Update October 2004

I updated this computer a couple of months ago to Fedora Core 1. The update was so that I could get the latest security patches. I had the same problem as before compiling NTP. It seemed that the Jupiter driver wasn't being linked into the ntpd daemon. Again I copied over the ntpd daemon that I had originally compiled on the Caldera 3.1 system and it worked. Go figure eh. I did get a little more determined this time however, and with a little digging I found that the install scripts weren't putting the binaries in the directories I expected them to be in. The Jupiter driver was indeed being compiled and linked so I replaced the old ntpd with the proper one. See Building the Jupiter driver for the Delorme Tripmate.

Another problem was the protective plastic "sandwich container" that I used for weather proofing the GPS receiver. It turns out that it was a disposable container and after a couple of years it disintegrated. All that was holding it together was the silicone I used to seal it. 

I replaced it with a similar container, but this time I used a standard (long term) container. You can see the residual silicone from the previous container.

I also updated the servers somewhat. I still had the other computer at the remote office connected as a peer. That seemed to cause some problems and this computer would eventually loose all the other server connections. I removed that connection and added a couple more. I tried to find a few closer  computers, but there is still a lot of delay.

     remote        refid         st t when poll reach   delay   offset  jitter
==============================================================================
LOCAL(0)         LOCAL(0)        10 l   41   64  377    0.000    0.000   0.008
*GPS_JUPITER(0)  .GPS.            0 l   35   64  377    0.000    0.550   0.485
+sue.cc.uregina. clock.uregina.c  2 u   22   64  177   59.003   -1.456  13.325
-dense.utcc.utor 128.100.200.166  2 u   35   64  377  103.629    1.038   0.120
+br3.umtnet.umt. ntp-p1.obspm.fr  2 u   38   64  377  100.366   -2.395   6.425
-caesar.cs.wisc. ben.cs.wisc.edu  2 u   35   64  377  126.759   -2.558   0.266

 The remote computer:

remote refid st t when poll reach delay offset jitter
==============================================================================
LOCAL(0)         LOCAL(0)        10 l   37   64  377    0.000    0.000   0.008
-time.nrc.ca     tac.nrc.ca       2 u  738 1024  377   62.998   -8.363   0.063
+ntp-2.gw.uiuc.e truechimer.cso.  2 u  633 1024  377  115.896    8.893   1.172
+ensb.cpsc.ucalg bonehed.lcs.mit  2 u  715 1024  377   73.928    3.950   0.790
*hssx-sktn-166-2 .GPS.            1 u  522 1024  377   49.232   -2.300   5.729

The closer computers (I'm in Saskatoon) don't seem to have the least delay.

Clockstats todate: 1,076,789 lines. With that many samples I wonder if the latitude and longitude are converging on the truth.

[ Back ] [ Up ]