Linearizing phase response

Started by August 23, 2007
Dear Group,

I have a 16th order Chebyshev type 2 low pass IIR filter and I'd like
add a couple of second order allpass stages to linearize the phase
response. Ideally, the resulting filter should have an equiripple
reponse around an ideal linear response.

I've googled quite extensively, but I can't find any information on
how to design the allpass sections.
Does anybody know of a techique?

Thanks,
Andrew

On Aug 24, 11:54 am, Andrew.J.Hunte...@gmail.com wrote:
> Dear Group, > > I have a 16th order Chebyshev type 2 low pass IIR filter and I'd like > add a couple of second order allpass stages to linearize the phase > response. Ideally, the resulting filter should have an equiripple > reponse around an ideal linear response. > > I've googled quite extensively, but I can't find any information on > how to design the allpass sections. > Does anybody know of a techique? > > Thanks, > Andrew
If you buy Filter Solutions the software does it for you and gives you teh circuit diagram as well!
On Aug 23, 7:54 pm, Andrew.J.Hunte...@gmail.com wrote:
> Dear Group, > > I have a 16th order Chebyshev type 2 low pass IIR filter and I'd like > add a couple of second order allpass stages to linearize the phase > response. Ideally, the resulting filter should have an equiripple > reponse around an ideal linear response. > > I've googled quite extensively, but I can't find any information on > how to design the allpass sections. > Does anybody know of a techique? > > Thanks, > Andrew
If you have the whole sequence in a file or memory, you can use something like Matlab's 'filtfilt' routine. If you don't you can approximate with a FIR and implement using fast convolution techniques. John
On 24 Aug., 01:54, Andrew.J.Hunte...@gmail.com wrote:
> Dear Group, > > I have a 16th order Chebyshev type 2 low pass IIR filter and I'd like > add a couple of second order allpass stages to linearize the phase > response. Ideally, the resulting filter should have an equiripple > reponse around an ideal linear response. > > I've googled quite extensively, but I can't find any information on > how to design the allpass sections.
Google -> iir filter phase equalization -> I'm feeling lucky!
In article <1187913274.631616.290380@m37g2000prh.googlegroups.com>, 
Andrew.J.Hunter74@gmail.com says...
> > >Dear Group, > >I have a 16th order Chebyshev type 2 low pass IIR filter and I'd like >add a couple of second order allpass stages to linearize the phase >response. Ideally, the resulting filter should have an equiripple >reponse around an ideal linear response. > >I've googled quite extensively, but I can't find any information on >how to design the allpass sections. >Does anybody know of a techique?
s/FILSYN (http://www.alkeng.com/sfilsyn.html) will do this. However, you may find that you need significantly more than two 2nd order allpass sections to get good group delay EQ, particularly if you want to equalize all the way to the edge of the passband. You can get a good feel for the magnitude of the problem by plotting the group delay vs. frequency of the original Cheby. If the group delay spikes over a narrow range around the cutoff frequency and its maximum value is much greater than its value at 0 Hz, then the required group delay equalizer is going to be high-order because it has to add a lot of group delay to all frequencies below the "spike" and it is going to have to be selective enough to incorporate the "spike" smoothly in the final equalized group delay curve.