[time-nuts] TIE measurement for arbitrary frequency oscillator

[Bob Camp]

Hi

Does your rubidium put out a pps? Do you have any counter capable of time interval (time channel 1 to channel 2) measurement?

Something like a 5334 is a pretty common item, and they are cheap if you need to get one. Feed the pps into the "start" channel. Feed the generator into the "stop" channel. The time interval measured will distribute over the period of the generator output. At first glance the data will look like junk. You need to post process it to "unwrap" the delta phase readings. Your data will be once a second and only good to a couple ns. 

Taking your original frequency, the period would be 135.83… ns.  A perfect setup would only report data in the 0 to 135.83 ns range. Real counters are never quite perfect, so you will probably not get 0 and you probably will get 136 or 137 ns. You might also get negative numbers. A lot depends on your counter. The first thing is to fold the data into the proper range. Next is to make a reasonable guess when you go from say 121 ns to 13 ns. That's the phase unwrapping process. Once you've done that, you have a phase record (delta phase once a second) that you can feed into Stable-32. There likely will be a bit more fiddling on real data, but that's the basics.

Bob

On Mar 19, 2013, at 6:08 PM, John Doering <johndoering57 at gmail.com> wrote:

> Bob,
> 
> I am a grad student working part time for a frequency generator
> distributor. For the most part, it's just buy/resell so we don't do
> production sized testing, rather specific tests when it's too expensive to
> outsource or we already have the necessary setup. However, we've had quite
> a few requests for TDEV & MTIE analysis, so we started looking into ways to
> do it in house for the future. I've been seeing a wealth of information on
> these boards, figured it was worth a shot to ask.
> 
> It's not a garage operation budget, but the size of the company doesn't
> permit for more expensive pieces of equipment either (like the symmetricoms
> that run for 25-30k)
> 
> I do most of the instrument and test application programming. For stability
> processing, we use stable32.
> 
> John
> _______________________________________________
> time-nuts mailing list -- time-nuts at febo.com
> To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
> and follow the instructions there.