How to change frequency range doubling of Complex-valued Chirp Signal Spectrogram to match the Real-Valued Chirp Spectrogram

When real-valued linear chirp is plotted in spectrogram, the frequency range is limited to the Nyquist rate frequency range.

However, when complex-valued linar chirp with the same parameters is plotted, the frequency range of the spectrogram doubles. Is there an option to limit the frequeny range to the Nyquist rate?

%%%% Sample Code %%%%

Fs=2E3; %sampling rate

f0=0; %initial frequency offset = 0

f1=0.5E3; %LFM sweeping frequency

t1=0.5;

t=0:1/Fs:t1;

%

%Real Valued Chirp

%

yr=chirp(t,f0,t1,f1,'real');

figure

spectrogram(yr,[],256,512,Fs,'yaxis');

title('UP Chirp-Real in Time-Freq ')

%

%Complex Valued Chirp

%

yc=chirp(t,f0,t1,f1,'complex');

figure

spectrogram(yc,[],256,512,Fs,'yaxis');

title('UP Chirp-Complex in Time-Freq ')

Answer 1

[Paul]

No, it appears there's no way to change that if you want to have spectrogram make the plot.

Can use ylim to alter the plot after the fact.

Fs=2E3;   %sampling rate
f0=0;     %initial frequency offset = 0  
f1=0.5E3; %LFM sweeping frequency
t1=0.5;
t=0:1/Fs:t1;
%
%Real Valued Chirp
%
yr=chirp(t,f0,t1,f1,'real');
figure
spectrogram(yr,[],256,512,Fs,'yaxis');
title('UP Chirp-Real in Time-Freq ')

%
%Complex Valued Chirp
%
yc=chirp(t,f0,t1,f1,'complex');
figure
spectrogram(yc,[],256,512,Fs,'yaxis');
title('UP Chirp-Complex in Time-Freq ')


ylim([0 1])

Or return the output arguments from spectrogram and then make the plot over whatever frequency range is desired.

But keep in mind that for complex signals the fully frequency range is typically required, i.e., there is no symmetry in the frequency domain as there would be for a real-valued signal.

[Taehwan Kim]

Thanks, Paul. The ylim command did not work, as you figured out. Would you have an example script to return the output arguement from spectrogram and push back the frequency range growth in the plot?

[Taehwan Kim]

The ylim does not work when complex chirp is either conjugated or reversed then the spectrogram is plotted. In these cases, the frequency range not only doubles but also the response is erroneously shifted to the edge of the highest sampling rate.

[Paul]

"The ylim command did not work,"

ylim worked exactly as advertised. It limited the range of the y-axis to be between 0 and 1 (kHz, which is Fs/2), just like spectrogram did itself for the real input.

According to spectrogram, the default freqrange for a real signal is 'onesided', which is what we see above, i.e., the frequency range goes from 0 -> Fs/2. But, becasuse the spectrum of a complex signal is not symmetric, the default freqrange for a complex signal is 'twosided', i.e., the frequency range covers 0 -> Fs. Or we can specify 'centered' to have the frequency range cover -Fs/2 -> Fs/2. According to that doc page, specifying 'onesided' as the freqrange for a complex input signal results in an error, which is why ylim is needed if you really want to limit the plot for the complex input to 0 -> Fs/2.

[Taehwan Kim]

Paul, Thanks! The 'centered' option solved two problems that I questioned to my satisfaction. First, it limited the complex chirp frequency range doubling that ylim could be also used. Second, it also limited the frequency shifting of conjugated or reversed complex chirp when 'centered' option was not entered.