• State TI Thinks Resolved
  • Locked Locked
  • Replies 6 replies
  • Answers 1 answer
  • Subscribers 46 subscribers
  • Views 729 views
  • Users 0 members are here
Support feedback
Options
Options
Related

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

ADC12DJ3200EVM: Why are there ripples in the frequency response of ADC12DJ3200EVM

Brandon Buckley
Prodigy 135 points
Part Number: ADC12DJ3200EVM
Other Parts Discussed in Thread: ADC12DJ3200, , ADC12DJ5200RF

Hello,

We are intending to use the ADC12DJ3200EVM to under sample signals above Nyquist, and are therefore interested in the full frequency response of the ADC. The specification of the ADC12DJ3200 claims usable input frequency range of >10 GHz. However, in Figure 14 of the attached data sheet for the ADC12DJ3200EVM eval board, there are strong ripples observed above 3GHz. We have also measured these ourselves in the lab. Such ripples are not observed in the data sheet for the ADC12DJ3200 alone.

Our concern is that these ripples are due to impedance mismatches on the EVM board at frequencies above 3GHz. If so, this will have negative impacts on our system performance.

We would like to get answers to the following questions:

1) What is the source of these ripples?

2) Is there a way we can reduce these ripples in the frequency response of our ADC12DJ3200EVM though a hardware fix or other modification?

3) Does TI offer other solutions, or can they recommend other eval boards, that have a smoother frequency response across the full frequency range of the ADC12DJ3200 (DC-10GHz)?

Thank you,

Brandon Buckley

tidud15.pdf

  • 0
    TI__Mastermind 41340 points

    Hi Brandon,

    The source of the ripples are the standing waves that are accumulated at the higher, GHz frequencies when the EVM is swept for passband flatness.

    The way reduce these ripples is somewhat empirical and requires a Tpad structure between the output of the balun and inputs of the converter. This seems to work best. See the attached picture for reference. The downside is that this adds loss on the frontend and more gain will be required on the primary of the balun or preceding stage.

    Lastly, per your comment on the 10GHz IF sampling. Full power BW is the converter's BW to function properly and acquire a sample accurately before it moves onto the next sample acquisition. In the datasheet the full power BW of the DJ3200 is 8GHz. There are some apps notes on the web that discuss these differences in terms.

    The user or sample BW of the converter is much less, I will check on this with design. My concern is that your application won't acquire accurate samples out at 10GHz. May want look into using a wideband T/H in front of the converter, but I realize this makes the system more complicated.

    Regards,

    Rob

  • 0 in reply to Rob Reeder
    Prodigy 135 points

    Hi Rob,

    I really appreciate your response.

    I have a few follow up questions:

    1) Are the standing waves due to impedance mismatches? If so, where? Between the ADC and the balun?

    2) How it the T-pad reducing the standing waves? Is it improving impedance matching of the balun?

    3) Is there a reason the T-pad was not included in the eval board? Are there boards available that have the T-pad and therefore reduced ripple?

    Thanks,

    Brandon

  • 0 in reply to Brandon Buckley
    TI__Mastermind 41340 points

    Brandon,

    See my answers below to your questions:

    1) Are the standing waves due to impedance mismatches? RR: yes If so, where? Between the ADC and the balun? RR: yes

    2) How it the T-pad reducing the standing waves? Is it improving impedance matching of the balun? RR: yes, that is correct, its an unmatched pad which attenuates the standing waves that go back and forth between the balun and input.

    3) Is there a reason the T-pad was not included in the eval board? RR: I am not sure, I am new to TI. I believe they didn't expect customers to use it this far out in frequency. I am getting some confirmation on that. Are there boards available that have the T-pad and therefore reduced ripple? RR: I don't think so, but again, I am asking for confirmation on this to see if anything is available.

    Hope this helps.

    Regards,

    Rob

  • 0 in reply to Rob Reeder
    TI__Mastermind 41340 points

    Hi Brandon,

    To get back with you on question 3, there are other boards that have the Tpad available between the output of the balun and input of the ADC. There are a couple ways this can be done. The ADC12DJ5200RF ADC part, has the same pinout and is pin compatible.

    I would recommend purchasing the 5200RF EVM and either using the 5200RF ADC to do your measurements, or have a DJ3200 ADC swapped out on the 5200RF EVM.

    Hopefully, that helps.

    Regards,

    Rob

  • 0 in reply to Rob Reeder
    Prodigy 135 points

    Hi Rob,

    That's great to hear. I looked at the datasheet online but couldn't find a frequency response curve. Do you know how much ripple suppression this new board achieves?

    Thanks!

    -Brandon

  • 0 in reply to Brandon Buckley
    TI__Mastermind 41340 points

    Hi Brandon,

    Some ripple suppression is seen, but probably not as much as you may want.

    What is the goal of the input network? 10GHz@-3dB? Passband flatness of 1dB? What is the max you can live with?

    Keep in mind the flatness of the frequency response on the DJ3200 in the datasheet is the ADC itself, after parasitics of the PCB are deembedded. Adding the input network on the frontend can change this response and you typically see the ripple effects from the standing wave, as we discussed, but I also, forgot to mention, the SMA input connector to the pcb as well and finally the balun itself.

    The balun on this frontend is a Marki BAL-0009SMG. Which is rated for 9GHz as it begins to rolloff its complex impedance changes become more radical and start to add to the rippling effect. A wider BW balun should help here too. On the DJ3200 EVM, there is a bypass option, this would allow the use of an external module balun.

    Regards,

    Rob