|Engineering Bulletin E-6:
Frequency Control with Quartz Crystals
At these high frequencies, careful consideration must be paid to the design and construction of the oscillator. Factors which are not serious at lower frequencies rapidly become important as the frequency is increased.
Not all tubes are satisfactory as crystal oscillators at frequencies greater than 18mc. With some tubes, especially the high-mu and pentode types, the crystal may be effectively shorted out by the high input capacity. Others, having a low feedback capacity and a large electrode spacing, do not operate efficiently. High frequency triode tubes, such as the 955, 6J5G, HY615, 6E6 and RK34, are the best for all-around performance. Pentodes, in general, are not to be recommended although some types can be employed in the Tri-tet or a modified Pierce circuit with fairly good results.
Parallel feed of the oscillator is seldom successful due to the difficulty of obtaining really good r.f. chokes. This means that the tuning condenser will be at a high potential and must be insulated from ground. The somewhat common arrangement of inserting a mica condenser in the tank circuit to block the d.c. voltage so that the tuning condenser can be grounded is not particularly satisfactory; mica condensers have appreciable losses at very high frequencies and, if used to carry circulating tank current, there may be a serious drop in power output.
All r.f. leads must, obviously, be short and direct. By-pass and tank condensers should be of the best quality. To minimize tank circuit losses, the coil should be self-supporting and wound with heavy copper wire or tubing. Use nothing smaller than number 12 wire.
The low plate impedance of the recommended triode tubes necessitates the use of a high-C tank for maximum power output. Along with the increased output, the high-C greatly improves the circuit stability; in fact, pentode stability is approached when the proper tank values are chosen. The cathode tank of the Tri-tet must also have a relatively high C, inasmuch as the oscillating portion is a triode.
Circuits designed for use with 18mc. to 30mc. crystals are shown in figures 13, 14, and 15. The circuits are basically conventional but all component values should be followed as these have been found to give the best output and stability. The oscillator tank inductances are specified for 10-meter crystals, but for other high frequencies, it is only necessary to choose appropriate coils. With the simple triode oscillator a 955 tube will provide about 1-3/4 watts output while approximately 2-1/2 watts can be obtained with the 6J5G. Either of these tubes will produce sufficient output to drive an 802 RK23, 807, RK39 or 6L6 tube as a buffer or doubler.
The dual-triode circuit is advantageous for frequency multiplying. As a matter of fact, a single RK34 with a 10-meter crystal, forms can excellent low power 5-meter transmitter. A 6E6 tube will supply an output of about 3 watts on 5 meters with a 10-meter crystal while the RK34 will give about 3-1/2 watts. The types 53 and 6A6 tubes are not comparable for output or performance.
An 802 or RK23 tube can be used in the Tri-tet circuit as shown. The output on 5-meters is approximately 2-1/2 watts with the 802 and 311/22 Watts with the RK23; a slightly greater output can be obtained by applying up to 45 volts positive to the suppressor grid. The 6L6 and 6L6G beam-power tubes are not particularly recommended because their poor internal shielding causes the development of parasitics which are difficult to eliminate.
For higher power output than obtainable
with a triode oscillator, an oscillator-multiplier circuit arrangement
similar to one of those previously described (figures 11 and 12)
might be used. By tuning the plate tank to the crystal frequency,
the resulting regeneration inherent in the circuit offsets some
of the circuit and tube losses such that pentode or tetrode tubes
can be employed with fairly satisfactory results. Frequency multiplying,
when using harmonic-type crystals, is, of course, not feasible
due to the fact that
such crystals will oscillate at their intended frequencies only when the plate tank is tuned accordingly.
In most cases, the use of a regenerative oscillator is best avoided where transmitter design permits the incorporation of a simple triode circuit. While satisfactory performance can be realized with a regenerative circuit, the complete elimination of self-oscillation and high crystal current is difficult, particularly with beam-power tubes.
The E-cut crystal, which is employed for frequencies above 23mc., has a frequency-temperature coefficient of +43 cycles/mc./°C. To avoid objectionable frequency drift where temperature control is not applied to the crystal, the oscillator tube should be operated at the lowest plate voltage consistent with required power output. This minimizes crystal heating and subsequent frequency drift. Where a low grid-drive tube such as the 807 or 6L6 follows the oscillator, an oscillator d.c. plate potential of 100 or 125 volts is adequate if efficient coupling exists between the two stages. With such a reduced plate voltage, direct crystal heating is quite low. Possible heating of the crystal from other sources should also be considered. It is well to locate the crystal in such a position in the transmitter that heat transfer from other components will be at a minimum.
Constructional details of practical ultrahigh frequency transmitters are given in the January, 1938 issue of QST in an article entitled, "56mc. Crystal-Control With 28mc. Crystals." Additional articles appeared in the April issue of the same publication.