[SciPy-User] SciPy and MATLAB give different results for 'buttord' function

[Renan Birck Pinheiro]

2015-01-21 14:44 GMT-02:00 Warren Weckesser <warren.weckesser at gmail.com>:
>
> When you design a Butterworth filter with buttord, there aren't enough
> degrees of freedom to meet all the design constraints exactly.  So
> there is a choice of which end of the transition region hits the
> constraints and which end is "over-designed".  The change made in
> 0.14.0 switched that choice from the stop-band edge to the pass-band
> edge.
>
> A picture will make it clear.  The attached script generates the
> attached plot.  (I changed Rp from 3 to 1.5.  -3 dB coincides with the
> gain at Wn, that's why your Wn was the same as Wp.)  The filters
> generated using either the old or new convention both satisfy the
> design constraints, so they are both correct.  With the new
> convention, the response just bumps against the constraint at the end
> of the pass-band.  With the old convention, the response hits the
> constraint the edge of the stop-band.


Hello,

This pretty much explains it. ​Thank you.

​Renan

-- 
Renan Birck Pinheiro - Chip Inside Tecnologia <http://www.chipinside.com.br>
Acad. Engenharia Elétrica <http://www.ufsm.br/cee> - UFSM
<http://www.ufsm.br> - Santa Maria, Brasil
http://renanbirck.blogspot.com - +55 55 91162798 / +55 55 99034839

*Talk is cheap, show me the code*. - Linus Torvalds
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.scipy.org/pipermail/scipy-user/attachments/20150121/23f74138/attachment.html>