[ntp:questions] NTP.ORG MEINBERG KEEP TIME ACCURATE to 10MS

unruh unruh at invalid.ca
Thu Nov 15 07:22:02 UTC 2012


On 2012-11-15, David Taylor <david-taylor at blueyonder.co.uk.invalid> wrote:
> On 14/11/2012 21:40, gbusenberg at yahoo.com wrote:
> []
>> Below ntpq -p from my 10.1.126.202 server which gets time from 10.1.126.204.  I see there is a lot of Jitter.  Is the offset of 259ms.
>>
>>       remote           refid      st t when poll reach   delay   offset  jitter
>> ==============================================================================
>> *10.1.126.204  20.200.0.1       3 u 1006 1024   77    0.043  -259.69 165.850
>>
>> driftfile "C:\Program Files (x86)\NTP\etc\ntp.drift"
>> server 10.1.126.204 iburst minpoll 4
> []
>
> In which case, there must be something grossly wrong.  The jitter I see 
> with LAN servers to LAN clients is usually well under a millisecond, and 
> on Windows Vista 0.977 milliseconds.  Even WAN servers only show 2 - 10 
> milliseconds (typically the low end of that range).  Offsets are in the 
> order of 0.5 milliseconds for LAN servers, and less than 10 milliseconds 
> for WAN servers.
>
> So for a LAN connection to show 259 milliseconds offset and 165 
> milliseconds of jitter suggests a problem.  Perhaps 10.1.126.204 is 
> really not very stable at all?  Perhaps, as was suggested, you are tying 
> too closely to the off-site server?  Perhaps the LAN is saturated?  I do 
> suggest that you consider a GPS/PPS for one of your local servers (or 
> even the PC if one is on 24 x 7 and has a serial port).  You could even 
> use a box as simple as the Raspberry Pi computer coupled with a GPS with 
> the total solution coming to around US $100 and providing within 50 
> microseconds accuracy.

Actually more like 2 microseconds accuracy should be possible, unless
the Raspberry Pi's servicing of the GPIO interrupt is really very bad. 

>
>    http://www.satsignal.eu/ntp/Raspberry-Pi-NTP.html

Nice page. 

>
> Stick it somewhere on your LAN where there is a view through a window!



More information about the questions mailing list