f



Accutime 2000 external clock

Hello,

    we're using an Accutime 2000 synchronization kit as external clock for
NTP 4.1.1. The "palisade" driver (127.127.29.x) gives us a time precision
about 5-10 ms. But we need 50 us at most.

    The kit is connected to serial port in our Linux box. This RS-232
connection carries TSIP protocol (Trimble Simple Interface Protocol). We
have tested it before with NMEA and got the same result. We do not use PPS
signal, as I understand isn't needed with TSIP.

    I would be pleased to get some feedback of people using this
synchronization kit. Do you achieve this time precision? do you use PPS
signal?

Abel Navarro


0
Abel
7/11/2003 8:20:22 AM
comp.protocols.time.ntp 4895 articles. 2 followers. Post Follow

3 Replies
532 Views

Similar Articles

[PageSpeed] 42

This is output from ntptime (i hope it helps):

ntp_gettime() returns code 0 (OK)
  time c2b94890.e9cc9000  Fri, Jul 11 2003 16:30:40.913, (.913278),
  maximum error 16205 us, estimated error 6724 us
ntp_adjtime() returns code 0 (OK)
  modes 0x0 (),
  offset -474.000 us, frequency 83.705 ppm, interval 4 s,
  maximum error 16205 us, estimated error 6724 us,
  status 0x1 (PLL),
  time constant 1, precision 1.000 us, tolerance 496 ppm,
  pps frequency 85.887 ppm, stability 0.000 ppm, jitter 0.000 us,
  intervals 0, jitter exceeded 0, stability exceeded 0, errors 0.

Abel


0
Abel
7/11/2003 2:31:41 PM
Hello again,

    i've found the problem. I miss to compile the Linux kernel with support
to PPS in serial port. I've compiled the PPS option in general tab, but that
wasn't enougth. Now I have 10 usec time precision.

Abel

"Abel Navarro" <anavarro@ac.upc.es> escribi� en el mensaje
news:bels27$6kf$1@defalla.upc.es...
> Hello,
>
>     we're using an Accutime 2000 synchronization kit as external clock for
> NTP 4.1.1. The "palisade" driver (127.127.29.x) gives us a time precision
> about 5-10 ms. But we need 50 us at most.
>
>     The kit is connected to serial port in our Linux box. This RS-232
> connection carries TSIP protocol (Trimble Simple Interface Protocol). We
> have tested it before with NMEA and got the same result. We do not use PPS
> signal, as I understand isn't needed with TSIP.
>
>     I would be pleased to get some feedback of people using this
> synchronization kit. Do you achieve this time precision? do you use PPS
> signal?
>
> Abel Navarro
>
>


0
Abel
7/12/2003 7:33:44 AM
"Abel Navarro" <anavarro@ac.upc.es> wrote in message
news:beodmp$reo$1@defalla.upc.es...

>  i've found the problem. I miss to compile the Linux kernel with support
> to PPS in serial port. I've compiled the PPS option in general tab, but
that
> wasn't enougth. Now I have 10 usec time precision.

    By the way, if you'd prefer to be able to run with a stock kernel, I
have a user-space PPS driver for Linux. It's not as good as the kernel
driver, providing an accuracy of 25 usec or so. It requires NTP be compiled
with the SHM reference clock driver, so that may mean a custom compile of
NTP.

http://shell.webchat.org/~jk/shm_linux_clock.c

    DS


0
David
7/15/2003 3:48:03 PM
Reply: