Building a "Jones Push-Pull Oscillator"

6A6 oscillator

The Search

Frank Jones (W6AJF) was a prolific west coast writer and radio experimenter. As author of several editions of 'The Radio Handbook', his name was one that most active hams in the '30s and later would have readily recognized. My own 1936 edition of the Jones Handbook is where I initially saw an intriguing circuit. It was one of many oscillators appearing in the handbook and although never named as such, it has come to be recognized by radio amateurs as the 'Jones Push-Pull Oscillator'. Its low-parts count, along with its pleasing symmetry, gives the circuit a unique and simple elegance ... one which I had always found interesting as a possible future project.

The circuit used a low-power 6A6 dual triode configured as a simple push-pull crystal oscillator to produce a few watts of RF. Earlier this winter I spent some time searching for a project that would inspire me enough to enjoy the radio-workshop during the upcoming cold, wet and dreary days ahead. I had another long look at the Jones Oscillator but eventually decided that the few watts produced by the 6A6 would likely not result in a transmitter that I would use very often. Anything less than typical QRP power levels can often make enjoyable contacts few and far between ... but the basic circuit still intrigued me!

When it comes to a favorite project building period, my heart lies in the 30s, particularly during the mid-30s transition between wooden breadboards and metal chassis. There was a brief period of time when these two materials were combined and, when transmitters in particular, where built on an aluminum plate which was then mounted on an attractive wooden 'breadboard' base. I imagine that the affordability of a single small aluminum plate versus a pricey manufactured steel chassis, was the motivation for this short-lived construction style. The finished results can often be quite attractive, having a somewhat 'Art-Deco' appearance ... but construction in this style is not an easy task and was probably one of the main reasons that it didn't last very long!

                   MOPA transmitter                    tri-tet

I had used this style of construction to build my Tri-Tet-Ten transmitter as well as an early '30s MOPA transmitter and although laborious, had been very pleased with the final results. Perhaps I could find inspiration for a Jones Oscillator of some form if combined with my favorite construction style ... and so the search began!

A search of my ARRL handbooks from the 30s revealed nothing up until the 1937 edition, even though pentodes and tetrodes had been around for many years. As well, crystals were still not widely used in the hungry 30's, and were probably not too affordable for most amateurs. This edition contained a generic description of a push-pull pentode crystal oscillator, similar to the Jones triode version ... but no projects using the circuit were described.

1937 Handbook

Having exhausted all of my own references, I turned to the 'American Radio History' website and began searching the pages of 'Radio', a monthly amateur radio magazine published in San Francisco and one to which Jones was a frequent contributor. It was the December '36 edition that grabbed my immediate attention and enthusiasm ... it looked like my search might be over!

Dec '36 Radio

The article described a simple, low-parts count, push-pull crystal-controlled power oscillator using a pair of 6L6s. Interestingly, either by oversight or by design, there was no author attribution to the article but I strongly suspect it was penned by Jones himself as the same design appeared the following year in Jones's own periodical (and now rare), 'The Radio Amateur Newcomer'. The circuit's mid-30's lineage would provide the perfect opportunity for me to utilize my favorite construction method once again!


***** Above images courtesy of SM7EQL *****

RCA had just introduced the 6L6 in July '36, for use in linear amplifier audio stages, and amateurs had quickly discovered its excellent capabilities when used in both oscillator and amplifier service at RF frequencies. The tube turned out to be an immediate winner and for the next two decades would be a popular choice by both amateurs and equipment manufacturers for use in audio and RF service. The 6L6's little brother (or sister?) and one of my favorite tubes, the 6V6, was introduced in December of the same year. Either tube would make a good candidate for use in the Jones transmitter.

                                    6V6                                     6L6


I usually begin any project construction by 'breadboarding' a quick lash-up of the circuit to check its performance. At this stage I'm also able to easily swap components and optimize values to learn a bit more about what makes the circuit tick.

                                    breadboard                  breadboard

I'll often make permanent changes in part values or even add or remove components where necessary. The original circuit as published in 'Radio' is shown below:

Radio article

After poking and prodding the circuit for several days, I ended up making some subtle mods.

1. The cathode bias resistor (200 ohms) was reduced to 100 ohms. Note that the original 'RADIO' schematic (above) has incorrectly shown the screen resistor as '100K'. This should be '10K'.

2. C3 was removed from the tank capacitor so that one side could be directly grounded, making for easier construction. There was no detectable difference in performance when making this change. As well, swapping the plate tank's dual capacitor tuning scheme for a single capacitor strapped across the entire plate coil worked just as well as the published design but I decided not to make the change which would have required floating the capacitor above ground and having high voltage on both stator and rotor plates. There is nothing that can be done about high voltage appearing on the tank coil, so one must always be aware to keep hands and fingers away from the plate tank when operating!

3. Reference is made for the need of a parasitic choke in one of the two plate leads in order to eliminate possible instability, not uncommon in early-run 6L6s. No instability was observed in my breadboarded model or in the final version, so RFC2 was eliminated. It seems possible that the original (1936) version of the 6L6 had been improved upon over the years so that my 1950's-manufactured tubes had a different internal architecture making parasitic generation less likely to occur.

4. With a link-coupled output, it's important to tune the link with a series-connected capacitor so that the transmitter can be coupled more effectively to a 50 ohm antenna system. Typically when adjusting the capacitor, the output along with the plate current will show a sharp rise when the optimum condition is reached. Tuning the link can often result in a doubling of output power.

In this particular circuit, the output coupling is quite important and if a point of ‘over-coupling is reached (too many turns on the link or link being too tightly coupled to the tank coil), the oscillator keying will become sluggish or even drop out of oscillation altogether. For my normal 3 or 4 turn links, I've found about 200pf is needed for 40m, while the 80m link requires an additional 700pf. I use an outboard 250pf variable mounted on a small base which allows me to use it with all of my vintage transmitters. When the extra capacitance is needed for 80m, it's a quick job to clip or solder the extra caps across the variable. The capacitor also acts as a simple antenna loading control, similar to what is found in a pi-network output, allowing more control of the final output power and load on the tube. Loading too heavily can result in loss of oscillation in these types of circuits.

                                    loading capacitor                  tank coil

A few days later and now satisfied with my changes, I clipped the 40m antenna on to the test bed and sent out a short CQ on 7121 just before sunset. I was hoping that perhaps someone nearby would copy and give me a keying report. I nearly fell off the chair when my CQ was answered by N2BE in New Jersey! I was even more astounded when he gave me a 589 report. With the test bed lash-up barely clamped in the bench vise along with a half dozen clip leads dangling in every direction, the 'magic' of radio was in full force ... from west coast to east coast on just a handful of shaky parts!


After gathering all of the needed components from my junk box, several parts placement options were considered and eventually modelled.

Parts Layout

Once finalized, a parts-placement plan was sketched along with a wiring plan. By choice, only a few of the prominent leads would be showing on the top surface while everything else went below.

Parts Layout

All of the parts are true to the era, including re-stuffed wax capacitors ... the only component that I had to fudge was the screen resistor. For this, a modern resistor was sanded and repainted using the body color codes found in resistors from the mid-30s.

                                stuffed caps                         antiqued resistor

A suitable breadboard base was made by edge-gluing two lengths of Philippine Mahogany that I had been storing for several years. Once dry, the laminated board was thickness-planed to 1/2" and then edge-routed. The routing and sanding were then followed by staining along with two coats of polyurethane satin spray.

                                breadboard                         breadboard

Once the plan was finalized and parts were measured, a drilling plan was drawn up. This was done on the aluminum plate which was covered with masking tape to protect it from scuffing.

The plate was then taped to the breadboard and all the needed holes were drilled through the plate and baseboard. Some of the baseboard holes were blind-tapped for 6-32 threads to avoid the clutter and appearance of nuts if the holes were through-drilled.

                                drilling                 plate

Once done, wiring the circuit began. Building in this style is tedious as each topside connection requires masking around the area to avoid scratching or splattering the aluminum with solder or flux. All wiring was salvaged from old radios, many that were (now regretfully) stripped when I was a teenager. Eventually the soldering was finished and the 'Jones' was ready for testing!


Thanks to the pre-built test bed, the finished circuit worked immediately and I was pleased with the overall outcome as it matched what I had visualized in my mind from the initial concept of a '36-style Jones Oscillator.



I originally had fears of cracking crystals with this circuit, especially when testing with 6L6s, and for that reason I used out of band crystals that would not be missed should they be damaged. I was surprised to see that the crystal current appeared to be very low as there is no obvious sign of stress or crystal heating such as frequency drift or sudden frequency jumps. Even a few tiny HC-49 crystals worked fine without cracking, although they were pretty chirpy.

Should you also build a Jones Oscillator, be sure to ground the metal shell of the 6L6s or 6V6s as this is reported to aid in reducing crystal current. It is also important to tune the output tank on the higher side of resonance (less C) as crystal current increases as you tune towards the lower side of resonance. Output will steadily increase as tank capacity is increased but so will crystal current. A point will be quickly reached, shortly after maximum output, where oscillation ceases altogether. There is a definite learning curve for optimum tuning when seeking best output versus the best CW note, typical of most grid-plate power oscillators.

As mentioned earlier, the 6V6 also works well in this circuit and although the power output is lower, keying is slightly better, no doubt because of lower crystal current.


Here are the two tubes compared:

6V6 vs 6L6

As is typical with crystal power oscillators, plate efficiency is rather dismal and I'm not sure how it could be improved. Perhaps, as in the original article, the 6L6s would be happier at 475-500VDC along with increased screen potential but I don't think there is any way that this circuit would meet the author's claim of 50W out. Measuring power accurately was not as easy for amateurs in the 30s as it is now and it was often just a case of guessing by seeing how bright a light bulb would glow. I suspect 50W of RF might light things a little differently than 50W of 60Hz AC. If the circuit did produce 50W, I suspect the keying, by today's standards, would be poor. On the other hand, maybe there's still plenty of room left for experimentation with this circuit.

I'm pleased with the overall outcome and operation of this project. Immediately upon completion, the 'Jones' was put to the test in the annual 'Novice Rig Roundup'. Over 50 contacts were completed, including many transcontinental QSOs with east coast stations on both 40 and 80m. One of my original goals has been met in that Jones's circuit readily provides enough useful output power to make good solid contacts.

Thanks Frank ... it's a keeper!