FFT analysis — mark length and size


I have a Signal, where I want measuring two first harmonics by 59,33 Hz < f1 < 83,49 Hz and f1*2 on the peak and . At the same signal 2 harmonics on the hill 34,25 Hz < f2 < 59,32 Hz.

what is the influence of (window)size?
How long should I mark my signal, what is difference?

I choose a 16384 window size and 2 seconds mark.
Thank you for help!

A single point of a waveform has no “frequency”, it only has “amplitude” at that point in time. “Frequency” only has meaning in relation to time (the number of cycles per second).

FFT analysis looks at blocks of samples, and analyzes the frequencies in each block.
The longer the block, the greater the frequency resolution.
You can think of this as looking at the waveform through a “frame” or “Window” of a given size, through which you see a block of samples. Then sliding the frame along the waveform and looking at the next snapshot, and so on.

The downside of using a very large “window size” (looking at big blocks of samples) is that each snapshot covers a long period of time.
The longer the block, the worse the time resolution.

Selecting the best window size is a question of finding an appropriate compromise between best frequency resolution, and best time resolution.