[time-nuts] phase noise/jitter
Don Collie jnr
merdon at xtra.co.nz
Mon Apr 28 11:35:42 EDT 2008
Hello Bruce and all,
OK, now I think I understand : the more phase noise on the VCO, the more jitter on the output of the divide by "N".
This output spectrum comprises *only* the PRF+ close-in noise sidebands [plus harmonics, and their noise sidebands]. Since the lowest frequency [ignoring close-in noise sidebands for a moment] present in the divide by "N"
logic is equal to its output frequency anyway, there are no frequencies present in the output that are below this, and because the loop rolls off at a frequency that is much less than this, the VCO is not modulated by any signal comming from the reference, or the divide by "N" logic ..The loop filter filters these signals out.This is how a PLL can "clean up" jittery signals. So I guess one way to improve PLL phase noise is to have as high a frequency going to the phase detector as possible, and to have as low a loop bandwidth as possible, [and put up with a long lock - in time].Having now filtered out most of the jitter, the problem becomes one of designing a VCO [or variable reluctance oscillator, as in a YIG], that produces little phase noise of itself. This must be where the designers get crafty!
Facinating stuff !,....................................................Don C.
More information about the time-nuts