[time-nuts] synchronizing a large number of weakly coupled oscillators

[jimlux]

Eugen Leitl wrote:
> On Wed, May 12, 2010 at 06:03:11AM -0700, jimlux wrote:

>> Allan had a paper where he had 8 inexpensive crystals coupled with 
>> mixers and a microcontroller.  I think it's called the Clever 
>> Temperature Compensated XO or something like that.
> 
> Presumably, a large population of cheap coupled oscillators could be
> rather accurate collectively.
> 

I think that there's a point of diminishing returns.. the uncertainties 
in the coupling start to dominate over the reduction in uncertainty from 
combining multiple clocks.  one of those N vs sqrt(N) sort of things.