# How to find peak or dip separation in frequency domain signal

19 views (last 30 days)
Ozan Erturk on 15 May 2020
Commented: Ozan Erturk on 31 Mar 2021
Hello all,
I have a frequency domain signal I got from a network analyzer. The signal (attached) has repetetive dips that are separated by some frequency (in this case it's about 5.5MHz). I would like to see the frequency separation through the data so I started using "findpeaks" by inverting the data making dips into peak, which sort of works. If I apply the findpeaks approach to an analytic function it's perfect. But since the data measured is noisy, even though tweaking findpeaks properties like minimum separation and prominance etc. can fail from time to time.
I am wondering if there is a more elegant approach to this by simply treaating this signal like a time domain signal and take the FFT. I am not sure how to map the frequency data from the measured data and map it to freqeuncy of the fft signal.
Findpeaks approach: (you can just load S_data.mat and run the code)
%% FSR with FindPeaks Method
f=S(:,1);
sig=S(:,2);
figure(1);
plot(f,sig); hold on
sig_neg=-sig;
[Dips, indexes] = findpeaks(sig_neg,f,'MinPeakDistance',4.3e6,'MinPeakProminence',.01);
Dip_freq=indexes;
figure(1);plot(Dip_freq,-Dips,'ok');
FSR=diff(Dip_freq); %find and plot the separation of each dip within the freq range
figure;
plot(Dip_freq(1:length(Dip_freq)-1),FSR*1e-6,'-ob');
xlabel('Hz');ylabel('FSR in MHz');

Sergio Yanez-Pagans on 28 Mar 2021
These might work for you:
Hope you find this useful!
Ozan Erturk on 31 Mar 2021
I was using some derivative of this but your code is put together much more nicely ! Thanks!

### Categories

Find more on Correlation and Convolution in Help Center and File Exchange

### Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!