[time-nuts] GPS orthodontics: sawteeth & hanging bridges

[Dr Bruce Griffiths]

Poul-Henning Kamp wrote:
> In message <458B03AA.4000209 at xtra.co.nz>, Dr Bruce Griffiths writes:
>
>   
>> If only one had access to the receiver firmware: the cure for "hanging 
>> bridges" etc is surely to add sufficient (rms noise ~ PPS positioning 
>> quantisation error) bandlimited gaussian noise to calculated PPS 
>> position before quantisation of the PPS position occurs in the hardware.
>>     
>
> The negative saw-tooth field _is_ the cure.
>
> As long as you apply the correction, you are in pretty amazing shape.
>
> The trouble is when you do not apply the correction.
>
> Curing that would take more involved hardware than the oncore has,
> so firmware access would not help you.
>
> The more involved hardware is what the CNS Clock is all about as far
> as I can tell.
>
> If you feed a PRS10, and it's firmware is new enough, you just need
> to read the serial data from the oncore, and send the correction
> to the PRS10 and everybody are happy.
>
>   
Poul-Henning

I am well aware that using the negative sawtooth cures most of the problems.
The proposal wasn't serious, it was an indication of a solution that 
didn't requires using a sawtooth correction, should that be useful to 
someone.

In other words it is conceivable that not all GPS timing receivers PPS 
outputs need suffer from hanging bridges even if no sawtooth correction 
is applied.
i.e. there may be timing receivers available that don't suffer from 
hanging bridge and other problems even without sawtooth correction.
However, none of the motorola/iLotus timing receivers have this 
favourable characteristic

The CNS clock merely uses the sawtooth correction to apply an analog 
correction to the PPS pulse position.
If one uses a processor of some kind in one GPSDOCXO then one can omit 
the analog correction and do it all digitally.

Bruce