[time-nuts] HP 5065a Rb ref update and question on time scale

Magnus Danielson magnus at rubidium.dyndns.org
Sat Oct 1 18:55:47 UTC 2011

Hi Paul,

On 01/10/11 16:05, paul swed wrote:
> Magnus,
> Indeed on my 5065a 8714 seems to be the right number. With the cfield I
> could trim to the gps just fine and hit the stability specification. By the
> table thats says I am -250 parts. OK, it works.


> But I would think I should be at 9999 or 0 offset by the table even though
> its not exactly zero.
> Second item is that if I adjust the lsd I would believe I should shift a
> bit. But it seems to have a much larger effect.
> As Corby said it doesn't actually matter.
> Just not making much sense though with your math maybe I need to run the
> numbers and see whats really going on. In that respect its much like the CS
> 5061standard.

What I presented was a scetchy thing. The thumb-wheels set the preset 
value which provides the divisor factor n. This control the time (in 200 
ns steps) for a sampler of the 5,3125 MHz oscillator. If you hit "good" 
rates then the sampler will sample the right point on the signal with 
good gain as result. If you hit a "bad" rate the sampling will beat over 
the oscillator and the gain will be poor if any.

> I am curious that maybe back in 1974 atomic time was one thing and then a
> correction was done that makes it -250 from what it was. Maybe a historical
> adjustment.

When the HP5065A was used to realize UTC, it was indeed intended to run 
in shifted frequency from the SI second. Also, recall that the cavity 
tuning of the rubidium cavity needs to be compensated.


More information about the time-nuts mailing list