[time-nuts] Time-nuttery at EFTF-2016

Magnus Danielson magnus at rubidium.dyndns.org
Wed Apr 6 23:09:39 UTC 2016


Fellow time-nuts,

Now a few days into the EFTF 2016 in York, I can report we are having 
some good fun. Attila get's to bounce his work against some good folks, 
and I get to bounce some of my ideas around too, with nice responce.

I did my presentation today, and it went really smooth. I held a 
presentation on the use of least-square methods to estimate phase, 
frequency and parabolic variance (PVAR), work I have done together with 
Francoise Vernotte and Enrico Rubiola. It was interesting as two counter 
manufactures have approached me wishing to implement this.
What I have done is to re-articulate the work of Francois and Enrico, 
but in a form more adaptable to accumulation in FPGA and then continue 
in software while maintaining the least-square properties as you extend 
the tau-length in further software decimation of the data. The algorithm 
I present allows for decimation and from the decimated blocks one can 
then produce any integer tau-multiple as one wish. This allows million 
of samples per second to be collected in FPGA and further processing at 
moderate rate and with much reduced memory needs. It avoids the biasing 
of deviation as otherwise often seen, I've spotted it in several posters 
and one presentation here. I wave my finger mumbling something about bad 
science! :) I naturally avoid doing such mistakes myself while 
presenting a solution that does it proper.
I got some good questions and in general good response. The real-time 
least-square estimation method that is the underlying method of what I 
do is quite useful in many contexts.

One interesting poster from Justervesenet (Norways National Meterology 
Institute) addressed some odd characteristics of the Keysight 53230A 
counter and the way it process data. It have shown several problems, and 
part of the research was done after borrowing a counter from a 
time-nuts. :) I found a USNO guy looking at it, thinking about it, and I 
brought Enrico and Francois to is, as they have been active in analyzing 
behavior of the lambda and omega counters.

Cheers,
Magnus



More information about the Time-nuts_lists.febo.com mailing list