Trading off amplifier linearity for better efficiency, why harmonic traps are needed, how push-pull architecture is used for broadband circuits, and the difficulties of its implementation at RF.
Nice! It was interesting to see this in RF, all my class B designs were for audio. In audio range it is normal to tinker with bias to have a complementary pair partially on at the crossover and then use feedback to control the harmonics. There used to be lots of arguments about whether you could hear the distortion from the varying transfer function even after feedback.
So with the balun you do not need to find complementary pair devices, making matching easier, which is nice. I suppose old vacuum tube amps did that even for audio when they did class B.
Shouldn't the push-pull output balun be split with a central tap to the power supply voltage? Your push-pull circuit connects everything to ground, no obvious source of power.
OIC, you have just left out the chokes on each side, for simplicity, referring back to see where power came from in the single-sided design. I guess that keeps the balun simple and they are likely the more expensive kind of inductor.
As I write these articles, I'm quite amazed how much can be learned from audio amplifier design techniques really, and then be applied to RF. Conceptually they are the same, but the details of implementation differ.
Yeah, I didn't want to complicate the push-pull diagram with bias and matching components. The main idea was to convey that the input half-sinusoids are amplified separately and then recombined at the output. I wanted to keep it light considering that push-pull is not used all that much in GHz-range PA applications due to issues with implementing good baluns.
Nice! It was interesting to see this in RF, all my class B designs were for audio. In audio range it is normal to tinker with bias to have a complementary pair partially on at the crossover and then use feedback to control the harmonics. There used to be lots of arguments about whether you could hear the distortion from the varying transfer function even after feedback.
So with the balun you do not need to find complementary pair devices, making matching easier, which is nice. I suppose old vacuum tube amps did that even for audio when they did class B.
Shouldn't the push-pull output balun be split with a central tap to the power supply voltage? Your push-pull circuit connects everything to ground, no obvious source of power.
OIC, you have just left out the chokes on each side, for simplicity, referring back to see where power came from in the single-sided design. I guess that keeps the balun simple and they are likely the more expensive kind of inductor.
As I write these articles, I'm quite amazed how much can be learned from audio amplifier design techniques really, and then be applied to RF. Conceptually they are the same, but the details of implementation differ.
Yeah, I didn't want to complicate the push-pull diagram with bias and matching components. The main idea was to convey that the input half-sinusoids are amplified separately and then recombined at the output. I wanted to keep it light considering that push-pull is not used all that much in GHz-range PA applications due to issues with implementing good baluns.
Thanks for your insightful comments as always!
Thank you. It is a very good explanation on amplifiers. I think you should mention that the expansion of the current is using Fourier series.