[ntp:questions] Garmin 18 LVC high offset and jitter

David Lord snews at lordynet.org
Mon Sep 19 10:56:23 UTC 2011


David Lord wrote:
> Miguel Gonçalves wrote:
>> Hi all!
>>
>> I am having a problem with a Dell server that has a Garmin 18 LVC 
>> attached.
>>
>> It's a FreeBSD 7.4 machine with the Garmin 18 LVC connected to a 
>> serial port
>> (PPS on DCD pin). The kernel has been compiled with the line "OPTIONS
>> PPS_SYNC". NTP is 4.2.4p5.
>>
>> Here's the ntp.conf file:
>>
>> server 127.127.20.0 mode 2 iburst prefer
>> fudge 127.127.20.0 flag3 1
>>
>> server ntp-p1.obspm.fr iburst
>> server ptbtime1.ptb.de iburst
>> server ntp.inrim.it iburst
>> server ntp1.oma.be iburst
>>
>> server canon.inria.fr iburst
>> server time.fu-berlin.de iburst
>> server chronos.csr.net iburst
>> server ntp1.nl.uu.net iburst
>>
>> statsdir /var/log/ntp/
>> filegen clockstats file clockstats type day enable
>> filegen peerstats file peerstats type day enable
>> filegen loopstats file loopstats type day enable
>> statistics clockstats peerstats loopstats
>>
>> logfile /var/log/ntp/messages
>>
>> restrict default kod nomodify notrap nopeer
>> restrict 127.0.0.1
>>
>> I've disabled ACPI in /boot/loader.conf (hint.acpi.0.disabled=1) but 
>> did not
>> disable serial FIFO.
>>
>> I am getting very high offsets and jitter for this reference clock:
>>
>>      remote           refid      st t when poll reach   delay   offset
>>  jitter
>> ============================================================================== 
>>
>> *GPS_NMEA(0)     .GPS.            0 l   40   64  377    0.000   -0.031
>> 0.043
>> +ntp-p1.obspm.fr .TS-3.           1 u   50   64  377   54.154    1.611
>> 1.843
>> -ptbtime1.ptb.de .DCFa.           1 u  119   64  376   70.069   -2.174
>> 1.283
>> -ntp.ien.it      .CTD.            1 u   35   64  377   65.682    2.037
>> 6.667
>> -ntp1.oma.be     .PPS.            1 u   26   64  377   61.911    2.280
>> 3.673
>> +canon.inria.fr  .GPSi.           1 u   30   64  377   54.626    1.442
>> 1.700
>> -time.fu-berlin. .PPS.            1 u   32   64  377   73.342   -0.292
>> 3.136
>> -chronos.csr.net .GPS.            1 u   32   64  377   85.049   -3.656
>> 5.096
>> -ntp1.nl.uu.net  .PPS.            1 u    2   64  377   67.085    3.435
>> 3.888
>>
>> tick# ntpdc -c kerninfo
>> pll offset:           -7.886e-06 s
>> pll frequency:        -6.729 ppm
>> maximum error:        0.030005 s
>> estimated error:      6e-06 s
>> status:               2107  pll ppsfreq ppstime ppssignal nano
>> pll time constant:    6
>> precision:            1e-09 s
>> frequency tolerance:  496 ppm
>> pps frequency:        -6.729 ppm
>> pps stability:        0.312 ppm
>> pps jitter:           4.03e-07 s
>> calibration interval: 256 s
>> calibration cycles:   152
>> jitter exceeded:      1029
>> stability exceeded:   0
>> calibration errors:   3
>>
>> It seems the internal clock of the machine is going nuts (small 
>> snippet of
>> loopstats file):
>>
>> 55822 70649.300 0.000037529 -6.306 0.000000816 0.139388 6
>> 55822 70714.307 0.000004089 -6.175 0.000001464 0.138376 6
>> 55822 70779.307 -0.000029839 -6.175 0.000013699 0.129439 6
>> 55822 70845.306 -0.000058655 -6.175 0.000000562 0.121079 6
>> 55822 70909.305 -0.000078382 -6.175 0.000000570 0.113259 6
>> 55822 70972.305 -0.000061801 -6.729 0.000001320 0.222533 6
>> 55822 71035.304 -0.000041884 -6.729 0.000000778 0.208161 6
>> 55822 71100.304 -0.000023847 -6.729 0.000000629 0.194717 6
>> 55822 71166.303 -0.000003148 -6.595 0.000001515 0.188165 6
>> 55822 71232.302 0.000006531 -6.595 0.000004666 0.176012 6
>>
> 
> NetBSD-5, ntpd 4.2.6p3
> 
> My time sync has blipped twice each evening at times when
> system load is high during log creation/rotation. Mostly
> I don't notice. Currently I'm using a Sure-Electronics
> but my Garmin 18x-LVC was the same. Blips are from < 15u
> up to over 40u.
> 
> peerstats.20110918
>        ident     mean    rms    max delay  dist  disp
> 127.127.20.2  -64.698 15.422 60.828 0.000 2.525 1.742
> 127.127.22.2    0.000  0.000  0.048 0.000 0.928 0.920
                              ^
Sorry my eyesight is bad. RMS should have been 0.008
and averaged over last 20 days is 0.0056 although packages
on that system have been upgraded and system rebooted
during that period.


David





More information about the questions mailing list