[ntp:questions] Indirect GPS time source options

Paul tik-tok at bodosom.net
Sun Mar 16 15:48:46 UTC 2014


On Thursday, March 13, 2014 1:52:51 PM UTC-4, Olivier Drouin wrote:

>Diversity is really what I'm looking for and
>I dont really need microsecond accuracy.

>I'll talk with the facility owner and do
>some tests with handheld GPS so I can verify what kind of signal I can get
>(and where).

I'm not sure what your goals and constraints are but ...

You should test with a high sensitivity timing GPS receiver (e.g. u-blox or
Resolution SMTx).  I get acceptable performance in an inner office off a
machine room on the ground floor of three-story building.  In my case
acceptable is 100s of microseconds.  If you can sometimes get a GPS signal
in your machine room (interior or exterior antenna) then run that into a
GPS disciplined oscillator which should give 10-100microsec./day drift when
there's no GPS signal -- O(10ms/yr) given stable temperature.

The high sensitivity + timing constraints let you stay locked with just a
single satellite fix.

There are also some time transfer tricks you could do to periodically
discipline an oscillator or you could just switch to a
Rubidium/quasi-Cesium based free-running oscillator that you "manually"
discipline if needed with NTP (e.g. approximate time2 from peer stats).

Some of these things you can buy, some you would have to assemble from
functional modules.

There was a discussion in the last 12 months or so on the Time-Nuts list
about doing time transfer into (really) RF free areas (i.e. secure bunkers).


More information about the questions mailing list