Oscillators 101: What Every Engineer Should Know about Crystal Controlled Oscillators

By Louis Bradshaw, Technical Support Engineer, Fox Electronics

Published in the November issue of Wireless Design & Development

This article is intended to provide basic information about crystal control oscillators for engineers. For the purposes of this paper, an oscillator may be defined as a device used to stabilize time-frequency generators, specifically those that provide the clock signals used by data processing equipment.

Oscillators have become nearly ubiquitous, because of the proliferation of microprocessor devices with built-in time-keeping capabilities. Despite this trend (or perhaps because of it) it is useful to summarize basic information about oscillators.

In its simplest form, an oscillator consists of two networks, an amplifier network and a feedback network. Figure 1.0 illustrates such a basic configuration.

Figure 1.0FX3141fig1.gif - 2993 Bytes

As might be expected, the amplifier network amplifies the strength of an applied signal. The feedback network receives that signal, corrects any out-of-phase condition that might exist and then returns the signal to the amplifier network and to the output stage of the oscillator. This process continues for as long as the oscillator is energized. Because of the nature of the devices in question, we may take for granted that an out-of-phase condition will exist at the input side of the feedback network.

Based on the above, it would appear that the feedback network is critical to the successful operation of our oscillator, and indeed, that is absolutely the case. While it is possible to use some combination of discrete components, such as an L-C-R network, as the phase-correction mechanism, experience has shown that this approach is seldom satisfactory. Such a circuit will have a very low "Q" value and will be quite large physically. However, its greatest shortcoming is the complete inability of such a network to maintain reasonable frequency stability over any significant temperature deviation.

Because the usual goal of an oscillator is to provide a specific, well-controlled frequency to the output stage, a precise means of phase-correction is required. Experience has shown that piezoelectric quartz crystal units are admirably suited to this task.

For the purpose of this discussion, we will assume that the Amplifier network of Figure 1.0 consists of an integrated circuit and that the feedback network consists of a quartz crystal. Two conditions must be satisfied before an oscillator circuit will operate as expected:

1. The loop gain around the circuit must be equal to one, or unity. This may be achieved either through the self-limiting characteristics of the amplifier network or through some external gain-control circuitry.

2. The net phase shift around the circuit must be equal to 2pn, where n is an integer, usually 1 or 2. In our present example, the quartz resonator is responsible for shifting the phase to the degree required to meet condition number 2. If the signal is so grossly out of phase that the frequency is far from the resonator's resonant frequency, the oscillator will not operate as desired, if at all.

A change in phase of the applied signal not only results in a change in frequency, but a change in reactance as well. While all components within the oscillator may cause a change in reactance to a greater or lesser degree, the reactance of a quartz crystal is so pronounced that the other components may be assumed to be of zero reactance.

Figure 2.0 illustrates the reactance curve of a quartz crystal.

Figure 2.0FX3141fig2.gif - 13781 Bytes

If we apply an alternating signal at a frequency close to the natural resonant frequency of the crystal, the crystal will be energized through the piezoelectric effect. As the frequency is increased, we will reach a point where the resistance of the crystal is

minimal and the current flow is maximal. This point is identified as the "series" resonant frequency, indicated as "fs" above. At this point the crystal is resonating at zero-phase.

If we continue to increase the frequency, we will reach yet another point of zero phase, known as the "anti" or "parallel" resonant point. At this point, the resistance of the crystal unit is maximal and the current flow is minimal. This frequency is inherently unstable and should never be selected as the frequency of operation for an oscillator. Figure 2.0 also illustrates a region identified as the "area of usual parallel resonance." For our purposes, we will consider a frequency within this area as being a "parallel" frequency.

Just as there are two frequencies of zero phase associated with a quartz crystal, there are two fundament types of oscillator. As with a crystal, the first of these is known as the Series Resonant type, illustrated in Figure 3.0.

Figure 3.0FX3141fig3.gif - 2770 Bytes

The series resonant circuit illustrated in Figure 3.0 is capable of oscillating at some non-crystal controlled frequency in the event of crystal failure. This circuit provides no means of frequency adjustment should it be required. This circuit is also known to be, on occasion, difficult to start. Further, a true series circuit is very difficult to achieve in production and therefore the output frequency usually differs, if only slightly, from the design parameter.

Figure 4.0FX3141fig4.gif - 2842 Bytes

The second type of circuit is the "Parallel Resonant" circuit, illustrated in Figure 4.0. You will note immediately the inclusion of two capacitors, referred to as CL1 and CL2. These capacitors are external to the quartz crystal and have the effect of increasing the resonant frequency above that of series resonance. Note especially that it is the oscillator frequency that is increased and not that of the crystal. This frequency is often referred to as the "parallel" frequency but this is in fact a misnomer. This frequency results from the inclusion of the capacitors and should properly be called the "Load Resonant" frequency. Because the capacitors increase the frequency to a point above the series resonant frequency, their value must be carefully calculated and specified.

In contrast to the series resonant circuit, the circuit illustrated in Figure 4.0 will not continue to oscillate in the event of crystal failure, and does provide a means of frequency adjustment if adjustment should be required. All oscillator circuits rely on noise as a means of starting oscillations; the parallel circuit provides additional noise through the medium of the capacitors. The so-called "Parallel Resonant" circuit is recommended in applications using integrated circuits.

This article provides only a brief overview of the basics of oscillator circuits. In any application involving the use of quartz crystal oscillators, the reader is strongly encouraged to discuss the requirements with his crystal vendor as early in the design stage as is practical.