spatical accuracy of continuous wavelet transform

I realise this is an anateur question but your help is much appreciated.

I wish to have spacio-temporal accuracy so am using a continuous wavelet ttransform. I have noticed that when using fft on a sin wave signal I get a nice spike around the frequency, but when doing a cwt (with cgau6 wavelet - a sin wave combined with a gaussian) I get a rather wide gausian around the frequency. This is not very usefull as it hides smaller frequency responces in the area when I apply the cwt to a real signal as the dominant frequency appears as a gaussian over a wide spacial area.

Why is this? Is there anyway to get a nice fine spike as in the fourier transorm?

thanks fin advance or your help
David