Ping Unruh -- Was Keeping a VERY accurate time (within a millisecond) - Networking

This is a discussion on Ping Unruh -- Was Keeping a VERY accurate time (within a millisecond) - Networking ; On 2008-09-06, Unruh wrote: > Ignoramus10032 writes: > >>I want my ubuntu servers to keep a VERY accurate time. That is, I want >>to have a clock accurate to less than a millisecond. > > Buy a Gamin 18LVC gps ...

+ Reply to Thread
Results 1 to 3 of 3

Thread: Ping Unruh -- Was Keeping a VERY accurate time (within a millisecond)

  1. Ping Unruh -- Was Keeping a VERY accurate time (within a millisecond)

    On 2008-09-06, Unruh wrote:
    > Ignoramus10032 writes:
    >
    >>I want my ubuntu servers to keep a VERY accurate time. That is, I want
    >>to have a clock accurate to less than a millisecond.

    >
    > Buy a Gamin 18LVC gps receiver (about $70). Wire it up to your
    > system and use one of the refclocks for that. (eg refclock_nmea)
    > (Although I wrote my own interrupt routine to grab the pps) It can
    > keep your clock accurate to about 3usec (that is usec not msec) You
    > do need to mount the gps where it can see the sky.


    Unruh, now that I have some understanding of the complexity of this
    issue, and have read a lot of stuff, I can understand your reference
    to that product.

    can you comment on Garmin GPS 18 LVC vs. Garmin GPS 18 PC?

    How did you wire those bare wires to your PC?

    Do I lose much by using a serial port and Garmin GPS 18 PC?

    i

  2. Re: Ping Unruh -- Was Keeping a VERY accurate time (within a millisecond)

    Ignoramus10032 writes:

    >On 2008-09-06, Unruh wrote:
    >> Ignoramus10032 writes:
    >>
    >>>I want my ubuntu servers to keep a VERY accurate time. That is, I want
    >>>to have a clock accurate to less than a millisecond.

    >>
    >> Buy a Gamin 18LVC gps receiver (about $70). Wire it up to your
    >> system and use one of the refclocks for that. (eg refclock_nmea)
    >> (Although I wrote my own interrupt routine to grab the pps) It can
    >> keep your clock accurate to about 3usec (that is usec not msec) You
    >> do need to mount the gps where it can see the sky.


    >Unruh, now that I have some understanding of the complexity of this
    >issue, and have read a lot of stuff, I can understand your reference
    >to that product.


    >can you comment on Garmin GPS 18 LVC vs. Garmin GPS 18 PC?


    ONLY the lvc has a PPS output. The PC only has a the nmea output to the
    serial port, from which you can get at best about 10ms accuracy. The LVC
    outputs a pulse on the second with an accuracy of 1usec. You can get GPS
    receivers which will output a pulse with 100ns accuracy but they tend to be
    much more expensive.

    There are a number of places on the we which will guide you to setting up
    the 18lvc to deliver usec accurate time to your computer.



    >How did you wire those bare wires to your PC?


    Well, I did it is a ratehr non-sandard way. I used a usb port to get the 5V
    to run the device, fed the nmea to the serial port and ran the PPS line to
    the parallel port. I then adapted a parallel port interrupt service routine
    to timestamp the parallel port interrupts, and put them out onto
    /dev/gpsint. I then adapted the refclock-shm driver from ntp to read
    /dev/gpsint and report the timestamps to ntp. Yes, that is a rather
    roundabout way of doing it. I had more faith in the parallel port interrupt
    serviceing than the serial. Anyway, the refclock nmea driver will read the
    nmea from teh serial port AND use the PPS on the serial port if you have
    it, so that is probably the better way to go.

    See for example
    http://time.qnan.org/
    http://www.linuxquestions.org/linux/...Signal_And_NTP

    However the main problem seems to be that you have to modify the kernel to
    give the pps. That was another reason I went with the shm driver and the
    parallel port interrupt driver.



    >Do I lose much by using a serial port and Garmin GPS 18 PC?


    About a factor of 5000 in accuracy.

    Now that may not be important for you. 10ms may be pleanty good enough.
    If not, the lvc is the only way to go.


  3. Re: Ping Unruh -- Was Keeping a VERY accurate time (within a millisecond)

    On 2008-09-07, Unruh wrote:
    > Ignoramus10032 writes:
    >
    >>On 2008-09-06, Unruh wrote:
    >>> Ignoramus10032 writes:
    >>>
    >>>>I want my ubuntu servers to keep a VERY accurate time. That is, I want
    >>>>to have a clock accurate to less than a millisecond.
    >>>
    >>> Buy a Gamin 18LVC gps receiver (about $70). Wire it up to your
    >>> system and use one of the refclocks for that. (eg refclock_nmea)
    >>> (Although I wrote my own interrupt routine to grab the pps) It can
    >>> keep your clock accurate to about 3usec (that is usec not msec) You
    >>> do need to mount the gps where it can see the sky.

    >
    >>Unruh, now that I have some understanding of the complexity of this
    >>issue, and have read a lot of stuff, I can understand your reference
    >>to that product.

    >
    >>can you comment on Garmin GPS 18 LVC vs. Garmin GPS 18 PC?

    >
    > ONLY the lvc has a PPS output. The PC only has a the nmea output to the
    > serial port, from which you can get at best about 10ms accuracy. The LVC
    > outputs a pulse on the second with an accuracy of 1usec. You can get GPS
    > receivers which will output a pulse with 100ns accuracy but they tend to be
    > much more expensive.
    >
    > There are a number of places on the we which will guide you to setting up
    > the 18lvc to deliver usec accurate time to your computer.
    >
    >
    >
    >>How did you wire those bare wires to your PC?

    >
    > Well, I did it is a ratehr non-sandard way. I used a usb port to get the 5V
    > to run the device, fed the nmea to the serial port and ran the PPS line to
    > the parallel port. I then adapted a parallel port interrupt service routine
    > to timestamp the parallel port interrupts, and put them out onto
    > /dev/gpsint. I then adapted the refclock-shm driver from ntp to read
    > /dev/gpsint and report the timestamps to ntp. Yes, that is a rather
    > roundabout way of doing it. I had more faith in the parallel port interrupt
    > serviceing than the serial. Anyway, the refclock nmea driver will read the
    > nmea from teh serial port AND use the PPS on the serial port if you have
    > it, so that is probably the better way to go.


    Unruh, thanks, I bought a Garmin 18 LVC an hour ago. Hopefully I will
    get it in a few days. I will experiment then. I wil first try at home
    and then will discuss it at work.

    i

    >
    > See for example
    > http://time.qnan.org/
    > http://www.linuxquestions.org/linux/...Signal_And_NTP
    >
    > However the main problem seems to be that you have to modify the kernel to
    > give the pps. That was another reason I went with the shm driver and the
    > parallel port interrupt driver.
    >
    >
    >
    >>Do I lose much by using a serial port and Garmin GPS 18 PC?

    >
    > About a factor of 5000 in accuracy.
    >
    > Now that may not be important for you. 10ms may be pleanty good enough.
    > If not, the lvc is the only way to go.
    >


    --
    Due to extreme spam originating from Google Groups, and their inattention
    to spammers, I and many others block all articles originating
    from Google Groups. If you want your postings to be seen by
    more readers you will need to find a different means of
    posting on Usenet.
    http://improve-usenet.org/

+ Reply to Thread