[time-nuts] Disciplining dual oscillators using a 3-corner hat

[Bruce Griffiths]

Brooke Clarke wrote:
> Hi Richard:
>
> I neat idea.
>
> I would think there are three key measurements:
> TI1 = GPS-A
> TI2 = GPS-B
> TI3 = A-B
> Of these A-B is the most precise and the other two with GPS involved are much 
> lower stability.  In your proposed method only the poor time intervals are 
> being measured.
>
> So I would think the approach would be make all three measurements at the same 
> time then in a perfect world you could solve three equations in three unknowns. 
>   But I suspect that the errors will not exactly balance.  So you could split 
> the closing error in two and assign half to A and half to B.
>
> Have Fun,
>
> Brooke Clarke
>   
Brooke

The 3 equations aren't independent so you cannot solve them for the 3 
unknowns.
However, in principle you can solve them for the statistical 
instabilities of the 3 sources provided that the 3 sources are 
statistically independent.

Bruce