In message <022101cc4970$b24a4b80$16dee280$@com>, "Jose Camara" writes: >After >one year of NTP queries, assume you have a 100ms jitter on the network time, >you could at most tweak your oscillator, based on past performance, to 6E-9. It is a lot more complicated than that, we need to talk allan-deviation here, not scalar numbers. The main problem here is that the 'default' NTPd software is not really written for something like this, and has attributes which makes it truly sucky for the task. If you want to do this, you want to write your own software and you want to give it an entirely different modus operandi. As with all oscillator discplining, what you are looking for is the so called "allan intercept" where the two sources allan deviation cross. With a NTP reference, its location varies depending on stochastic network properties, which depends what's between the server and you. If you control the network topology (as in: Can make sure there is no other traffic), you're fine, normal PLL style stuff works. If you don't control the network topology, the RTT between you and the server becomes a BIG problem, because the fundamental NTP assumption that it is symmetric is almost always wrong. You can average over long tau's, but then your ISP upgrades their routed and a systematic change in RTT screws your integrator over for several weeks. Alternatively, if your LO is stable enough (=Rb/Cs), you can operate on first derivative of the RTT, which turns the routed upgrade into a single spiky sample, but the cost is an overall higher noise in your error signal. -- Poul-Henning Kamp | UNIX since Zilog Zeus 3.20 phk@FreeBSD.ORG | TCP/IP since RFC 956 FreeBSD committer | BSD since 4.3-tahoe Never attribute to malice what can adequately be explained by incompetence.

Poul and others? As usual, I suddenly had a thought ripple across my otherwise placid cortex, and have forgotten if there was a previous answer on this thread. What does a NTP server look like on the network? There are now several small net appliances/eval boards available. I run a program called NMEAtime on my Windoze computers that currently get info from a net NTP server. I also have some simple net appliances. So, using my GPS 10 MHZ or 1 sec signals, or Rb 10 MHZ, how might I generate a local NTP server? I hope I don't put this awkwardly; what is the protocol? What does the supplicant client see and how does it ask for service? Best to all of you, Don Poul-Henning Kamp > In message <022101cc4970$b24a4b80$16dee280$@com>, "Jose Camara" writes: > >>After >>one year of NTP queries, assume you have a 100ms jitter on the network >> time, >>you could at most tweak your oscillator, based on past performance, to >> 6E-9. > > It is a lot more complicated than that, we need to talk allan-deviation > here, not scalar numbers. > > The main problem here is that the 'default' NTPd software is not really > written for something like this, and has attributes which makes it > truly sucky for the task. > > If you want to do this, you want to write your own software and you > want to give it an entirely different modus operandi. > > As with all oscillator discplining, what you are looking for is > the so called "allan intercept" where the two sources allan deviation > cross. > > With a NTP reference, its location varies depending on stochastic > network properties, which depends what's between the server and you. > > If you control the network topology (as in: Can make sure there is > no other traffic), you're fine, normal PLL style stuff works. > > If you don't control the network topology, the RTT between you and > the server becomes a BIG problem, because the fundamental NTP > assumption that it is symmetric is almost always wrong. > > You can average over long tau's, but then your ISP upgrades their > routed and a systematic change in RTT screws your integrator over > for several weeks. > > Alternatively, if your LO is stable enough (=Rb/Cs), you can operate > on first derivative of the RTT, which turns the routed upgrade into > a single spiky sample, but the cost is an overall higher noise in > your error signal. > > -- > Poul-Henning Kamp | UNIX since Zilog Zeus 3.20 > phk@FreeBSD.ORG | TCP/IP since RFC 956 > FreeBSD committer | BSD since 4.3-tahoe > Never attribute to malice what can adequately be explained by > incompetence. > > _______________________________________________ > 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. > -- "Neither the voice of authority nor the weight of reason and argument are as significant as experiment, for thence comes quiet to the mind." R. Bacon "If you don't know what it is, don't poke it." Ghost in the Shell Dr. Don Latham AJ7LL Six Mile Systems LLP 17850 Six Mile Road POB 134 Huson, MT, 59846 VOX 406-626-4304 www.lightningforensics.com www.sixmilesystems.com

Hi Don, 1PPS is useful. 10MHz is not direcly useful. You also need some kind of timecode, telling ntpd which second the 1pps indicated. The software side is normally ntpd configured with one of its drivers (called refclock) for the GPS protocol your receiver has. Again the most used is probably refclock_NMEA. (http://www.eecis.udel.edu/~mills/ntp/html/drivers/driver20.html) look if there is anything else fitting your receivers. http://www.eecis.udel.edu/~mills/ntp/html/refclock.html Then wire a (preferably) serial cable with rx, tx and gnd in the usual places, add your 1PPS on DCD (pin 1). Depending on your serial port you could get away without levelshifting the 1pps to rs232-levels. Last but not least, for OS select your favorite unix derivative. -- Björn > Poul and others? > As usual, I suddenly had a thought ripple across my otherwise placid > cortex, and have forgotten if there was a previous answer on this > thread. What does a NTP server look like on the network? There are now > several small net appliances/eval boards available. I run a program > called NMEAtime on my Windoze computers that currently get info from a > net NTP server. I also have some simple net appliances. So, using my GPS > 10 MHZ or 1 sec signals, or Rb 10 MHZ, how might I generate a local NTP > server? > I hope I don't put this awkwardly; what is the protocol? What does the > supplicant client see and how does it ask for service? > Best to all of you, > Don > > > Poul-Henning Kamp >> In message <022101cc4970$b24a4b80$16dee280$@com>, "Jose Camara" writes: >> >>>After >>>one year of NTP queries, assume you have a 100ms jitter on the network >>> time, >>>you could at most tweak your oscillator, based on past performance, to >>> 6E-9. >> >> It is a lot more complicated than that, we need to talk allan-deviation >> here, not scalar numbers. >> >> The main problem here is that the 'default' NTPd software is not really >> written for something like this, and has attributes which makes it >> truly sucky for the task. >> >> If you want to do this, you want to write your own software and you >> want to give it an entirely different modus operandi. >> >> As with all oscillator discplining, what you are looking for is >> the so called "allan intercept" where the two sources allan deviation >> cross. >> >> With a NTP reference, its location varies depending on stochastic >> network properties, which depends what's between the server and you. >> >> If you control the network topology (as in: Can make sure there is >> no other traffic), you're fine, normal PLL style stuff works. >> >> If you don't control the network topology, the RTT between you and >> the server becomes a BIG problem, because the fundamental NTP >> assumption that it is symmetric is almost always wrong. >> >> You can average over long tau's, but then your ISP upgrades their >> routed and a systematic change in RTT screws your integrator over >> for several weeks. >> >> Alternatively, if your LO is stable enough (=Rb/Cs), you can operate >> on first derivative of the RTT, which turns the routed upgrade into >> a single spiky sample, but the cost is an overall higher noise in >> your error signal. >> >> -- >> Poul-Henning Kamp | UNIX since Zilog Zeus 3.20 >> phk@FreeBSD.ORG | TCP/IP since RFC 956 >> FreeBSD committer | BSD since 4.3-tahoe >> Never attribute to malice what can adequately be explained by >> incompetence. >> >> _______________________________________________ >> 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. >> > > > -- > "Neither the voice of authority nor the weight of reason and argument > are as significant as experiment, for thence comes quiet to the mind." > R. Bacon > "If you don't know what it is, don't poke it." > Ghost in the Shell > > > Dr. Don Latham AJ7LL > Six Mile Systems LLP > 17850 Six Mile Road > POB 134 > Huson, MT, 59846 > VOX 406-626-4304 > www.lightningforensics.com > www.sixmilesystems.com > > > _______________________________________________ > 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. >

http://www.eecis.udel.edu/~mills/ntp/html/refclock.html

Björn

Hi Bjorn: Thanks very much for the sources! I was afraid I was not clear when I sent my question(s). I want to generate a network server rather than a client. I looked up man ntpd, and it seems at first glance to be a Linux/Unix client that will sync up a local computer. I want to generate a network server that ntpd or other clients can access. Perhaps I did not read closely enough, and if not, I apologize. Don bg@lysator.liu.se > Hi Don, > > 1PPS is useful. 10MHz is not direcly useful. You also need some kind of > timecode, telling ntpd which second the 1pps indicated. > > The software side is normally ntpd configured with one of its drivers > (called refclock) for the GPS protocol your receiver has. Again the most > used is probably refclock_NMEA. > (http://www.eecis.udel.edu/~mills/ntp/html/drivers/driver20.html) look > if > there is anything else fitting your receivers. > > http://www.eecis.udel.edu/~mills/ntp/html/refclock.html > > Then wire a (preferably) serial cable with rx, tx and gnd in the usual > places, add your 1PPS on DCD (pin 1). Depending on your serial port you > could get away without levelshifting the 1pps to rs232-levels. > > Last but not least, for OS select your favorite unix derivative. > > -- > > Björn > >> Poul and others? >> As usual, I suddenly had a thought ripple across my otherwise placid >> cortex, and have forgotten if there was a previous answer on this >> thread. What does a NTP server look like on the network? There are now >> several small net appliances/eval boards available. I run a program >> called NMEAtime on my Windoze computers that currently get info from a >> net NTP server. I also have some simple net appliances. So, using my >> GPS >> 10 MHZ or 1 sec signals, or Rb 10 MHZ, how might I generate a local >> NTP >> server? >> I hope I don't put this awkwardly; what is the protocol? What does the >> supplicant client see and how does it ask for service? >> Best to all of you, >> Don >> >> >> Poul-Henning Kamp >>> In message <022101cc4970$b24a4b80$16dee280$@com>, "Jose Camara" >>> writes: >>> >>>>After >>>>one year of NTP queries, assume you have a 100ms jitter on the >>>> network >>>> time, >>>>you could at most tweak your oscillator, based on past performance, >>>> to >>>> 6E-9. >>> >>> It is a lot more complicated than that, we need to talk >>> allan-deviation >>> here, not scalar numbers. >>> >>> The main problem here is that the 'default' NTPd software is not >>> really >>> written for something like this, and has attributes which makes it >>> truly sucky for the task. >>> >>> If you want to do this, you want to write your own software and you >>> want to give it an entirely different modus operandi. >>> >>> As with all oscillator discplining, what you are looking for is >>> the so called "allan intercept" where the two sources allan deviation >>> cross. >>> >>> With a NTP reference, its location varies depending on stochastic >>> network properties, which depends what's between the server and you. >>> >>> If you control the network topology (as in: Can make sure there is >>> no other traffic), you're fine, normal PLL style stuff works. >>> >>> If you don't control the network topology, the RTT between you and >>> the server becomes a BIG problem, because the fundamental NTP >>> assumption that it is symmetric is almost always wrong. >>> >>> You can average over long tau's, but then your ISP upgrades their >>> routed and a systematic change in RTT screws your integrator over >>> for several weeks. >>> >>> Alternatively, if your LO is stable enough (=Rb/Cs), you can operate >>> on first derivative of the RTT, which turns the routed upgrade into >>> a single spiky sample, but the cost is an overall higher noise in >>> your error signal. >>> >>> -- >>> Poul-Henning Kamp | UNIX since Zilog Zeus 3.20 >>> phk@FreeBSD.ORG | TCP/IP since RFC 956 >>> FreeBSD committer | BSD since 4.3-tahoe >>> Never attribute to malice what can adequately be explained by >>> incompetence. >>> >>> _______________________________________________ >>> 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. >>> >> >> >> -- >> "Neither the voice of authority nor the weight of reason and argument >> are as significant as experiment, for thence comes quiet to the mind." >> R. Bacon >> "If you don't know what it is, don't poke it." >> Ghost in the Shell >> >> >> Dr. Don Latham AJ7LL >> Six Mile Systems LLP >> 17850 Six Mile Road >> POB 134 >> Huson, MT, 59846 >> VOX 406-626-4304 >> www.lightningforensics.com >> www.sixmilesystems.com >> >> >> _______________________________________________ >> 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. > > -- "Neither the voice of authority nor the weight of reason and argument are as significant as experiment, for thence comes quiet to the mind." R. Bacon "If you don't know what it is, don't poke it." Ghost in the Shell Dr. Don Latham AJ7LL Six Mile Systems LLP 17850 Six Mile Road POB 134 Huson, MT, 59846 VOX 406-626-4304 www.lightningforensics.com www.sixmilesystems.com

 sudo ln -s /dev/ttyACM0 /dev/gps0

 server 127.127.20.0 mode 81

/etc/init.d/ntp restart

http://www.eecis.udel.edu/~mills/ntp/html/refclock.html

Björn

Hi Don, ntpd is both server and client. For a ublox with usb i did the following a few days ago. sudo ln -s /dev/ttyACM0 /dev/gps0 add this to the /etc/ntp.conf server 127.127.20.0 mode 81 and then /etc/init.d/ntp restart This particular setup without 1pps gives a lousy time (some 150ms late with 50ms jitter). I apologize if I was not clear enough in my previous answer. -- Björn > Hi Bjorn: Thanks very much for the sources! I was afraid I was not clear > when I sent my question(s). I want to generate a network server rather > than a client. I looked up man ntpd, and it seems at first glance to be > a Linux/Unix client that will sync up a local computer. I want to > generate a network server that ntpd or other clients can access. Perhaps > I did not read closely enough, and if not, I apologize. > Don > > bg@lysator.liu.se >> Hi Don, >> >> 1PPS is useful. 10MHz is not direcly useful. You also need some kind of >> timecode, telling ntpd which second the 1pps indicated. >> >> The software side is normally ntpd configured with one of its drivers >> (called refclock) for the GPS protocol your receiver has. Again the most >> used is probably refclock_NMEA. >> (http://www.eecis.udel.edu/~mills/ntp/html/drivers/driver20.html) look >> if >> there is anything else fitting your receivers. >> >> http://www.eecis.udel.edu/~mills/ntp/html/refclock.html >> >> Then wire a (preferably) serial cable with rx, tx and gnd in the usual >> places, add your 1PPS on DCD (pin 1). Depending on your serial port you >> could get away without levelshifting the 1pps to rs232-levels. >> >> Last but not least, for OS select your favorite unix derivative. >> >> -- >> >> Björn >> >>> Poul and others? >>> As usual, I suddenly had a thought ripple across my otherwise placid >>> cortex, and have forgotten if there was a previous answer on this >>> thread. What does a NTP server look like on the network? There are now >>> several small net appliances/eval boards available. I run a program >>> called NMEAtime on my Windoze computers that currently get info from a >>> net NTP server. I also have some simple net appliances. So, using my >>> GPS >>> 10 MHZ or 1 sec signals, or Rb 10 MHZ, how might I generate a local >>> NTP >>> server? >>> I hope I don't put this awkwardly; what is the protocol? What does the >>> supplicant client see and how does it ask for service? >>> Best to all of you, >>> Don >>> >>> >>> Poul-Henning Kamp >>>> In message <022101cc4970$b24a4b80$16dee280$@com>, "Jose Camara" >>>> writes: >>>> >>>>>After >>>>>one year of NTP queries, assume you have a 100ms jitter on the >>>>> network >>>>> time, >>>>>you could at most tweak your oscillator, based on past performance, >>>>> to >>>>> 6E-9. >>>> >>>> It is a lot more complicated than that, we need to talk >>>> allan-deviation >>>> here, not scalar numbers. >>>> >>>> The main problem here is that the 'default' NTPd software is not >>>> really >>>> written for something like this, and has attributes which makes it >>>> truly sucky for the task. >>>> >>>> If you want to do this, you want to write your own software and you >>>> want to give it an entirely different modus operandi. >>>> >>>> As with all oscillator discplining, what you are looking for is >>>> the so called "allan intercept" where the two sources allan deviation >>>> cross. >>>> >>>> With a NTP reference, its location varies depending on stochastic >>>> network properties, which depends what's between the server and you. >>>> >>>> If you control the network topology (as in: Can make sure there is >>>> no other traffic), you're fine, normal PLL style stuff works. >>>> >>>> If you don't control the network topology, the RTT between you and >>>> the server becomes a BIG problem, because the fundamental NTP >>>> assumption that it is symmetric is almost always wrong. >>>> >>>> You can average over long tau's, but then your ISP upgrades their >>>> routed and a systematic change in RTT screws your integrator over >>>> for several weeks. >>>> >>>> Alternatively, if your LO is stable enough (=Rb/Cs), you can operate >>>> on first derivative of the RTT, which turns the routed upgrade into >>>> a single spiky sample, but the cost is an overall higher noise in >>>> your error signal. >>>> >>>> -- >>>> Poul-Henning Kamp | UNIX since Zilog Zeus 3.20 >>>> phk@FreeBSD.ORG | TCP/IP since RFC 956 >>>> FreeBSD committer | BSD since 4.3-tahoe >>>> Never attribute to malice what can adequately be explained by >>>> incompetence. >>>> >>>> _______________________________________________ >>>> 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. >>>> >>> >>> >>> -- >>> "Neither the voice of authority nor the weight of reason and argument >>> are as significant as experiment, for thence comes quiet to the mind." >>> R. Bacon >>> "If you don't know what it is, don't poke it." >>> Ghost in the Shell >>> >>> >>> Dr. Don Latham AJ7LL >>> Six Mile Systems LLP >>> 17850 Six Mile Road >>> POB 134 >>> Huson, MT, 59846 >>> VOX 406-626-4304 >>> www.lightningforensics.com >>> www.sixmilesystems.com >>> >>> >>> _______________________________________________ >>> 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. >> >> > > > -- > "Neither the voice of authority nor the weight of reason and argument > are as significant as experiment, for thence comes quiet to the mind." > R. Bacon > "If you don't know what it is, don't poke it." > Ghost in the Shell > > > Dr. Don Latham AJ7LL > Six Mile Systems LLP > 17850 Six Mile Road > POB 134 > Huson, MT, 59846 > VOX 406-626-4304 > www.lightningforensics.com > www.sixmilesystems.com > > > _______________________________________________ > 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. >

> Indeed thats why I was saying choose 3 servers. > Now I see in the thread that you can pick pools of servers so thats > good. > Then average what they say the time is and drive oscillator. > Wonder if you could look towards the stratum 1 servers. > But that said I could easily believe that it might ot be any better then > wwv. > Good thread. Things to learn as we all seek a backup to GPS I assume. > Regards > Paul > WB8TSL Paul, be aware that three servers is not the best choice, as if one fails you are left with two servers, and you don't know which is correct. IIRC four, five and seven servers allow for the failure of 1, 2 or 3 of them, but do choose more than three. 73, David GM8ARV

Hi The standard NTP software does a "best of the bunch" selection when given a large number of servers. What it considers best and what you would pick likely are not the same thing. You stick at a few ms of error rather than getting say a 10X improvement with 100 servers. Custom code, custom hardware, stable routing - you can do *much* better. Of the three, routing is normally the biggest source of error. If it's not you have a very unusual ISP / connection / server. Bob On Jul 24, 2011, at 12:50 AM, "David J Taylor" <david-taylor@blueyonder.co.uk> wrote: >> Indeed thats why I was saying choose 3 servers. >> Now I see in the thread that you can pick pools of servers so thats good. >> Then average what they say the time is and drive oscillator. >> Wonder if you could look towards the stratum 1 servers. >> But that said I could easily believe that it might ot be any better then >> wwv. >> Good thread. Things to learn as we all seek a backup to GPS I assume. >> Regards >> Paul >> WB8TSL > > Paul, be aware that three servers is not the best choice, as if one fails you are left with two servers, and you don't know which is correct. IIRC four, five and seven servers allow for the failure of 1, 2 or 3 of them, but do choose more than three. > > 73, > David > GM8ARV > > _______________________________________________ > 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.