Leaky PSD when using sampling rate that is not a power of 2

[jdpigeon]

Just noticed this when debugging problems that arose when moving from an EEG device sampling at 256hz to one at 1000hz.

It seems that the PSD returned by

const fft = new FFT(bins, samplingRate);
        fft.forward(samples);
        return Array.from(fft.spectrum);

Is wrong when the sampling rate of the signal is not a power of 2 (e.g. 256, 1024). With a sampling rate of 1000hz I'm getting PSDs that are far too "leaky" and spread out to represent my test.

I actually don't notice any change in the PSD output at all when I change the samplingRate parameter, which makes me think that the implementation might just be inferring sampling rate from the number of bins? It would explain why it only works when my signal's sampling rate is equal to the number of bins in the FFT