• State Resolved
  • Locked Locked
  • Replies 4 replies
  • Subscribers 39 subscribers
  • Views 1406 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.

On Group delay

Sira Rao1
Intellectual 260 points

0827.HPF.TSC

Dear Members,

I thought I had understood group delay conceptually, but in looking at the Group delay vs frequency plot generated by TINA on the attached schematic, I am confused.

The phase response of this High pass filter is almost flat in its stop band and in its pass band. In its transition band, it varies. Group delay, the scaled negative slope of the phase, has a similar looking curve.

In the stop band, where the phase response is flat, the group delay is constant and about 50us. My question is, why is it not equal to 0? After all, the slope is 0. The group delay is 0 in the pass band, where the slope is 0.

A low pass filter exhibits similar behavior. The group delay is non-zero in the pass band where the slope is 0, and 0 in the stop band where the slope is 0.

Thanks for any guidance.

Sira

  • 0
    TI__Expert 3555 points

    Sira,

    I am looking into this to give you a good detailed answer. Thank you for the TINA simulation file. This makes things easier. I am curious about two things. 1.) what software did you use to generate this filter, 2.) what are your filter specifications (corner frequency, approximation type [butterworhth, bessel, etc).

  • 0 in reply to Bonnie Baker - WEBENCH Design Center
    Intellectual 260 points

    Bonnie,

    The filter specifications don't really matter for me at this point. I am interested in getting the Group delay concept down firmly.

    If I recall correctly though, it is a 10kHz high pass Bessel Multiple Feedback filter.

    Thanks,

    Sira

  • 0
    TI__Mastermind 40295 points

    Sira,

    TINA's default for plotting is to use logarithmic frequency spacing.This can make estimating slopes - which use linear spacing - kind of tricky.
    In the TINA phase plot,  make the frequency spacing linear and zoom in on the frequency span of 10Hz to 5 kHz.

    The graphic below shows the resulting plot:


    The cursors show the phase decreases by 84 degrees, and is approximately linear over the 5kHz frequency span. Normalizing the delta-phase by 360 and dividing by the 5k frequency span gives 47usec, which is pretty close to the 50us you mentioned.

    I hope this helps. Please let me know if you have any questions.
    Regards,
    John

  • 0 in reply to John Miller - Sensing
    Intellectual 260 points

    John,

    Thanks for the insightful answer. I now get it. In fact, I suspected the log scale might be the issue, but I didn't pursue it enough to arrive at the answer myself.

    Thanks again,

    Sira