What i need is to design a delyiannis band-pass filter with Q enhancement technique.

The theoretical analysis of this technique is referred here in pages 107-112.

For those who don't know how to design delyiannis band pass-filer here is an example: enter image description here

And here is the expected bode plot enter image description here

In my case, i need to design a filter with f0 = 825 Hz and Q = 36.

Although i do the same process like the above i don't get the expected results for my case.

Here is my circuit:

enter image description here

And here is the bode plot (where 825Hz is not on the top as expected)

enter image description here

  • \$\begingroup\$ Try use this analog.com/designtools/en/filterwizard \$\endgroup\$
    – G36
    Mar 2, 2019 at 15:42
  • \$\begingroup\$ Well, either you've made a mistake or the original author of the paper made a mistake or both of you have made mistakes. Try following the guidelines to recreate the filter in the worked example and see what happens. And, in the future, when you say things like "is not at the top as expected", nobody here knows where the top is because you haven't stated it. Neither does you bode plot have any reference axes. \$\endgroup\$
    – Andy aka
    Mar 3, 2019 at 16:23
  • 1
    \$\begingroup\$ Beside a possible mistake, as a general comment, if you want to plot the response of a high-Q filter, you have to increase the amount of points in your ac analysis. Push it to 1000 rather than 100 and it should help. \$\endgroup\$ Mar 3, 2019 at 21:28
  • \$\begingroup\$ @VerbalKint THANK YOU SO MUCH. Changing the resolution points from 100 to 1000 solved the problem. You can post the answer to accept it \$\endgroup\$ Mar 4, 2019 at 21:21
  • \$\begingroup\$ Glad if I could help! These things are usually not disclosed in the user manual : ) \$\endgroup\$ Mar 10, 2019 at 13:14

1 Answer 1


When plotting high-\$Q\$ transfer functions with any graphical viewer whether data are coming from a SPICE engine or a mathematical solver, it is important to maintain an adequate number of points per decade. 100 is usually enough for 1st-order responses but as soon as \$Q\$ is way above 1, you can potentially miss the peak if granularity is not good enough. For this purpose, you can push the number of points per decade to 1000 or even more if necessary.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.