Dismiss Notice
Join Physics Forums Today!
The friendliest, high quality science and math community on the planet! Everyone who loves science is here!

How to select the PLL parameters

  1. Sep 14, 2016 #1
    The voltage waveform of the PLL circuit VCOin is shown(The PLL can lock).I need to get the stable voltage.If I change C18 from 0.1uF to 10uF, PLL cann't lock phase,the VCOin is always 5V,How to select C18/R5/R6 of the filter.How to calculate?
     

    Attached Files:

    • pll.jpg
      pll.jpg
      File size:
      35.7 KB
      Views:
      71
  2. jcsd
  3. Sep 16, 2016 #2

    jim hardy

    User Avatar
    Science Advisor
    Gold Member
    2016 Award

  4. Sep 17, 2016 #3
    First,I illustrate the application of this circuit,as shown in Figure 1.It is called capacitive liquid level

    detection.When the needle touches the liquid level,the capacitance value will become larger.So using phase locked loop

    to detect the change of capacitance.My schematic diagram is shown in Figure 2.Network named signal is connected to the

    needle.Normally,the voltage of TP1 is 2.5v,if the neddle touches the liquid level,the voltage of TP1 will change,as

    shown in Figure 3.However, When the needles are filled with ionized water, the voltage will become unstable,as shown

    in Figure 4.The frequency of the TP1 point voltage is 50Hz.It will affect the judgement of the waveform.Now I have two

    solutions,one is change the low pass filter parameter of PLL,so it can filter out the interference of 50hz, but to

    consider the timeliness of change.Another is to add a notch filter.where to put the notch filter,I want to put it in

    the output of OP U4,so the signal is passed through the notch filter before it is passed to MCU.
    Now my question is that which will be better,and how to make it.If you have some advice,I will appreciate it very

    much.Thank you very much.
     

    Attached Files:

  5. Sep 17, 2016 #4

    analogdesign

    User Avatar
    Science Advisor

    Jim,

    The old Signetics app note is great, but I really think the best free resource for PLL design is the book PLL Performance, Simulation, and Design by National Semiconductor (now TI). It is comprehensive, easy to understand, and goes step-by-step with easy to understand explanations and design procedures. It was my go-to book when I designed an integrated PLL a few years ago.

    http://www.ti.com/tool/pll_book
     
  6. Sep 17, 2016 #5

    jim hardy

    User Avatar
    Science Advisor
    Gold Member
    2016 Award

    At first glance it looks excellent ! Thank you analogdesign,

    i'll make the transition from sail to steam yet.....

    saved a copy.

    old jim
     
  7. Sep 17, 2016 #6

    jim hardy

    User Avatar
    Science Advisor
    Gold Member
    2016 Award

    Well slowing the loop filter reduces capture range.

    plllooptc.jpg

    I dont quite understand what you are up to.
    Are you shifting the pll center frequency and trying to lock that to output of whatever is (U1 divided by U3) ?
     
  8. Sep 17, 2016 #7

    Averagesupernova

    User Avatar
    Science Advisor
    Gold Member

    I would assume he allows the capacitance to change which in turn detunes the oscillator in which case the PLL compensates to bring it back on frequency. Watching the VCO voltage is tells just what the capacitance is doing. I know some about PLL but there is plenty I do not know also. I know that the loop filter is a critical little bugger. I would say control system theory (PID and such) will get you where you want to be with PLL. A wise engineer once told me that to troubleshoot a PLL often it is desirable to break the loop. Concerning your 50 hertz noise, is it an oscillation within the loop or is it external? Are you in a part of the world that uses 50 hertz instead of 60 hertz for power?
     
  9. Sep 17, 2016 #8
    I just want the VCOin to be DC Voltage. I don't understand why this voltage could fluctuate? (PS:my AC power is 50 HZ).
     
  10. Sep 17, 2016 #9

    jim hardy

    User Avatar
    Science Advisor
    Gold Member
    2016 Award

    I feel not very well informed.
    What is the frequency you are trying to lock to, i think at HEF4024 Q3 ?
    Is fig 3 a scope trace when pll can lock, 0.1uf, and fig 4 when it fails to lock, 10 uf ? What are sweep speeds ?

    Could be you're picking up 50 hz at pins C1A and C1B from your probe. It looks from figure 1 rather like an antenna. I think those pins both need DC isolation and 50 hz isolation(if you can get it) from signal common.
    First thing i'd do is set scope trigger to LINE, if trace stands still then your 50 hz is power line frequency not loop natural frequency. I suspect strongly that's what you'll find.

    Next i'd try hooking my probe up like this so it's got some Z in series at low frequency

    pllLiquidDetector.jpg
    500 pf at 50hz is 6 megohms, what is it at your operating frequency ?

    Next thing i'd try is, since your frequency setting resistors R3 and R19 go to signal common and so does your probe capacitance, try connecting your probe to them in turn.

    Got to push at this thing and figure out how it behaves.
     
  11. Sep 17, 2016 #10
    Jim Hardy,Thank you so much.
    I'm trying to lock to HEF4024 Q3,the frequency is 375khz. But if the needles are filled with ionized water,the PLL can not lock, as shown in figure. It is strange that if I put the ground of the machine to the earth(the real ground),no any changes,the PLL can lock again,put it off the earth,unlock again.
    I will connect a capacitor between needles and C10 on your suggestion.Right now I don't have 500pf,is it ok to use 200pf?And you said that"Next thing i'd try is, since your frequency setting resistors R3 and R19 go to signal common and so does your probe capacitance, try connecting your probe to them in turn". I can't see your point.Is the probe capacitance to be C10?what's the meaning of "in turn".
     

    Attached Files:

  12. Sep 18, 2016 #11

    jim hardy

    User Avatar
    Science Advisor
    Gold Member
    2016 Award

    pllLiquidDetector2.jpg
    Any voltage between them is injected right into your PLL's C1A ,if i read correctly the image attached to post #1.
    Might be that 'earthing' the signal common on circuit board removes any voltage between it and the beaker of deionized water.

    Of course, we're looking for cause-effect.

    Hmmm wait a second here, you said ionized water not de-ionized. I'm guilty of inattention to detail.
    De-ionized water is nonconductive and has dielectric constant perhaps 80. I dont know what is ionized water, but surely it's conductive .
    So when you said 'capacitive' i jumped to the conclusion you were sensing very pure water. That's why i suggested series capacitance.
    If that's not so, your probe when wetted may experience more change in resistance than in capacitance.
    Furthermore i'd wager it makes a few tens or hundreds of millivolts DC by galvanic action.

    No, for this experiment it's to alter the value of R3 or R19&R20
    Water that's not deionized is pretty conductive.
    If your probe when wetted sees more resistance change than capacitance change, it'd make sense to let it affect the frequency setting resistor instead of the frequency setting capacitor, wouldn't it ?



    First connect probe at pin R1 so that when wetted it decreases resistance from that pin to common, as if value of R3 were changed
    then try it at pin R2 as if changing R19&R20 .


    I think you'll have to connect those two "grounds" in my altered sketch above.
    "Ground" is a much abused term. Usually it means 'signal common' or 'zero volt reference point' as in right part of your drawing . It means "earth" only if physically connected to earth, which clearly in an airplane or rocket it's not.
    Is your beaker connected to earth by a wire or metal counter top ? Is your "signal common" connected to earth through the oscilloscope probe ? Is there any voltage between them?

    Keep us posted?
     
    Last edited: Sep 18, 2016
  13. Sep 18, 2016 #12

    NascentOxygen

    User Avatar

    Staff: Mentor

    Is that sensor plate coated with plastic insulation? If not, it will be sensitive to R as well as C. Is it a pin or a plate?

    Is your connection of the sensor plate directly to the IC pin correct? This evades any filtering; it may be better to connect the sensor across the timing capacitor C18. disregard
     
    Last edited: Sep 18, 2016
  14. Sep 18, 2016 #13
    Thank you a lot.Figure 5.1 shows the neddle connects to the circuit and the tube. The gnd of circuit is connected to

    the gnd of machine.
    Today I try to change C18(the low pass filter cap) from 0.1uf to 1uf.The waveform of VCOin can be stable,as shown in

    Figure 5.2,the figure 5.2 shows that the waveform of VCOin before and after touching the liquid level.But the response time will be longer.This will cause the needle to overshoot at the level of the liquid.Now solved a problem and lead to another problem.I am going to crash.
    Thank you again, Jim.
     

    Attached Files:

    • 1.jpg
      1.jpg
      File size:
      40.8 KB
      Views:
      49
  15. Sep 18, 2016 #14
    it may be better to connect the sensor across the timing capacitor C18,you mean connect the neddle to the C18? what is the sensor?
     
  16. Sep 18, 2016 #15

    NascentOxygen

    User Avatar

    Staff: Mentor

    Sorry, I misinterpreted your diagram.

    I can't identify where your needle connects into the PLL. But I'm wondering why you are using [apparently] a needle and not a plate to come into contact with the liquid's surface?
     
  17. Sep 18, 2016 #16

    jim hardy

    User Avatar
    Science Advisor
    Gold Member
    2016 Award

    Looks to me like you've solved your problem. It appears to never lose lock ?
    Fig 5.3 also locks . I am unable to read what is the time per division on your 'scope
    but there's no more line frequency.
    That "hash" at VCO in is at your operating frequency hundreds of khz?
    If you measure at capacitor C18, other side of R5, you'll see DC without the hash. Then you'll be able to reduce C18 for faster loop response.
    I would try a 0.01uf C18, too. It's only 1 ohm at your frequency. That'll speed up the loop, and you can begin exploring those closed loop equations.

    I hope you'll try wiring your needle to R3 and R19 , just to see what happens.
     
  18. Sep 18, 2016 #17
    no,i don't solve my problem.using 0.1uf (C18) ,the VCOin can not be stable, the frequency of the hash is 50HZ,you can watch the figure 4 at 3# where I posted.If using 1uf,the VCOin have no hash,but the rising time will be 5 times before. It is unacceptable.

    It looks like Figure a,attached in the annex.

    To change the circuit will have a little bit of trouble,so I want to know why to do this?
     
  19. Sep 18, 2016 #18
    Attachment
     

    Attached Files:

    • 2.png
      2.png
      File size:
      13.3 KB
      Views:
      50
  20. Sep 18, 2016 #19

    jim hardy

    User Avatar
    Science Advisor
    Gold Member
    2016 Award

    Well then , what's figure 5.3 attached to post 13 ? It says C18=0.1 uf and there's no 50hz.

    upload_2016-9-18_19-0-45.png

    i am confused.
     
  21. Sep 18, 2016 #20
    Sorry about that,because i put the ground of the machine to the earth.
     
Know someone interested in this topic? Share this thread via Reddit, Google+, Twitter, or Facebook

Have something to add?
Draft saved Draft deleted



Similar Discussions: How to select the PLL parameters
Loading...