[time-nuts] Z3801A - notes on the hp58533a GPSDO
shoppa at trailing-edge.com
Sun Jan 14 15:43:34 EST 2007
"David I. Emery" <die at dieconsulting.com> wrote:
> I see your point, an architecture with the VCXO derived 1 PPS
> output NOT coherent with the 10 MHz would be somewhat unnatural and hard
> to use in a system and presumably virtually ALL GPSDOs do phase lock the
> 10 MHz to the GPS time base rather than just frequency locking to it.
> There are, however, some subtle differences between a loop
> optimized to make the best estimate of 10 MHz possible given the OCXO
> used and one optimized to make the 1 PPS the best possible approximation
> to UTC.
Most GPSDO manuals say that if lock is lost and when lock is recovered
recovered it is found that drift is greater than some margin, then the
phase relation is not preserved. The system "jumps" to the new
phase and there will not be the exact 10M:1 ratio that otherwise
Usually this margin is many microseconds of drift, typically
accumulated over a day or more.
More information about the time-nuts