The long term time dependence of a frequency source's frequency is often called frequency aging. Since other long term changes can occur, such as changes in the elements of the equivalent circuit of a crystal resonator, and in a crystal oscillator's input power ("input power aging" is defined in MIL-O-55310 [1]), clarity of expression requires that the particular parameter of interest be specifically stated. In the remainder of this paper, "aging" will mean frequency aging, unless otherwise specified.
"Aging" and "drift" have occasionally been used interchangeably in the frequency control literature. However, in 1990, recognizing the "need for common terminology for the unambiguous specification and description of frequency and time standard systems," the CCIR adopted a glossary of terms and definitions .
According to this glossary, aging is "the systematic change in frequency with time due to internal changes in the oscillator." Added to the definition is: "Note - It is the frequency change with time when factors external to the oscillator (environment, power supply, etc.) are kept constant." Drift is defined as "the systematic change in frequency with time of an oscillator." Drift is due to aging plus changes in the environment and other factors external to the oscillator. Aging is what one specifies and what one measures during oscillator evaluation. Drift is what one observes in an application. For example, the drift of an oscillator in a spacecraft is due to (the algebraic sum of) aging and frequency changes due to radiation, temperature changes in the spacecraft, and power supply changes.
The CCIR definitions of aging and drift are now incorporated into the military specifications for crystal oscillators, MIL-O-55310 [1], and have been recommended by the IEEE Standards Coordinating Committee 27 on Time and Frequency for inclusion in the next edition of the IEEE Standard Dictionary of Electrical and Electronics Terms.
Random changes of frequency with time, called short term stability (or, more correctly, short term instability), are characterized in the time domain by the two-sample deviation (also called the square-root of the Allan variance), and in the frequency domain by the various measures of phase noise, as defined in IEEE Standard 1139-1988 [3].
The very accurate and precise measurement of frequency allows the observation of very small changes in a resonator. It is generally true that the crystal resonance frequency is a more sensitive measure of the state of the resonator system than other measurements that can be made. It has therefore been very difficult to apply measurements other than frequency to studies of the nature of the aging process, particularly for low-aging devices.
Many aging measurements have been reported, but few have included a detailed scientific or statistical study of the aging processes. Our understanding of resonator aging processes is often based on indirect evidence gained from other fields (such as material science, and the science of solid surfaces), and on process developments that seemed "sensible" for general reasons, and which were followed by an evaluation of the resulting aging. For high aging rate resonators, advanced surface science measurements seem to support the use of the "sensible" processes by qualitatively correlating with the measured aging.
The "sensible" processes generally include suitable crystal surface preparation, high level of cleanliness during assembly of the resonator, reasonably well controlled crystal mounting and processing, and hermetically sealing the resonator into a clean enclosure. High temperature in-process baking is often used at some stage of resonator fabrication, such as after mounting, or before final frequency adjustment or sealing. Sometimes a burn-in after sealing is used to "preage" the resonator before shipment, and to test for processing deviations.
Precision crystal resonators need to be protected from operating system environments by sealing in an appropriate package. Typically, these packages have been glass or metal. For these glass or metal packaged resonators the resulting resonator system is very complex, in different ways. Mechanical, chemical, and electrical interactions between the package, the enclosed materials, and the resonator all can cause aging.
In a crystal oscillator, in addition to the aging of the resonator, aging of some of the electrical elements (for example, series inductance or load capacitance) can also change the oscillator's frequency. Changes in the shape and configuration of the metal leads, and deformation of circuit boards and enclosures can also produce aging.
Some of the aging of crystal filters is probably associated with the aging of associated electrical elements. Otherwise, crystal filters are subject to the same aging processes as other types of resonators.
The following sections of this paper contain discussions and references to published reports on the topics that are relevant to our understanding of the aging of resonators, crystal oscillators, and crystal filters. The emphasis is on reviewing aging processes, and the literature since 1983. The pre-1983 literature was reviewed by Gerber [4].