Frequency Control with Quartz Crystals
|
Engineering Bulletin E-6: Frequency Control with Quartz Crystals |
An oscillating quartz crystal is a mechanically vibrating body. Internal stresses are present and heat is developed as a result of the motion. If the vibration amplitude is permitted to become great, the stresses can reach a value sufficient to shatter the crystal and, thereby, destroy its oscillating properties. The shattering is a physical rupture of the quartz and is brought about by the crystal literally tearing itself to pieces under the extreme stresses set up by the vibrations. Typically, the rupture appears as a ragged crack, or series of cracks, in the crystal. In some instances, especially with harmonic-type crystals, the fracture may occur at a single point as though the crystal had been punctured by high voltage.
The heat developed by an oscillating crystal is the direct result of frictional losses. Heating is undesirable for it causes the crystal temperature to change while the crystal is oscillating. The change in temperature brings about a corresponding frequency shift such that the frequency will 'drift' as the crystal warms up. Naturally, the amount of frequency drift is determined by the frequency-temperature coefficient of the crystal and by the final operating temperature attained.
Crystals having a high frequency-temperature coefficient are best stabilized by employing automatic temperature control but this, of course, increases the cost of the transmitter. If temperature control is not used, the crystal should be operated at a low amplitude of vibration and the holder should have good heat dissipating abilities. A simple, but effective, expedient is to mount the crystal holder with the heat dissipating surface in contact with the metal chassis of the transmitter or in contact with a metal block, preferably of copper or aluminum. Where the heat dissipating surface is in electrical contact with one crystal electrode, that electrode should be at ground potential.
At any given frequency, the vibration amplitude of a crystal is a direct function of the radio frequency voltage which it develops, or of the radio frequency voltage applied to it (excitation). The amplitude is also a function of the current through the crystal but only directly so under conditions of constant phase angle between the current and the exciting voltage. The phase angle varies between different types of circuits and, also, with the individual conditions in any one circuit. The error introduced by change in phase angle is small, however, and may be neglected for all practical purposes. Since accurate measurement of radio-frequency voltages is generally inconvenient, it is accepted practice to rate quartz oscillating crystals for power limits by a statement of the maximum safe crystal current.
In frequency multiplying circuits where there is a cathode tank or condenser which carries currents at both the fundamental and harmonic frequencies, regeneration at harmonic frequencies is obtained. As the crystal circuit then carries currents both at the fundamental and harmonic frequencies, the crystal current will be somewhat higher than if only the fundamental current were present. The harmonic current does not contribute to the crystal excitation and the current reading will, therefore, infer a greater amplitude of vibration than actually exists. For practical purposes, it is fortunate that the crystal current reading is increased by the presence of the harmonic current; if the current .actually flowing is assumed to fully indicate the excitation to the crystal, it is certain that the crystal is not being excited in excess of the indications.
The presence of parasitic oscillations in an oscillator will also increase the reading of the crystal current. Parasitics are not only undesirable from the standpoint of stability and efficiency but, also, because it is possible, under severe conditions, for the parasitics to become sufficiently intense to fracture the crystal.
The operating crystal current, or more correctly, the crystal excitation, will vary considerably between oscillators of different types and also between oscillators of apparently identical construction. It is best practice, therefore, especially when trying out new circuits, to check the crystal current with a thermomilliammeter. The circuit operating conditions should then be set such that the crystal current will not exceed the maximum safe value under any possible condition of operation.
If a thermomilliammeter is not available, a fair approximation of the crystal current can be made by connecting a low current radio dial lamp in series with the crystal. Knowing the characteristics of the particular lamp in use, the current can be estimated from the brilliancy of the filament.
Standard radio dial lamps having ratings of 6.3 volts, 0.15 ampere, and 2 volts, 0.06 ampere, are recommended for checking crystal current. The 2-volt type is especially advantageous because of its rapid breakdown when the normal rated current is exceeded. By using one 2-volt lamp with crystals rated under 100 ma. and two 2-volt lamps in parallel for crystals over 100 ma., there will be some protection against excessive current. It is a good rule to use a single 2-volt lamp with any crystal rated at 60 ma. or more, at least when making preliminary tests or adjustments.
Figure 7 shows how the light developed by the lamp filament varies with current for the two recommended types of lamps. At the bottom point of the curves, representing 0.1% of normal light, the filaments will be very dull red in considerably subdued light. If the current is reduced a little more, the filaments become nonluminous.
(Data furnished through the courtesy of General Electric Company and Westinghouse Lamp Company)
A common misconception is that the brilliancy varies directly as the current; that is, at one-half normal brilliancy the current is one-half the rated value. An inspection of the curves will readily show the extreme error of this assumption. Under conditions of subdued daylight, the 2-volt lamps show a dull red glow at about 41 ma. (0.041 ampere) while the 6.3-volt lamps reach this condition at about 75 ma. (0.075 ampere). Half brilliancy, as judged by the eye, occurs at about 52 ma. with the 2-volt lamps and 118 ma. with the 6.3-volt (0.15 ampere) lamps. Under steady current conditions, the Filament will burn out at approximately 100 ma. with the 2-volt series and 250 ma. with the 6.3-volt series. It should be realized that these current values stated are subject to variation and are not absolute; the characteristics of individual lamps are not identical and the estimation of brilliancy by the human eye is subject to considerable error.
Reasonably accurate measurements can be made by comparing the brilliancy of the filament directly against the brilliancy of a similar lamp connected in series with a millimeter and a source of variable voltage. By adjusting the variable voltage until the brilliancy of the two lamps is identical, the radio-frequency current will be equal, assuming identical lamps and no radio-frequency bypassing, to the reading of the milliammeter. This is a good procedure to follow when first using lamp indicators as it will teach the operator how to estimate the current by a direct observation of the filament brilliancy.
While pilot lamps serve as an economical and effective substitute for a thermomilliammeter, these lamps must not be considered as foolproof devices in the same class as thermomilliammeters and fuses. The characteristics of individual lamps vary and there will always be some bypassing of the radio-frequency current around the lamp filament due to stray circuit capacities appearing in parallel with it. To keep these capacities at a minimum, it is essential that the leads to the lamp be as short and direct as possible; that they be well separated and not twisted; and that they be soldered directly to the lamp base without the use of a socket.
The lamps will, if properly chosen and installed, offer some protection against excessive crystal current. They are not, however, perfectly reliable; the breaking point of the filaments varies with individual lamps and, most important, the actual current for failure is dependent on the nature of the current itself. If conditions are such that the current is rising at a relatively slow rate, the current required for rupture will be close to the figures previously stated and the lamp will open the circuit. Should the current be rising at a high rate, a much greater value can be reached before the filament ruptures and there is every possibility that the crystal will be fractured before the lamp has a chance to burn out. Conditions of this latter type will occur when a crystal is first plugged in a circuit having excessive feedback, when a radio-frequency surge is fed back into the oscillator stage, during the tuning process in a circuit with too much feedback, or during keying of an oscillator which has excessive feedback or strong parasitics.
With some transmitters, in which the oscillator is keyed for radiotelegraphy, the added resistance of the lamp may affect the ability of the oscillator to be keyed at high speeds. If this occurs, the lamp should be shorted out during transmissions.
It is always best practice in conventional triode, tetrode or pentode crystal oscillators to operate the circuit such that the crystal current is within the maximum safe rating with no load on the oscillator. The amount of feedback to the crystal is controlled among other factors, by the radio frequency voltage across the oscillator tank. At no load this voltage is maximum and, therefore, the crystal excitation and current will be greatest. If the crystal current is well under the maximum safe rating with no load on the oscillator, there will be little chance of its becoming excessive with any degree of loading.
The crystal current does not vary in the same manner with the Tri-tet circuit. With the plate tank tuned to the crystal frequency, the crystal current will increase as the oscillator is loaded and will be maximum at full load. When, however, the plate tank is tuned to some harmonic of the crystal, the crystal current will not vary widely from the no-load value under any degree of loading.
