[time-nuts] Divide by five

Bob Camp kb8tq at n1k.org
Mon Nov 10 19:00:19 EST 2014


Hi

> On Nov 10, 2014, at 6:33 PM, Magnus Danielson <magnus at rubidium.se> wrote:
> 
> Bob,
> 
> On 11/10/2014 01:17 PM, Bob Camp wrote:
>> 
>>> On Nov 10, 2014, at 2:49 AM, Magnus Danielson <magnus at rubidium.dyndns.org> wrote:
>>> 
>>> 
>>> 
>>> On 11/09/2014 07:11 PM, Bob Camp wrote:
>>>> Hi
>>>> 
>>>> That may (or may not) give you the best ADEV on the output. My guess is that the filtering algorithm will need to be a bit more complex. NTP’s aim is mainly to throw out bad clocks and pick one as best. We would more likely want to combine the outputs and use all of the good clocks we have. The idea is to improve on the ADEV of the *best* source you have available.
>>> 
>>> The aim is to remove false-tickers and then build the best ensemble of the remaining sources and weigh them according to stability.
>>> 
>>> It seems this goal is not very well met in practice, but the theory foundation is pretty good.
>> 
>> The intent of NTP is great. The implementation is targeted at the real NTP world. A set of good clocks that all are equally good simply is not what happens in the real NTP world. They don’t address it because it does not happen often enough to matter.
> 
> Who said equally good? Rather, they are sufficiently near each other so that some weighted average can be formed.

My original example that started this whole sub sub thread … I was trying very hard to keep things simple.

Bob

> 
> However, some of the issues is in getting "close enough".
> 
> Cheers,
> Magnus



More information about the time-nuts mailing list