This summer’s (2022) radio project began when I came across 'The Type 53 as an Harmonic Oscillator' by R. Anderson (VK3WY) in the January ‘36 issue of Australia’s AMATEUR RADIO monthly.
The circuit used either a 53 or a 6A6, a tube that I’ve always been interested in exploring but had never got around to doing it. The 53 was originally released in 1933 as a 2.5V filament dual triode (common cathode) designed for audio service. Hot on its heels in 1934 was its equivalent, the 6A6 with a 6.3V filament.
I had long-considered building a circuit often associated with this tube, the 6A6 push-pull crystal oscillator, heavily promoted in the mid-30s by Frank Jones (W6AJF) via his RADIO HANDBOOKS and RADIO magazine. Being a push-pull circuit, it was pretty much limited to the lower bands, mainly 160 - 40m, since multiplying with this circuit to any of the crystal’s higher harmonic frequencies is not possible.
With the anticipation of good HF conditions from Solar Cycle 25 soon to be upon us, I had almost given up hope of utilizing the 6A6 on any of the higher bands … until I saw this circuit. Although it was described as the possible first-stage oscillator in a multiband transmitter, it seemed to me that with a little tweaking it might be used as a stand-alone single-tube transmitter for the higher bands … just what I was looking for!
As seen above, the schematic drawing in the Australian magazine followed a somewhat different style than what was commonly used in North American publications. It showed one triode section of the 53 being used as a crystal oscillator to drive the second triode section as a harmonic-rich class C amplifier. Being suggested as a first-stage transmitter oscillator for the low bands (with an 80m crystal doubling and tripling), I wondered if it would provide any useful power on 20, 15 and 10m when used with a 40m crystal? There was only one way to find out. My summer project had began!
My usual method for exploring vintage tube circuits is to build a test-bed version on a well-drilled aluminum plate. After gathering the needed parts from my junk box, I was soon generating RF harmonic energy on the upper bands ... but not nearly as much as claimed in the magazine article!
The initial test-bed version produced only 750mW output on 20m and even less when tripling and quadrupling to 15 and 10m. The two 6A6 tubes that I had in my junk box appeared to be pretty weak performers and a couple of NOS 6A6s and 53s were obtained from a local tube seller. Greater output was obtained with the new tubes but still not as much as indicated in the published article.
However, VK3WY's article described a circuit using an 80m crystal doubling to 40m, to produce 6.4W out while quadrupling to 20m produced 3.6W. This was a considerable difference! Certainly the higher frequency demands of my own model would lead to higher losses, particularly when quadrupling to 10m, but I was still surprised to see the numbers so low. Could it be that the 1933 tube, designed for audio service, was just not capable of handling RF at these higher frequencies?
Exactly how the power was being measured by the article’s author was not indicated. Many articles from the 30s describe observing the level of brightness of a small bulb of known wattage and comparing this to what is seen at full brilliance. I've always been skeptical about the validity of stated output powers as this method would include harmonic energy as well as the desired frequency.
In harmonic-rich circuits like this, 'output power' is actually the sum total of the fundamental frequency plus all of the harmonic energy. To determine the true output on a given band, the output can be fed into a 50 ohm non-reactive resonant load for that particular band. A suitable load could be a resonant antenna, tuned to represent a resistive load (1:1 match). This will ensure that any harmonic energy shows up as reflected power. Using a normal 50 ohm dummy load will not indicate out of band harmonics as reflected power since it is not a resonant load. It's also vital that the correct harmonic has been selected by initially tuning the oscillator's output with a calibrated wave-meter or RF ‘sniffer’.
Nevertheless, I was determined to try and squeeze a little more power if I could, reasoning that a watt or two would be enough to have some worthwhile success during Cycle 25’s peak years.
About halfway through the test-bed project I discovered an earlier version of this same circuit, published one month earlier than VK3WY’s article, in the December ‘35 QST ... "Oscillators Using 14-Mc. Quartz Crystals" by Bliley's J.M. Wolfskill.
The author claimed 2W output on 10m and 3/4W on 5 meters but he was doubling and tripling with a 20m crystal. Nowadays, suitable larger style 20m crystals are as rare as hen's teeth and an unaffordable luxury when found.
Since I've had little interest in multistage transmitters of the early 30s, I was unfamiliar with what was popularly used as a first stage oscillator / multiplier … further research revealed that, much to my surprise, it was this very same 53 / 6A6 harmonic oscillator! I was now seeing it everywhere!
It should be noted that the Tri-tet oscillator had also been developed by this time but had yet to garner widespread popularity, possibly because of its unfair reputation as a ‘crystal-cracker’. It wasn’t until the amazing beam-power 6L6 tetrode was introduced in 1936 that the Tri-tet became the hands-down favorite oscillator stage for the next decade, ending the 6A6's three-year run of popularity.
Things were changing quickly in the ham radio world of the 30s as advancements in vacuum tube technology spurred rapid growth in both transmitter and receiver developments. What a great time to have been a ham!
I later ran across what may have been the very first published appearance of this circuit. It was found in the September ‘35 RADIO magazine. This article described an experimental setup to test the oscillator at various plate voltage levels and circuit constants. Although not indicated, I suspect this was authored by Frank Jones, a big fan of the 6A6.
Even though the test-bed circuit was not producing a lot of output, I decided to procede with a finished build, on the basis that previous projects had always produced higher outputs than the test-bed versions. I attribute this to the test-bed's myriad of clip leads and unwanted stray coupling.
Several parts placement options were considered and eventually a suitable full-size layout was drawn so that final construction could begin.
Thanks to the earlier test-bed shakedown, the final version worked immediately with no problems.
Built on a breadboard, 'platform-style' from B.C. Douglas Fir, most of the parts are true to the era, including the three re-stuffed wax capacitors ... the only component that I had to fudge was the grid leak resistor. For this, a modern resistor was sanded and repainted using the body color codes found on resistors from the 30s.
The final version puts out a harmonic-free 4W on 20m, 1.5 W on 15m and 3/4W on 10m. This doesn’t sound like much, but for this old 1935 design and what was likely the first dual-triode manufactured (for audio not RF!), I’m pleased. Proof of concept is in the results and so far, in just a few evenings of operation, the tiny rig has not disappointed in the ‘fun’ department.
As of this writing, a half-dozen Europeans have been worked on 20m, (all on CQs), 17 states on 15m, and when the F2 MUF finally reached 10m in early September, two W4s were easily worked (NC and SC) on that band!
The increased power output was a result of a few changes in component values as well as the elimination of clip-lead stray coupling, shorter connections, better parts placement and a dedicated ground bus wire running across the bottom of the circuit. As well, the antenna coupling link seems fairly influential on output power, both in its placement and in the number of turns. As is often the case for link-coupled outputs, a 150pfd variable capacitor in series with the link produces a small output peak on each band when tuning for forward power.
The oscillator to amplifier coupling capacitor (C4) was increased from 50pf to 100pf. The grid leak (R2) was reduced from 47K to 15K. A .002mfd bypass capacitor (C6) was added to the amplifier's plate tank circuit. Some articles mentioned seeing greater output when tripling and quadrupling by introducing some regeneration in the amplifier stage. This was accomplished via a small variable capacitor (Cr) between the grid and the plate of the amplifier section. Although I did see a slight increase in the 10m output during the test-bed phase, no improvement with regeneration was noted in the final version but I may still play with this feature.
I'm very much looking forward to using this little transmitter during Cycle 25's peak winter months, especially on 15 and 10m. It doesn't take much power when these bands are at their best.
Please keep an ear out for me ... I'll be a little chirpy on ten!