[time-nuts] is there a cheap and simple way to measure OCXOs?

John Miles jmiles at pop.net
Wed May 30 15:11:19 UTC 2012


> Unfortunately, although i have a reasonable park of measurement
instruments,
> none of them are in the precision range that i'd need to characterize
> the 8663's.
> 
> I have three oscilloscopes (analog 20MHz, digital 200MHz and 1GHz),
> a cheap handheld frequency counter and a couple of multimeters of
> various quality. I also have a FE-5680A at hand as frequency reference.
> 
> Any ideas what i could do to measure phase noise and frequency without
> buying a lot of stuff? Soldering is not a problem though.

Cheap, yes... cheap and simple, no.  

Since you have multiple copies of the oscillator in question, the cheapest
way to measure PN would be to use a couple of isolation amplifiers, a mixer,
and an FFT analyzer of some sort, rigged together in a quadrature-PLL
arrangement.  You might have a look at the app notes on www.wenzel.com as
well as the other references on PN measurement near the bottom of the page
at www.ke5fx.com/stability.htm .  

Most documents on quadrature PLLs conveniently neglect to mention the need
for isolation amps to avoid injection-locking.  Meanwhile, given that you're
measuring two stable OCXOs, I wouldn't bother with a PLL.  Just let them
warm up, then use the scope to tweak them into phase quadrature long enough
to run the measurement.  

On the other hand, you will most likely need a low-noise amp between the
mixer and signal analyzer.  Wenzel and other authors use fairly elaborate
hybrid discrete-IC amplifiers for this sort of thing, but I would try an
ordinary LT1028 first.  The 8663 OCXOs are not ULN-class parts.

For short-term stability measurement the cheapest solution is almost
certainly a tight PLL of some kind, discussed in the archives at some
length.  For long-term measurement beyond ~100 seconds, a high-performance
counter is fine by itself.  

Your FE-5680A will be noisier than the OCXOs at tau intervals shorter than
about 10 seconds, so you don't need to worry about how to measure stability
below that.  Suggest looking for a good HP 5370A/B counter, and use it with
TimeLab or a similar graphing tool.  

-- john, KE5FX





More information about the time-nuts mailing list