Diodes instead of or across emitter resistors

[jez]

I read a letter to Electronics World magazine from Jim Bongiorno just the other day in which he claimed that crossover distortion in power amps had been cured many years ago by the use of diodes in place of the emitter resistors in the Marantz Model 15 power amp designed by Sid Smith and that he had later improved on this by using a combination of schottky diodes and resistors in a later design of his own....
In my many years of involvement in audio electronics I can't recall ever seeing this topology and a google search of the subject has come to nought!
I'm assuming that if there was really any mileage in this it would be commonplace by now!?
Anyone familiar with the idea and care to comment on it?

[Conrad Hoffman]

Not a clue if it's worth doing, but I remember my ancient RCA transistor data book from the 1970's showed diodes in one of the their little audio power amps in the applications section in the back. The circuit worked well enough, but not close to what we can do today.

[simon7000]

They blew up output transistors! Diodes have the same thermal issues as the transistors.

[DF96]

This is not the first time that Bongiorno has claimed that crossover distortion was cured years ago. If it was that simple everyone would be doing it.

To eliminate crossover distortion you need the transconductance of the output stage to be roughly constant, irrespective of signal level. This means that at the quiescent point (where both sides are conducting) the gm of each side needs to be about half of the gm of one side at full signal (when only one side is conducting). It is difficult to see how this can be achieved if the gm changes with signal, as a diode would seem to imply.

Alternatively, you run at a high quiescent current and accept central gm-doubling and switching distortion at some other signal level. Some people prefer this.

[Elvee]

Originally Posted by jez

I read a letter to Electronics World magazine from Jim Bongiorno just the other day in which he claimed that crossover distortion in power amps had been cured many years ago by the use of diodes in place of the emitter resistors in the Marantz Model 15 power amp designed by Sid Smith and that he had later improved on this by using a combination of schottky diodes and resistors in a later design of his own....
In my many years of involvement in audio electronics I can't recall ever seeing this topology and a google search of the subject has come to nought!
I'm assuming that if there was really any mileage in this it would be commonplace by now!?
Anyone familiar with the idea and care to comment on it?

Probably something like John Broskie's class AC.
Make a search with these terms, it might answer your question.

There could be other reasons: Baxandall diode fe.

Or something completely different: in the Circlophone©, resistors serve purely as current sensors, and have no effect on the transfer function.
They thus have a relatively large resistance, to maximize the sensivity, and they are shunted by schottky's at high current where they play no role, to avoid wasted power.

There might be other reasons....

[john_ellis]

My two pennorth: diodes are highly exponential just like transistor bases. Therefore, to remain even approximately linear, they need to be biased on, slightly, just the same as a transistor. If not, either they will cause a high non-linearity when they do (for the case where diodes are in parallel with the emitter resistor) or the output stage is not biased enough (in the case where only diodes are used) and the output stage will be operating in Class B as opposed to Class AB and have a high crossover distortion anyway.

Using diodes it seemed to me was a crude trick to try to control quiescent current at the expense of linearity.

[DF96]

Yes, exactly. To a first approximation the diode has the same response as the thing you are trying to correct. It is a bit like adding an extra BJT to a Darlington, except that this one has a beta of 1.

I don't do much SS now, but some years ago I had a look at p-p outputs (specifically the complementary pair). The biggest problem is thermal stability. Setting the quiescent current for low crossover is easy - you can even calculate the optimum value. The trick is keeping it correct as everything heats up.

I wrote it up for Wireless World about 10 years ago. Bongiorno wrote in and said that it was all unnecessary as the problem had been solved years earlier.

[ingenieus]

Does anybody have a schematic of the Marantz Model 15. In the meantime we can probably that the emitter resistors in question are the ones attached to the output devices. Sounds intriguing, but if it worked then why isn't everybody doing it today?

[jcx]

there are a few related techniques, it helps to know a few search terms to get started

try the forum Goolge Search button with
Tanaka, Pass;
square law output;
non switching output

New Class A, Super-A, Non-Switching : need a revival ?

low crossover-distortion class AB output stage topologies

New class A biasing (with non-switching class AB overflow)

[transistormarkj]

Conceivably he was talking about the "Baxandall diode" that was used in the 1970's, to reduce (not eliminate) crossover distortion in quasi-complementary output stages. It's D1 in the attached figure (from JLH's "The Art of Linear Electronics").