[ntp:questions] Compensating DCF77 radio signal propagation delay

Xisco Lladó timekeeper at bitic.net
Sun Nov 16 17:23:44 UTC 2003


Hello.

I am using a low cost DCF77 receiver at home to reach stratum 1. It is 
connected to one of my Linux boxes via serial port.

server 127.127.8.0 mode 14

tom:~# ntpq -c rv
status=02f4 leap_none, sync_lf_clock, 15 events, event_peer/strat_chg,
version="ntpd 4.1.2a at 1:4.1.2a-1 Thu Oct  2 10:41:07 CEST 2003 (1)",
processor="i686", system="Linux2.4.20", leap=00, stratum=1,
precision=-17, rootdelay=0.000, rootdispersion=483.678, peer=44084,
refid=DCFa, reftime=c3622fd6.4043d46b  Sun, Nov 16 2003 18:18:14.251,
poll=6, clock=c3622feb.f44b09e9  Sun, Nov 16 2003 18:18:35.954, state=4,
offset=-3.751, frequency=-51.589, jitter=8.704, stability=0.616

It really is not the most precise configuration, but it is enough for me.
You can reach this machine on pdn.nucli.net (also included in 
es.pool.ntp.org).

To minimice delays (in a multi service box), I have set the serial port 
to be low_latency

tom:~# setserial /dev/refclock-0
/dev/refclock-0, UART: 16550A, Port: 0x02f8, IRQ: 3, Flags: low_latency

I also am thinking in patching the kernel with low latency and 
preemptive patches (but I don't know if it is good for NTP).

At this moment, the server sincronices to the radio clock, but it is far 
away of the other servers about 11-12ms (quite stable). The minpoll for 
all the external servers is fixed to 10 in configuration file.

tom:~# ntpq -p
      remote           refid      st t when poll reach   delay   offset 
  jitter
==============================================================================
*GENERIC(0)      .DCFa.           0 l   45   64  377    0.000   -0.709 
  0.362
  192.168.1.255   0.0.0.0         16 -    -   64    0    0.000    0.000 
4000.00
  192.168.2.255   0.0.0.0         16 -    -   64    0    0.000    0.000 
4000.00
  192.168.3.255   0.0.0.0         16 u    -   64    0    0.000    0.000 
4000.00
  LOCAL(0)        LOCAL(0)        10 l    8   64  377    0.000    0.000 
   0.008
-hora.rediris.es .GPS.            1 u  851 1024  377   63.388  -11.282 
  3.338
-hora.roa.es     .GPS.            1 u  847 1024  377  144.720  -29.732 
21.807
+tictac.ci.uv.es .PPS.            1 u  844 1024  377   79.580   -7.531 
  1.120
+chronos.cru.fr  .GPS.            1 u  848 1024  377  100.563  -11.213 
  1.616
-ntp0.ja.net     .MSF.            1 u  845 1024  377   95.464  -11.459 
  6.368
-lluna           hora.rediris.es  2 u  814 1024  377    0.355  -27.986 
  2.282
#ns.keso.fi      tick.usnogps.na  2 u  844 1024  377  243.332   -0.005 
29.018
-130.206.130.95  swisstime.ee.et  2 u  825 1024  353   72.988  -12.810 
  3.940
#65.125.233.206  hora.rediris.es  2 u  806 1024  377  177.191  -17.134 
  2.956
#bergantells     time.nist.gov    2 u  813 1024  377  247.103   12.706 
  5.740
#ns.saard.net    murgon.cs.mu.OZ  2 u  839 1024  377  424.542  -12.133 
  4.016
-dns.univ-lyon1. chronos.cru.fr   2 u  855 1024  377  111.339  -14.887 
42.232
-merlin.karpo.cz ntp1-rz.rrze.un  2 u  851 1024  377  157.671   -9.897 
  0.970
#ntp1.belbone.be ntp0.belbone.be  2 u  850 1024  377  103.373  -21.197 
  0.918

I supose it is, in part, due to propagation delay of the radio signal 
from transmiter (Germany) to Mallorca (Balearic Islands - Spain) where I 
  am situated. (about 1265km).

This propagation delay varies during day and seasons but I think it 
could be compensated with a mean value of it.

Do you know any method to calculate it? I would like to fix it with 
fudge parameter.

Thank you.
Xisco.




More information about the questions mailing list