[ntp:questions] NTP sever on an isolated Network

Unruh unruh-spam at physics.ubc.ca
Sat Jan 10 03:22:31 UTC 2009


Unruh <unruh-spam at physics.ubc.ca> writes:

>leituras at gmail.com (Diego Ramos) writes:

>>All right. I'll show those options to my manager, but I think he will prefer
>>the USB Option as we don't need to be extremely accurate.

>Some consider 1ms extremely accurate. Some consider it extremely
>inaccurate.( Mine runs about 500 times better than this).
> Both would consider 1s pretty crude. 
>But then again, some consider running a mile in 12 min the height of
>accomplishment.




>>Once more, I'm really glad for all the help I'm getting here. I'm stuck with
>>this problem for a long time and the help I got here saved me a lot of days.


>Actually I withdraw my reccomendation of the USB unit. It seems that Garmin
>uses a proprietary protocol on this interface, and I am not sure that there
>exists a driver for this. It is not NMEA. I think one needs to do a more
>thorough research than I have done. 

>Thus, either get a 18PC and supply the 12V it needs from some other source
>(It comes with a cigarette lighter type input which you can plug into a car
>lighter output. However your manager might get upset if you told him you
>have to buy a car as well and park it in your office). or think that the
>18LVC is probably the best bet still depite the need to wire it up. There
>are a number of places on teh web that will teach you how to do that. 

There is a set of programs on sourceforge:
garmin_gps and garminusb2nmea
which are claimed to be able to use the usb version. Now one post found on
google re the use of garmin_gps with the gpsd program which can link up
with ntpd. How well it works I have no idea. The garmin_gps driver is
supposed to be a part of the kernels after 2.6.8 already (try
modprobe garmin_gps 
and see if it finds it or says no such module)

I have also seen one post that claims taht the garminusb2nmea is not well
programmed and can get confused.

I have used neither and have no idea how well they work.


>For anyone with a modicum of soldering experience, hooking up an LVC is
>fast and simple. 




>>Thanks a lot.

>>Diego

>>On Fri, Jan 9, 2009 at 10:22 PM, Unruh <unruh-spam at physics.ubc.ca> wrote:

>>> leituras at gmail.com (Diego Ramos) writes:
>>>
>>> >> To sync to a second any gps receiver will do. The Garmin 18PC or USB
>>> will
>>> >> do  and will get you down to about 5ms accuracy with some tuning. The
>>> PPS
>>> >>  on the 18LVC will do 1000 times better ( again $50-100 range)
>>> >>
>>> >> On the 18PC you have to supply the, I think it is 12 V from somewhere
>>> else.
>>> >> On the USB it will supply its own voltage if I read correctly.
>>> >>
>>> >>
>>> >But I'll be able to use the Garmin 18 USB and xntpd? Which refclock I'll
>>> >use?
>>>
>>> nmea
>>>
>>> The gps nmea signal message includes the time. It takes about 200ms for
>>> that nmea sentence to be sent, but that can be roughly  zeroed out by an
>>> appropriate
>>> fudge to the nmea driver in the ntp.conf file. The usb port  is a bit
>>> flakey in its timing etc, but that will onely be a few millisec. Thus if
>>> you are happy with 5ms or so, the nmea driver and the GPS usb should be
>>> fine. (someone may have done it so could refine that 5 ms figure to be an
>>> accurate estimate of the time error you could get.) It is certainly much
>>> better than 100ms (1/10)
>>>
>>> Why use xntpd? That is ancient.  Use the latest ntpd 4.2.4p6
>>>
>>>
>>> _______________________________________________
>>> questions mailing list
>>> questions at lists.ntp.org
>>> https://lists.ntp.org/mailman/listinfo/questions
>>>




More information about the questions mailing list