[time-nuts] nubie querie

[Tom Van Baak]

> Even defining when the sand timer is "done" is not a real
> simple thing. Waiting for that very last particle to drop may
> not be the best approach. 
> 
> Bob

Correct. Marking time with an hour glass is not that different
from marking time with a 1PPS. Each signal has a rise time;
one picks the appropriate live trigger level or sampled slope
waveform model to minimize jitter.

Waiting for the last grain is like waiting for the last millivolt of
a TTL 1PPS pulse; no one does that.

Note that [this] hourglass interval has a standard deviation on
the order of 10 seconds. So my initial goal is 1 second timing
resolution, which turns out to be pretty easy to do optically.

Whether sand or cesium, phase comparators have a minimum
resolution. While lower resolution is better, if the ADEV of the
DUT is too far above the ADEV of the comparator then that
resolution is wasted. So detection to a granularity of 1 second
is sufficient for this application.

/tvb