Is this a way to evaluate phase group delay ?
Posted:
Tue Mar 14, 2023 2:02 pm
by Tepeix
Just randomly, trying to compare some phase i get this.
is this a way to check if a system have linear group delay ?
It seams to works but i'm not 100% sure.. Also the graphic must be not in a log view to see a linear function, but it make more hard to clearly see witch frequency will be the more phase distorted..
Re: Is this a way to evaluate phase group delay ?
Posted:
Sun Apr 09, 2023 6:20 pm
by Tepeix
Hum, some while after i discover that the module of the precedent post is a little stupid..
If i understand well, the bode plot i use was in log mode.
To see if the phase is linear we could switch to linear mode which help to see if there's a curvature;)
But now here's a virtual linear delay representation.
(i hope it works and present reality and not something else?)
I plan to use it to try to implement a 2th order allpass interpolation.
(One thing i learn about this is that the 2th order could correct the phase of some frequency, but at the price of making worse other's. (not 100% sure of this but almost) generally the 2th order will make the <1,x/2 nyquist better but >1,x/2 nyquist worse.. ...EDIT Thinking more about this, it's maybe wrong, the phase will be less linear at the end but each frequency will have less additional delay ?)
So providing a fractional delay it give the virtually perfect (i hope) phase response and make possible to compare with a true delay phase response.