Filter Phase

[Rob]

Hello,

I need to ask some basic questions which I am not able to find good
answers for.
When I look at filter programs and other information regarding filter
responses I see phase and gain plots. Could someone tell me how phase
affects the filter, or how it relates?
Also what is a good phase response vs a bad phase response?
How can I tell if the filter is unstable?

Thank you,
Rob

 

[Phil Allison]

"Rob"

I need to ask some basic questions which I am not able to find good
answers for.
When I look at filter programs and other information regarding filter
responses I see phase and gain plots. Could someone tell me how phase
affects the filter, or how it relates?

** A phase plot shows how input and output sines waves will differ in phase
at any frequency.

Most fairly simple filters are "minimum phase" circuits - true when
there are no time delays or "all pass filters" included in the filter
network.

Such simple filters have the property that *phase follows gain* - which
means that if the gain plot is known, the phase plot will correspond to it
and cannot be varied without altering the shape of the gain plot.

Also what is a good phase response vs a bad phase response?

** As in life, good and bad are subjective judgements - it depends
entirely on the application.

But for all "minimum phase" circuits, you are stuck with the phase response
for any given gain curve.

How can I tell if the filter is unstable?

** Unstable filters oscillate or are continually on the verge of
oscillation.

Simulation software should be able to warn when this is likely.



....... Phil

 

[Bob Masta]

Hello,

I need to ask some basic questions which I am not able to find good
answers for.
When I look at filter programs and other information regarding filter
responses I see phase and gain plots. Could someone tell me how phase
affects the filter, or how it relates?
Also what is a good phase response vs a bad phase response?
How can I tell if the filter is unstable?

Thank you,
Rob

To expand slightly on Phil's good explanation, note that
analog filters have a frequency-dependent phase shift.
Because different frequency components are shifted
by different amounts, the shape of the filtered waveform
can change (besides the rounding or whatever that is
due to removing high frequencies, etc). Typically you see
this as overshoot and/or ringing in the step response.

That's not an issue with single pole (one RC section active) filters,
but as you get to higher orders you have the option of trading
different response features around. For a given order, you
can have a sharper cutoff if you are willing to accept some
ripple in the frequency response and ringing in the step
response.

So-called "linear phase" designs have sacrifice
frequency response sharpness to get nice waveforms.
They esentially are just a delayed version of the input,
with some high frequencies removed by the filter. They
get their name from (you guessed it!) the phase plot is
a more-or-less straight line in the region of interest.
Linear phase is one of the virues of FIR digital filters,
by the way. Just as in the analog world, this comes
at a price of many more stages for the same cutoff sharpness,
compared to simpler IIR filters.

Best regards,


Bob Masta

DAQARTA v3.50
Data AcQuisition And Real-Time Analysis
www.daqarta.com
Scope, Spectrum, Spectrogram, Signal Generator
Science with your sound card!

 

[Phil Allison]

"Bob Masta"

To expand slightly on Phil's good explanation,

** Courtesy noted.




.... Phil