[time-nuts] Low noise frequency multiplication

[David I. Emery]

On Thu, Mar 01, 2007 at 08:55:58PM -0500, SAIDJACK at aol.com wrote:
>  
> this is not necessarily as complicated as you mention. If you use a simple  
> Exor gate as the phase comparator with the 100MHz (or 1GHz) divided down to  
> 10MHz, then there is no dependency between the loop bandwidth to the  frequency 
> of the VCXO. It will simply lock, no matter how slow the loop  bandwidth (of 
> course the bandwidth has to be faster than thermal, and aging  effects on the 
> VCXO since this is what you would like to compensate for with the  PLL).

	It will only lock if the beat note frequency at the 10 MHz phase
comparator output  with the worst case initial 100 MHz or 1 GHz VCXO
cold start error is more or less inside the loop bandwidth.   The beat
frequency obviously goes up as the initial VXCO error gets larger...

	If the loop bandwidth is 25 Hz this implies maybe 500 to 1 KHz
max cold error to ensure reliable lock at 100 MHz depending on loop gain
and LPF order. 1 KHz is only around 10 ppm and for typical VXCO parts
might require some sort of tweak in order to ensure the VCXO is within
lock range OR a more complex circuit with wider loop bandwidth before
lock or some form of acquisition aid such as sweep.

	And if you are talking - as the original poster was - about
under a 1 Hz loop bandwidth things get worse.

> You do have to make sure that the VCXO adjustment range is sufficient to  
> allow it to always lock to the reference, taking long-term crystal aging etc  
> into account.

	Most certainly.

> The cleanup filter is a simple second-order 25Hz cut-off RC, and it only  has 
> to remove the 10MHz phase comparison output from the Exor gate. No  
> relationship between loop bandwidth and output frequency.

	No, but there is a relationship between close in noise sidebands
around the VXCO and loop bandwidth.   Inside the loop bandwidth their
amplitude will be determined by the multiplied reference noise mostly,
outside the loop bandwidth mostly by the native phase noise of the VCXO.
In some cases the VCXO may have better close in phase noise than the
reference (multiplied by the ratio between it and the VCXO) and you may
want a  narrow loop bandwidth - which my point was might make ensuring
reliable start up phase lock a complication.   If one can live with a
wider loop than that is not a problem.

	Sorry to be unclear...

-- 
   Dave Emery N1PRE,  die at dieconsulting.com  DIE Consulting, Weston, Mass 02493
"An empty zombie mind with a forlorn barely readable weatherbeaten
'For Rent' sign still vainly flapping outside on the weed encrusted pole - in 
celebration of what could have been, but wasn't and is not to be now either."