Filter construction kit

[Nubeat7]

thanks myco, thats great! now i have new inspiration to study the filter world - and a new chance to understand it one day

[Urnsoft]

Great modules ,very useful .
I like the way you used to call the methods .

http://yehar.com/blog/?p=121

definitely goes to my must read folder.

Thanks Myco ,I'm learning a lot .

[MyCo]

I wrote down some of my current knowledge about filter design, so when I forget it I can find my way back into this topic. I wanted to use the least amount of math, but it turned you need a let of it to describe what's going on.

The file is in my first post of this thread

[stw]

I wrote down some of my current knowledge about filter design, so when I forget it I can find my way back into this topic. I wanted to use the least amount of math, but it turned you need a let of it to describe what's going on.

There is still a lot of stuff missing, eg. normalization, calculating filter response, higher order filters...

This is really cool MyCo.
Thanks for sharing!

[trogluddite]

Many thanks MyCo.

As you might have noticed, English is not my native language. Be patient!

No worries on that score - it is very well written. You've done a great job of keeping it concise and only explaining what us coders need to know.
Bravo!

[philton]

thank you myco, atm i think "this stuff is too high for me" but i am sure i will better understand it after i read that stuff

[MyCo]

I would like to know, if someone who has never learned filter design could understand that and use it. I tried to keep it as straight forward as possible, but maybe I forgot something important so that someone who never learned it before can't follow my explanations.

[infuzion]

Awesome stuff; thanks for sharing!

[tester]

I would like to know, if someone who has never learned filter design could understand that and use it. I tried to keep it as straight forward as possible, but maybe I forgot something important so that someone who never learned it before can't follow my explanations.

Guess this is addressed to me

I think it's a good direction, but not yet enough to do something on my own.

Things like real-life advantages and disadvantages are missing. "Go shopping" like:

...difference between those two is, that when you give a FIR filter an input of a fixed length, then the output has a fixed length, too. The output of an IIR filter can be infinitely long. The advantage of an IIR filter is, that you can get very good results in less processing steps, than with a FIR filter...

...so what's the practical difference outside programming? Why should I use this one instead of that one?

he sample processing in an IIR filter is always done by one and the same formula, but it can be modified to
fit your needs.

What do you mean by "my needs"?

...To keep it simple, in the further text I’ll only use 3 input coefficients...

Why ? What is the meaning of these coeffs in practical terms and simple words? In other words - why should I take this one?

*

Other things might be useful - a descriptive way of understanding the unrolling formulas. I see long and maybe simple equation, but help me to memorize it's meaning.

"okay, there are two parts inside that formula. one is related to coeffs "b" and second to coeffs "a". basically you (substract?) part with coeffs "a" from the part with coeffs "b". now in regards to these parts, keep it like this. when you go step by step from b1, b2, bn, you multiplyu it by... with coeffs "a" part - you go the same way. now - if you use a0-a3, then you use the same amount of "b" coefs (is there any "because"?)".

So, from modular and simple understanding of the formula you go deeper into the details. I don't need to remember the details as a whole, because these details suit simpler lego bricks. If I understand what is there and why it is there (or what else could be there), then I might be able to create such formula on my own. And then - translate it into programming language.

*

The same approach "from bricks to details" could help further I guess. Finding right simplifications, right bricks. Their context (reason to remember things) and details inside.

[MyCo]

...difference between those two is, that when you give a FIR filter an input of a fixed length, then the output has a fixed length, too. The output of an IIR filter can be infinitely long. The advantage of an IIR filter is, that you can get very good results in less processing steps, than with a FIR filter...

...so what's the practical difference outside programming? Why should I use this one instead of that one?

Outside programming, it is only used in math, and you don't want to go in this direction, believe me
FIR is usefull, when you have a fixed input signal length, and you want to stop the filter processing after the filter completed its work. An IIR filter never completes, it runs and runs forever, unless you use a break condition, eg. When my lowpass filter outputs values lower than 10^-10, stop processing.

he sample processing in an IIR filter is always done by one and the same formula, but it can be modified to
fit your needs.

What do you mean by "my needs"?

That's a good point, but hard to explain, before you've already read the rest of the document.

...To keep it simple, in the further text I’ll only use 3 input coefficients...

Why ? What is the meaning of these coeffs in practical terms and simple words? In other words - why should I take this one?

I limit it, because it's common. Later I want to include higher order calculations in the "appendix". But when I would use 10 coeffs to describe everything, the formulars would get 100 times bigger

Other things might be useful - a descriptive way of understanding the unrolling formulas. I see long and maybe simple equation, but help me to memorize it's meaning.

hm... ok... so you don't know what the big greek letters mean? I could explain that too, but that's just midlevel math.