reshaping data using fft and iff

I am carrying out fft on data sample as shown below and padding with zeros, and plot my frequency spectrum which is fine.

y=fft(x,NFFT)/L

when I carry out the inverse as shown below the reconstructed signal is fine except for the fact that the new signal length has increased by the number of zeros I have padded by, where the additional signal is a row of zeros added to the signal.

s=L*ifft(y,NFFT)

How would I use the ifft to reconstruct the signal so that there is no additional zeros added to the signal? As I do not want to simply chop the zeros off.

Many Thanks Anthony

[Adam]

Why don't you want to simply chop the zeros off? That is exactly what I do.

[Anthony]

in this simple example I can but I have other plans that does not allow such a simple operation.

[Adam]

You need to include the most complex example you want to solve in your question then as I have never found a case yet using the fft where I cannot simply trim off the extra padding from the result.

[Anthony]

I also wish to rescale the data length without recalling it with the zeros attached. So I cannot then remove the zeros after.

[Adam]

I still don't fully understand what the issue is. Trim the ifft result down to the size of your original signal as the first thing you do after ifft and then do any rescaling or whatever else you want to on a signal-length result.

[Anthony]

I wish to rescale using the ifft, rather than in the time domain.

[Adam]

Do you have an example you can include?