Filter construction kit

[tester]

MyCo - some time ago (was it 15-20 years ago) - I was teaching some math and physics to people who did not knew how to approach it. They were able to pass the exams with high notes. What I've learned is. Even easy math looks complicated if you don't "split" it into something else. If I see an equation, that is very long and is made of many symbols (no matter how simple they are) - if I don't have basic "flow" idea of what is going on there, then I wont get it. Either because there are too many tiny things to remember at once (remember these similar pictures in which you had to find minor differences?), or because these things are so abstract, that I forget them right after I switched my attention to another pieces. It's like this. If I have a device with two gauges, then first what it's important is - "this gauge is for speed, and this for amount". Then - "okay, this gauge is linear, and that one is not (which means...)". Then "there are four signifficant settings for these two gauges". And then - "okay, here are the equations for these gauges". If you have a device with two abstract formulas with many symbols inside, then you probably wont use these gauges first place... It's like looking inside FS/SM schematic - too many links will force you to think more about one thing, but you can't focus on another (remember that movie with basketball exchange and person wearing a monkey suit walking through the center of the scene? many people don't see the monkey).

Outside programming, filter is something, that produces sound from another sound. What is important for me, is loike this:

- how the slope behaves (shaprness and edge amplification), and how it sounds
- what happens with phase of my audio - how it changes over the spectra
- what is stable and what is unstable
- how it will react to slow/fast modulation
- how it will react to dynamics (adaptation over time; i.e. is it better for constant amplitude, or for dynamic impulses?)
- what about accuracy in various parts of the spectra?
- is it better for higher or lover spectra or else?
- how these above informations refer to that formula

Btw, you see what are the "needs"; these are very general things, but they will define my reason to remember this and not that formula. Context.

...plus what you wrote - "need a condition to stop", and "these coeffs are not like gauges for speed and amount - they are math related tiny things and they have no physical representation in terms of simple relationship".

Regarding the size of the formula - no matter what size is - if the concept is always the same - then everyone can make big formulas on their own.

Regarding greek symbols - generally I see what they refer to, but if you ask me about the details - I would have to go through wiki or something like that. But you don't need to explain the greek equation (in mid level technical school - we saw these symbols maybe few times?); more imporant are the unrolling parts in their context (attentional motivation, why should I remember this or that), and then the details inside, what to pay attention to.

You asked if this is easy for someone who never developed a filter. I never developed a filter, but my skills allow me to do things like this. So answering to your concern - I would define first - who is the "target reader". FS/SM allows to develop apps without any specific programming/DSP/math skills or knowledge; it's like lego bricks and it requires really basic understanding of things. Who among them - is the target reader?

*

Another example. You started about "filter response", but at that moment - I should have a practical context behind that. "In practical terms, filter response is..." ? And here your reader stops to follow de to lack of tactile contact with abstraction. Sure, if I'm lucky, google will find something easy to understand, but as you know - google finds very different things on different computers.

[MyCo]

Well, I can't explain everything. This is just a spare time project, and writing is not my favorite work.

I've added the chapter "Filter normalization", the file is in my first post in this thread.

[tester]

Don't worry. I'm not complaining. It took me some years, to polish my self in order to write usable texts. It's not an easy task, especially when you deal with technical articles and so diversified audience.

[MyCo]

Documentation Update
- Fixed a bug in the documentation (filter code wasn't right).

Toolbox Update
- Added Filter cascading module
- Added RBJ calculation
- Added Biquad stream module
- Updated z-Plane view
- Added some tests for the changes above

Downloads are in the first post.

[MyCo]

Toolbox Update
- Added Butterworth pole/zero calculation
- Added Log-Frequency Response calculation
- Added multi-pole/zero to coeffs transformation
- Added Phase Response unwrapping
- Added some tests for the changes above

Downloads are in the first post.

[tester]

Nice development there.

[trogluddite]

Great work, MyCo, many thanks.

[infuzion]

Well, I can't explain everything. This is just a spare time project, and writing is not my favorite work.

Typical programmer; low on communication

[tester]

Inf - are you kidding? MyCo is very responsive, one of my favorites on this and SM board

[MyCo]

Typical programmer; low on communication

Yeah, but how about you? Try writing in a foreign language... try German

Documentation Update
- Fixed a bug in the coeff calculation

Toolbox Update
- Added Graph module
- Extended Biquad calculation, to support 2 Real-Axis poles/zeros
- Change coeff calculation on multi-pole/zero transformation, so that the least amount of coeffs are generated
- Added Butterworth audio test (it sucks!)

Downloads are in the first post.