and FDDC input exceeds max limit of 1500 MSPS.   The "spikes" observed in the waveform suggest a timing hand-off issue.  Solution is to operate with 4x decimation to avoid issue.

    2) For large scale input signals at high input frequencies, the noise spectral density of the AD9082's ADC will show a hump.  This can be even be seen in the datasheet plot shown below for FIN=1.8 GHz (no decimation).  The mechanism causing this "noise hump" is discussed in UG1578.  To minimize this hump, it is important to set the odd or even   "Nyquist Zone" parameter according to UG1578.


    3)  and 4)                  Refer to explanation in 1) since CDDC = 2x.

  • Hello,

    thumbs up, thanks!!!
    For 1), when I re-read the  UG1578 information yesterday, I saw a paragraph in it, and I understood what was going on.
    For 2), the noise floor of the collected data has a hump. After reading your reply, I went to see the API and configuration of the Nyquist zone, and found that the description of the API is different from the description of the manual  UG1578. I always thought that my configuration was correct。
    Thank you very much!!!