Push-Pull Amplifier vs Push-Me Pull-You Octave

[thehallofshields]

I get the basic idea of Push-Pull amplifiers and crossover distortion, but I don't understand it how the Escobedo Push-Me Pull-You does its rectification.





Is it working as a phase-splitter, and transistor is acting a diode? Does it amplify? How is the base biased?
Can anyone lend me a helping... brain.. so that I can better understand this?

[whoisalhedges]

I'm no expert, (and PLEASE correct me if I'm wrong!) but:

Don't get hung up on the name. It's from Doctor Doolittle, it has nothing to do with P-P amplification.
Q2 and Q3 are biased at the voltage at Q1's collector, plus the voltage coming off the power rail via the 10k and 500k resistors to their bases.
The rectification comes via negative feedback - it comes off the collector of Q2 to its base via the 500k; and from the emitter to base on Q3... those two signals are 180 degrees from each other. Smoosh them together (technical term) at the junction of both emitters, and blammo! You have a doubled frequency.

The pros can probably explain it better, but if you're really waiting with bated breath to hear something, hopefully this helps a little.

[anotherjim]

I think it's full wave rectifier style frequency doubling.

At the zero crossing, Q2 & Q3 are both partly on via the 500k resistors, but the output from the lower 10k will be near ground.

As the input from Q1 rises, Q2 turns on more while Q3 turns off. Output V rises via Q1 and follows the positive signal half cycle.

As signal falls below the zero crossing, going negative, Q3 turns on more while Q2 starts to turn off. This has the same effect as it did for the positive, the output V rises, but for the negative going half cycle.

Both Q2 & 3 are isolated from Q1 by capacitors, so Q1 only provides pre-amplification.

[PRR]

> basic idea of Push-Pull
> Escobedo Pushme Pullyou

Forget push-pull. The pushme-pullyou turns all push-pull thinking on its head.

Study the Cathodyne. (Named for use in tubes, but can be done with transistors.)

In conventional form you bias the grid/base near 1/4 of supply and get two push-pull outputs.

Move the grid/base slowly higher. At half supply the phase-split action quits because plate and cathode slam into each other. Go further, and both cathode and plate move together in-phase.

Now do it again but with a P-type device and all polarity reversed.

Now connect both forms together.

> lend me a helping... brain..

No. It took me a half-hour to get a glimmer, I hurt too much to sit and type my thoughts. I'm also observing that "NO" other circuit uses this topology. My advice is to build it and enjoy, or not.

FWIW: pencil says the lower 10K sits at 4V and the bottom of the upper 10K sits at 5V. Each base sits 0.6V up from emitter.

Gain of unity, except when it is less. I think for very small signals, the two cathodynes cancel.

[thehallofshields]

Gain of unity, except when it is less. I think for very small signals, the two cathodynes cancel.

Just like the transistor-as-phase-splitter used in the Green Ringer.

No. It took me a half-hour to get a glimmer, I hurt too much to sit and type my thoughts. I'm also observing that "NO" other circuit uses this topology. My advice is to build it and enjoy, or not.

Thanks for lending some brain power Paul. I don't know what I'd do without it sometimes.

I've got this on my breadboard, and it does its job pretty well, but I was hoping there was some untapped potential here.

[PRR]

At low levels it gives a remarkably smooth-looking "sine" at twice the input freq. At 1KHz in, but the output analyzed as a 2KHz wave, it is 9% THD, nearly all 2nd (4KHz). The 2KHz is 20dB up on the 1KHz output, and the strong overtones are spaced 2KHz, so it "sounds like 2KHz". In this realm the "gain" is like 0.7.

Basically the zero crossing, ideally a hard corner, is smoothed by transistor lack-of-gain at low current. That's exponential, not sine, but close enough to trick you.

You could probably do the same with a CT transformer and two diodes biased to the edge of conduction. I think that is a Scrambler; the same but plain cathodyne instead of transformer is another well-known octave-up.

At 500mV input it is getting a little sharp, grating. Above 1V input it gets wonky and shows other-phase break-through as is expected in an over-driven cathodyne.

[whoisalhedges]

Mind = blown... I totally forgot that Q3 is PNP - so yeah, the signal is coming off both collectors. So my guess was totally wrong (not surprising) and cathodyne inversion never crossed my mind.

It's kind of fun to be in this place where I've read a *lot* and have a bunch of information banging around in my brain, but without context for most of it; but every now and then somebody puts 2 and 2 together for me and I have a "eureka" moment. Right now I have a reasonable grasp on BJT amplification and am working on FETs... I was hoping on moving on to modulation next; but maybe I should look to valves. The operating conditions are different for different devices, but the theory.... Maybe I should buy a broken amp, discharge the filter caps, and dig in.

I guess the point is (it's not to hijack the thread, I swear I have a point!) that this kind of knowledge and experience really builds on itself. You learn 10 facts, but don't necessarily know how to put them together; but then you learn something else - and all of a sudden, it all comes into view. In my very short time on this forum, I've seen this over and over again: the theory brought into practice in a way that can never be accomplished by memorizing any number of applications of Ohm's law (though most of us do need to bone up on theory more, it can only help).

So, awesome. Time to learn more about cathodynes - so far, I'm only familiar with them as phase inverters (which yeah, is also what's going on here) for some old Orange amps - and I wanna say Fender Princetons, too, but I might be misremembering....

[thehallofshields]

I'm still messing with this thing.
As expected, the signal pulled from the + junction is identical from the signal from the -.

Is it theoretically possible to get any gain out of this stage, or is it limited by its topology?

[R.G.]

It's limited by its topology, at least as far as getting a prominent octave. The two facing transistors have an emitter and a collector tied together. So at the output, one has a gain of unity (i.e. it's an emitter follower) and the other is a common emitter stage, forced to a gain of one because it has equal collector and emitter resistors. You could get more gain from the collector side, but you'd have to muck up the resistor ratios to get it, and the common collector/emitter follower would have a gain of one no matter what you did.

So it's limited by the topology.

Notice that there are two places to take off the octave? And both have the same issues/performance/gain, just opposite polarities of their signals.

[thehallofshields]

I threw a 10k pot between the positive path and the PNP/Q3.
It works like the Scramblers' Texture control. -Not as smooth of a taper, but workable.

[PRR]

> possible to get any gain

Gain is cheap. Add another stage where you like.