29 October 2022

Ageing and Precision Crystal Oscillators

Often when purchasing a precision oscillator such as an Oven Controlled Crystal Oscillator (OCXO), the phrase ‘after 30 days continuous ageing’ will be present. This phrase is a catch-all to allow for green age on a crystal within the crystal oscillator in which it is employed.  Does this mean that specifications are guaranteed after thirty days continuous ageing?

Surprisingly, the answer is both yes and no. If the oscillator is powered continuously and is not powered down after the thirty days, then generally the answer is yes.  If, however, the oscillator is powered down, even after a continuous thirty days of ageing, the answer is no. To understand why, we need to look at the principle causes of ageing within crystals.

There are two primary causes of ageing within quartz crystals:

1.     Crystal resonator frequency is determined by the crystal mass, including electrode mass. Contamination of the surfaces of a crystal causes mass increases which cause the resonator frequency to be lower.  These contaminants include, but are not exclusive to, outgassing of the conductive epoxies utilised in the manufacturing process, water vapour and other manufacturing contaminates. As these contaminants are products of the manufacturing process, they are not affected by the level of clean room or laminar process employed by the manufacturer.  When viewed under a microscope both the quartz and the metal electrodes are found to be porous and can therefore harbour contaminates within these porous cavities. Crystal manufacturers will seek to minimise these contaminants by utilising high temperature vacuum bakes and other cleaning processes, such as ozone cleaning.

2.     Mechanical stress introduced through the manufacturing process is the second major cause of ageing. These stresses cause crystal lattice deformations which cause frequency changes in the mechanical structure. Some cuts of quartz are less prone to stress changes, typically the SC cut as employed in higher performance oven-controlled oscillators.

In the thirty days initial ageing period, crystal contaminants, especially with a new crystal that is heated within an oven-controlled environment, move and redistribute themselves throughout the package. This can occur through several processes such as the mechanical resonance of the crystal under power, evaporation or condensation. The result is a rapid shift of frequency whilst the contaminants re-stabilise themselves. Mechanical stress, introduced through the lapping, polishing and mounting stages along with general handling are all sealed within the crystal when the resonator is taken up to temperature in an oven-controlled environment when these stresses are gradually annealed. Through a thirty-day ageing period, the resonator and mounts will gradually anneal to a preferred stress position. This shows as a frequency change in the resonator due to the crystalline lattice structure of quartz.

If the oscillator is powered off for any length of time after ageing, crystal contaminants will again move when a resonator cools and is no longer under mechanical vibration. The contaminants will seek a new equilibrium level. The amount of re-distribution and movement of the contaminants will depend on both the amount of time the unit has been powered on for and the amount of time the unit is powered down for. An oscillator that has been powered on for several years will have a relatively small amount of re-distribution. Whereas an oscillator that has been powered on for only thirty days but is then powered off for several months will have a large amount of contaminant re-distribution.

With mechanical stresses, an OCXO that is powered off will acquire stresses in the crystal as the quartz cools relative to the quartz mounts. This will have the effect of re-stressing the crystal and the amount of re-introduced stress will be proportionate to the amount of off time.

In conclusion, for precision manufactured crystal blanks within an OCXO aged continuously under power for thirty days and continuing to be powered, then specifications with the caveat after thirty days continuous ageing are valid. However, if the unit is powered down for any length of time, then the crystal within the oscillator will return to the ageing curve to some degree and the ageing will need to be at least partially re-done. It should be noted however that in an oscillator that has been running for several years, the amount of return on the ageing curve for any power off period will be minimal.