The idea is that both sides of the balanced amp (positive and negative) generate the same harmonics, so they will cancel out. The odd order ones won't, because they're shifted in phase compared to the base signal - so you'd either cancel noise and even order, or just odd order and add noise. Generally the former is better.

It's the same principle as push-pull design, but can be done without the switching and associated crossover distortion, because both ends are running all the time - and it produces less added odd order if elements are very well matched - because it's not necessarily dependent on timing. However since nothing is perfect, there will be some remaining noise and distortion.

The best way to get rid of harmonic distortion is a current mirror (preferably two of those in a balanced configuration), but again, nothing is perfect. Good opamps have these already anyway and integrated is way better than discrete at matching, which is required for that part of a circuit.

If you require math, go read a nice EE book, there are a few around. There's e.g. one presentation by Pete Millet from HeadRoom floating around on the net.
Do note that when he says "push-pull creates more odd order distortion" it's actually relative - they typically aren't too much worse than single ended at this.

The odd order ones won't, because they're shifted in phase compared to the base signal

with odd order distortion, the gain curve is "S" shaped.

Thanks guys, that makes perfectly good sense to me.

If you require math, go read a nice EE book

I teach high school math and am trying to write my doctorate in ME... I'm going to avoid as many textbooks as possible thanks   



How about Jan Meier's concept of "active balanced ground"?... http://www.meier-audio.homepage.t-online.de/grounds.htm
Would that be effective in taking out even order harmonics? Or any other advantages/disadvantages present in that configuration?

I built one myself as a sort of proof of concept thing and I thought it sounded good, but it really wasn't much more than a fancy cmoy with jiggered inputs.

The proper term for balanced as it was used for a long time before headphiles went crazy involves CMR (Common Mode Rejection). http://en.wikipedia.org/wiki/Common-mode_rejection_ratio

This is different than running separate amps and inverting signals to reduce noise for each channel and bridging their power.

Neither topology is a guarantee that one will sound better than the other.  Understanding the theory is helpful but there are tons of other factors to consider that may even be more important.  Just remember, all designs are compromises and no 'benefit' ever comes for free.


Don't forget, the O2 has very low distortion, so....

Yup, I was just pointing out how chasing lower distortion might not be the answer to everything and it may have costs when you do.

Another question on output power of balanced amps. I commonly see the points mentioned of higher voltage swing and slew rate, and higher power output as a result of the higher voltage... but what if each side of the amp is already running near its current limit? I see people saying that it doubles/quads the power output, but it should be the same shouldn't it?