Oscillators are critical components in RF and digital designs. Although oscillator circuitry may already be provided as part of an integrated circuit (IC), a product-design engineer may still be required to select the crystal resonator and external capacitors. To achieve final product success, it is important that the designer has a basic understanding of how an oscillator and crystal function to select the correct crystal and external capacitors for the device. Poor crystal selection can lead to a product that does not operate properly, fails prematurely, or will not operate over the intended voltage and temperature range. What follows is a basic explanation of crystal resonators and crystal oscillators. Armed with this basic knowledge, a product design engineer will have the insight to select the correct crystal and external capacitors for an IC. Designers will also obtain an understanding of the interrelationships of the various circuits available that make up an oscillator circuit and what to consider when working with RF ICs and commercially available microcontrollers.

Reduced to its simplest components, a crystal oscillator consists of an amplifier and a filter operating in a positive feedback loop (Fig. 1). To begin oscillation, the circuit must satisfy the following Barkhausen criteria: 1) the loop gain exceeds unity at the resonant frequency, and 2) the phase shift around the loop is n2π radians (where n is an integer. Intuitively, it can be seen that the amplifier provides the gain for the first criteria. The amplifier is inverting, causing a π rad (180-deg.) phase shift to meet the requirements of the second criterion. The filter block provides an additional π rad phase shift for a total of 2-π rad (360 deg.) around the entire loop. By design, the filter block inherently provides the phase shift in addition to providing a coupling network to and from the amplifier. The filter block also sets the frequency of oscillation, using a tuned circuit (inductor and capacitor) or crystal.

Operation of an oscillator is generally broken up into two phases: startup and steady-state operation. An oscillator must start itself with no external stimulus. When the power is first applied, voltage changes in the bias network result in voltage changes in the filter network. These voltage changes excite the natural frequency of the filter network and signal buildup begins and the signal developed in the filter network is small. Positive feedback and excess gain in the amplifier continuously increases the signal until the nonlinearity of the amplifier limits the loop gain to unity. At this point, the oscillator enters steady-state operation; the time from power on to steady-state operation is the oscillator start-up time.

An oscillator's steady-state operation is governed by the amplifier and the tuned circuit of the filter block. Loop gain steadies at unity due to the non-linearity of the amplifier. The tuned circuit reactance will adjust itself to match the Barkhausen phase requirement of 2-π rads. During steady-state operation, the main concerns are the power output and loading of the tuned circuit. The amplifier circuit is typically implemented with a bipolar junction transistor (BJT) or metal-oxide-semiconductor field-effect transistor (MOSFET). The linear characteristics of the transistor determine the starting conditions of the oscillation while the nonlinear characteristics determine an oscillator's operating point.

The filter block sets the frequency that the oscillator will operate, which is accomplished by using an inductive-capacitive (LC) tuned circuit or crystal. Figure 2 depicts a basic shunt-C coupled LC-series resonator that provides phase shift and a coupling network. Since an inverting amplifier is being used, the filter block must provide a µ-rad (180-deg.) phase shift to satisfy the second Barkhausen criteria. There is an unlimited number of circuit combinations for oscillators. Numerous circuits take on the name of their inventors (i.e., Butler, Clapp, Colpitts, Hartley, Meacham, Miller, Seiler, and Pierce) and, in turn, many of these circuits are derivatives of one another. Readers should not worry about a particular oscillator's nomenclature, but should focus on operating principles.1 No one circuit is universally suitable for all applications2 and the choice of oscillator circuit depends on device requirements.

The next step in developing a crystal oscillator is to add circuitry to the simplified oscillator block diagram shown in Fig. 1. Figure 3 shows a simplified oscillator circuit drawn with only the RF components, no biasing resistors, and no ground connection. The inverting amplifier is implemented with a single transistor and the feedback mechanism depends upon which ground reference is chosen. Of the numerous oscillator types, the three most common ones are Pierce, Colpitts, and Clapp configurations, and each consists of common circuitry except that the RF ground points are at different locations. The type of oscillator commonly employed in ICs is the Pierce configuration (Fig. 4). It has many desirable characteristics, including the capability of operating over a wide range of frequencies with very good short-term stability.3 Although inductors and capacitors are convenient for use in oscillator-tuned circuits, the primary disadvantage of this type of oscillator is the tendency to drift with changes in temperature, power-supply voltage, or mechanical vibrations. Setting the frequency of an LC oscillator requires precise manual tuning.