• State
  • Locked Locked
  • Replies 5 replies
  • Subscribers 21 subscribers
  • Views 914 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: Spurs in phase noise measurement

Zichuan Zhou
Prodigy 40 points
Part Number: LMX2595EVM
Other Parts Discussed in Thread: LMK61PD0A2, LMX2595

Hi

I have a question about LMX2595EVM. We are trying to use LMX2595EVM to generate 400MHz clock. We are using external clock for reference clock and its frequency is 100MHz. 

However, we saw different behaviour using different 100MHz reference clock. In this case we used 1) Wenzel (100MHz, sine wave) and 2) LMK (LMK61PD0A2, which is embeded on LMX2595EVM control board, 100MHz square wave). We compared the phase noise of 400MHz clock generated by LMX2595EVM and the result is shown below. 

We found that even though 100MHz Wenzel (blue) has much lower phase noise than 100MHz LMK (orange), the phase noise of 400MHz triggered by Wenzel has lots of spurs and 400MHz generated by LMK looks very clean. 

The measured waveform of two 100MHz reference clock is shown below. The measured peak-to-peak voltage of Wenzel and LMK are 400mV and 800mV, respectively (which is above the minimum required voltage for LMX2595). 

The LMX2595 PLL setting is also attached. 

Can you help us solving this problem? We are expecting 400MHz generated by Wenzel should be better than LMK because 100MHz Wenzel has better phase noise. However, we are not sure where the spurs in phase noise measurement come from. Is there any way to remove the spurs? 

Regards

Zichuan Zhou

  • 0
    TI__Guru* 97040 points

    Hi Zichuan,

    I have no idea why there are so many spurs with the Wenzel clock. Could this be a power supply issue? Is there way to increase the voltage swing of the Wenzel clock?

  • 0 in reply to Noel Fung
    Prodigy 40 points

    Hi Noel,

    Thanks for your reply. I have tried using a low noise power supply and increasing the Wenzel power. The Wenzel power is now 3dBm. 

    We also tried to add a low pass filter (with 150MHz bandwidth) after LMK61PD0A2. The waveform of LMK61PD0A2 after filter becomes a sine wave as shown below:

    Then we compared the phase noise of 400MHz generated by LMX2595 under different conditions: 

     

    Blue curve: Using Wenzel and old power supply (same plot as in previous post), Orange: Using LMK without low pass filter (square wave, same plot as in previous post), Green curve: Using Wenzel and low noise power supply and Red curve: Using LMK with low pass filter. 

    Basically, here is what we found: 1) using low noise power supply suppress the power of spurs and 2) square waveform also causes some spurs. I am wondering if the waveform of input reference clock causes any performance degradation? I assume as long as the transition is clear, LMX2595 would be happy with either sine wave or square wave. Am I correct? 

    Regards

    Zichuan Zhou

  • 0 in reply to Zichuan Zhou
    TI__Guru* 97040 points

    Hi Zichuan,

    Sine wave is not an issue, especially when the frequency is high. I also use 100MHz sine wave in my lab, I never see issue. 

    Looks like you have great improvement with the Wenzel clock using a low noise power supply. Have you checked the grounding? Were you using a SMA I-join to connect the Wenzel and the EVM?

  • 0 in reply to Noel Fung
    Prodigy 40 points

    Hi Noel,

    Thanks for your reply. Yes, I agree, the spurs from Wenzel do seems very confusing. Since we get some improvement by using a low noise power supply, we were guessing whether these spurs might come from power supply? Therefore, we added another capacaitor between power supply and Wenzel to further reduced noise, however, the phase noise measurement result looks similar. 

    In addition to this, I also checked grounding, we believe grounding is not an issue. The case of Wenzel is connected to the ground. In terms of SMA cable, we did suspect the funny spurs come from dodgy cables. To verify that, we used same SMA cable to connect LMK/Wenzel to LMX2595 and same cable to connect LMX2595 output to phase noise analyzer. However, as discussed above, we didn't see spurs when using LMK (without low pass filter). That suggests spurs are not from cable. I am bit confused about what you mean by SMA I-join, can you clarify that? 

    What really confuses me is that we did see some spurs when we added low pass filter after LMK. Does it have anything to do with evaluation broad of LMX2595? I made some modification to the broad so that we can use external reference clock (I basically follow the steps from TI website). Is it possible that power supply of LMX2595 evaluation broad has some noise and this causes spurs? Or is there anything I can do to optimize LMX2595 setting? 

    Regards

    Zichuan Zhou

  • 0 in reply to Zichuan Zhou
    TI__Guru* 97040 points

    Hi Zichuan,

    SMA I-join is a connector with both ends having the same gender, I think it comes with the EVM kit.

    By default, the EVM is configured with single-ended input with a 50Ω shunt right after the SMA connector. This configuration matches well with the Reference Pro board (LMK61PD0A2) and RF signal generator.

    The EVM requires external power supply, I believe you have a good power supply because there is no spurs when the EVM is working with the Reference Pro board.

    The spurs is definitely not generated from the EVM since you are working at an integer channel, so there is nothing wrong with the software setting.

    I am afraid I don't have any guess for you at the moment, feel free to post again if you have further findings.