Re: Linear Phase via 'Phase Cloning' New Method

westocl wrote:

Should be a Cheby with .2dB ripple. (coef from MATLAB) 100 points with
some delay. =A0I havent checked the data just generated it. =A0Hope its
ap=
prox
linear phase.

[snipped FIR filter coefficients]

Ok westocl, that's not bad. You have designed a lowpass FIR filter
with chebyshev type magnitude behaviour and reasonably flat group
delay in the passband. So with your method, can you design FIR filtes
with arbitrary given magnitude response, such that the group delay in
the passband is flat (within some given tolerance)?

If that is the case, and you can actually save coefficients compared
to standard linear-phase FIR design programs for equivalent magnitude
response, your method would be interesting (allowing to trade FIR
order for nonlinear-phase response in the stopband). There are some
FIR design algorithms that allow you to do that (Differential
Evolution comes to mind), and you would have to shoot out your
algorithm against those as well.

Good luck,
Andor

Andor, the quick answer is i 'think so'. I made the last one wiggle like a
cheby. My whole aim in this was never to design filters, i was looking for
a way to rescue a singnal from noise by exploiting the difference in phase
from a 'clean signal' and one with a phase fluctuation due to noise and my
approach to that solution involved having a steep cutoff and linear
phase... The FIRs you find in books had too much delay, and i needed to
exploit institaneous phase differences... then i stumbled on this. I have
no clue what differential evolution is but i will do some research.
meawhile check this one out. Beats a traditional FIR huh? It has a cutoff
of approx .36; Tell me what you think.

About the group delay... i think so again. i dont know how much i should
dig in, but i 'guess' its possible to introduce a 'negative' group delay,
...one that would cancel a channel fading..

0.000412103661964
0.00832873345355
-0.0118548715753
0.000696931761247
0.021302439686684
-0.031792586435677
0.004841066431734
0.059792050371327
-0.116313121676602
0.068716050083537
0.650590932304718
0.443903173393508
-0.098720553331659
-0.035192222200803
0.062521910133592
-0.030855462348474
-0.007263136504605
0.021124080639311
-0.011831940688703
-0.002618296651644
0.008363895060847
-0.00466465206965
-0.001251838708
0.003519818747037
-0.001848826596328
-0.000650951555946
0.001521826410226
-0.000731038505205
-0.000341229051687
0.000666714374177
-0.000286674979199
-0.000176963995876
0.00029416186866
-0.000110697569329
-0.000090823937431

.

Relevant Pages

  • Re: Testing FIR filter design
    ... Without learning resources i am not blindly designing a FIR ... resources about the FIR filter design and verification... ... To verify if an apple is "sweet" we can use tongue sensor ... What "property" of of a filter is important in your design? ...
    (comp.dsp)
  • Re: Testing FIR filter design
    ... Without learning resources i am not blindly designing a FIR ... resources about the FIR filter design and verification... ...
    (comp.dsp)
  • Re: crackling sound after FIR filtration
    ... The FIR filter for crossover has to be ridiculously ... It will also introduce the significant amount of delay, ... However the cost of the DSP which is able to compute several hundred FIR taps in the real time is a big fraction of the cost of the amp. ...
    (comp.dsp)
  • Re: phase compensation using allpass
    ... delay to an existing filter such that the total delay is constant with ... This means the allpass delay is maximum where the filter ... the one and only time i ever needed to use APF to help out an IIR ... the 2-dim array of FIR coefs might be ...
    (comp.dsp)
  • Re: Linear Phase via Phase Cloning New Method
    ... [snipped FIR filter coefficients] ... You have designed a lowpass FIR filter ... delay in the passband. ... FIR design algorithms that allow you to do that (Differential ...
    (comp.dsp)
Loading