Interpolated phase data will not unwrap

I would like to unwrap some phase data after it is interpolated, but it's not working.

The results of this plot should look similar, but they don't.

freqHz = [20.51 21.97 23.44 24.9 26.37 27.83 29.3 30.76 32.23 33.69 35.16 36.62 38.09 39.55 41.02 42.48 43.95 45.41 46.88 48.34 49.8 51.27 52.73 54.2 55.66 57.13 58.59 60.06 61.52 62.99];
phaseDeg = [142.97 120.69 90.32 67.07 50.38 28.25 5.72 -15.26 -29.47 -49.75 -68.85 -93.73 -119.85 -136.02 -145.41 -161.94 -179.65 174 155.32 132.26 112.73 94.44 78.02 64.64 58.87 62.91 66.56 51.86 41.99 36.01];
freqHzTarget = [20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50];
phaseDegInterpolated = interp1(phaseDeg, freqHz, freqHzTarget);
phaseDegUnwrapped = unwrap(deg2rad(phaseDeg));
phaseDegInterpolatedUnwrapped = unwrap(deg2rad(phaseDegInterpolated));
semilogx(freqHz, phaseDegUnwrapped, freqHzTarget, phaseDegInterpolatedUnwrapped)

Answer 1

[Star Strider]

The arguments to the interp1 call are incorrect. The first two should be reversed, since the interpolation is over frequency.

Try this —

freqHz = [20.51 21.97 23.44 24.9 26.37 27.83 29.3 30.76 32.23 33.69 35.16 36.62 38.09 39.55 41.02 42.48 43.95 45.41 46.88 48.34 49.8 51.27 52.73 54.2 55.66 57.13 58.59 60.06 61.52 62.99];
phaseDeg = [142.97 120.69 90.32 67.07 50.38 28.25 5.72 -15.26 -29.47 -49.75 -68.85 -93.73 -119.85 -136.02 -145.41 -161.94 -179.65 174 155.32 132.26 112.73 94.44 78.02 64.64 58.87 62.91 66.56 51.86 41.99 36.01];
freqHzTarget = [20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50];
% phaseDegInterpolated = interp1(phaseDeg, freqHz, freqHzTarget);     % Original
phaseDegInterpolated = interp1(freqHz, phaseDeg, freqHzTarget);     % Corrected
phaseDegUnwrapped = unwrap(deg2rad(phaseDeg));
phaseDegInterpolatedUnwrapped = unwrap(deg2rad(phaseDegInterpolated));

figure
semilogx(freqHz, phaseDegUnwrapped, freqHzTarget, phaseDegInterpolatedUnwrapped)

This appears to be the desired result!

.

[Nathan Lively]

Thanks so much Star. You are totally right.

I hate to see that error at 45Hz. Maybe it is actually better to unwrap before interpolation.

[Nathan Lively]

Oh, I remember why I switched the variables in interp1. Because it throws this error: Sample points must be unique.

I don't want to run the unique function on it because I need all of the values. It's totally normal for some neighbors to be the same value. ¯\_(ツ)_/¯

[Star Strider]

As always, my pleasure!

Thank you!

‘Maybe it is actually better to unwrap before interpolation.’

I agree. That would remove the ‘spike’ discontinuity, since nothing like it is present in the original vector.

..