Design Of Peaking Filter

Hi

I am designing a graphic equalizer for ISO frequency starting from 31.5 Hz 63Hz ... 16000Hz
I decided to use a 48000Hz as a sampling frequency. In Short for graphics equalizer if my slider position is on 0dB I should pass the signal as is.

Problem is when I design a filter for [ 31.1, 63, 125, 250, 500] @ 48KHz poles of the filters are very close to unit circle and can affect each other. Really not good numbers!
I used peak filter design from mat lab filter design toolbox.

So I want to design a good transfer functions so that I can get good pole zero plot at these lower frequency. And will be happy is somebody pour good valid thoughts on this problem.

"In Short how to design transfer function for peaking filter hows pole zeros are good"!

Classical analog graphic equalizers are using low order filters as far as I'm aware of.
From the shown frequency characteristics, this is also true for their DSP based successors.
So simple IIR filters should serve the purpose.

But I don't understand your considerations related to pole zero locations. Generally, filters
for audio DSP have to deal with a large f0/fs ratio, requiring a high coefficient resolution
and an internal wordlength surplus.

Thanks for response.
Yes, I understand that, but I was wondering, with some good design of the transfer function modeling, I can come over the problem caused using mat lab peak filter. (As I said, numbers get are really bad and might not fit in 24 bit and if I use DF-II intermediate sate might required more that double precision).
Can you suggest a good design methodology for the same. Also is there any other work around for large f0/fs ratio.

-Sam

I am confused with the design methodology for the Graphic EQ.

Method 1 : connect 'Peaking filters' in Parallel and then pass signal through each Peaking filter. Scale output of each peak filter by 1/N (N number of bands) and them together.
Method 2: connect 'Bandpass Filters' in parallel, pass signal through each filter, then apply gain to each filter and add all the filters output together.

Method 2 because of the bandpass filters there might produce more phase distortion than peaking filters because very high stop band attenuation. So Method 1 seems to be a good option.
Please share some block diagram of the Graphic-EQ design in terms of the filter connectivity and filter type.

Is this a correct statement : Peaking filters are better than Band Pass filters in terms of 'Phase Distortion'

Sam

I never designed equalizers in the digital domain. I remember to have seen reference designs from FPGA vendors, but I don't know where.

Basically, a digital EQ would have a similar signal flow as an analog equivalent, I think.

P.S.: There seems to be a problem in the new board software with inserted images. I'll try again.