[time-nuts] 10MHz Rubidium reference source for frequency counter
csteinmetz at yandex.com
Sat Nov 1 17:25:19 EDT 2014
> > Yes, using TI mode is essential for getting down to the counter's limits.
>What's going on there? It's just a divide, right? Is the firmware not smart
>enough to do get enough precision?
>Do all counters have that problem?
The raw TI data have all of the benefit of the interpolation done by
the basic TI counter. To get frequency (or period), the counter
divides the TI into (or by) the event count, which is an integer that
is more granular than the interpolated TI result. This reduces
precision (not necessarily by a huge amount, since it's typically a
big integer) and has the potential to introduce periodic anomalies
into the data.
To my knowledge, all TI-based counters behave this way.
More information about the time-nuts