|Engineering Bulletin E-6:
Frequency Control with Quartz Crystals
Excitation is the most important consideration in the application of quartz crystals for frequency control of oscillators having appreciable power output. A quartz crystal can be applied to any type of oscillator circuit with any type of tube as long as the crystal excitation is kept within reasonable limits. Or, in other words, if the maximum rated crystal current is not exceeded under any possible condition of operation. This does not necessarily infer that it is possible to use high power oscillator tubes and obtain unusually large power outputs. The conditions for safe crystal current generally will be such that the power output will be no greater than obtainable with smaller tubes.
In testing a crystal oscillator circuit, especially when the excitation characteristics of that particular circuit are not well known, always make preliminary adjustments with reduced voltages. The crystal current should be measured under these conditions and, if sufficiently low, the voltages can then be raised to the desired values or to such values at which the crystal current approaches the maximum safe rating; whichever is the limiting factor. It is always best practice to set the operating conditions for the lowest crystal current consistent with required power output.
Figure 23a indicates the effects of tuning a crystal oscillator (except the Tri-tet). As the crystal goes into oscillation there will be a pronounced drop in the plate current. Maximum output will occur at the point of least plate current but operation should be between points B and C point A is unstable and, if the circuit is operated under that condition, erratic performance will result. When cathode bias is used, the plate current, under load, may rise with tuning and exceed the non-oscillating value. If this occurs, operation should be between the equivalent points B and C on the corresponding rising plate current curve.
The crystal oscillator portion of the Tri-tet circuit will show a characteristic tuning curve somewhat different from the conventional circuit. This tuning curve is shown in figure 23b. When first placing the circuit in operation, it should be tuned approximately to point A. After the plate tank circuit has been tuned to the desired harmonic, the cathode tank should be retuned for greatest output in the plate circuit regardless of the actual plate current.
Tubes such as the 802, RK23, 807 and RK37 which have a very low internal plate-to-grid capacity may require the use of external feedback to bring about sufficient excitation of the crystal, especially at the lower frequencies. This is usually accomplished by connecting a capacity of 2 mmf. to 10 mmf. between the control grid and the plate of the tube. Such a capacity, however, should be used only when necessary and with considerable care. Add the smallest amount of capacity which is consistent with good performance only after all other circuit values are found to be correct and in proper working order.
The 6L6 is preferable to the 6L6G as a crystal oscillator. Lowest crystal current with good output is obtained when the metal shell is connected to the cathode pin directly at the tube socket.
When using beam-power tubes in Tri-tet and conventional tetrode or pentode oscillator circuits at the higher crystal frequencies, a considerable reduction in crystal current can be obtained by the simple expedient of connecting a 50 mmf. to 100 mmf. condenser in series with the crystal. Most tubes of this type are easily over-driven due to their high power sensitivity; the condenser reduces the excitation with no appreciable loss in power output. If the capacity is too small the crystal will stop oscillating, while too much capacity will be ineffectual.
When using variable frequency crystal units, such as the Bliley VF1 or VF2, the oscillator power output normally will drop off as the frequency is varied over the adjustable range from the lowest to the highest values. The actual amount of power output variation which can be encountered may be as high as 250/0 in some circuit arrangements. In conventional triode, tetrode or pentode oscillators, the power output can be made to approach a constant value by the use of a relatively high C tank. With the Pierce oscillator-multiplier arrangement (figure 11), power output constancy can be improved through grid-to-cathode feedback as pointed out in the discussion of that circuit.
Harmonic generating power oscillators usually make use of a pentode (or tetrode) tube as a combination triode crystal oscillator and pentode frequency multiplier. It should be remembered that the development of harmonics in such circuits is basically dependent on the choice of circuit conditions to bring about a distorted output. Only in instances where the crystal has a very low activity does the crystal affect the harmonic generation. The most foolproof harmonic generator is a low-power crystal oscillator driving a beam-power tube. A simple low-voltage oscillator, using a tube such as the 6C5, 6J5G or 6F6, driving a 6L6, 807, RK39 or RK49 frequency multiplier is an excellent combination.
It has been pointed out that the frequency of an oscillating quartz crystal can be altered by changing its effective equivalent electrical network through application of external reactance. This fact is often useful in instances where it is advisable to slightly change the frequency of a 'fixed' frequency crystal unit. In triode, tetrode or pentode oscillators of the tuned-plate crystal-grid type, the circuit frequency can be lowered by connecting capacity directly in parallel with the crystal. It is desirable that an air-condenser be used and it should, naturally, be variable. The maximum amount of possible frequency lowering varies with frequency, type of holder and circuit characteristics; it amounts to about 200 cycles/second at 800kc., 1500 cycles/ second cat 4000kc., 800 cycles/second at 6500kc. and 250 cycles/second at 10,000kc. The effect of the added capacity, is to lower the effective crystal activity as the capacity is increased. In any event, 75 mmf. represents the maximum usable capacity at any crystal frequency; at high frequencies 25 mmf. is maximum.
When a variable air-gap holder is used in services such as broadcast where very close frequency adjustment is necessary, it is sometimes advantageous to connect a very small variable air condenser across the crystal. The air-gap adjustment then serves as a coarse frequency adjustment whereas the condenser acts as a trimmer for final frequency setting.
In Pierce oscillators, frequency variation is best obtained by connecting reactance directly in series with the crystal. The use of capacity will raise the frequency while inductance has the opposite effect. Also, changing the value of the grid-cathode feedback capacity will influence the frequency. A capacity in parallel with the crystal likewise will cause a frequency change due to the fact that the crystal does not work purely at its natural resonant frequency; the use of such capacity, however, very rapidly decreases the effective crystal activity.
Occasionally, kinks will be found in various radio periodicals to the effect that the frequency of a crystal can be lowered by coating its faces with India ink, iodine or some other material. The effectiveness of such a process is dependent upon the fact that the coating dampens the crystal and increases its effective mass; the result is a lowering of frequency. Oppositely, the frequency can be raised by inserting a piece of paper between the crystal and one of its electrodes to, in effect, create an air-gap. In either case, the total amount of possible frequency change is quite limited and is accompanied by a rapid decrease in activity. If the crystal is subjected to relatively high excitation, arcing might occur finally causing fracturing as a result of the concentrated heat of the arc. Also, with some crystals, erratic performance and encouragement of a frequency jump can occur. Simple reasoning dictates that altering crystal frequency by physical means as described should be applied only as an emergency measure where a special situation demands such action.
As explained in previous sections, excessive excitation will fracture a quartz crystal rendering it useless. The following are the outstanding sources of excessive excitation: (1) high tube voltages, (2) too much bias; grid-leak, cathode or combinations of both, (3) insufficient bypassing of the screen-grid circuit, (4) stray oscillator plate-to-grid feedback brought about by improper circuit layout, (5) the existence of strong parasitics in the oscillator, (6) operating straight through on the crystal frequency in the Tri-tet circuit with poor internally shielded tubes such as the 59, 47, 42, 6L6, 6F6, (7) improper inter-stage shielding bringing about undesirable coupling between the oscillator and some other stage of the transmitter, (8) feedback into the oscillator stage brought about by self-oscillation in one of the buffer stages or the final amplifier, (9) improper circuit values with oscillators in which the crystal feedback is considerably dependent on circuit adjustments, (10) failing to place the band switch in its proper position with the Bi-Push Exciter, or, (11) in certain instances, by removing the plate voltage from a Pierce Oscillator employing an untuned tank. This is a unique situation in that the buffer stage, which follows the oscillator, can act as a crystal controlled oscillator. The crystal is effectively connected between the control grid and ground of the buffer tube and, if the buffer plate tank is tuned to the crystal, that stage can act as a crystal oscillator when the voltage is removed from the oscillator stage proper. If conditions in the buffer stage are 'incorrect for a crystal oscillator, the crystal may be fractured.
A somewhat common cause of crystal fracturing is self-oscillation in a buffer or amplifier stage (item No. 8), particularly in the stage following the crystal oscillator. The damaging self-oscillation may occur during initial neutralization if plate voltage is applied to the neutralized stage before the adjustment has been perfectly made. Likewise, it might occur during operation if neutralization is not complete. Also, it might be caused by poor layout in a tetrode or pentode stage not normally requiring neutralization to prevent oscillation. As a general rule, it is recommended that plate voltage should not be applied to any stage working at the crystal frequency until the usual tests indicate proper neutralization and full freedom from self-oscillation.
If a transmitter is to be built on the basis of published constructional information, the crystal oscillator portion should be carefully checked with respect to the crystal frequencies intended to be used. This precaution is advisable in view of the fact that some circuits, designed on the basis of certain oscillator frequencies, are not satisfactory for crystals at higher frequencies. For instance, a certain oscillator which functions well with 80-meter amateur crystals, can have characteristics such as to cause fracturing of 40-meter crystals due to excessive feedback at the higher frequencies. Particular attention, in this respect, should be paid to Pierce and modified-Pierce oscillators. If harmonic type crystals might be employed for some frequencies (all present Bliley crystals for frequencies above 11mc. are harmonically operated), the oscillator should be capable of causing the crystals to work at the correct frequencies.
A large number of amateurs attempt to work close to the edge of the various amateur bands to obtain certain operating advantages. When choosing a crystal frequency for such purposes, there are several considerations which are important:
1. The frequency of any crystal is somewhat dependent upon the characteristics of the circuit in which it is used. Variations, under operating conditions, may be as great as .03% from the laboratory calibration, depending upon the particular oscillator arrangement. The calibrated crystal frequency should be such that a possible difference of ± .03% will not place the actual operating frequency outside of the band limits.
2. The frequency of any crystal will be affected by its temperature. All Bliley Amateur Crystal Units are calibrated at approximately 80°F. Therefore, make allowance for frequency drift due to other possible crystal operating temperatures as a result of direct crystal heating, heating by transmitter components and varying ambient temperatures. Low temperature-coefficient crystals can have a positive or negative drift depending on the characteristics of the individual crystal. If the sign of the drift is unknown, assume that the drift will be towards the band edge.
3. The Federal Communications Commission requires that all modulation frequencies be within the band limits. In addition to allowances as in 1 and 2, leave sufficient frequency difference to accommodate any side bands. Allow cat least 4kc. for radiotelephony and approximately 500 cycles for radiotelegraphy.
4. Edge of band operation should be attempted only when the station is equipped with a means for accurately measuring the operating frequency.
5. For working extremely close to a band edge, the use of a variable frequency crystal unit is to be preferred. All possible effects of circuit characteristics and operating temperature on crystal frequency can then be offset by direct frequency adjustment. Under such conditions, the limitation for proximity to the band edge depends to the largest extent upon the accuracy to which the absolute frequency can be measured.
Quartz crystals are, fundamentally, devices for the purpose of frequency control and stabilization. While modern crystals will control a considerable amount of power, best operation and frequency stability can be obtained only when the oscillator is operated lightly loaded and under conditions which bring about very low crystal current.