[time-nuts] Phase Noise/Jitter/SNR/N to Slope ratio etc
jmiles at pop.net
Tue Apr 29 02:33:39 EDT 2008
> So to summarise : To make a synthesiser`s phase noise low :
> - Apply the KIS principle [Keep It Simple]
> - Use high speed [non-saturating?] logic rather than low
CMOS is actually among the cleanest logic families for digital PLLs these
days. ECL has always been among the worst. It was used for years because
loop designers had no other choice if they needed high speed.
> - The VCO resonant circuit should be of the highest
> "Q" possible - Why?
Higher Q means narrower bandwidth. Losses in the tank circuit lower the Q
and raise the oscillator's noise floor.
Another way of putting it: in any low-noise circuit you don't want more gain
than necessary. That includes oscillators. Lower Q, higher loss... more
gain available to amplify any noise present.
Wikipedia probably has an entry for the "Leeson equation," which is worth
> - Use a YIG in preference to Varicaps, and use back-to-
> back matched varicaps if possible.
There is nothing magical about YIG oscillators except for their linearity
and the possibility of multi-octave coverage. They are difficult and
expensive parts to design with; you don't use them unless you have no other
> - Use great care with PSU, and earth returns between the
> digital [phase detector] circuits, and analogue
> circuits [ VCO].
> -The VCO supply Voltage needs to be very clean, and
> - Minimise the tuning range of the VCO where possible
> to minimise the effect of any jitter that *does*
> reach the VCO through the loop filter.
The core reason for this has to do with the fact that varactor diodes have
relatively-bad Q, compared to fixed capacitors and low-loss inductors. The
less influence the varactor has over the overall tank capacitance, the lower
the tuning range... and the higher the overall Q.
It's true that oscillators with narrow tuning ranges are less susceptible to
external noise from opamps, logic gates, resistors, and the like, but that's
really a second-order effect if the rest of the oscillator is of quality
-- john, KE5FX
More information about the time-nuts