Ripple when plotting Transfer Function

I'm making measurements of amplifiers using a NI USB-4431 DAQ card. I'm routing a chirp signal from an audio device at 48kHz sample rate, through a Hi-Fidelity analog amplifier, adjusted each time to output 1-5Watts through an 8ohm resistor. I record the chirp output of the amplifier using the DAQ.

I am trying to plot the transfer function between the chirp at 1W and 2W, as a test to validate my code. I expect to see a flat line with some gain. If I can validate that I am plotting this correctly, I can proceed to plot the transfer functions for some other amplifiers that I have measured in a similar way.

This is my code:

 load Rec_data
 load Chirp_data

 fs = 48000;
 new_fs = 4800;
 nfft = 2048;

 %Downsampling with a 13th order Chebychev low-pass filter 
 %since I am only interested in maximum frequency of 1200 Hz. 
 %I have tried 2400, but I believe I still have aliasing.

 Al_data = decimate(Rec_data,fs/new_fs,13);
 Al_Chirp = decimate(Chirp_data,fs/new_fs,13);

 % Windowing and FFT.
 window1 = hanning(length(Al_data));
 window2 = hanning(length(Al_Chirp));

 Al_data = window1' .* Al_data;
 Al_Chirp = window2' .* Al_Chirp;

 otpt = fft(Al_data,nfft);
 ipt = fft(Al_Chirp,nfft);

 %Transfer function h given by
 h = otpt./ipt;

 dF = new_fs/nfft;
 f = 1:dF:new_fs; 

 %plotting with a 3rd order median filter (as done in other literature that I am referencing.
 plot(f,medfilt1(20*log(abs(h)),3));

The transfer function has a weird ripple in it, that gets exaggerated when I try different settings for nfft, or new_fs. Lower values of nfft give me a very jagged graph too.

Answer 1

[Aditya TB]

Ok, Turns out this was a result of not zero padding the signal for the FFT, because the measured data was a 10 second signal sampled at 48kHz, and I was using far too small an FFT size.