IIR filter question

Hi Guys,

   Thanks much for all your posts so far.

Your words are very interesting!!

Regards,
[-Rick-]

I can only speak within my area of professional expertise, which is
professional audio; and admittedly this is a small area within the
realm of DSP.  That said, IIR filters are still used far more
frequently than FIR.  I can think of a few reasons off the top of my
head:
1.  IIR filters can be easily designed such that they have an analog
counterpart, with which the user is familiar.
2.  IIR filters are more easily adjustable in real-time.
3.  IIR filters are more efficient at low frequencies (< ~0.1*fs),
where much of audio processing takes place.

Brian Neunaber
QSC Audio Products
www.qscaudio.com

Brian Neunaber wrote:
> I can only speak within my area of professional expertise, which is
> professional audio; and admittedly this is a small area within the
> realm of DSP.  That said, IIR filters are still used far more
> frequently than FIR.  I can think of a few reasons off the top of my
> head:
> 1.  IIR filters can be easily designed such that they have an analog
> counterpart, with which the user is familiar.

A light shines ;]
Suspect is cause of many of my questions said to be not meaningful and 
why I just "don't comprehend" many answers.

The *ONLY* filters I've had hands on experience with were analog.
Hey, opamps were advertised as "chopper stabilized".

Will have to start rereading Mr. Lyon's tome with that in mind [If I can 
where I left it this time ;]





> 2.  IIR filters are more easily adjustable in real-time.
> 3.  IIR filters are more efficient at low frequencies (< ~0.1*fs),
> where much of audio processing takes place.
> 
> Brian Neunaber
> QSC Audio Products
> www.qscaudio.com
> 

Brian Neunaber skrev:
> I can only speak within my area of professional expertise, which is
> professional audio; and admittedly this is a small area within the
> realm of DSP.  That said, IIR filters are still used far more
> frequently than FIR.  I can think of a few reasons off the top of my
> head:
> 1.  IIR filters can be easily designed such that they have an analog
> counterpart, with which the user is familiar.

Once upon a time this was true. These days (or rather, for the
last 15-20 years), few students who are assigned the task of
designing discrete-time IIR filters based on analog prototypes
will have the experience necessary to be able to exploit
the scarce keywords contemporary DSP texts provide.

> 2.  IIR filters are more easily adjustable in real-time.
> 3.  IIR filters are more efficient at low frequencies (< ~0.1*fs),
> where much of audio processing takes place.

Very good reasons for getting to know IIR filters more intimately.
Just too bad no current DSP authors seem to see it as worth their
while to lay down the necessary foundations. 

Rune

Hello Rafael,


>>the author implies that IIR (recursive)
>>filters aren't as popular nowadays as they were in
>>the past (say 10-20 years ago).
> 
> 
> Probably goes back to the implication in 
> old books that IIRs are better because they 
> need less multiplications then FIR and 
> since multiplications are timeconsuming
> ( shift & add ) so FIRs are no use.
> "20-30 years ago" would be ok on that
> line of argument.
> 

It's still a valid argument even though WDFs are encroaching. Look at 
the MSP430 series. A HW multiplier is like ordering leather seats. Comes 
at a steep increase in cost and when you need very small footprints like 
the F2xxx series you won't be able to buy one with a HW multiplier for 
quite a while. Then you are back to assembler and shift-add.

Regards, Joerg

http://www.analogconsultants.com

Brian Neunaber wrote:
> I can only speak within my area of professional expertise, which is
> professional audio; and admittedly this is a small area within the
> realm of DSP.

i really don't know diddley about trends and i certainly cannot use my
judgement of what should be more popular because it's more often than
not wrong.  what i *think* should be popular (because i think it's a
"better mousetrap") is almost always *not* what is popular.

>  That said, IIR filters are still used far more frequently than FIR.

in audio, right Brian?  dunno if i would make such a statement more
broadly.  and i dunno if i would use the term "far more".  but i think
i agree.

also, i dunno if straight filtering and EQ is done in audio (in which i
would agree is more often IIR than FIR) more often than sample rate
conversion (or some other resampling) in which i think FIR is the way
that is implemented.

but there are many effects, such as phasing (uses APF), reverb, and
high-Q filtering used in audio which nearly has to be done with IIRs.

>  I can think of a few reasons off the top of my
> head:
> 1.  IIR filters can be easily designed such that they have an analog
> counterpart, with which the user is familiar.

i guess you can sample the analog impulse response, sorta like the
"inpulse invariant" method of IIR design, but just hack off the IIR and
call it an FIR.  that's pretty easy to design.  not so sure it's a
"good" design.

> 2.  IIR filters are more easily adjustable in real-time.

i guess so, if the FIR is pretty long.

> 3.  IIR filters are more efficient at low frequencies (< ~0.1*fs),
> where much of audio processing takes place.

yup, FIRs for bass have to be pretty long.  one problem with IIRs down
there is that the coefficient precision can be a little problematic
because some coefs have cos(w0) in them and the difference between
cos(w0) and 1 (where all the information is) gets very poor precision,
even with floating-point, unless you do something about it.

r b-j

On 6 Apr 2006 08:32:45 -0700, "Brian Neunaber" <neunaber@gmail.com>
wrote:

>I can only speak within my area of professional expertise, which is
>professional audio; and admittedly this is a small area within the
>realm of DSP.  That said, IIR filters are still used far more
>frequently than FIR.  I can think of a few reasons off the top of my
>head:
>1.  IIR filters can be easily designed such that they have an analog
>counterpart, with which the user is familiar.
>2.  IIR filters are more easily adjustable in real-time.
>3.  IIR filters are more efficient at low frequencies (< ~0.1*fs),
>where much of audio processing takes place.
>
>Brian Neunaber

Hi Brian,

   Looking at your Item# 2, the author of the article 
that I'm reviewing states that IIR filters are "difficult, 
if not impossible" to use for adaptive filtering.

Is he exaggerating the difficulty in using IIR filters 
for adaptive filtering?

Thanks,
[-Rick-]
 

On Thu, 06 Apr 2006 19:52:23 GMT, R.Lyons@_BOGUS_ieee.org (Rick Lyons)
wrote:

>On 6 Apr 2006 08:32:45 -0700, "Brian Neunaber" <neunaber@gmail.com>
>wrote:
>
>>I can only speak within my area of professional expertise, which is
>>professional audio; and admittedly this is a small area within the
>>realm of DSP.  That said, IIR filters are still used far more
>>frequently than FIR.  I can think of a few reasons off the top of my
>>head:
>>1.  IIR filters can be easily designed such that they have an analog
>>counterpart, with which the user is familiar.
>>2.  IIR filters are more easily adjustable in real-time.
>>3.  IIR filters are more efficient at low frequencies (< ~0.1*fs),
>>where much of audio processing takes place.
>>
>>Brian Neunaber
>
>Hi Brian,
>
>   Looking at your Item# 2, the author of the article 
>that I'm reviewing states that IIR filters are "difficult, 
>if not impossible" to use for adaptive filtering.
>
>Is he exaggerating the difficulty in using IIR filters 
>for adaptive filtering?

I have done quite a bit of adaptive filtering work with FIRs. They are
extremely easy to get to work and always stable. IIRs have been
certainly difficult to use for adaptive filtering because they can be
made unstable inadvertently. But there has been a recent upsurge in
IIR adaptive filtering research and it is certainly doable. Difficult
yes but definitely not impossible, no where near impossible actually.
And if you consider the amount of hardware you're saying (low power is
the order of the day) it's certainly worth the effort.

Rick Lyons wrote:
> On 6 Apr 2006 08:32:45 -0700, "Brian Neunaber" <neunaber@gmail.com>
> wrote:
> 
> 
>>I can only speak within my area of professional expertise, which is
>>professional audio; and admittedly this is a small area within the
>>realm of DSP.  That said, IIR filters are still used far more
>>frequently than FIR.  I can think of a few reasons off the top of my
>>head:
>>1.  IIR filters can be easily designed such that they have an analog
>>counterpart, with which the user is familiar.
>>2.  IIR filters are more easily adjustable in real-time.
>>3.  IIR filters are more efficient at low frequencies (< ~0.1*fs),
>>where much of audio processing takes place.
>>
>>Brian Neunaber
> 
> 
> Hi Brian,
> 
>    Looking at your Item# 2, the author of the article 
> that I'm reviewing states that IIR filters are "difficult, 
> if not impossible" to use for adaptive filtering.
> 
> Is he exaggerating the difficulty in using IIR filters 
> for adaptive filtering?

It sounds like this guy is doing a lot of hand waving he would find 
difficult to support with hard evidence. Adaptive IIRs seem to be widely 
used these days. I've never tried using them, and they do seem to be a 
PITA to get right. They do, however, seem to work well enough, when care 
is taken over stability, to ship product.

Regards,
Steve

Steve Underwood wrote:
>
> It sounds like this guy is doing a lot of hand waving he would find
> difficult to support with hard evidence. Adaptive IIRs seem to be widely
> used these days. I've never tried using them, and they do seem to be a
> PITA to get right. They do, however, seem to work well enough, when care
> is taken over stability, to ship product.

lotsa Gugel hits when you type in "IIR Adaptive Filter".

the philosophy appears to let the output feedback states be additional
input states to a sorta pseudo-FIR and do the same LMS downward
gradient song-&-dance that the regular old LMS adaptive FIR does.

r b-j