FFT & IFFT with data with large number of samples

I got data from an oscilloscope. Sampling frequency is 10 Mega Samples per second. The signal generated has 500002 samples. The fundamental is at 60 HZ, noise with a mean value of zero is present. I would like to use FFT to see the frequency spectrum, and then use IFFT to filter out the noise. The number of samples is so high that all methods I am using are unsatisfactory. I need a sequence of commands or a script to solve this.

[Walter Roberson]

Please clarify what you mean by "unsatisfactory" ? Are you talking about the amount of time it takes?

Answer 1

[Vieniava]

FFT/IFFT filtering is not appropriate in your case, because of poor spectral resolution. The best resolution you could achieve is ca. 10e6/500e3=20 Hz but notice that your fundamental is only 60Hz. If your fundamental is not exactly 60 HZ, e.g. 63 Hz, the output, filtered signal would have 60Hz (to be more precisly would have (k-1)*Fs/NFFT Hz, where k is number of bin with peak, Fs-sampling frequency, NFFT - fft/ifft length), because of averaging nature of FFT/IFFT.

[Walter Roberson]

Good point, that's only 1/20th of a second worth of data, only 3 cycles of 60 Hz. If that is the fundamental, then the overtones are going to be nearly undetectable.