A Key to Proper Frequency Control in Portable Equipment

by Jim Northcutt, Engineer, Fox Electronics

Appeared in the May 1999 issue of ECN

Designers of portable equipment have a tough job. The marketplace demands equipment that exhibits low power consumption (for extended battery life) and is robust enough to survive the rigors of day-to-day handling, including the impact of being accidentally dropped. Equipment must perform properly "as-is", even in environmental extremes. Improved performance with reduced size is always an important goal. Finally, the designer must specify readily available components -- which meet and exceed the design requirements of the finished product - to promote an efficient manufacturing process.

This article will discuss the challenge of maintaining a constant operating frequency in spite of fluctuating temperatures - one of the keys to proper performance of portable equipment - through the use of temperature compensated crystal oscillators (TCXOs).

Crystal controlled oscillators (XOs) have long been used to regulate operating frequencies in portable equipment. However, their frequency drifts when the ambient temperature changes. Typically, a Temperature Compensated Crystal Oscillator, or TCXO, is used to eliminate (or at least limit) this frequency change. However, simply recognizing that a TCXO is required is not enough. Designers must specify the device's operating characteristics - including temperature range and the desired degree of compensation. But, just specifying the temperature range isn't enough, either.

In a perfect world, we could specify an oscillator having a temperature range between -55�C and +125�C, and a frequency compensation of +/-0.01 PPM�and actually have the device perform accordingly. But, the real world is constrained by the laws of physics and electricity, and we can't always specify an absolutely perfect oscillator for a given application. That's why equipment design engineers should always discuss their requirements with the oscillator vendor -- at the earliest possible stage of the design process. They can then determine exactly how the oscillator will really operate, and design and plan accordingly.

Oscillator Basics

A brief review of the theory of oscillators will help simplify the specification of a TCXO.

In its simplest form, an oscillator consists of an amplifier network and a phase correction network. In order to start and sustain oscillation, the loop gain around the circuit must be greater than unity, and the phase shift of the signal must be equal to 2npi.gif - 872 Bytes, where n is an integer, such as 0,1,2, or 3.

The phase correction network includes a quartz crystal, usually functioning in the inductive portion of the complex impedance plane. When operated in this manner, the crystal is said to be in load capacitive form. A quartz crystal exhibits excellent phase-compensating characteristics, and so is an acceptable frequency-determining device. However, designers must contend with its tendency toward frequency drift when the temperature changes.

Perhaps the best way to solve the frequency drift over temperature change dilemma is to utilize a compensating network. A compensating network consists of a temperature-sensing device coupled to a reactive component in the feedback path of the oscillator circuit. This component modifies the output frequency. Although such a network will be helpful, it also complicates matters, even though the technology of compensation is mature. Oscillator size must be increased to accommodate these extra components. Power consumption increases slightly because the components also draw power.

There are two primary types of frequency compensation networks; a third type is a hybridization of the first two. Each of these methods deserves consideration. Understanding these approaches to oscillator design will help the equipment designer make an informed decision when specifying a TCXO.

Thermistor-Resistor Networks
Method one uses a thermistor-resistor network, in which temperature indirectly varies the capacitance of a component that is part of the capacitive load seen across the crystal, and thus affects the frequency. An unstable supply voltage will change the output frequency. A voltage reference is required to stabilize the output frequency, and this will increase current consumption.

Also, the presence of noise in the output section of the voltage reference will modulate the frequency of the oscillator output. Therefore, a low-noise voltage reference is required, adding to the overall cost. Additional PCB area will also be needed to accommodate these extra components, making the oscillator larger.

Obviously, this approach runs counter to the design engineer's goals of reduced cost, low current consumption and reduced size.

The Direct Method of Temperature Compensation
The second method of temperature compensation (usually called the "direct" method) relies on capacitors that use rated temperature coefficients to replace the temperature-stable capacitors commonly found in the crystal's feedback network. By testing an uncompensated oscillator over a specific temperature range, the oscillator designer can determine the capacitive change that is required to compensate the frequency. Using this value, the oscillator designer can then determine the temperature coefficient that must be exhibited by the capacitor.

In practice, the temperature coefficient is usually determined by placing capacitors with different temperature coefficients in parallel with each other. This technique is effective over only a linear portion of the crystal's frequency/temperature curve, but output stability on the order of +/-0.5ppm can be achieved. Compared to method one, overall size is close to equal and power consumption is not impacted. The cost is increased simply because this approach is test intensive and the angle of cut of the quartz plate must be rigidly controlled. This control involves considerable testing throughout the manufacturing cycle. When this approach is utilized, the results are excellent. The impact on cost varies from vendor to vendor.

xtalfig4.gif - 9116 Bytes Maintaining a constant operating frequency in crystal oscillators is essential for properly functioning portable electronic equipment.

"Hybrid" Temperature Compensation
The "hybrid" method combines the best aspects of the two previously discussed methods. Thermistors are placed in parallel with capacitors that have both positive and negative temperature coefficients. By properly choosing the thermistor value and temperature curve, designers can introduce or remove a required capacitor from the circuit. This method allows frequency compensation over a wider temperature range, since (unlike the direct method), it can be used to compensate a non-linear crystal frequency/temperature characteristic. The "hybrid" method is, however, somewhat difficult to implement. The difficulty arises because components must be selected carefully and extensively tested. In many cases, this becomes almost a "test-and-fit-by-hand" approach rather than a mass production approach. It is excellent for very precise oscillators.

801b_p.gif - 7279 Bytes Fox Electronics' FOX801BH is an SMD temperature controlled crystal oscillator 
that is suitable for a variety of cost effective applications which require excellent 
frequency stability.

A Question of Temperature: Available Ranges vs. Desired Ranges
Because quartz crystals' frequency drift does not change in parallel with temperature changes, compensating a crystal's change in frequency outside the linear region is difficult to achieve and costly to implement. The basic reason for the difficulty and expense is that the "A-T Cut" crystal has a cubic frequency versus temperature shape. The exact curve traced by a specific crystal is dependent on the angle at which the piece of quartz was cut from the stone. Angles can be "positive" or "negative" when compared to a reference angle, the value of which caries according to the mode of operation and the crystal blank geometry.

The frequency changes exhibited by "negative" angles will be quite large per degree of temperature change. They are therefore extremely difficult to compensate. "Positive" angles typically increase in frequency as the temperature increases. However, at a certain temperature, the frequency will begin decreasing. The point of change is called the "low temperature turn point". As the temperature increases further, the frequency will continue to decrease until a certain temperature is reached. At this point (known as the "upper temperature turn point"), the frequency again begins to increase.

Depending on the exact value of the angle in use, the two temperatures can be widely separated. The usual practice is to select an angle that has turn point temperature equal to (or very nearly equal to) the temperature extremes over which compensation is desired. In such a case, the frequency will be decreasing over the temperature range and can therefore be compensated more easily with a minimum of required components. Utilizing the angle of cut such that the frequency change/temperature relationship is linear, simply minimizes the number of compensating components required, as well as the amount of testing.

Furthermore, none of the available techniques will compensate all crystals over all temperature ranges; the degree of compensation within a given temperature range is limited by the choice of oscillator design. It's still up to the designer to select the most practical approach for a given application.

In general, narrow temperature ranges present opportunities for excellent frequency compensation, while wider temperature ranges impose larger frequency deviations. A temperature range between -20� and +70�C is easily managed; a temperature range between -40� to +85� C is more difficult to manage, but is still practical. The degree to which the frequency may be compensated over these ranges is determined by the compensating technique used by the oscillator manufacturer.

Maintaining a constant operating frequency in crystal oscillators is essential for properly functioning portable electronic equipment. Because the performance of these oscillators can be adversely affected by changes in temperature, designers must rely on frequency compensation.

There are three commonly used methods to achieve frequency compensation, but each has its particular set of assets and liabilities; no single method will be suitable for every application. By consulting with oscillator manufacturers very early in the design process, design engineers will increase their chances of obtaining a nearly ideal oscillator.