Distorsion mechanisms

[cuelist]

Here's something for the experts...

A differential input pair, if done properly, reduces 2nd order harmonic distorsion generated by the input circuit compared to a single transistor solution. In the single transistor input configuration, the 2nd order harmonic distorsion dominates.

It seems to be an established opinion (fact?) that 2nd order harmonic distorsion is more "musical" than 3rd order (or any odd order for that matter).
Now, so I wonder; with a differential input pair having 'done away' with 2nd order distorsion, thus revealing the 3rd order, is that really what you would want from a subjective point of view. Does that really sound better? The grand total % of distorsion goes down, but to what purpose?

And - does the 2nd order distorsion effectively "mask" the 3rd order component in a single transistor input?

NOTE: this disregards the other stages of an amp that may well be more of an issue from a distorsion point of view.

 

[Guest]

[quote author="cuelist"]A differential input pair, if done properly, reduces 2nd order harmonic distorsion generated by the input circuit compared to a single transistor solution. [/quote]

I can not imaginate why?
Odd harmonics wanishes if output signal is symetric respectively
to time axe.
It is property of push-pull circuit, H-bridges, etc.
Maybe symetrical differential pair with symetrical input and symetrical
output can have this behavior but differential pair working with
asymetric input signal behaves like one transistor.

Properties of differential pair depends on working schematics.

xvlk

 

[cuelist]

[quote author="xvlk"]I can not imaginate why?[/quote]

Perhaps it is hard to image but it seems to be an established fact. There is a very good explanation of this in one of Self's books. I can find it and quote from it later but it is largely due to the push-pull nature of an input differential keeping the transistors more firmly in the linear region (less gm variation).

 

[bcarso]

Fully diff drive is not required for the reduction in second. Good discussions can be found, one in particular in Feucht's book.

As to the question of third masked by second, that's one for your ears I think. I like clean, and I like avoiding the (almost) inevitable d.c. shift with circuits that produce 2nd, while conjecturing that this oomph sound is what some people need to hear to make things seem like hell's foundations are quivering.

Rich May (just retired, ex- of many companies including Sumo and Miller & Kreisel) really hates second, and thinks clean third is usually o.k. But the discussions in the Boyk/Sussman paper I posted a link to a while ago and subsequent experiments by Larrchild indicate that such nonlinearites in the presence of feedback lead to some really obnoxious high-order IM stuff that isn't there otherwise. I think this consideration is far more important when you are processing complex textures, as we are most of the time.

I noticed Erno Borbely recently joined the forum. Perhaps we can get him to comment? Erno? :grin:

 

[cuelist]

Quoting from Self's "Self on Audio" book, page 139:

"The use of a single input transistor sometimes seems attractive where the amplifier is capacitor-coupled or has a separate DC servo; it at least promises strict economy. However, the snag is that this singleton configuration has no way to cancel the second-harmonics generated by its strongly curved exponential Vin/Iout characteristic."

This is part of a long chapter, no way to reproduce it all here.

 

[Guest]

[quote author="cuelist"]However, the snag is that this singleton configuration has no way to cancel the second-harmonics generated by its strongly curved exponential Vin/Iout characteristic."
[/quote]
Why? The some linearizing diodes which are used in OTA controlled gain
amplifiers can works with one transistor.

Yes, stability, blablabla, but DC servo works here also.

Good noise figure is everything. Someone can give here THD of ear.
some 2 percent at 80 dB?
Fiction? there are psychoacoustic methods how to measure it.
xvlk

 

[cuelist]

[quote author="xvlk"]Yes, stability, blablabla, but DC servo works here also. Good noise figure is everything. Someone can give here THD of ear. some 2 percent at 80 dB? Fiction? there are psychoacoustic methods how to measure it.[/quote]

:shock: What are you talking about?? Did you even bother to read the initial posting from me?