• State TI Thinks Resolved
  • Locked Locked
  • Replies 7 replies
  • Answers 4 answers
  • Subscribers 20 subscribers
  • Views 382 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.

LMX2594: Level of phase noise the LMX2594 can track from the external oscillator

Danilo A.
Mastermind 24705 points
Part Number: LMX2594

Hi Team,

Our customer is working with the LMX2594 in combination with a very high quality (very good long term stability) external reference oscillator. He would like to know at what level of phase noise (e.g. -135 dBc/Hz at 10 Hz) the LMX2594 can track from the external oscillator over long periods before the synthesizer itself becomes the long term stability limitation.

Regards,

Danilo

  • 0
    TI__Guru* 81400 points

    Hi Danilo,

    From the design point of view, the synthesizer's phase noise would not degrade much with aging. 

  • 0 in reply to Noel Fung
    TI__Mastermind 24705 points

    Hi Noel,

    Thank you for your response.

    The customer have used the PLLatinum software to simulate the phase noise behavior of the LMX2594 for the eval board configuration. For phase noise very close to the carrier (e.g. 10 Hz offset), the phase noise of the PLL/VCO synthesizer is generally directly driven by the response of the external clock (e.g. 10 MHz crystal oscillator). He plan to use a very expensive, high quality reference oscillator with phase noise of -135 dBc/Hz at 10 Hz offset in a RF multiplication design. Based on PLLatinum, an oscillator of that quality cannot be leveraged by the LMX2594 because the simulation indicates the PLL contribution limits the 10 Hz phase noise to about -70 dBc/Hz.

    Can you please confirm if the LMX2594 will not be able to take full advantage of such a low phase noise crystal reference?

    Regards,

    Danilo

  • 0 in reply to Danilo A.
    TI__Guru* 81400 points

    Hi Danilo,

    Now I understand your question in the first post :)

    What is the output frequency?

    Given the input clock is 10MHz and the phase noise is -135dBc/Hz at 10Hz offset, the phase noise will become -135 + 20*log(output frequency / input clock frequency). For example, if output frequency is 1GHz, then the phase noise due to the reference clock at 1GHz will be -95dBc/Hz at 10Hz offset. 

    From PLL sim, the phase noise of LMX2594 at 1GHz will be -98dBc/Hz at 10Hz offset. So eventually, the phase noise at 10Hz offset will be the sum of reference clock and synthesizer, this is about -93.23dBc/Hz.

    In summary, we have to "scale" the phase noise of the reference clock to the output frequency with 20*log(output frequency / input clock frequency) in the overall phase noise calculation.

  • 0 in reply to Noel Fung
    TI__Mastermind 24705 points

    Hi Noel,

    I have forwarded your response to the customer and here is feedback.

    Sorry I should have included the design frequency (8 GHz). In the interest of conveying the full details, I have attached my PLLatinum design file so you can more clearly follow the characteristic I am inquiring about. My question is regarding the various phase noise contributions reported in the left/bottom panel plot window, more specifically the OSC and PLL contributions. I believe I have correctly modelled the -135 dBc @ 10 Hz phase noise of my 10 MHz oscillator. Hopefully you can see from my design file (and associated plot also attached here) that the phase noise of my 10 MHz oscillator is not the dominant phase noise source at 10 Hz as I was expecting. Rather, the 10 Hz phase noise is dominated by the PLL noise contribution and it is this component I am trying to better understand. I was hoping to see that the phase noise at offsets < 100 Hz would be dominated by my oscillator so I'm wondering what level of control there is over the PLL component. Can you offer some insight?

    8GHzsim.sim

    Regards,

    Danilo

  • 0 in reply to Danilo A.
    TI__Guru* 81400 points

    Hi Danilo,

    The phase noise at 10Hz offset is actually dominated by the reference clock. The customer is seeing the opposite because they use "Metric" to model the reference clock.

    In below is the simulation comparison using "Metric" and "Load file". I created a text file for the phase noise of the reference clock. This method returns the best accurate simulation. 

    Below 10kHz offset, the phase noise of LMX2594 is dominated by the 1/f noise of the PLL. This noise is determined by the process and design, and is not software configurable. 

  • 0 in reply to Noel Fung
    TI__Mastermind 24705 points

    Hi Noel,

    According to the customer, he don't fully understand the "metric" input to specify the oscillator model. The documentation on the pllatinum software appears to be limited to the little pop-up boxes in the GUI. If there is more documentation (e.g. a pdf document) that would be helpful.

    Regards,

    Danilo

  • 0 in reply to Danilo A.
    TI__Guru* 81400 points

    Hi Danilo,

    Unfortunately, we don't have the user's guide for this tool.