Why the resonance frequency is different in time domain versus frequency domain in LTspice ?

Discussion in 'Test & Measurement Forum' started by Ghoncheh, Oct 2, 2017.

  1. Ghoncheh

    Thread Starter New Member

    Oct 2, 2017
    7
    0
    Using frequency analysis shown in the plot for this LC circuit I found the resonance frequency of it (138.88 MHz), But when I switched to the transient analysis with a sine pulse at that resonance frequency, I saw ringing on the plot.
    I don't understand what causes this issue. I tweaked the sine pulse frequency until I got the ring up and ring down without that ringing artifact at 138.1 MHz. What has caused this frequency difference?
    If I calculate the time constant for the ring down based on this time domain response, I would see it is much shorter than what I expected from the Q factor of the coil, what is the reason behind that?

    I have attached all of my LTspice plots.

    Thank you so much,
    Ghoncheh
     
  2. crutschow

    Expert

    Mar 14, 2008
    21,755
    6,270
    Post your .asc file.
     
  3. Papabravo

    Expert

    Feb 24, 2006
    12,043
    2,591
    This may be an impertinent question, but why would you expect the transient and the steady state analysis to produce the same result?
    Both schematics have the .tran 70u statement un-commented in them, so what are we actually looking at?
     
  4. Bordodynov

    Well-Known Member

    May 20, 2015
    2,110
    643
    For circuits with high Q-factor, it is necessary to increase the accuracy
    Draft1__.png
     
    Ghoncheh likes this.
  5. Ghoncheh

    Thread Starter New Member

    Oct 2, 2017
    7
    0
    I thought when the drive pulse length is long enough in the time domain simulation, I have steady state situation like the frequency domain case. So I expect that the resonant frequency of the circuit stays the same. But maybe I'm wrong?
    Sorry for the confusion, I have commented out the .trans 70u statement for getting the 1st plot which is in the frequency domain. The two un-commented statements correspond to the other time domain responses.
     
  6. Papabravo

    Expert

    Feb 24, 2006
    12,043
    2,591
    I see what you are saying now. I always thought that resonance was not a kick it once and it goes on for a long time, like for example an under damped mass spring system which is the equivalent of an LC circuit. Resonance requires a periodic forcing function and involves passing the energy back and forth between two elements, like a capacitor and an inductor, or a mass and a spring.
     
  7. Ghoncheh

    Thread Starter New Member

    Oct 2, 2017
    7
    0
    I uploaded the .asc file also.
     
  8. crutschow

    Expert

    Mar 14, 2008
    21,755
    6,270
    Did you include the 1mΩ default value for the inductor series resistance?

    I could find no reason for the slight difference between the simulated resonant frequency found in the time domain as compared to the frequency domain.
     
  9. Papabravo

    Expert

    Feb 24, 2006
    12,043
    2,591
    The more I look at these two simulations convinces me that there is a difference in the initial conditions. What exactly do you mean by a "sine pulse" at the resonant frequency, and how is that initial condition different from the AC analysis initial condition where you sweep the frequency from 130 MREG to 150 MEG?
    Again -- why would you expect them to be the same?
     
  10. crutschow

    Expert

    Mar 14, 2008
    21,755
    6,270
    In my time domain simulation I started the oscillations with a short 1ns pulse and then measured the free-running oscillation frequency after that.
    It was slightly lower than the frequency domain simulation gave (which matched the calculated resonant frequency).
    I see no reason that they shouldn't be the same.
     
  11. Papabravo

    Expert

    Feb 24, 2006
    12,043
    2,591
    Would that be a square pulse, with negligible rise and fall times?
    I would also be interested to see if the transient response varies with different integration methods.
     
  12. Bordodynov

    Well-Known Member

    May 20, 2015
    2,110
    643
    Schemes with high Q are very inertial. In fact, the oscillatory circuit in one oscillation period takes on very little energy. The standard errors for any Spice program are not sufficient in this case. I increased the accuracy and everything fell into place. Look at my picture and see that I have greatly reduced the error.
     
  13. Ghoncheh

    Thread Starter New Member

    Oct 2, 2017
    7
    0

    You are right! Thank you so much, I increased the accuracy as you suggested and it doesn't have those weird ringings at the rise up anymore.
     
  14. Ghoncheh

    Thread Starter New Member

    Oct 2, 2017
    7
    0
    No, I didn't include any series resistance to my circuit.
     
  15. Ghoncheh

    Thread Starter New Member

    Oct 2, 2017
    7
    0
    Thanks for thinking about my question, I meant the voltage source which has sine wave format at the frequency which corresponds to the resonance frequency of the coil found with the frequency domain analysis.
    I just expected not to see the ringings when I am driving the circuit with its own resonance frequency. By changing the accuracy as Bordodynov suggested, the issue has resolved.
     
  16. Ghoncheh

    Thread Starter New Member

    Oct 2, 2017
    7
    0
    Thanks for thinking about my question, the issue was in the accuracy of the time domain simulation.
     
Loading...