maTheMatic said:
Hi,guys
I am a little confused about definition of clock/oscillator
stability. it is x ppm over some temperature range in common spec and
there usally another aging spec in addion to stability, so I assume the
stability was not very related with the measuring duration. but
recently I am read some material about atom clock which used in gps,
and they specify the stability with a duration(short or long) and it
seemed they assuming some unchanged temperature. so can any one help
me clarify the concept or recommend some reference?
x-posted to and followups set to sci.electronics.basics
Oscillator specs and crystal specs (often used as the frequency setting
element) vary a little.
Crystal specs usually state their stability across temperature, loading
capacitance, cal tolerance (initial accuracy) and long term ageing.
Long term ageing occurs simply because a crystal is an
electromechanical device. One I have in front of me specifies those
things as:
Stability: 50ppm from -40C to +85C
Initial accuracy: 30ppm
Loading capacitance : 7pF
Ageing : 3ppm max first year
There isn't any more on ageing, as this device is not expected to be
used in a reference oscillator.
As you mentioned, there is a temperature issue.
For very accurate clocks, we run the crystal in a temperature
stabilised oven with an AFC (automatic frequency control) loop, which
maintains the temperature such that the oscillator maintains it's
frequency very accurately.
This is known as a TCXO (Temperature Compensated Crystal Oscillator),
and we can get very accurate ( << 1ppm) accuracy and stability from
such devices.
Also keep in mind that an oscillator is more than just the crystal;
it's the crystal plus amplifier plus control loop circuitry (for really
accurate devices amongst other things) and when specifying an
oscillator, the entire circuit must be taken into account.
Cesium (atomic) clocks use the natural resonance of cesium (which
starts as liquid and is heated in an oven to it's gaseous state) to
lock in a crystal oscillator. By using cesium as the reference feedback
element, accuracies of 0.00000002ppm (about 2 E-14) have been achieved
for reference units (such as the one used at NIST).
This one, incidentally, is known as a Primary Standard or a Laboratory
standard.
Secondary standards are more commonly used.
You can read all about cesium clocks :
http://tycho.usno.navy.mil/cesium.html
Cheers
PeteS