Active filters allow more options in the filter shape. One example is a notch which can be placed at the reference frequency to greatly reduce the spurious sidebands at that frequency on the output.
I once saw an application note by Motorola for a synthesizer for an aviation band radio local oscillator which had three notch filters to remove the first three reference spurs. The output spectrum was very low in phase noise and spurious spectral lines.
Active filters allow more options in the filter shape. One example is a notch which can be placed at the reference frequency to greatly reduce the spurious sidebands at that frequency on the output.
I once saw an application note by Motorola for a synthesizer for an aviation band radio local oscillator which had three notch filters to remove the first three reference spurs. The output spectrum was very low in phase noise and spurious spectral lines.
my 2 cents:
for what i know PLL is one of those designs that prefer passive filters. That's because just after the filter you have the VCO. On the imput of the VCO you would like to have the most possible constant value (no oscillations at all neither double pole frequency response), otherwise you will see spuriouses on the spectrum. For some designs I've seen some solutions made by active filters followed by an RC backend to avoid these phenomenons. They seem to work pretty fine.