| View previous topic :: View next topic |
| Author |
Message |
primiano
Guest
|
 Posted: Sat Nov 15, 2008 10:48 pm Post subject: Low Freq Highpass FIR filter problem |
 |
|
Hi,
I'm trying to model acoustic (20hz-20khz) crossover filters using FIR
filters (Fs: 48000 Hz)
Now, I have succesfully "approximated" the most filters (tipically LC
Butterworh 2nd order filters) using 40-50 taps FIRs.
A problem comes up when I try to made an highpass (or bandpass) FIR filter
for the midrange.
It should pass at 300-500 hz (and stop the frequencies below)
But it seems that designing an higpass FIR that cuts at so low frequencies
is an impossibile mission...
I worked all the day with Matlab FDATool but all I could design were
"horrible" filters (with high ripples in the passband) that require no less
than 200-300 taps
Can anyone give me an hint about the problem?
Am i doing something wrong?
Thanks in advance,
Primiano Tucci
--
http://www.primianotucci.com |
|
| |
|
 Back to top |
Tim Wescott
Guest
|
| Posted: Sat Nov 15, 2008 11:22 pm Post subject: Re: Low Freq Highpass FIR filter problem |
|
|
On Sat, 15 Nov 2008 10:48:53 -0600, primiano wrote:
| Quote: | Hi,
I'm trying to model acoustic (20hz-20khz) crossover filters using FIR
filters (Fs: 48000 Hz)
Now, I have succesfully "approximated" the most filters (tipically LC
Butterworh 2nd order filters) using 40-50 taps FIRs.
A problem comes up when I try to made an highpass (or bandpass) FIR
filter for the midrange.
It should pass at 300-500 hz (and stop the frequencies below) But it
seems that designing an higpass FIR that cuts at so low frequencies is
an impossibile mission...
I worked all the day with Matlab FDATool but all I could design were
"horrible" filters (with high ripples in the passband) that require no
less than 200-300 taps
Can anyone give me an hint about the problem? Am i doing something
wrong?
Thanks in advance,
Primiano Tucci
|
Your filter order depends in large part on the ratio of the cutoff
frequency and the sampling rate, and must get larger if you want things
like steep skirts and/or flat passband response.
So I'm not surprised.
I don't do this often enough to help you out with the exact filter order
that you need, but you have a 160:1 ratio between the sampling rate and
the desired cutoff, I would expect that you'd have to go at least 5 times
higher than that to get something that sounds nice, particularly
considering that your cutoff is smack in the middle of the human hearing
range.
I'd be looking at longer filters, or IIR filters.
--
Tim Wescott
Wescott Design Services
http://www.wescottdesign.com
Do you need to implement control loops in software?
"Applied Control Theory for Embedded Systems" gives you just what it says.
See details at http://www.wescottdesign.com/actfes/actfes.html |
|
| |
|
| Back to top |
Jerry Avins
Guest
|
| Posted: Sun Nov 16, 2008 12:40 am Post subject: Re: Low Freq Highpass FIR filter problem |
|
|
Tim Wescott wrote:
| Quote: | On Sat, 15 Nov 2008 10:48:53 -0600, primiano wrote:
Hi,
I'm trying to model acoustic (20hz-20khz) crossover filters using FIR
filters (Fs: 48000 Hz)
Now, I have succesfully "approximated" the most filters (tipically LC
Butterworh 2nd order filters) using 40-50 taps FIRs.
A problem comes up when I try to made an highpass (or bandpass) FIR
filter for the midrange.
It should pass at 300-500 hz (and stop the frequencies below) But it
seems that designing an higpass FIR that cuts at so low frequencies is
an impossibile mission...
I worked all the day with Matlab FDATool but all I could design were
"horrible" filters (with high ripples in the passband) that require no
less than 200-300 taps
Can anyone give me an hint about the problem? Am i doing something
wrong?
Thanks in advance,
Primiano Tucci
Your filter order depends in large part on the ratio of the cutoff
frequency and the sampling rate, and must get larger if you want things
like steep skirts and/or flat passband response.
So I'm not surprised.
I don't do this often enough to help you out with the exact filter order
that you need, but you have a 160:1 ratio between the sampling rate and
the desired cutoff, I would expect that you'd have to go at least 5 times
higher than that to get something that sounds nice, particularly
considering that your cutoff is smack in the middle of the human hearing
range.
I'd be looking at longer filters, or IIR filters.
|
Sharp skirts will make the transient response suffer badly. 80 dB/decade
is about as much as is tolerable. The phase at crossover is important.
Zero phase shift is inappropriate if the speakers are physically separated.
Jerry
--
Engineering is the art of making what you want from things you can get.
¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯ |
|
| |
|
| Back to top |
Richard Dobson
Guest
|
| Posted: Mon Nov 17, 2008 10:47 pm Post subject: Re: Low Freq Highpass FIR filter problem |
|
|
Jerry Avins wrote:
...
| Quote: | Sharp skirts will make the transient response suffer badly. 80 dB/decade
is about as much as is tolerable. The phase at crossover is important.
Zero phase shift is inappropriate if the speakers are physically separated.
|
Meridian (celebrated for their high-end dsp-controlled speakers)
concluded that minimum-phase was the preferred solution in most cases.
This was recently mentioned in passing on the sursound list. The only
reference I have found is:
www.audiosignal.co.uk/Resources/Digital_room_equalisation_A4.pdf
Richard Dobson |
|
| |
|
| Back to top |
Jerry Avins
Guest
|
| Posted: Mon Nov 17, 2008 11:59 pm Post subject: Re: Low Freq Highpass FIR filter problem |
|
|
Richard Dobson wrote:
| Quote: | Jerry Avins wrote:
..
Sharp skirts will make the transient response suffer badly. 80
dB/decade is about as much as is tolerable. The phase at crossover is
important. Zero phase shift is inappropriate if the speakers are
physically separated.
Meridian (celebrated for their high-end dsp-controlled speakers)
concluded that minimum-phase was the preferred solution in most cases.
This was recently mentioned in passing on the sursound list. The only
reference I have found is:
www.audiosignal.co.uk/Resources/Digital_room_equalisation_A4.pdf
Richard Dobson
|
That's been my unfounded opinion for a long time. One of the best
excuses I know for a ridiculously high sample rate is that
impulse-invariant crossovers then conform closely to minimum phase.
Jerry
--
Engineering is the art of making what you want from things you can get.
¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯ |
|
| |
|
| Back to top |
|
