• State TI Thinks Resolved
  • Locked Locked
  • Replies 10 replies
  • Answers 1 answer
  • Subscribers 21 subscribers
  • Views 681 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.

LMX2595: Kvco affecting +85C performance for VCO7?

Dylan g
Prodigy 170 points
Part Number: LMX2595

I am currently using the LMX2595 at +85C in Full Assist mode with a calibration ran at +85C

We are seeing a problem when tuning to the end of the VCO7 core where on an SSA, the frequency looks very noisy (not unlocked just poor looking). I can provide a graph of it at +85C once I obtain, I've attached what it looks like at +25C

At hot, the "settled" frequency is much noisier (think zoomed in)

I was wondering if the Kvco for this core would affect it and cause that, a couple months ago I was encountering Kvco problems at -40C and thought that this issue might be related: e2e.ti.com/.../lmx2595-calibration-at-25c-does-not-lock-at--40c-for-vco4

  • 0
    TI__Guru** 108270 points

    Hi Dylan,

    If you can provide the phase noise plot, it would be easier for us to analyze.

    You said the calibration was taken at hot, could you provide the calibration data?

  • 0 in reply to Noel Fung
    Prodigy 170 points

    Hey Noel

    I can get a phase noise plot, will be able to provide that later today

    The PLL plot will be for 3mA, since I am briefly switching in and out of 6mA to improve my settling time as well as switching in and out of wideband to narrowband loop filter to mitigate integer boundary spurs

    For the calibration data, do you want just the N-Divider, VCO Core, CAPCTRL, and Numerator values for the two frequencies I am hopping between?

    Additionally, when stepping down the temperature the noise gets "smaller", as well as jumping to lower frequencies (on the magnitude of 10s-100s of MHz)

    Also when doing the math for the Kvco I saw that, from the datasheet, a jump of that magnitude would take me from a Kvco of 73 to a Kvco of 238.29

  • 0 in reply to Noel Fung
    Prodigy 170 points

    Do you think you would be able to provide a Kv curve over temperature for this part? Similar to how you provided them for me at -40C

    Also, one more thing we would be interesting in knowing is if there are any other settings we can set the charge pump to besides the 0,3,6,9,12,15mA provided in the register?

    Thanks

  • 0 in reply to Noel Fung
    Prodigy 170 points

    here is the settling time at +85C and the phase noise at +85C

  • 0 in reply to Dylan g
    TI__Guru** 108270 points

    Hi Dylan,

    I will get some test data.

  • 0 in reply to Noel Fung
    TI__Guru** 108270 points

    Hi Dylan,

    Frequency Kvco at room Kvco at 85C
    13.9GHz 206MHz/V 185MHz/V
    14.5GHz 193MHz/V 148MHz/V
    15GHz 222MHz/V 200MHz/V

    When I measure transient response, I will adjust VBW accordingly, especially at "Narrow band", otherwise it will be too noisy to read the data.

    For example,

    Were you using the same VBW (or something similar, depending on the test equipment) for 25C and 85C measurement?

    Below is the transient measurement at 15GHz at 25C and 85C, VBW = 200kHz. There is no obvious difference.

    25C

    85C

    Below is the VCO parameters I got with calibration at 25C and 85C, the values are close and make sense. How does your data look like?

        15GHz  
    Temp VCO capcode daciset
     25C 7 18 238
     85C 7 18 291

  • 0 in reply to Noel Fung
    Prodigy 170 points

    Thank you for the data Noel that is very helpful

    I turned on a smoothing function on my E5052B SSA and it got rid of the noise, but I'm not sure how much that helps my performance since it's smoothing it (lipstick on a pig?)

    Additionally, do you think having such a high numerator value (denom = 65535, num = 65207) affects this?

    Here are my values, they look pretty close to yours:

    14.999GHz
    Temp VCO capcode daciset
    25C 7 20 236
    85C 7 21 291
  • 0 in reply to Dylan g
    TI__Guru** 108270 points

    Hi Dylan,

    With E5052 narrow band transient measurement, we need to set the proper "Frequency Range". Setup --> Freq Range.

    This is not a smoothing to the DUT's performance, it is a smoothing to the test equipment. This smoothing is similar to spectrum anlayzer. When we use Full Span, the noise floor is very high because RES BW and VBW are big. With 10kHz span, for example, we need to use (usually automatic) a smaller RES BW and VBW so that we can see the real noise floor of the signal.

    Your VCO parameters is normal, that means the poor phase noise is not due to wrong parameters.

    Denominator setting may affect phase noise and you should see the effect in the phase noise measurement. If the phase noise plot is similar at 25C and 85C, then we should see similar transient response at these temperatures.

  • 0 in reply to Noel Fung
    Prodigy 170 points

    That's what I thought looking at my parameters, with smoothing turned on it does have a good transient response so I'm attributing the noise to perhaps sensitivity in my loop or measurement

    When I go up to 15GHz I use an external doubler, the noise instantly disappears and the signal is good which my hypothesis was due to the fact that I am now using VCO1 at 7.5GHz rather than a close close boundary for VCO7 

  • 0 in reply to Dylan g
    TI__Guru** 108270 points

    Hi Dylan,

    the VCO parameters look fine, we still have margin to go higher frequency. You can try, full assist mode, make VCO higher than 15GHz, reduce capcode, I think is should lock.

    There will be some phase noise difference between using VCO1 at 7.5GHz + external doubler and VCO7 at 15GHz, as the VCO phase noise is better at 7.5GHz. I don't expect the difference is big. From simulation, jitter is only worst by 5fs.