N Fractional PLL spectrum

Question

I'm studying the different types of N-fractional PLLs based on switching between two or more frequencies to get the desired frequency.

However, I'm facing difficulties to understand how the output spectrum can be basically a predominant tone on the desired frequency. If we think of a, let's say, 3rd order Delta-Sigma modulator to control de divider, we would switch between (N-4)Fref, (N-3)Fref... (N+3)Fref, hence the output spectrum will be basically sinc functions centered on those frequencies. Of course, because of the sinc shape, the spectrum will contain also frequencies in the desired one but at a much lower level. So, how do we get our desired frequency to be the predominant?

Put in more simple terms with a traditional first order sigma delta (i.e., a counter) without compensation: the spectrum will consist of two sinc functions centred at NxFref and (N+1)Fref (basically an FSK signal), how do we get the desired frequency?

Thanks.

Answer 1

I think you might be over-thinking the problem.

With a fraction N PLL, the divisor is modified during the process of phase comparison to obtain a value of N that is numerically an integer plus some fraction of an integer. The analogue input to the VCO still uses a low pass filter and it is that filter that keeps the VCO output stable over the period. Hence the output is stable but the divisor now can be much smaller and this can lead to much less phase jitter compered to an integer N PLL.