[time-nuts] Traceable to NIST...??
magnus at rubidium.dyndns.org
Mon Jun 28 18:27:29 EDT 2010
On 06/28/2010 11:36 PM, Greg Burnett wrote:
> Here's a newer article by Michael Lombardi:
> Note that a measurement uncertainty analysis is a required component of
> traceability. Therefore for the GPSDO to be traceable to NIST, someone
> (either the manufacturer or the user), must perform an uncertainty analysis.
Correct, you need a uncertainty analysis, but for traceability you
cannot make one at factory... that will only be an estimate. For
traceability you need to record the deviations in the chain and make
estimates for the error contribution in each step. You make the
traceability by recording the deviations and uncertainty to the previous
step in the hierarchy, which needs to be traceable in itself.
You may or may not adjust to correction, traceability as such is not
about adjustments, but it may be a practical way to get the corrections
into the device.
For legal traceability a number of complementing things plays in. The
simple thing is much more paperwork.
I think hobbyists at best reaches attempted traceability. It may be good
enough for the hobbyist, for some arbitrary value of good enough which
may be the challenge for us.
Getting the frequency sufficiently correct over time isn't too hard.
Frequency/time stability is, and time biases is.
More information about the time-nuts