HELP PLEASE::: NTP Garmin GPS 18lvc with PPS andgpsd - NTP

This is a discussion on HELP PLEASE::: NTP Garmin GPS 18lvc with PPS andgpsd - NTP ; Hey guys, Could anyone help me with Garmin GPS 18lvc device that comes with PPS output. I've been trying to configure the device to achieve about 1ms accuracy. The specification of this device states it could achieve about 1ns accuracy. ...

+ Reply to Thread
Results 1 to 12 of 12

Thread: HELP PLEASE::: NTP Garmin GPS 18lvc with PPS andgpsd

  1. HELP PLEASE::: NTP Garmin GPS 18lvc with PPS andgpsd

    Hey guys,

    Could anyone help me with Garmin GPS 18lvc device that comes with PPS
    output. I've been trying to configure the device to achieve about 1ms
    accuracy. The specification of this device states it could achieve about 1ns
    accuracy. Im quite puzzled as to why i can't achieve such accuracy. My
    software include:

    ntp-4.2.2p3-RC
    Debian Kernel 2.6.16-2-486
    gpsd 2.33

    My content of ntp.conf is:
    -------------------------------------------------------------
    server 127.127.28.0 minpoll 4 maxpoll 4
    fudge 127.127.28.0 time1 0.320 refid GPS

    server 127.127.28.1 minpoll 4 maxpoll 4 prefer
    fudge 127.127.28.1 refid GPS1
    ------------------------------------------------------------

    My command line arguments for gpsd is:
    gpsd /dev/ttyS0 -n -D 5 -N 2

    When i run ntpq the output i get is:
    ntpq> peers
    remote refid st t when poll reach delay offset jitter
    ======================================
    *SHM(0) .GPS. 0 l 4 16 377 0.000 -6.006 6.758
    SHM(1) .GPS1. 0 l - 16 0 0.000 0.000 0.004

    ntpq> assoc

    ind assID status conf reach auth condition last_event cnt
    =======================================
    1 1576 9624 yes yes none sys.peer reachable 2
    2 1577 8015 yes yes none reject clock expt 1

    ntpq> rl
    assID=0 status=0964 leap_none, sync_telephone, 6 events,
    event_peer/strat_chg,
    version="ntpd 4.2.2p3-RC@1.1581 Tue Jul 25 02:07:02 UTC 2006 (1)",
    processor="i686", system="Linux/2.6.16-2-486", leap=00, stratum=1,
    precision=-18, rootdelay=0.000, rootdispersion=15.597, peer=1576,
    refid=GPS, reftime=c877d40c.174f6c36 Mon, Jul 31 2006 10:57:48.091,
    poll=4, clock=c877d418.085b5148 Mon, Jul 31 2006 10:58:00.032, state=4,
    offset=-6.472, frequency=1.799, jitter=8.704, noise=4.119,
    stability=2.464

    So im unable to achieve accuracy anywhere around 1ns, Some times i get
    offset about 8ms. There seesm to be something weird about:
    remote refid st t when poll reach delay offset jitter
    ==========================================
    SHM(1) .GPS1. 0 l - 16 0 0.000 0.000 0.004

    and when i envoke the ntpq > assoc with the information about SHM(1).

    If anyone can offer me some assistants and help it would be greatly
    appreciated. Thanks in advance.

    Cheers,
    Sleek
    _______________________________________________
    questions mailing list
    questions@lists.ntp.isc.org
    https://lists.ntp.isc.org/mailman/listinfo/questions


  2. Re: HELP PLEASE::: NTP Garmin GPS 18lvc with PPS andgpsd

    sleek daelite wrote:
    > Hey guys,
    >
    > Could anyone help me with Garmin GPS 18lvc device that comes with PPS
    > output. I've been trying to configure the device to achieve about 1ms
    > accuracy. The specification of this device states it could achieve
    > about 1ns accuracy. Im quite puzzled as to why i can't achieve such
    > accuracy. My software include:
    >
    > ntp-4.2.2p3-RC
    > Debian Kernel 2.6.16-2-486
    > gpsd 2.33


    I believe that those who get the best accuracies use FreeBSD.

    I would have thought that 1ms was quite easy to achieve, providing your
    GPS has a good view of the whole sky hemishpere. Even my own system gets
    10 - 20us and it can hardly see half the sky!

    http://www.david-taylor.myby.co.uk/mrtg/pixie_ntp.html

    http://www.david-taylor.myby.co.uk/n...SD-GPS-PPS.htm

    and that's just using an old Pentium 133 with 48MB memory....

    David



  3. Re: HELP PLEASE::: NTP Garmin GPS 18lvc with PPS and gpsd

    On 2006-07-31, sleek daelite wrote:

    > Could anyone help me with Garmin GPS 18lvc device that comes with PPS
    > output. I've been trying to configure the device to achieve about 1ms
    > accuracy. The specification of this device states it could achieve
    > about 1ns accuracy.


    According to http://www.garmin.com/products/gps18oem/spec.html the PPS
    output is accurate to 1 microsecond. That's 1us.

    > ntp-4.2.2p3-RC
    > Debian Kernel 2.6.16-2-486


    I use Debian for most of my systems. But found that my Soekris 4801
    clock was jumping a considerable amount ( > +- 20 ms) with a Linux
    kernel. For that hardware I had to use FreeBSD.

    There is a PPSkit patch available for some Linux kernels. It can provide
    better PPS performance.

    > gpsd 2.33
    >
    > My content of ntp.conf is:
    > -------------------------------------------------------------
    > server 127.127.28.0 minpoll 4 maxpoll 4
    > fudge 127.127.28.0 time1 0.320 refid GPS


    The GPS-18lvc is supported by the Type 20 Generic NMEA GPS Receiver
    (NMEA). This driver includes PPS support. I use this for my lvc:

    | # GPS+PPS using Atomized Generic NMEA driver
    | server 127.127.20.0 mode 2 minpoll 4 maxpoll 4 prefer
    | fudge 127.127.20.0 flag3 1 # Enable Kernel PPS

    > My command line arguments for gpsd is:
    > gpsd /dev/ttyS0 -n -D 5 -N 2


    Why do you need gpsd?

    --
    Steve Kostecke
    NTP Public Services Project - http://ntp.isc.org/

  4. Re: HELP PLEASE::: NTP Garmin GPS 18lvc with PPS and gpsd

    Hi,

    for better performance I recommmend FreeBSD with your receiver. I have
    two Garmin 18LVC receivers connected to three FreeBSD boxes. You may
    found mrtg results at:

    https://ecoca.eed.usv.ro/mrtg/time_s..._internet.html

    ntp1 - PII/266MHz with 192MB RAM
    ntp2 - AMD K6/III 400MHz with 256MB RAM
    ntp3 - P1/200MHz with 80MB RAM

    All are FreeBSD 5.4 with the kernel compiled with PPS_Enable option.

    Note: All three offsets are temperature dependent (the best is the PII
    which is a Fujitsu-Siemens PC).For ntp2 I made a temperature
    compensated mini-enclosure for the Quartz oscillator on the motherboard
    (better results than before).

    The offsets displayed on this page are measured in microseconds
    (actually the offset between the GPS and the kernel time plus 50
    microseconds - for better displaying both negative and positive
    offsets).


  5. Re: HELP PLEASE::: NTP Garmin GPS 18lvc with PPS and gpsd

    Eugen COCA wrote:
    []
    > https://ecoca.eed.usv.ro/mrtg/time_s..._internet.html


    For your interest:

    - the security certificate for this page appears to be issued by a
    non-trusted company

    - the page contains a mixture of secure and insecure items

    Interesting data, though!

    David



  6. Re: HELP PLEASE::: NTP Garmin GPS 18lvc with PPS and gpsd

    David J Taylor wrote:
    >
    > For your interest:
    >
    > - the security certificate for this page appears to be issued by a
    > non-trusted company


    Yes. You may visit the link only if you want. Push the "No" button when
    asked for certificate validation and it's gone. Anyway, I'm looking for
    a sponsor to buy a real SSL certificate for my site

    > - the page contains a mixture of secure and insecure items
    >


    Yes. The graphs from the pool.ntp.org system are hard-linked to their
    original location.


    > Interesting data, though!
    >


    Thank you.

    As a "time keeping fan" I spent a lot of time reading, implementing,
    setting and talking with other people on this subject. I think it is
    really a very interesting subject and will be as long as we have to
    have watches on our hands (I have a very good watch on my hand too -
    drifting about 280-300 miliseconds per day).

    In the near future there will be a rubidium time source installed on my
    site, so I hope the graphs will look much better than now.


  7. Re: HELP PLEASE::: NTP Garmin GPS 18lvc with PPS and gpsd

    Eugen,

    Have you seen:

    http://ntp.isc.org/bin/view/Support/...ControllingNTP

    in particular the 2nd section: RRDTOOL notes?

    It might be of some help - or not.

    H

  8. Re: HELP PLEASE::: NTP Garmin GPS 18lvc with PPS and gpsd

    I sent that link so you could see how I got rrdtool to properly graph
    negative numbers.

    H

  9. Re: HELP PLEASE::: NTP Garmin GPS 18lvc with PPS and gpsd

    On 2006-08-03, Eugen COCA wrote:

    > for better performance I recommmend FreeBSD with your receiver. I have
    > two Garmin 18LVC receivers connected to three FreeBSD boxes. You may
    > found mrtg results at:
    >
    > https://ecoca.eed.usv.ro/mrtg/time_s..._internet.html


    https://ecoca.eed.usv.ro/mrtg/ntp1usvro_offset.html
    https://ecoca.eed.usv.ro/mrtg/ntp2usvro_offset.html
    https://ecoca.eed.usv.ro/mrtg/ntp3usvro_offset.html

    The scaling (0-100us) makes it difficult to see what is really
    happening. If you use RRDTool instead of MRTG you can plot negative
    values. So the graphs can be auto-scaled to show reasonable detail.

    Here's an Cricket plot of the RRDTool data for :my soekris+gps18lvc:

    http://ntp.isc.org/pub/Users/SteveKo...h-20060803.png

    Here's the loopstats for roughly the same period:

    http://ntp.isc.org/pub/Users/SteveKo...p-20060803.png

    --
    Steve Kostecke
    NTP Public Services Project - http://ntp.isc.org/

  10. Re: HELP PLEASE::: NTP Garmin GPS 18lvc with PPS and gpsd

    May I install RRDTool on Win2k ? Are there any usefull links ?


  11. Re: HELP PLEASE::: NTP Garmin GPS 18lvc with PPS and gpsd


  12. Re: HELP PLEASE::: NTP Garmin GPS 18lvc with PPS and gpsd

    I modified mrtg in order to better display the graphs. For those
    interested in, pictures looks better than ever

    I've created a "special" graph for the main server:

    https://ecoca.eed.usv.ro/mrtg/ntp1usvro_offset_10.html

    by adding only 10 microseconds to the offset. As the clock sync works
    great (thanks to NTP developers), there is not very much extra
    information to see, comparing to the + 50 microseconds graph.


+ Reply to Thread