Considering the small number of time nuts  this issue is taking on leap second dimensions. I wonder how many commercial systems have had down time due to this? 
I'll have to dig out and test all my sleeping receivers to see which have turned into boat anchors.

Le 14 août 2013 à 09:44, David Malone a écrit : > On Sun, Aug 11, 2013 at 03:12:00PM +0200, Magnus Danielson wrote: >> I'm sure that the NTP drivers can be hacked to make necessary >> adjustments without too much code. > > I seem to have been caught by the same time warp (or a similar one) > on a GPS unit that I've been using with our NTP server since 1999. > I doubt I will be able to update the firmware, so I've made the > change shown below to the NTP NMEA refclock. It assumes that your > GPS unit might be slow by a multiple of 1024 weeks, and trys to get > the timestamp within 512 weeks of the current system time before > feeding it to NTP. Considering the small number of time nuts this issue is taking on leap second dimensions. I wonder how many commercial systems have had down time due to this? I'll have to dig out and test all my sleeping receivers to see which have turned into boat anchors.

David.

David, Well Done, your patch has breathed life into some refclocks I thought had reached the end of the road. --marki -----Original Message----- From: time-nuts-bounces@febo.com [mailto:time-nuts-bounces@febo.com] On Behalf Of David Malone Sent: Wednesday, 14 August 2013 5:44 PM To: Magnus Danielson Cc: time-nuts@febo.com Subject: Re: [time-nuts] Warped back to 1993 On Sun, Aug 11, 2013 at 03:12:00PM +0200, Magnus Danielson wrote: > I'm sure that the NTP drivers can be hacked to make necessary > adjustments without too much code. I seem to have been caught by the same time warp (or a similar one) on a GPS unit that I've been using with our NTP server since 1999. I doubt I will be able to update the firmware, so I've made the change shown below to the NTP NMEA refclock. It assumes that your GPS unit might be slow by a multiple of 1024 weeks, and trys to get the timestamp within 512 weeks of the current system time before feeding it to NTP. The patch seems to work for me, though it may not be pedantically correct. Hal might have some comments on if it could easily be improved. It might be an interesting option to have in the NMEA driver, but it does seem a litle hacky. David. --- refclock_nmea.c.orig 2010-11-10 03:38:22.000000000 +0000 +++ refclock_nmea.c 2013-08-13 20:05:44.000000000 +0100 @@ -979,6 +1076,8 @@ date.yearday = 0; /* make sure it's not used */ DTOLFP(pp->nsec * 1.0e-9, &reftime); reftime.l_ui += caltontp(&date); + while (reftime.l_i + 512*7*86400 < rd_timestamp.l_i) + reftime.l_i += 1024*7*86400; /* $GPZDG postprocessing first... */ if (NMEA_GPZDG == sentence) { _______________________________________________ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there.

Considering the small number of time nuts  this issue is taking on leap second dimensions. I wonder how many commercial systems have had down time due to this? 
I'll have to dig out and test all my sleeping receivers to see which have turned into boat anchors.

Hi Commercial system: 1) GPS is giving wrong time, NTP is still running (as a slave) 2) GPS is broke 3) Manufacturer says they won't fix / patch the obsolete GPS 4) Buy new GPS on company credit card Whole process likely takes less than an hour. Bob On Aug 14, 2013, at 3:58 AM, mc235960 <mc235960@gmail.com> wrote: > > Le 14 août 2013 à 09:44, David Malone a écrit : > >> On Sun, Aug 11, 2013 at 03:12:00PM +0200, Magnus Danielson wrote: >>> I'm sure that the NTP drivers can be hacked to make necessary >>> adjustments without too much code. >> >> I seem to have been caught by the same time warp (or a similar one) >> on a GPS unit that I've been using with our NTP server since 1999. >> I doubt I will be able to update the firmware, so I've made the >> change shown below to the NTP NMEA refclock. It assumes that your >> GPS unit might be slow by a multiple of 1024 weeks, and trys to get >> the timestamp within 512 weeks of the current system time before >> feeding it to NTP. > > Considering the small number of time nuts this issue is taking on leap second dimensions. I wonder how many commercial systems have had down time due to this? > I'll have to dig out and test all my sleeping receivers to see which have turned into boat anchors. > > > _______________________________________________ > time-nuts mailing list -- time-nuts@febo.com > To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts > and follow the instructions there.

On 08/14/2013 12:25 PM, Bob Camp wrote: > Hi > > Commercial system: > > 1) GPS is giving wrong time, NTP is still running (as a slave) > 2) GPS is broke > 3) Manufacturer says they won't fix / patch the obsolete GPS > 4) Buy new GPS on company credit card > > Whole process likely takes less than an hour. So true. When the GPS is broke, re-fill with credit card. I've seen this happen to a national network. :P Took more than an hour thought. Cheers, Magnus

David.

Mark and David, On 08/14/2013 12:24 PM, Mark C. Stephens wrote: > David, Well Done, your patch has breathed life into some refclocks I thought had reached the end of the road. It should be noted that they way GPS works, most if not all GPS receivers will process the data properly including leap seconds information, so it will only be this GPS-time to normal time conversion which will fail. It's not like it will go back to 1993 leap-second wise also, it simply does not have that information. That's why the mod looks sound to me. Cheers, Magnus > > --marki > > -----Original Message----- > From: time-nuts-bounces@febo.com [mailto:time-nuts-bounces@febo.com] On Behalf Of David Malone > Sent: Wednesday, 14 August 2013 5:44 PM > To: Magnus Danielson > Cc: time-nuts@febo.com > Subject: Re: [time-nuts] Warped back to 1993 > > On Sun, Aug 11, 2013 at 03:12:00PM +0200, Magnus Danielson wrote: >> I'm sure that the NTP drivers can be hacked to make necessary >> adjustments without too much code. > I seem to have been caught by the same time warp (or a similar one) on a GPS unit that I've been using with our NTP server since 1999. > I doubt I will be able to update the firmware, so I've made the change shown below to the NTP NMEA refclock. It assumes that your GPS unit might be slow by a multiple of 1024 weeks, and trys to get the timestamp within 512 weeks of the current system time before feeding it to NTP. > > The patch seems to work for me, though it may not be pedantically correct. Hal might have some comments on if it could easily be improved. It might be an interesting option to have in the NMEA driver, but it does seem a litle hacky. > > David. > > > --- refclock_nmea.c.orig 2010-11-10 03:38:22.000000000 +0000 > +++ refclock_nmea.c 2013-08-13 20:05:44.000000000 +0100 > @@ -979,6 +1076,8 @@ > date.yearday = 0; /* make sure it's not used */ > DTOLFP(pp->nsec * 1.0e-9, &reftime); > reftime.l_ui += caltontp(&date); > + while (reftime.l_i + 512*7*86400 < rd_timestamp.l_i) > + reftime.l_i += 1024*7*86400; > > /* $GPZDG postprocessing first... */ > if (NMEA_GPZDG == sentence) { > _______________________________________________ > time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts > and follow the instructions there. > >

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 It looks like it affects truetime XL-DC's I just noticed it at my house as well :( On 8/11/13 07:09:40, Magnus Danielson wrote: > If Furuno can supply the FW to update with, we would be more than > happy. > > A slight alteration of the code would suffice... > > if (week < 729) week += 2048; else week += 1024; > > Which would keep them floating for another 1024 weeks. If they then > let us know where the two constats is located so we could modify it > again ourselves we be very happy. > > Cheers, Magnus > > On 08/11/2013 01:29 PM, Azelio Boriani wrote: >> OK, good, found the bug. Now, iwe wish it were possible to >> download the firmware, make the correction and then upload... >> >> On Sun, Aug 11, 2013 at 1:14 PM, Magnus Danielson >> <magnus@rubidium.dyndns.org> wrote: >>> On 08/11/2013 05:18 AM, Mark C. Stephens wrote: >>>> Okay this is what worked for me: >>>> >>>> 1. Removed power and antenna. 2. apply power with no >>>> antenna. 3. send :GPS:INIT:DATE 2007,08,11 4. plug antenna >>>> back in. >>>> >>>> For some reason if I used the correct date, the Z3815A warped >>>> back to 1993. >>>> >>>> But I am curious why did this happen today? >>> GPS weeks begin on sundays. Today is first day of week 1753: >>> >>> http://csrc.ucsd.edu/scripts/convertDate.cgi?time=2013+08+11 >>> >>> Going back 1024 days gives you week 729 day 0, which occurs on >>> 1993 12 26: >>> http://csrc.ucsd.edu/scripts/convertDate.cgi?time=1993+12+26 >>> >>> A simple way to compensate for the lack of bits is to assume >>> wrapping occurs, so week numbers lower than som value actually >>> lacks 1024 weeks. A trivial code like this fixes this: >>> >>> if (week < 729) week += 1024; >>> >>> Brilliant until you reach week 1753. >>> >>> I've seen this happen at 500 and 512. >>> >>> Cheers, Magnus >>> >>>> >>>> --marki >>>> >>>> -----Original Message----- From: time-nuts-bounces@febo.com >>>> [mailto:time-nuts-bounces@febo.com] On Behalf Of Mark C. >>>> Stephens Sent: Sunday, 11 August 2013 1:06 PM To: Discussion >>>> of precise time and frequency measurement Subject: Re: >>>> [time-nuts] Warped back to 1993 >>>> >>>> Hal, I can't get it to take, I keep getting E-350 and the >>>> time does not change. Did you unplug the antenna or anything >>>> while you changed date? >>>> >>>> >>>> >>>> >>>> >>>> -----Original Message----- From: time-nuts-bounces@febo.com >>>> [mailto:time-nuts-bounces@febo.com] On Behalf Of Hal Murray >>>> Sent: Sunday, 11 August 2013 12:54 PM To: Discussion of >>>> precise time and frequency measurement Subject: Re: >>>> [time-nuts] Warped back to 1993 >>>> >>>> >>>> marks@non-stop.com.au said: >>>>> A Z3805A, a Z3815A and 58534A integrated timing antenna all >>>>> think it's 26 Dec 1993. What happened?! >>>>> _______________________________________________ >>>> Is that off by 1024 weeks? (Looks close, but I haven't >>>> checked the details.) >>>> >>>> There is a week field in the GPS data stream. It's only 10 >>>> bits. >>>> >>>> I had that problem on a Z3801A. It did the right thing after >>>> I told it the date. :GPS:INIT:DATE 2011,12,26 >>>> >>>> >>>> >>> _______________________________________________ time-nuts >>> mailing list -- time-nuts@febo.com To unsubscribe, go to >>> https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and >>> follow the instructions there. >> _______________________________________________ time-nuts mailing >> list -- time-nuts@febo.com To unsubscribe, go to >> https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and >> follow the instructions there. > > _______________________________________________ time-nuts mailing > list -- time-nuts@febo.com To unsubscribe, go to > https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow > the instructions there. > -----BEGIN PGP SIGNATURE----- Version: GnuPG/MacGPG2 v2.0.17 (Darwin) Comment: GPGTools - http://gpgtools.org Comment: Using GnuPG with Thunderbird - http://www.enigmail.net/ iQEcBAEBAgAGBQJSFRYKAAoJEHJ/lMQe1SM00eQIAKdWyxF+S02ggU6QV6Y8koTk uZgSsJdbJDi2jyq/m1mcN7ywqj2QuOQY87r+Hqcqcz6Ym0D3MinINOKggszUYXDA HGH3thUZt086LtD+znUgQbDIrctzRdnQldVcx8ueEb1TIq+9pT1Y71J42LcxLmjl zaP7QM/ol9tEYZro2ARQHqztE/HwEcpXm7ESLCaFu3av0thGLaON89gdcToRdKOp KmMMtVNr2A6onlYX+OLqDDZcfcLulM4wvhZxuqwQVMDaUAlb8jyhKoKCgHdUsZd1 BNPZqOw2gkJPjyz13XaS7wE51OjKsgJmYrLPE+KzoSetBuqApkMdrCOQgR+jKE4= =pEef -----END PGP SIGNATURE-----

This seems to impact some, but not all, XL-DC units. At least one of mine was impacted, some of the others were not. I think it depends on firmware version. If you check and set the f68 "Set Year" option this will probably let you get things back to displaying the correct date. At least it worked for the one XL-DC I have which I noted having year roll-over issues earlier... -- Russell Tammy A Wisdom wrote: > -----BEGIN PGP SIGNED MESSAGE----- > Hash: SHA1 > > It looks like it affects truetime XL-DC's I just noticed it at my > house as well :( >