Book: The Art of VA Filter Design 2.1.2

[Kraku]

This DSP and plug-in development forum is the only one where people discuss about the PDF format in a topic related to an e-book about DSP

And that's a great thing. As the book is meant to reach as wide audience as possible, the "usability/readability" of the book becomes one of the main focuses. The easier the book is "to use", the more audience it will get in a longer run.

And if the book keeps spreading throughout the internet, Vadim will likely have even more motivation to keep adding interesting content into the book. Then we get even more goodies

[Z1202]

And if the book keeps spreading throughout the internet, Vadim will likely have even more motivation to keep adding interesting content into the book. Then we get even more goodies

So far I'm working on the motivation to reorganize the references according to the forum suggestions

[aciddose]

This has been mostly solved in printed books by finding an OK balance between repeating the formula/graph and just referring to it.

I've edited this post to remove redundant nonsense and to be honest... I don't care enough either way.

Short version: redundancy = bad.

[Musicisbest]

Nice, I love to see this kind of info, thanks.

[Ivan_C]

[Z1202]

...youtube link....

Thanks Ivan! I was waiting for this, but you beat me to it

[Ivan_C]

You're welcome, it was a pleasure to meet you there

[Z1202]

You're welcome, it was a pleasure to meet you there

Same here, Sir

[evv]

Great resource, thanks Vadim.

I might have found a typo (or I might be misunderstanding sth!)

On version 2.1.0 of the book, page 21, in the middle of the last paragraph on subsection "Rolloff", the text says "The same statement applies for w->0 if Nz0 > 0, ...". Should this be "Nz0 = 0" instead? It's talking about the cases where the rolloff is 0dB/oct, and given that the rolloff speed at w->0 is 6 Nz0 dB/oct, it would make more sense to me if the above sentence was "Nz0 = 0" instead of "Nz0 > 0".

Thanks!

[Z1202]

I might have found a typo (or I might be misunderstanding sth!)

On version 2.1.0 of the book, page 21, in the middle of the last paragraph on subsection "Rolloff", the text says "The same statement applies for w->0 if Nz0 > 0, ...". Should this be "Nz0 = 0" instead? It's talking about the cases where the rolloff is 0dB/oct, and given that the rolloff speed at w->0 is 6 Nz0 dB/oct, it would make more sense to me if the above sentence was "Nz0 = 0" instead of "Nz0 > 0".

Your understanding is correct. Thanks!

[Z1202]

While we're on the subject, I'd like to refine the statement of the footnote 15 on the same page. Of course, the direction at which a zero at s=0 or s=inf can be approached is not fully arbitrary, but is always a multiple of 90 degrees and so is the phase response at 0 and inf.

[mystran]

While we're on the subject, I'd like to refine the statement of the footnote 15 on the same page. Of course, the direction at which a zero at s=0 or s=inf can be approached is not fully arbitrary, but is always a multiple of 90 degrees and so is the phase response at 0 and inf.

This is actually one of my favorite "weird" cases, since you have the -90 degrees per transfer function zero in theory, but unless you have a fixed-point digital filter, component tolerances or floating-point rounding errors usually tend to move the zeroes ever so slightly, so you end up with a multiple of 180 anyway... even though you might not even be able to measure it if the error is small enough, because getting a true DC reading would (literally) take forever.

[Z1202]

While we're on the subject, I'd like to refine the statement of the footnote 15 on the same page. Of course, the direction at which a zero at s=0 or s=inf can be approached is not fully arbitrary, but is always a multiple of 90 degrees and so is the phase response at 0 and inf.

This is actually one of my favorite "weird" cases, since you have the -90 degrees per transfer function zero in theory, but unless you have a fixed-point digital filter, component tolerances or floating-point rounding errors usually tend to move the zeroes ever so slightly, so you end up with a multiple of 180 anyway... even though you might not even be able to measure it if the error is small enough, because getting a true DC reading would (literally) take forever.

What I was mostly getting at is that the phase response smoothly approaches a multiple of 90 degrees, which for most practical purposes can be considered true (as you mentioned, the error is quite difficult to measure).

[mystran]

While we're on the subject, I'd like to refine the statement of the footnote 15 on the same page. Of course, the direction at which a zero at s=0 or s=inf can be approached is not fully arbitrary, but is always a multiple of 90 degrees and so is the phase response at 0 and inf.

This is actually one of my favorite "weird" cases, since you have the -90 degrees per transfer function zero in theory, but unless you have a fixed-point digital filter, component tolerances or floating-point rounding errors usually tend to move the zeroes ever so slightly, so you end up with a multiple of 180 anyway... even though you might not even be able to measure it if the error is small enough, because getting a true DC reading would (literally) take forever.

What I was mostly getting at is that the phase response smoothly approaches a multiple of 90 degrees, which for most practical purposes can be considered true (as you mentioned, the error is quite difficult to measure).

I have some old EQ code that uses a fitting method that can theoretically run into infinities if it ever samples an exact zero from the transfer function... so the code in question just designs all notches to be at -300dB (or something like that) and everything works perfectly fine and you really won't notice anything using any stock analyser.

[Z1202]

A minor bugfix update
https://www.native-instruments.com/file ... _2.1.2.pdf