The Ecology of the Oscillator-Only Transmitter

An oscillator-only transmitter sounds like the simplest possible means of getting on the radio air—until you consider the challenges it presents as part of the system that is a transmitting and receiving station:

    The problem of an adequate crystal-controlled transmitter for the newcomer who wishes to come on the air for the first time on c.w. has never been very effectively solved for the amateur who has but a small amount of capital to invest in the first rig. The majority of the beginners now coming on the air use some type of a keyed crystal oscillator coupled directly to the antenna system. We all well know the results of this kind of operation. If the crystal is anything but the very best and most active type, the arrangement is capable of almost no usable power output. If the crystal happens to be an especially active one, it will be possible to obtain a respectable amount of power output but bad keying chirps are likely to be present, dots are very often not present if rapid keying is used, or the rig may frequently become temperamental after having lain idle for a few hours or days and refuse to key at all until it is retuned.
    Then, if the crystal does operate well and key cleanly, there is always the temptation to raise the plate voltage to increase the power output. This may be fine until the antenna accidently or intentionally becomes uncoupled some fine day with the result that we will usually be in the immediate market for another crystal "as close as possible to the frequency of the last one."
—Jack Rothman, W6KFQ, "The Newcomer's Special," Radio, October 1938, pages 38–40 and 66.

    The usual beginner's transmitter is built around a single tube, following the premise that more than one stage will make the first step too complicated or too expensive. The frequent result is a transmitter whose coupling to the antenna circuit is critical in adjustment for a compromise between maximum output and reliable, chirp-free keying. To make matters worse, an attempt is usually made to make up for the lack of additional power stages by running the oscillator at high-power input, incurring the danger of crystal fracture as well. Almost everyone who has had experience with high-power oscillators eventually comes around to the oft-voiced conclusion that more troubles would be avoided if the oscillator, be it crystal or self-controlled, is treated as a frequency-control unit with power output of decidedly secondary importance.—Don Mix, W1TS, "A Fool-Proof Rig for 80 and 40 meters," QST, June 1941, pages 20–23.

    If you've ever juggled both a single-tube transmitter and an antenna, you'll know what this is all about before we begin. But if you haven't, and are searching for something "simple," look before you leap! Yes, those single-tube 6L6 rigs look awfully simple on paper. And they are just as simple to build. If you've ever operated one, however, you know that simplicity ends right there. Oscillators have a way of misbehaving when coupled to an antenna, and experience has shown that the only way to eliminate all possibility of trouble is to isolate the oscillator from the antenna.—Richard M. Smith, W1FTX, "A Beginner's CW Transmitter," QST, May 1948, pages 25–30.

    The keying of an oscillator is something to be avoided if you want to have a signal free from "yoops" and "chirps." Unfortunately, however, the oscillator must be keyed for break-in operation if you want to work near your own frequency, which seems to be the only way to work anyone these days. Most crystal oscillators do not key well unles care is exercised in adjusting their tuning. The Tri-tet and grid-plate circuits key well without critical adjustment as long as the output circuit is tuned to a harmonic of the crystal frequency. However, most crystal oscillators in 3.5-Mc. transmitters are being operating at the crystal fundamental frequency and under this condition the regenerative circuits have little advantage, so far as keying is concerned, over the simple triode or tetrode circuits unless a very well-screened tube is used. The usual 6V6 and 6L6 do not fall into this category. With any of these circuits, oscillation ceases or is erratic whenever the plate is tuned near resonance where best output is obtained. To obtain good keying characteristics at the fundamental, the plate circuit must be tuned so far off resonance on the low-capacitance side that the useful output from doubling oscillators is often less at the fundamental than at the second harmonic.
    To adjust such a circuit for clean keying, it isn't sufficient to hold the key closed and tune the plate tank to the point on the edge of oscillation where the output is greatest. With such an adjustment a loaded oscillator seldom will key well, if starts again at all, once the key is opened. The only way to adjust the oscillator tuning is to listen to the signal while it is being keyed as the plate circuit is tuned to the point where the circuit keys well, regardless of the output. It is impossible to determine this point through meter readings. Depending upon how heavily the oscillator is loaded, it may be necessary to detune the plate circuit considerably to avoid chirps.
—Donald Mix, W1TS, "Unstable Signals," QST, August 1946, pages 25–30, 126.

Talking Points

The operating frequency of an oscillator is affected by loading.

Output variation with loading—including variation to the extreme of no output, as can happen when heavy loading reduces feedback below the point required to sustain oscillation—can make adjustment of antenna matching difficult to impossible.

Limited options are available for keying waveshaping.

In crystal-controlled single-tube transmitters, signal quality (frequency stability and keying waveshaping) may vary with frequency multiplication and with the particular crystal used.

In crystal-controlled single-tube transmitters, output power may vary with the particular crystal used.

In non-tri-tet crystal oscillators, loss or drastic reduction of output loading may destroy the crystal. Ensuring that this cannot happen when the antenna is intentionally disconnected from the transmitter—such as may be true when receiving incoming signals—requires either that the transmitter be connected to a dummy antenna or have its keying disabled during receiving periods.

Running the oscillator at less than full power for spotting—simultaneously listening to your own transmitted signal alongside other, incoming signals—is difficult.

    The c.w. man on 80 meters is not a dx man, at least for the moment, and is more concerned with operating ease and efficiency than with anything else. He usually goes up on 80 to work U.S.N.R. or A.R.R.S. drills, to keep skeds on traffic nets on spot frequencies, or to just plain chew the fat, without being smothered with QRM or dropped like a hot potato after reports are exchanged.
    Now anybody who has done much operating on 80-meter c.w. knows that 25 to 50 watts in an antenna worthy of the name will do just about as good a job as a kilowatt if there isn't somebody else with more power smack on your frequency. For this reason many c.w. amateurs do not attempt to put their regular transmitter on 80 meters but instead use a keyed crystal oscillator which is left all tuned up "sitting in the corner" and ready to go at a moment's notice.
—W. W. Smith, W6BCX, "Strictly 80 Meters," Radio, January 1941, pages 90–93 and 162.

Chambers's "A Compact Portable-Emergency Transmitter" (April 1941 QST)

This transmitter consists of a grid-plate (aperiodic-anode tri-tet) oscillator circuit coupled to the antenna via a pi impedance-transformation network (Figure 1).

Circuit of Chambers Transmitter, April 1941 QST
Figure 1—Unusually for a pre-World War 2 circuit, Vern Chambers's beginner's/emergency transmitter, a cathode-keyed Colpitts crystal oscillator, used a pi network (ignore the output link; that was optional) for output coupling. The pi became standard after the war because the advent of TV and TVI made the pi's low-pass-filtering action valuable and because coaxial cable became much more widely used as an antenna feedline. The use of cathode bias (developed by voltage drop across R2), uncommon in an era when hams strove to get every last watt out of a circuit, is a good practice because it protects the tube from overload when it's keyed but fails to oscillate. (C6 is unnecessary, by the way; the RF-ground end of C3 can be connected directly to chassis. The capacitive voltage divider that makes this oscillator a Colpitts comprises the tube's grid-to-cathode capacity and C3.) This revised drawing corrects the original schematic's misconnected screen–R3–R4 arrangement, in which R4 was connected directly across the +B line. With 250 volts on the tube plate, even with a 6L6 in use this circuit would have put out no more than 5 watts or so—the output power level we now refer to as QRP.

Results of Experiments with the Chambers Transmitter

Unshielded 12V6GT with 105 V on screen: 1.3 Wo (watts output) at 7.01 MHz, excellent keying.
Unshielded 12V6GT with 150 V on screen: 4.5 Wo at 80 meters, excellent keying.
1631 (12-V-heater metal 6L6, shell grounded to chassis) with 105 V on screen: 2.4 Wo at 7.01 MHz, excellent keying.
1631 with 105 V on screen: 4.3 to 4.5 Wo at 80 meters, excellent keying.
1631 with 213 V on screen: 9.0 to 9.3 Wo at 80 meters, excellent keying.
1631 with 213 V on screen: 6.9 Wo at 7.01 MHz, good keying (slight yoop detectable).
1631 with 150 V on screen: 3.7 Wo at 7.01 MHz, good keying (slight yoop detectable, but less than with 213 V on screen).
1631 with 150 V on screen: 6.0 to 6.2 Wo at 80 meters, excellent keying.
12A6 (shell grounded) with 150 V on screen: 2.7 to 2.8 Wo at 80 meters, excellent keying.
Unshielded 6AK6 with 105 V on screen and 22 pF/100 pF feedback: 1.5 W at 3.52 MHz with 3.52-MHz crystal, excellent keying.
Unshielded 6AK6 with 105 V on screen and 22 pF/100 pF feedback: 0.4 W at 7.04 MHz with 3.52-MHz crystal, excellent keying.
Unshielded 6AK6 with 105 V on screen and 22 pF/100 pF feedback: 0.77 W at 7.12 MHz with 7.12-MHz crystal, good to excellent keying.
Unshielded 6AK6 with 150 V on screen and 22 pF/100 pF feedback: 1.7 W at 3.52 MHz with 3.52-MHz crystal, excellent keying.
Unshielded 6AK6 with 150 V on screen and 22 pF/100 pF feedback: 0.77 W at 7.12 MHz with 7.12-MHz crystal, good keying.

In all cases, the plate voltage was 360, key down, and screen voltage was VR-tube-regulated. The plate inductor used was that of the Mighty Midget. The RF-voltmeter detector diode was a 1N4148 (silicon). 80-meter crystals were FT-171Bs; the 40-meter crystal was housed in a large, rectangular black bakelite case, likely with a quartz element of a size on par with that of the FT-171Bs. Although the grid-plate circuit can be used for fundamental and harmonic output with some tubes—actually not a good idea because of the low-pass nature of the pi network—with the 1631, 80-meter crystals will not oscillate with the plate tuned to 40 meters, indicating the importance of feedback from the plate circuit in this ostensibly electron-coupled triode-tetrode circuit. (I did not evaluate the usability of 80-meter crystals at 40 meters with the 12V6GT and 12A6.)

Optimization of the output network for each tube type would likely increase power output over the values shown above while maintaining the ratio of the band-to-band output differences for a given tube. Optimization of the feedback capacitor values—in the original Chambers circuit, only the tube's grid-to-cathode capacitance serves as the upper capacitor in the divider—would likely improve output and/or keying, especially at 40 meters.

Additional Single-Tube-Oscillator Experiments, Various Tubes and Topologies

The 17JQ6 Beam Power Tube as a Power Crystal Oscillator

The xJQ6 (6JQ6, 12JQ6, 17JQ6) beam power tube, intended for vertical-deflection-amplifier service, is attractive for RF power oscillator service because it has low grid-plate capacitance (0.32 pF, unshielded, as compared to 0.3 pF for the 5763/6417) and a relatively high-power cathode (its heater operates at 7.56 W [6.3 V at 1.2 A, and so on]), and because its beam-forming plates are not internally connected to its cathode (allowing better decoupling of oscillation feedback from output tuning in triode-tetrode topologies). A curiosity: The xJQ6 includes an integral beam-plates-to-cathode diode intended to regulate the beam-plates-to-cathode differential at about 5 volts, but because beam plates are normally operated at ground potential (negative with respect to the cathode) or wired to the cathode (therefore operating at the same potential as the cathode) in RF oscillator and RF power amplifier service, the integral diode will therefore be inoperative when the xJQ6 is used in most radio-transmitter applications.

In an experimental grid-plate (aperiodic tri-tet) power crystal oscillator at AB2WH, a 17JQ6 produced 6.0 Wo at 80 meters (at 11.8 Wi, for an efficiency of 51%). The circuit used a 22 pF/220 pF feedback divider; 340 ohms of cathode bias resistance; a 22-kilohm grid leak (with no series RF choke); a plate voltage of 360 (key down); and a screen voltage of 150. The output network was a parallel-LC circuit (in ac parallel with dc feed through a 2.5-mH choke) using a 10.6-µH solenoid (17 turns of #24 enameled copper wire, 1.25 inches in diameter, closewound [0.5 inch long]) with output coupling via a 3-turn link series-tuned with an 1100-pF variable capacitor. Up to 13.3 Wo was available with 180 V on the tube's screen and no cathode bias resistance, but keying was slightly to moderately yoopy at this power level. In further experiments at 80 meters with the aperiodic tri-tet at W9VES, a 17JQ6 configured as above but with a pi output network with a 13-µH solenoidal plug-in inductor produced 6 to 6.3 Wo at a screen voltage of 105 and 8 to 8.4 Wo at a screen voltage of 150. Keying was yoop-free in both cases.

Reconfiguring the circuit as a frequency-doubling (3.5-to-7-MHz) tuned-anode tri-tet oscillator (cathode tank, 3.3 µH in parallel with 220 pF; plate output matching, via a pi network based on a 6.5-µH inductor wound on a T-130-2 toroidal powdered-iron core) netted 6.2 Wo (3.55-MHz crystal) to 7.8 Wo (3.51-MHz crystal) with good to excellent keying. (In most "good keying" cases, keying could be improved to "excellent" by detuning the output network away from exact second-harmonic resonance.) The screen voltage was 180, the cathode bias resistance was 340 ohms, and the grid leak resistance was 22 kilohms. (Apparently, a higher screen voltage can be used in this doubling tri-tet configuration [compared to 150 V for acceptably good keying in the grid-plate configuration described above] because the crystal is subjected to relatively little output-to-input feedback when the output circuit is tuned to twice the resonant frequency of the crystal.)

Experiments in producing 40-meter output from 40-meter crystals are planned.

[to be continued...]

Revised June 12, 2015. Copyright © 2007–2015 by David Newkirk ( All rights reserved.