Quuestion about signal dissection

[rtarbell]

Question:

I have two identicle amplifiers. I want to take a signal and "chop it in half"; that is to say that I want all of the positive part of the signal to go to one amplifier, and all of the negative part of the signal to go to the other amplifier. Their outputs will be summed up to drive a simple speaker.

My idea here is to design an output stage using all NPN or N-channel devices for better device matching without using the quasi-complementary approach. I know that tube amps use phase splitters or transformers with center taps, but I'm trying to get a completely DC coupled amp from first to last stage.

I'm in a brain lock as to how to design such a device...

[Eva]

That's not practical.

[lndm]

Impractical but not impossible, but I'm not going to be the first to prove it .

If you don't want complimentary symmetry, and you don't want quasi complimentary, why not try class A?

[Workhorse]

Originally posted by Eva
That's not practical.

No EVA, Its practical in his "IMAGINATION"



Cheers,
K a n w a r

[ifrythings]

Well I (and a few others I believe) don't understand why you'd want to do this????

A diode will clip any positive or negative singal above .6V if placed correctly... but I don't think you'd want to use that aproche.

P.S. I think all your going to do is make the worst cross-over distorstion amp ever, but if you really are bored prove me wrong.

oh, good luck keeping you wave semectrical across two amps.

[lndm]

Perhaps you might read Douglas Self's words. I believe he understands distortions produced in amps, particularly along the lines of your concerns, very well indeed. He might put your concerns to rest.

http://www.dself.dsl.pipex.com/ampins/dipa/dipa.htm

If not, Nelson Pass has written about a cascode power stage which I guess you may find interesting.

[ilimzn]

Thinking about a dual coil speaker? Alternatively, you could use a transformer to 'combine' the halves, or, a circlotron circuit (one of the configurations used to do what you want with vacuum tubes).

[lumanauw]

MusicalFidelity A1 seems to be 2 amps, one for "push" and one for "pull". But they are not all Nch though

[traderbam]

Hi Lumanauw,
Well I personnally like this kind of lateral thinking...too many ideas around here are deep inside "the box".

The perennial problem is developing a bidirectional current with uni-directional devices. The tube fraternity have solved this a million times over, so why not borrow some of their ideas?

Single-ended always works well but is wasteful of power. Circlotron has come up with some natty ways to minimize wasted power by using inductors as current sources.

If you want push-pull and don't want a transformer I know no satisfactory method that avoids p channel devices. That doesn't mean you can't invent one!

[jan.didden]

Such a design has been published in WW years ago, I think by Sandman. Used current driven diodes (to overcome the 0.6V threshold) to route the pos polarity to the pos half of the amp, and vice versa. I'll see if i can find it.

Anyway, you would still need two complete amps. The amp that is NOT amplifying the signal still has to sink the speaker current ...

Jan Didden