• State Resolved
  • Locked Locked
  • Replies 7 replies
  • Answers 1 answer
  • Subscribers 19 subscribers
  • Views 391 views
  • Users 0 members are here
Support feedback
Options
Options
Similar topics

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.

LMX2595EVM: Why does Sigma-Delta modulator add varying levels of phase noise to the LMX2595 when different OCXOs are used as reference clock?

Mark
Prodigy 47 points
Part Number: LMX2595EVM
Other Parts Discussed in Thread: LMX2595

Hello. We are testing the LMX2595EVM with two 100MHz OCXOs.  The frequency of reference clock  is 100 MHz and Osc_doubler of LMX2595EVM is enabled.

We found that when the LMX2595 is in integer mode and two OCXOs are used as reference sources, the phase noise of the LMX2595 measured at frequency offsets above 1kHz is basically the same. However, in fractional mode, enabling the Sigma-Delta modulator results in the degradation of phase noise for both sources, but to different extents. From the results, it appears that the Sigma-Delta modulator introduces different levels of phase noise depending on the reference source used, with less phase noise added for better reference sources. We would like to confirm if the Sigma-Delta modulator does indeed exhibit this characteristic.

Measured phase noise of two OCXOs and LMX2595 are as follows.

The configurations of TICS pro for the LMX2595 in integer and fractional mode are shown in the following figures.

  • 0
    TI__Mastermind 27581 points

    Mark,

    These are interesting observations that I have not seen before.  

    In general, the delta sigma modulator has it's own noise floor that is independent of the PLL 1/f noise and figure of merit.  For this close in delta sigma noise floor, this is directly related to how linear the charge pump is.  As the noise floor of the delta sigma modulator is far higher than the input reference, it makes me think that this has nothing to do with this input reference.  What I think is perhaps your lower noise input reference might have a faster slew rate which somehow impacts this delta sigma modulator noise.

    Regards,
    Dean

  • 0 in reply to Dean Banerjee
    Prodigy 47 points

    Hi,

    Thank you Dean a lot for your helpful reply. I have not previously paid attention to the difference in slew rate between the two reference oscillators and I conducted another experiment to verify your hypothesis.

    The results of the experiment confirmed that the OCXO-B with lower phase noise does indeed have a faster slew rate. As the input reference slew rate decreases, the phase noise of the PLL increases. 

    However, I think that the difference in slew rate between the two OCXOs may not be the main influence factor contributing to the difference in phase noise between the two PLLs.  I  reduce the slew rate of the OCXO by lowering the power output to the OSCin port of the lmx2595, and when the slew rate is lower than that of OCXO-A, the phase noise of OCXO-B+lmx2595 is still significantly lower than that of OCXO-A+lmx2595.

    Both OCXOs output sine waves, so I calculated the slew rate using the following equation. Slew Rate = 2*pi*frequency*Vpp. Different attenuators are used in the lmx2595 input reference path to reduce the output power of OCXO-B.

    The output power and slew rate under different conditions are shown in the table below and measured phase noise curves under different conditions are shown in the figure below.

    Are there any other possible reasons for the output phase noise difference between the two PLLs?

  • +1 in reply to Mark
    TI__Mastermind 27581 points

    Mark,

    What your measurements are suggesting is that noise to the input of the PLL is amplified by the delta sigma modulator, which is what you originally suggested.

    Now in the case of using an ultra-low noise reference, the PLL input path still has noise, so if this theory is true, this noise would be multiplied up.  But as this noise is lower, maybe it gets dominated by something else.  But if you put an input reference that has higher noise, maybe it gets multiplied up to a level that dominates other noise (like the N divider noise).

    If there is a ultra-clean input, but the slew rate  is varied, this impacts the noise of the input path, so for low enough slew rates, this would theoretically impact the delta sigma noise.

    I haven't done any experiments to prove that noise of the input is amplified by the delta sigma modulator, but your measurements are beginning to make me believe that this may be the case.

    Regards,

    Dean

  • 0 in reply to Dean Banerjee
    Prodigy 47 points

    Thank you, Dean. It is a very useful reply! 

    Can I understand it this way, that as the phase noise of the OCXOs is sufficiently low, the differences in phase noise between the two PLLs are not only due to the phase noise difference between two OCXOs I used, but there may also be other influence factors? If I understand it correctly, I think what you're saying makes a lot of sense.

    Currently, we are selecting an OCXO as the reference for  LMX2595 in our products, and based on the experimental results, OCXO-B is undoubtedly the best choice. However, while we primarily considered the phase noise of the OCXOs when selecting a reference, this experiment suggests that we should also consider other factors in our selection process. 

    For these two OCXOs, besides their slew rates, are there any other possible influence factors that could cause differences in LMX2595 phase noise? If you know any other possible reasons or have any other ideas about the issue to share, we would greatly appreciate it and be glad to conduct several experiments to investigate this issue. 

    Sincerely,

    Mark

  • +1 in reply to Mark
    TI__Mastermind 27581 points

    Mark,

    I tried to replicate this in the lab, but I did not see it.  Either the phase noise of the OSC dominated totally, or was dominated by the LMX part.

    It would seem that if the OSC noise is too low, it is dominated by the input path of the PLL.  If it is too high, we see the raw OSC noise go through.  But if the OSC noise is just right, it can mix down.  Note that with the delta sigma modulator, far out phase noise can mix down and create this noise.

  • 0 in reply to Dean Banerjee
    Prodigy 47 points

    Hi,

    I think your reply would make sense. However, perhaps due to my lack of understanding of the working principle of the sigma-delta modulator, I don't fully comprehend the specific reasons that lead to this.

    Thank you Dean for your patiently responding and teaching me a lot about sigma-delta modulator through your replies. However, I also realize that I know very little about sigma-delta modulator and I will learn more about it in the future.

  • 0 in reply to Mark
    TI__Mastermind 27581 points

    Mark,

    Glad to be of help.  Also, feel free to check out my website at DeanBanerjeePLL.com.

    Regards,

    Dean