(analog audio) filters for dummies.

Thread Starter

BobbyTheD

Joined Mar 10, 2016
23
hi forum-

i'm not much of an engineer or mathematician, but i've become interested in filters for audio purposes. i've read a bit and built a bit, but there are some things that evade me. i was hoping some folks here could explain a few things so that a simpleton might understand them. i will review the ESP page more: http://sound.westhost.com/articles/active-filters.htm

let me see if i'm on the right track - please don't hesitate to correct me:

firstly, my understanding is that there are 5 primary "ideal" categories of analog filters. low-pass, band-pass, high-pass, band-reject (notch), and all-pass.

there is a lot of talk of "filter responses" - these responses are graphed according to higher-level mathematic theory - and they include things like the chebyshev, butterworth, and bessel. i think that these responses can be applied to any of the primary categories (with the possible exception of all-pass), but i'm not totally sure about that point. is there a compilation and summary of all the most popular responses? are there definitive translations of these responses into topologies?

now, these "responses" are neither active nor passive. they're conceptual. i believe they are derived via "transfer function", which involves "polynomials" - but i don't really understand how that stuff works...is there an ultra-basic introduction to that kind of material?

from there, active and passive filters generally refer to specific circuit topologies... there are many. i'm unclear which things are "fixed" and which things are "theoretical." for instance, are there multiple topologies of a state-variable filter?

i notice a lot of the equations for these circuits deal with q and cutoff frequency..usually a ratio-metric relationship...but i am curious about how a filter designer might begin to deal with things like transient response and "ripple" in her or his work.

thanks everyone - i apologize for the broadness.
 

Papabravo

Joined Feb 24, 2006
14,214
...is there an ultra-basic introduction to that kind of material?

The ultra-basic introduction involves the differential calculus. If you don't understand that, then the rest of filter specification and design is going to remain a mystery. Transfer functions require the understanding of the Laplace Transform and how it changes a time domain problem into a frequency domain problem. Lastly the methods of solving Linear Differential Equations are required to tie things up in a neat package.

I can tell you that there are simple algebraic formulas you can apply that will tell you the proper "order" for the lowpass prototype filter given a minimum attenuation in the passband and a maximum attenuation in the stopband. I can also tell you that there are algebraic methods to scale and transform a lowpass prototype into a highpas, bandpass, and bandreject filter. AFAIK the allpass filter is just a wire.
 

atferrari

Joined Jan 6, 2004
4,069
i apologize for the broadness.
Hola Bobby,

Diferential equations aside, let me suggest this exercise: take that OP of yours and think of it as given for YOU to answer the questions.

I probably know less than you about that and many (many!) other subjects related to electronics but when it comes to ask questions about a particular subject, do the broad search by yourself. Nobody is going to post any lengthy explanation here. Come here for specific doubts. It works.

If you reread precisely the page you linked in your post, you could see that some of your questions could be answered. Other simple (and effective way) is to read the basics in more than one application note or particular article. From the top of my head: National Semiconductors, Microchips, Analog Devices, Maxim, run some ANs where you could start. Google Bonnie Baker.

Maybe one day, you could be able to comment (based on own experience) an all pass filter being a simple wire. Or is it the the other way round?
 

Attachments

Last edited:

Thread Starter

BobbyTheD

Joined Mar 10, 2016
23
My sense was that I summarized my understanding of the article and expressed the areas where I still felt the doubts. I think I misunderstood the forum's purpose. My apologies.
 

crutschow

Joined Mar 14, 2008
25,250
My sense was that I summarized my understanding of the article and expressed the areas where I still felt the doubts. I think I misunderstood the forum's purpose. My apologies.
The purpose of this forum is to answer posted questions related to electronics.
I realize that we sometimes (perhaps often :rolleyes:) digress from the original questions, but that's normal.
So did none of our comments or atferrari's references help answer your questions?
 

AnalogKid

Joined Aug 1, 2013
8,534
To answer one of your questions, yes. For some filter topologies, simply changing resistor and capacitor values can create filters with different rolloff characteristics (Butterworth, Bessel, etc.) at exactly the same corner frequency.

ak
 

Thread Starter

BobbyTheD

Joined Mar 10, 2016
23
Oh, no, there were several helpful responses and the reference sheets are very useful- I just got the impression that it was bad form to use this thread to hash out my misconceptions.
 

Veracohr

Joined Jan 3, 2011
720
Butterworth, Bessel, etc. are also called "approximations", because they approximate an ideal filter (while optimizing certain characteristics). I think they only apply to low pass and high pass. If the response is partly defined by passband behavior it wouldn't make much sense for a bandpass filter would it? Unless it was a wideband filter, which is just a low pass and high pass cascaded.

You can calculate the response of a filter without knowing Laplace transforms or differential equations if you're willing to take a few things for granted and not ask why. I learned both within the last couple years but there's no way I could do either off the top of my head any more. But I can still find a transfer function regardless.
 

Kermit2

Joined Feb 5, 2010
4,162
Most circuit designs need adjustment when theory collides with reality. You design for freq X, and then must change component values when you build the circuit and find its cutoff at freq Y. Component layout and PCB layout introduce many inductive and capacitive values that are mostly missing from a schematic.
Whatever you calculate will almost always be different than what you actually get when you build it. It may be very close but almost never exactly what the math says.
Knowing theory is good but seldom does it prove absolutely accurate. Experimentation and experience are also major factors in a good circuit design.
 

AnalogKid

Joined Aug 1, 2013
8,534
I just got the impression that it was bad form to use this thread to hash out my misconceptions.
Not at all. Don't let the grouchy minority put you off.

Active filter design is about as analog-y as it gets, definitely toward the deep end of the pool. But ask a decent question and you'll get a decent answer.

ak
 

SLK001

Joined Nov 29, 2011
1,548
For what you are doing, learning the theory makes no sense. There are plenty of online filter calculators and books and articles that will not only design your circuit, but will also plot your responses. For "audio", I'm assuming that you mean 20-20kHz. Are you designing a cross-over network for speakers?

Butterworth filters are maximally flat in the passband, but at the expense of shallower slope in the roll-off. Chebyshev filters have a steeper roll-off, but at the expense of ripple in the passband.

Perhaps if you could share with us what your ultimate goal is, more directed help can be garnered.
 

Thread Starter

BobbyTheD

Joined Mar 10, 2016
23
Well, I'm essentially interested in exploring sounds that can be made with electronics. My ultimate goal, I think - is to design filters that misbehave more than the standard fare. I'm interested in the *worst* transient response and the *most* passband ripple, in particular. The subgoal was to just understand what I was reading about in all these application notes and overviews - and I think I'm getting closer.
 

Papabravo

Joined Feb 24, 2006
14,214
I see where you are coming from and it is my belief that you may discover interesting audio effects by messing around with filters, but the theory of what the filters are doing will probably provide no insight at all into what you find interesting or objectionable. As an alternative you may find it easier to do your experimentation in the digital domain. Trying to adjust components in a physical filter will probably not satisfy your curiosity, especially since analog filters need to be designed for a specific purpose rather than just messing around.
 
Top