Forums

How to find phase shift of signal through a system

Started by karanbanthia November 7, 2012
Hi,
  Signal from the sensor passes through the following blocks in my system -

Ac Coupling circuit, Anti-Aliasing filter and then goes to DSP controller
where it is low-pass filtered twice.
I want to calculate the total phase shift introduced by the system at
frequency 50Hz. Following are the required data : 

                              Phase Delay                Group Delay
                 
Anti-aliasing filter    -10.603 degrees (at 50Hz)     595.264 usec (at
50Hz)

low-pass filter-1            -0.93 degrees/Hz           2.5838 msec
(in DSP controller)

low-pass filter-2            -7.4415 degrees/Hz         20.6706 msec 
(in DSP controller)

Which factors need to be considered and how can the total phase shift be
calculated? Please help!

Karan M. Banthia.
Its been a week now and no one has replied till now... Please help
experts!!

>Hi, > Signal from the sensor passes through the following blocks in my system
-
> >Ac Coupling circuit, Anti-Aliasing filter and then goes to DSP controller >where it is low-pass filtered twice. >I want to calculate the total phase shift introduced by the system at >frequency 50Hz. Following are the required data : > > Phase Delay Group Delay > >Anti-aliasing filter -10.603 degrees (at 50Hz) 595.264 usec (at >50Hz) > >low-pass filter-1 -0.93 degrees/Hz 2.5838 msec >(in DSP controller) > >low-pass filter-2 -7.4415 degrees/Hz 20.6706 msec >(in DSP controller) > >Which factors need to be considered and how can the total phase shift be >calculated? Please help! > >Karan M. Banthia. >
On Sat, 10 Nov 2012 09:19:25 -0600, karanbanthia wrote:
(top posting fixed)
>>Hi, >> Signal from the sensor passes through the following blocks in my >> system > - >> >>Ac Coupling circuit, Anti-Aliasing filter and then goes to DSP >>controller where it is low-pass filtered twice. >>I want to calculate the total phase shift introduced by the system at >>frequency 50Hz. Following are the required data : >> >> Phase Delay Group Delay >> >>Anti-aliasing filter -10.603 degrees (at 50Hz) 595.264 usec (at >>50Hz) >> >>low-pass filter-1 -0.93 degrees/Hz 2.5838 msec (in >>DSP controller) >> >>low-pass filter-2 -7.4415 degrees/Hz 20.6706 msec (in >>DSP controller) >> >>Which factors need to be considered and how can the total phase shift be >>calculated? Please help! >> >>Karan M. Banthia. >> > Its been a week now and no one has replied till now... Please help > experts!! >
No one has answered because the question is malformed, and it looks an awful lot like homework. We always tend to approach malformed questions hesitantly, and at the end of long poles. And we _don't_ answer homework questions directly, because we _don't_ want more incompetent engineering managers in the world. The reason it looks like homework is because it is laid out somewhat artificially, it contains assumptions about closing control loops that someone with DSP knowledge and little control systems chops would likely make (i.e., a prof), and because if you know how to get the intermediate answers you have, then you know how to get the final answer. In short, it's a contrived exercise to get you to go from sufficient data to an answer, but it's not a question that you'd come up with in the workplace and then have trouble answering. The reason that it looks malformed is because it gives the phase shift in the correct dimensions for the anti-aliasing filter, but in different dimensions for the low-pass filters. The specifications as given would make sense if the two low-pass filters were symmetrical FIR filters, but using FIR filters inside a control loop is generally quite unwise. (For that matter, entirely unlike audio applications, one should generally avoid using an anti-alias filter in a control loop. There are times that they're necessary, and using an anti-alias filter is not, in general, as dumb as trying to close your loop with FIR filters, but it's still not generally recommended: http://www.wescottdesign.com/articles/Sampling/sampling.pdf) So -- what's the class, and how far have you gotten toward answering the question? If you know the frequency of interest, and you know the phase shift per Hertz, then how might you find the phase shift at the frequency of interest? If you look at the given group delay for the FIR filters, and you look at the given phase shift slope, are the numbers consistent? What equation do you use to figure this out? -- Tim Wescott Control system and signal processing consulting www.wescottdesign.com