The One That Started it All
The Cavalli-Jones headphone amplifier came in a series of steps that started when I built my Fender Tone Stackheadphone project. This amp was an experiment where I wanted to try out a few things, including the White Cathode Follower as an OTL tube output stage. I did some research on the WCF and discovered some really good information on John Broskie's brilliant Tube Cad Journal.
After completing the amp I asked Chu Moy of Headwize if he would be interested in an article on this amp. Chu expressed interest and we began to work on the article. Chu noticed that the WCF output stage in my amp was the same as the output stage in the Morgan Jones amp that had already been written up in a other Headwize article. Chu asked me if the Broskie optimization (see the Design section) that I had used would also apply to the MJ amp. Of course it did and so Chu further asked me to optimize the MJ for better performance. Which we did.
The amp with these changes became known as the Optimized MJ. Several headwizers built it and got good results.
As I began to collaborate with other headwizers on WCF output stage amplifiers it became clear that in order to handle headphones ranging from 32Ω to 300Ω it was necessary to apply different amounts of NFB. To accomplish this I introduced a shunt feedback network into the Optimized MJ amplifier and then made the NFB adjustable using a pot connected as a rheostat. This amp is called the Cavalli-Jones Amplifier. Many headwizers have built the Cavalli-Jones with good success.
The Original Morgan Jones Earmax
Morgan Jones, author of the book Valve Amplifiers, reverse engineered a small, portable tube headphone amplifier called the Earmax. Several individuals used Morgan's schematic to build their own headphone amplifiers and, eventually, the idea made its way to Headwize. The original Morgan Jones Earmax schematic had these valus:
In the reverse engineering process Morgan was not attempting to improve on the Earmax, but there is a significant flaw in the Earmax design. This flaw is in the value of the plate resitor for the WCF output stage. It is too big.
The Optimized Morgan Jones Earmax
In order for the WCF to optimally deliver current to the headphones the two triodes need to be operating in a balanced push-pull mode. This enables them to work in opposite phase, each drawing the same magnitude of current thus doubling the current delivery to the load. John Broskie, in a very early article from The Tube Cad Jounal, did some experiments with the WCF and determined that the best value for the plate resistor is Rp = 1/Gm, where Gm is the transconductance of the triode. Originally John asserted that this result is good for any value of the load resistor. After some prompting from Morgan Jones I actually calculated the value of Rp as a function of the load. You can read this article here. I mentioned my results to Broskie who then re-did his calculations to get the correct result.
Optimizing the push-pull action of the WCF leads to the values shown in this new schematic. I made a few other changes to get better operating points for the Optimized MJ amplifier.
The Cavalli-Jones Amplifier
The final step in the evolution of the amp was the realization that it could benefit from some amount of negative feedback. Headphones are a punishingly low impedance for an OTL tube topology. Negative feedback reduces the distortion and lowers the output impedance of the amplifier. But, as I began to study this problem, I realized that no single value of NFB would work equally well for low Z and high Z headphones. The simple solution was to add a variable feedback mechanism controlled using another potentiometer. The schematic for this change is:
The Cavalli-Jones is a simple amplifier that provides good performance and the ability to adjust its behavior depending on the headphones being used. Quite a few Cavalli-Jones amplifiers have been built since the optimizations were published, but this basic tube headphone amplifier has been replaced by better ones.