On Saturday, 8. September 2012, 10:26:36 Martin Burnicki wrote: > To the original poster (sorry I don't see his real name anywhere): > > In your initial email you wrote: > > remote refid st t when poll reach delay offset > jitter > ===================================================================== > xGENERIC(0) .DCFa. 0 l 20 64 377 0.000 -11.980 > 7.579 *ntp2.rrze.uni-e .PPS. 1 u 129 256 377 16.380 > 4.990 1.111 +time.fu-berlin. .PPS. 1 u 116 256 377 23.770 > 4.349 1.287 -ntp1.as34288.ne .PPS. 1 u 174 256 377 > 20.779 5.490 0.911 -ptbtime1.ptb.de .PTB. 1 u 35 64 > 377 27.754 7.891 3.220 > > The offsets are from -12 to +5 or +8, so this means the difference is about > 17 to 20 ms which need to be added to fudge time1 (or subtracted, I'm not > sure out of my head). > > Anyway, the default fudge time1 value may not be 0, so you should first set > fudge time1 0, let ntpd settle, check the difference between the offsets, > and then set the fudge time 1 to the correct value, with the appropriate > sign, e.g. +0.018 or -0.018 with the numbers above. > > Martin
Martin, thanks for this good explanation. Starting with fudge time1 0 helped me to find the right direction. Before I played around with time1 values, but they make it only worse. Now my DCF77 clock is the main source that is used and I can fine tune time1 to the optimum. Thanks again and best regards, Klaus _______________________________________________ pool mailing list [email protected] http://lists.ntp.org/listinfo/pool
