[time-nuts] Clock accuracy

[Bob kb8tq]

Hi

Depending on what you do or don’t consider GPS is somewhere in the 3 to 30 ns 
range without corrections and better than that with corrections. Your GPSDO has some
internal “stuff” that also gets into the act. 

Indeed if you went to a dual band receiver and used any of the many free correction 
services, you could get well below 3 ns. 

Over your 100 year span, 30 ns comes out to about 1x10^-17.

Bob

> On Jul 19, 2019, at 6:20 PM, donald collie <donaldbcollie at gmail.com> wrote:
> 
> Thanks to all who replied! It looks like the antioxidants will win and the
> clock will fail before the 100 years are up. Assuming the "accuracy" of the
> GPSDO is 1 part in 10^12 then the inaccuracy after 100 years will be up to
> : 60x60x24x365.25x100x1x10^-12= 3ms [approximately] - which is probably
> good enough for an old fella. I have to admit that I have an ulterior
> motive for asking this question : I wanted to know what sort of long term
> accuracy I could expect from the GPS constellation - looks as if 1 part in
> 10 to the 12th is about right.
> Cheers!...............................................................................Donald
> C.
> 
> On Sat, Jul 20, 2019 at 7:01 AM Bob kb8tq <kb8tq at n1k.org> wrote:
> 
>> Hi
>> 
>> The simple answer is that your clock is locked directly to a set of time
>> sources built
>> into the GPS satellites. Those sources are corrected by ground stations
>> via comparison
>> to NRL and NIST (and indirectly other sources as well). The various ground
>> reference
>> time systems get measured and evaluated to form what we call the right
>> time. This
>> is done by BIH in Paris. That process also keeps NRL and NIST “in sync”
>> with the correct time.
>> 
>> Since everything is locked together, there really isn’t any long term
>> drift. As long as
>> everything is functioning (and the PPS is from GPS not some random
>> divider) you
>> should be “on time” to within 100 ns pretty much forever. The time
>> involved could
>> be GPS time or UTC depending on how you associate time stamps with your
>> PPS edges.
>> 
>> If indeed something goes wrong with GPS ( as unfortunately happened to
>> Galileo
>> very recently), your time could be just about anything if the error is
>> undetected. If
>> it is detected, your will go into holdover. The drift then depends very
>> much on just
>> what “Trimble” you have inside your setup. 10 us a day for the first day
>> is not an
>> uncommon number to see. Since it’s really frequency drift rather than time
>> drift,
>> the second day will be worse and it just goes downhill from there.
>> 
>> If your PPS *is* from some random divider off of (say) 10 MHz, then every
>> time power
>> goes out, it will come back up at a random point in the second. If you
>> punch
>> a button to “sync” it, you will only be able to move it in 100 ns steps (
>> the period
>> of 10 MHz). If the 10 MHz edge is “right on” with GPS that’s fine. If it’s
>> off by some
>> random amount ….. not so fine.
>> 
>> This gets into a vary basic gotcha: A typical GPSDO *does* get the output
>> PPS from
>> the 10 MHz. The PPS output direct from a GPS module probably is closer to
>> “on time”
>> that the GPS PPS. It will bounce around a lot more, but it likely is
>> closer to being correct.
>> 
>> Lots of twists and turns …...
>> 
>> Bob
>> 
>>> On Jul 19, 2019, at 1:17 AM, donald collie <donaldbcollie at gmail.com>
>> wrote:
>>> 
>>> Without wanting to show my ignorance by confusing accuracy, and
>> precision,
>>> etc, would some kind person please answer the following : Let me explain
>> -
>>> I have my prototype GPS diciplined [ Trimble inside] standard frequency
>>> source connected to both a divide by 5,2,5 and 2 producing all the
>>> reference frequencies necessary for the various bits of equipment in my
>>> workshop, AND the 1pps
>>> output connected to a 7474 "T" flipflop and thence via a 100uF capacitor
>> to
>>> a modified $10 analogue wall clock. Can anybody tell me this : If I live
>>> another 100 years [Let`s say I take antioxidants ;-)  ] what sort of
>> error
>>> should I expect in this clock? [I know that it`s better than 1 second per
>>> day]
>>> _______________________________________________
>>> time-nuts mailing list -- time-nuts at lists.febo.com
>>> To unsubscribe, go to
>> http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com
>>> and follow the instructions there.
>> 
>> 
>> _______________________________________________
>> time-nuts mailing list -- time-nuts at lists.febo.com
>> To unsubscribe, go to
>> http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com
>> and follow the instructions there.
>> 
> _______________________________________________
> time-nuts mailing list -- time-nuts at lists.febo.com
> To unsubscribe, go to http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com
> and follow the instructions there.