Phase Measurement

[rpursley8]

Message Edited by rpursley8 on 12-28-2005 12:02 PM

[rpursley8]

You connected a dbl numeric indicator to the dynamic data type, and it appears that this gives you the last element of the dbl array as an output, while the other indicator is a dbl array and this gives you all of the data out.

[ResearchGuy]

Randall,

After integrating the phase vi into my data acquisition application program and testing it, I notice that the phase changes when the sampling rate is changed. But (correct me if I am wrong) from my knowledge, I think phase is "independent" of sampling rate and it shouldn't be affected at all. Do you know why this is happening? Would appreciate any pointers or suggestions to correct this problem. Thanks.

[rpursley8]

As I mentioned earlier, I think you need to sample at a high enough rate to get an accurate representation of the phase measurement.  Your acquisition is essentially a phase modulated signal and its bandwidth can be higher than you think depending on how quickly the phase changes.

[WantToGolf]

Nyquist theory suggests that you must sample at least 2x the frequency of interest.

f(Nyquist) = fs / 2

[rpursley8]

That's the problem.  With phase modulation, the bandwidth is determined by the how quickly the phase changes and the full swing of the phase.  Since I don't know either parameter, I cannot determine fs +/- fm.  Also, if these parameter are too large, distortion will occur because phase modulated signals have wide bandwidths (many sidebands).

[ResearchGuy]

Randall,

As I understand, phase is independent of I/P frequency right? But for a set I/P frequency one does observe fluctuations in phase which depends on the sampling rate. So I need to experiement with the sampling rate and pick a sampling rate where in I can observe the least fluctuation in phase. Was this what you mean't in reply to my previous email.

[rpursley8]

Yes.  I would guess that any fluctuations in phase where it should be constant are due to sampling error (probably undersampling).

[ResearchGuy]

Randall,

Adding to your previous reply, I don't think in reality one can totally get rid of these minor fluctuations, right?

[rpursley8]

I don't really know anything about what you are measuring.  Is it truly a phase measurement or is it a time delay measurement?  A true phase measurement would mean that the measured phase would be independent of input frequency.  A time delay measurement would not.

My responses were referring to true errors in the acquired output due to sampling frequency.  There will always be noise in any measurement so you will have to live with that.  What does the errors you are seeing look like?  Are they repeatable from trial to trial or do they change?