Prev: partical directed coherence
Next: Basic concatenation question
From: david t53 on 7 Jul 2010 04:50
With filterbuilder, arbritary response (or fdesign.arbmag),
I get the following error if I have more than one band:
"The frequency bands 1 and 2 must be separated by a don't care region"

how do I specify a "don't care region"?


Regards,
David
From: david t53 on 7 Jul 2010 05:41
"david t53" <david_t53(a)sogetthis.com> wrote in message <i11f40$m2$1(a)fred.mathworks.com>...
> With filterbuilder, arbritary response (or fdesign.arbmag),
> I get the following error if I have more than one band:
> "The frequency bands 1 and 2 must be separated by a don't care region"
>
> how do I specify a "don't care region"?
>
>
> Regards,
> David

or what is the meaning of "band" in this case?
(it seems clear that it is not the same meaning as in a graphical equalizer, that can have 30 "bands")
an example of using fdesign.arbmag with more than one band?
From: Wayne King on 7 Jul 2010 06:36
"david t53" <david_t53(a)sogetthis.com> wrote in message <i11i3j$84a$1(a)fred.mathworks.com>...
> "david t53" <david_t53(a)sogetthis.com> wrote in message <i11f40$m2$1(a)fred.mathworks.com>...
> > With filterbuilder, arbritary response (or fdesign.arbmag),
> > I get the following error if I have more than one band:
> > "The frequency bands 1 and 2 must be separated by a don't care region"
> >
> > how do I specify a "don't care region"?
> >
> >
> > Regards,
> > David
>
> or what is the meaning of "band" in this case?
> (it seems clear that it is not the same meaning as in a graphical equalizer, that can have 30 "bands")
> an example of using fdesign.arbmag with more than one band?

Hi David,

fdesign.arbmag supports both single- and multi-band designs. I take it from the error you are getting that you are using a multi-band design. Let's take the case of a lowpass filter: the power of the multi-band design is that you can specify one response for the passband and one for the stopband. The "don't care" region is the transition region between the passband and stopband. By saying basically, I DON'T CARE what happens in the transition region, the algorithm is able to converge to a solution with better behavior in both the individual bands (passband and stopband). Of course, in order to do that, your passband and stopbands have to be disjoint. To specify these, just create your frequency vector and assign part of it to region 1 (passband) and another (disjoint) part to region two (stopband). For example:

% Using normalized frequency

F = [0 .25 .3 .4 .401 .5 .501 .6 .601 .7 .701 .8 .801 .9 .901 1];
F1 = F(1:3);
F2 = F(4:end);

If this doesn't make sense, then write back and include the code for your filter design.

Hope that helps,
Wayne
From: david t53 on 7 Jul 2010 08:11
> F = [0 .25 .3 .4 .401 .5 .501 .6 .601 .7 .701 .8 .801 .9 .901 1];
> F1 = F(1:3);
> F2 = F(4:end);

Ok, I see, the "don't care region" in your example is between F1(length(F1)) and F2(1) (.3 and .4).

Thanks,
David
 | 
Pages: 1
Prev: partical directed coherence
Next: Basic concatenation question