[sdiy] new to the list... already have a question :)

[Magnus Danielson]

From: Robert Pietrusko <bobby at csounds.com>
Subject: 
Date: Tue, 30 Dec 2003 07:25:47 -0800 (PST)
Message-ID: <20031230152547.45391.qmail at web41312.mail.yahoo.com>

> Hi Magnus,

Hi Bobby,

You've lost the subject-line, so I recreated it. Make sure your emails to the
list got a subject-line, M'key?

> The best reference I have found, and the one I am
> using, is the old Van Valkenburg text "Analog Filter
> Design," which has been a bible for any other filters
> I've designed.  In this case I am using mainly
> chapters 12 and 13.
> 
> The Gioffi equations in chapter 12 are used to compute
> resistor and capacitor values for a Boctor active
> filter circuit from a 2nd order transfer function. 

Well, I just don't have that book. I did dig up something on the web thought,
so now I've seen it.

> The result is a high-pass notch. I'm fairly certain I
> can't get this response out of a Sallen-Key design.

Sallen-Key is nice for "clean" responces in the traditional sense, but Cauer
stuff is not as straigh on. Sallen-Key is not my choice for large-pole stuff
that you are looking at anyway.

> I guess another more practical question is,  if most
> people are not familiar with this approach and I am
> having a hard time finding material, am I barking up
> the wrong tree?

I would consider doing either a multiple state-variable filter design,
with 2-pole/2-zero per section. It is very easy to design and the dimensioing
formulas is really a childsplay relative to what you are looking at.

I would however consider doing a ortonormal leapfrog ladder. It is really a
good choice since you both have a low sensitivity to individual components and
good dynamic properties. When doing a 10th order filter I think you should pay
attention to those details. I have the relevant ortonormal leapfrog ladder
paper lying around here somewhere. As I recall it (it is after all some 10
years ago I read it), it was not that horrifying to dimension once you've
read it. Basically any leapfrog filter contains multiple feedback loops and it
takes some experience to make fault analysis on it. It does become good filters
in the end.

> What is a more "standard" method of realizing
> Cauer/Elliptic filters using opamps? or even, a more
> standard method of realizing high-pass notch functions
> that are peaky at the cutoff frequency?

State-variable filters isn't a bad selection many times, but if you want to
press the most cream out of it, you should have a large selection of filter
layouts and choose depending on each requirement. The above mentioned methods
is fairly good and would probably do the trick for you most of the times, when
doing FIXED filters.

> as is  always the case for me, it's not the filter
> design, but its analog implementation that is causing
> the trouble. :) ugh!

Start of in the SVF-field then. It is farily easy, well known and really quite
straight-forward. They also have a theoretic beutty to them, but then are not
as good as you can do it if you make more specific choices, but may very well
do the trick for you.

> I'd be happy to message in more detail off list to
> spare the inboxes of other list members.

Feel free! ;O)

However, I know for a fact that there is several lurkers out there that tend to
read threads like this in the quite and just sucking up knowledge even if it is
over their heads. I've got comments from some of these that they read ALL of my
posts! Lately my production of quality comments have been low though, but you
got me going on one of my favorite topics, so... ;O)

Cheers,
Magnus - a.k.a. Captain Signaltheory - your continous-time superhero!