Short-term stability oscillator

[alcatraz1972]

I am looking for comparison between different types of oscillators as crystal, saw, and crystal with temperature compensation.

What I need to know, which kind is better in short-term stability? how much stable there are? I want to do time-og-flight measurement using RF, but i need a good reference clock, stable in short-term and large term?



Please If someone can give any information

Thanks in advance!

 

[sergio mariotti]

The simplest method is using a frequency meter. Regardeless for the causes of instability, just record freq. Set the gate lenght in the 10 sec. range.

A more advanced method is find the causes. Find δTemp/δFreq., δVolt/δFreq. etc.
Need for a climatic chamber. Youll' got an idea about dependences and drift.

The most advanced, and the most difficult to do is the Allan Variance. Too long to explain here.

 

[echo47]

The most advanced, and the most difficult to do is the Allan Variance. Too long to explain here.

That's the good way.
Here's an explanation straight from the horse's mouth:
http://www.allanstime.com/AllanVariance/index.html
http://www.allanstime.com/Publications/DWA/Science_Timekeeping/TheScienceOfTimekeeping.pdf

You didn't quantify your "short term" and "large term". Maybe an ovenized crystal oscillator OCXO would fit your needs. For example, here is the stability spec of the venerable HP 10811A/B. (I have a scanned manual in case anyone wants it.)

Averaging Time    Stability
  (seconds)    (Allan Deviation)
---------------------------------
   0.001           1.5e-10
   0.01            1.5e-11
   0.1             5.0e-12
   1.0             5.0e-12
   10.0            5.0e-12
   100.0           1.0e-11

If you need that sort of stability over longer duration, then look into a Rubidium oscillator. I'm happy with my SRS PRS10. You can download the very nice manual:
http://www.thinksrs.com/products/PRS10.htm

I've never heard of using a SAW oscillator as a precision frequency reference.

Somewhat interesting:
**broken link removed**