ntpd and GPS

[Tom Storey]

Ok, it was nothing like what I was thinking. Turns out my GPS didnt  
have a fix on anything. It was getting signals, but no fix. :-)

Now that I have it mounted on a pole outside, hey presto:

building# ntpq -p
      remote           refid      st t when poll reach   delay    
offset  jitter
= 
= 
= 
= 
= 
= 
========================================================================
+resolv.internod 128.250.33.242   2 u   50   64  377   18.121    
-1.127  14.549
*sparky.services 131.203.16.6     2 u   56   64  377   21.275    
-9.704  15.235
  GPS_NMEA(0)     .GPS.            0 l    3   64    1    0.000   
-442.78   0.002

Cheers for your help. I'll keep tweaking it now until I get it working  
just right (jitter is incrementing and Im sure thats not a good thing).

Tom


On 20/09/2008, at 8:09 AM, Bob Johnson wrote:

> On 9/19/08, Tom Storey <tom at snnap.net> wrote:
>> Hi all,
>>
>> Ive been toying with setting up my old Garmin GPS12 as a reference  
>> for a
>> server (FreeBSD 6.2) running ntpd, but Ive run into an issue.
>>
>
> Is it possible the issue isn't what you think it is?
>
>> Ive searched around a bit and cant find an answer, perhaps because  
>> there
>> isnt one.
>
> I once (years ago) had a Garmin GPS working with ntpd, so it's
> reasonable to believe it can be done again, unless support for that
> capability was dropped (which I doubt). Unfortunately, it was long
> enough ago that I don't remember what I did. It's possible I used the
> 1 PPS output without NMEA sentences, but that's not my recollection.
>
>>
>> Is there any way I can set ntpd to expect a $GPRMC string every 2  
>> seconds,
>> which is the frequency at which the GPS12 transmits them?
>>
>
>> Alternatively, does anyone know how to make the GPS12 transmit a  
>> $GPRMC
>> string every second?
>
> I'm almost certain you can't. The complete set of all NMEA sentences
> takes more than one second at the default 4800 baud, so IIRC it
> outputs sentences only on odd seconds, and perhaps the older units are
> too slow to compute a fix once per second. Two things that may work
> around this are to turn off everything except the GPRMC sentence:
>
> $PGRMO,,2
> $PGRMO,GPRMC,1
>
> and perhaps free up some CPU time (for faster position calculation) by
> (oddly enough) reducing the output data rate to 1200 bps:
>
> $PGRMC,,,,,,,,,,1,
>
> but I don't think that will actually work. To go back to 4800 bps, use
> 3 instead of 1. I think there are 11 commas after the "C" in that
> command, but my eyes aren't so sharp any more.
>
> There is a Linux driver for the Garmin proprietary protocol. Don't
> know if it is distributed in a FreeBSD version. Try
> http://jensar.us/~bob/garmin/
>
>>
>> If there is a better place I can post this, please let me know.
>
> This is probably a good place for this question, but if you don't get
> a better answer, try the archives of the "time nuts" mailing list
> https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts which
> unfortunately appears to be down right now. General info about that
> group is at http://www.leapsecond.com/time-nuts.htm
>
> If that yields nothing, you might post your question to the Time Nuts
> list, time-nuts @ febo.com. It is probably a FAQ for them, but they
> will be polite about it. And I had hoped to once again stick an old
> Garmin on an NTP server, so I'll be curious to know if this turns out
> to be insurmountable.
>
> Good luck,
>
> -- Bob Johnson
>   fbsdlists at gmail.com