[time-nuts] New to Time Synching hardware - needing some advice
albertson.chris at gmail.com
Thu Dec 27 19:35:14 UTC 2012
Yes or even more compllex setups.
NTP can work with any number of clocks. In fact the old joke that
goes "A man with one clock knows what time it is, but a man with two
is never sure of the time." This is 100% true. NTP can work with
three or five PPS clocks and will compare them and stop listening to
the ones that do not agree with the others (It uses a fairly
sophisticated method to select the subset of conected clocks to use)
So if you are going to rotate Rb units get four of them and always
have one outside on the GPS being adjusted and three inside and then
you have a good check on the Rb units and saw out worst clock. This
way you have zero down time as you will always have two good clocks
even after an automated tai over. Kind of like RAID disks
if this is going inside some RF shielded room or submarine the sub
captain is going to want a way to detect a failure. I think you need
three clocks as a minimum for that. Then every month you surface and
swap ut the worst of you three clocks. A truly paranoid timekeeper
would use a system of five clocks.
Also using five clocks is common. Many GPS clients will use five
Internet time servers from the pool, works the same way with three GPS
or Rb clocks.
On Thu, Dec 27, 2012 at 10:48 AM, Gabs Ricalde <gsricalde at gmail.com> wrote:
> It could be possible to use two Rb GPSDOs, one providing the PPS and
> another disciplined by GPS, then rotate them every month.
> On Fri, Dec 28, 2012 at 2:41 AM, Attila Kinali <attila at kinali.ch> wrote:
>> On Thu, 27 Dec 2012 10:19:46 -0800
>> Chris Albertson <albertson.chris at gmail.com> wrote:
>>> Yes you could us something like one of the low-cost rubidium
>>> oscilators to supply a pulse per second to Linux NTP and it would work
>>> for some time before the error was as large as 10 mS.
>>> One missing requirement is how long you need be without GPS or
>>> networking. Obviously you can't stay out of contact with the world
>>> forever but it is very easy and cheap if you only need a day. Harder
>>> if you need a month and expensive if you are talking about years.
>> Actually, a couple of months is still quite simple.
>> The Rb approace gives you at least something in the range of 10^-9 stability,
>> long term. Which gets you into the ballpark of 100 days (with 10ms). If you
>> temperature stabilize the Rb you should get down to 10^-11, which would
>> be 30 years. At least theoretically, aging is AFAIK larger than this.
>> But a couple of years should be still possible without too much effort.
>> IIRC the FE5680 sell now for 200USD. Some heatsink and temperature control
>> come maybe at 50-100USD. Some equipment to measure where the PPS lies
>> relative to UTC and a powersupply that can get you 15V from mains and
>> a battery (to get the Rb where the computer is) would also be necessary.
>> Alternatively, put the Rb next to the computer in question, hook a powerfull
>> linedriver to the PPS output, and a wire that goes all the way out where
>> you have GPS reception and can measure the PPS pulse. After measurement,
>> you can remove that cable.
>> Ofcourse this only works, if you can lay a cable temporarily. If you are in a
>> EMP secured room, the only way to get the pulse measured is the approach
>> with the traveling Rb.
>> Attila Kinali
>> There is no secret ingredient
>> -- Po, Kung Fu Panda
>> time-nuts mailing list -- time-nuts at 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 at febo.com
> To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
> and follow the instructions there.
Redondo Beach, California
More information about the time-nuts