+ Post New Thread
Results 1 to 12 of 12
  1. #1
    Member level 2
    Points: 1,741, Level: 9

    Join Date
    Dec 2006
    Location
    Egypt
    Posts
    43
    Helped
    9 / 9
    Points
    1,741
    Level
    9

    How to deal with the sweep of Kvco on PLL system design?

    Hi All

    After I design my PLL with VCO gain Kvco=333.MHz/volts and design my system on that after I draw changing of Kvco with input voltage I found it change from Kvco(min)=60 MHz/volt to Kvco(max)=600 MHz/volts with non-Linear curve as shown in the figure.

    1- Is this curve is OK or it is a bad design

    2- How I can deal with the swep of Kvco on system design and its effect on stability, Lock time, Phase noise ,..........

    Note:
    Freq Vs Vcontrol in blue
    Kvco Vs Vcontrol in red



    Thanx
    regards
    Salem

  2. #2
    Newbie level 1
    Points: 1,010, Level: 7

    Join Date
    Apr 2007
    Posts
    1
    Helped
    0 / 0
    Points
    1,010
    Level
    7

    gain of vco

    please could you tell me how to probe kvco curve by hspice
    just like your fugure
    thank~~~



    •   Alt19th September 2007, 04:20

      advertising

        
       

  3. #3
    Full Member level 2
    Points: 1,795, Level: 9
    RDRyan's Avatar
    Join Date
    May 2007
    Location
    China
    Posts
    133
    Helped
    19 / 19
    Points
    1,795
    Level
    9

    hspice vco gain

    in PLL project design, Kvco is not linear. sometimes it is OK, if the curve of Kvco is monotone.
    In the considering of system stability, you should check when using the max Kvco and min Kvco, the PLL system should be both stable.

    Ryan


    1 members found this post helpful.

  4. #4
    Junior Member level 2
    Points: 1,268, Level: 8

    Join Date
    Aug 2006
    Posts
    22
    Helped
    6 / 6
    Points
    1,268
    Level
    8

    what do you mean by vco has a linear tuning curve

    Hi,Salem:

    First of all,I think there is no problem of your simulated result.

    But, in practice,Kvcomax:Kvcomin=10:1 is definitly not a good choice.It will cause the PLL bandwith to vary 10x.What is more,If you run full process corner simulation,the Kvcomax:Kvcomin will be even worse.

    I think you should make your VCO run at the middle part of your F-V curve,this will reduce Kvcomax:Kvcomin to less than 2:1,which is a better choice.If the frequency range is not covered, you can use more VCO bands.

    Remember:The linearer of KVCO,the easier the system design.

    Best wishes!


    1 members found this post helpful.

    •   Alt19th September 2007, 15:10

      advertising

        
       

  5. #5
    Member level 2
    Points: 1,741, Level: 9

    Join Date
    Dec 2006
    Location
    Egypt
    Posts
    43
    Helped
    9 / 9
    Points
    1,741
    Level
    9

    how to compute vco gain

    Quote Originally Posted by lineage050505
    please could you tell me how to probe kvco curve by hspice
    just like your fugure
    thank~~~
    I don't make it using hspice. Unfortunately I don't have any idea about it.
    I used the Specter simulator under cadence tools.

    I plot this curve using the pss simulator you can read more about it in the documentation manuals of cadence.

    I Hope that help you



    •   Alt19th September 2007, 21:08

      advertising

        
       

  6. #6
    Junior Member level 2
    Points: 1,268, Level: 8

    Join Date
    Aug 2006
    Posts
    22
    Helped
    6 / 6
    Points
    1,268
    Level
    8

    kvco curve

    lineage050505,

    you can use .measure statement in Hspice to measure the period and then the frequency of the VCO after .tran simulation.

    Best regards!

    chenmy



  7. #7
    Advanced Member level 1
    Points: 3,875, Level: 14
    jecyhale's Avatar
    Join Date
    Feb 2008
    Location
    China
    Posts
    403
    Helped
    54 / 54
    Points
    3,875
    Level
    14

    vco gain measure hspice

    what type vco you used?



  8. #8
    Member level 2
    Points: 1,741, Level: 9

    Join Date
    Dec 2006
    Location
    Egypt
    Posts
    43
    Helped
    9 / 9
    Points
    1,741
    Level
    9

    measure hspice vco

    Quote Originally Posted by jecyhale
    what type vco you used?
    It was LC-Oscillator.



  9. #9
    Junior Member level 2
    Points: 1,599, Level: 9

    Join Date
    Sep 2004
    Posts
    22
    Helped
    2 / 2
    Points
    1,599
    Level
    9

    pll hspice kvco

    It seems that the VCO gain is not low, and the linearity is not very good. This is not good for PLL system, and also the phase noise performance will not easy to control.


    1 members found this post helpful.

  10. #10
    Member level 2
    Points: 1,577, Level: 9

    Join Date
    May 2006
    Posts
    47
    Helped
    8 / 8
    Points
    1,577
    Level
    9

    constant kvco

    What you can do is to use discrete VCO tuning together with your continuous tuning. In this case you won't have to sweep the whole KVCO curve, instead you will have sub-bands each with linear acceptable KVCO variation. The drawback is that you will need another loop to control the discrete part of your VCO (A frequency counter can do the job)


    1 members found this post helpful.

  11. #11
    Member level 3
    Points: 1,694, Level: 9

    Join Date
    Feb 2006
    Posts
    60
    Helped
    2 / 2
    Points
    1,694
    Level
    9

    vco corner simulation

    How to control the LC vco gain?
    if we can control it, we can linearize it?
    so, what are affecting parameters of Kvco?

    thanx in advance
    MohamedAbouzied



  12. #12
    Member level 2
    Points: 1,741, Level: 9

    Join Date
    Dec 2006
    Location
    Egypt
    Posts
    43
    Helped
    9 / 9
    Points
    1,741
    Level
    9

    reduce vco gain

    I think the Kvco is affected mainly by the efficiency of your varactor. If the capacitance of the varactor linearly changes with the applied volt, This will make your vco more linear. Also decreasing your frequency range makes the vco more linear.

    The most important thing here I want to ask to you is that how linear you want your vco. I mean what is range of your Kvco variation your system required. As you know you will not get a constant Kvco this is an ideal case. All you want is a Kvco with an acceptable range which will not affect your system loop stability, Loop frequency, phase margin and the lock time.

    As an example of the above curve the Kvco change from 60MHz/volt to 600MHz/Volt. So you have to go back to your system and simulate it on the value of 60MHz/volt and also the value of 600MHz/volt if your system works well in both cases and satisfy the required specs that mean in between this 2 values the system is OK and this range of Kvco is acceptable and if not you have to go back and either change your system parameters by doing some changes in the loop filter or back to your vco and try to change its Kvco linearity.

    Another approach you may use to make the KVCO constant is to make a VCO with multiple varactors each one with the same KVCO and a small range to make it linear. In this case you need a control circuit to control which range(varactor) to use. It may be a frequency counter like "mbyoussef"suggest or any other approach you may think about.

    Best Regrds,
    Salem



+ Post New Thread
Please login