I'm working my way through Bowick's RF Circuit Design. In the resonant circuit chapter he does a great job of describing how component Q affects circuit Q, and how loaded Q works. There, you set the bandwidth of a band pass filter by deliberately changing its quality.

But in the filter chapter describing filter prototypes, the design procedure is quite different. First you select a normalized prototype based on the attenuation and passband ripple needed. Then you transform if needed, then scale. But the component Q criteria are left at this table:

quality for filter type

The lower the component Q, the wider the bandwidth and the shallower the rolloff attenuation. However, it was not explained by exactly how much. To calculate total Q, I can presume that I could take the Q of the last shunt segment and combine it with the Q of the second last series segment using Qtot = 1/(1/Q1+1/Q2). But I'm not sure how to then combine that with the next shunt Q. Or I could try to transform all resistance out of the filter into one series resistor. Not sure which approach to take. So that's my first question.

My other questions are:

  • Where do these figures in the table above for minimum Q come from?
  • How does component Q affect the actual Butterworth parameters?
  • How do I adjust the Butterworth parameters to account for component Q, rather than just having to follow the general guidelines above?
  • If an ideal Butterworth has no resistance, how is it possible to have a wide bandwidth?
  • \$\begingroup\$ Presumably he showed you how to do the analysis in the earlier chapter. But you want him to spoon feed you in the later chapter. Who's copping out? \$\endgroup\$ – The Photon May 7 '18 at 22:42
  • \$\begingroup\$ This looks like the Q needed for n>10th order filter. I have many tools to choose any n and all filter types or search my answers for Bessel for details compared to Butterworth \$\endgroup\$ – Tony Stewart Sunnyskyguy EE75 May 7 '18 at 23:38
  • \$\begingroup\$ Cheby. Filters have steepest skirts demanding staggered high Q’s while Bessel has lowest Q with staggered f for maximally flat group delay and thus lower sensitivity \$\endgroup\$ – Tony Stewart Sunnyskyguy EE75 May 7 '18 at 23:41

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.