Very low distortion, error corrected LM3886 amplifier.

[temperance]

It has to do with the phase of the 2nd harmonic.

Are you aware the phase of those harmonics are phsae locked with the fundamental? If you don't believe me, try squaring a sine wave. Or clip a waveform with two diodes and look at the result with an oscilloscope and question what you are seeing.

Attached is an LT spice simulation. Run the sumulation and press ctrl L at the end. It  pops up a window showing you the results of an FFT. Just ignore the 4...9 harmonic because this is just noise.

Here is the FFT for the non linear transform:

Harmonic   Frequency    Fourier    Normalized    Phase     Normalized
 Number      [Hz]      Component    Component   [degree]   Phase [deg]
    1       1.000e+03    1.150e+00    1.000e+00      90.00°       0.00°
    2       2.000e+03    1.998e-01    1.738e-01     180.00°      90.00°
    3       3.000e+03    4.989e-02    4.340e-02     -90.02°    -180.02°

And here is the FFT for the diode clipped waveform:
Harmonic   Frequency    Fourier    Normalized    Phase     Normalized
 Number      [Hz]      Component    Component   [degree]   Phase [deg]
    1       1.000e+03    7.833e-01    1.000e+00      90.00°       0.00°
    2       2.000e+03    1.956e-05    2.496e-05     100.90°      10.91°
    3       3.000e+03    1.025e-01    1.308e-01      90.01°       0.01°

No matter how many positive/negative supplies a power amp has or not has or how many times you invert the phases with transformers, the phase relation between the signals stays the same. Imagine what would happen if this would not be the case.

[Kokoriantz]

A non linear amplifier can have a variable gain, either increasing or decreasing with the voltage or the current.

[temperance]

A non linear amplifier can have a variable gain, either increasing or decreasing with the voltage or the current.

That's why negative feedback was invented...

[Kokoriantz]

Thanks to NFB, the amount of non linearity becomes adjustable. If the output Z depends upon the feedback,  the non linear element will modulate it.

[temperance]

This isn't going anywhere. Some beating around the bush.

[Andy Chee]

It has to do with the phase of the 2nd harmonic. The voice goes backwards or forward.

Please show a simulation of;

a) in-phase 2nd harmonic, and
b) out-of-phase 2nd harmonic.

Even better, demonstrate this with real hardware measurements.

[Kokoriantz]

At your service.

[temperance]

Those are just two weirdly drawn buffers and they have nothing to do with what you stated earlier.

you need to inverse the input phase with a transformer and reverse the speaker wires.

[Kokoriantz]

Simulate it on LT one of the buffers with inverting and non inverting inputs and outputs.
Below is Marantz 9, this option implemented.

[temperance]

That's a switch to invert the phase. Your buffer does not invert.

[Kokoriantz]

A decade ago I have experimented with negative impedance load to unload the main power amp. It is a universal external element that can be hooked on the output and double the power.
https://www.diyaudio.com/community/threads/30watt-class-a-phoenician-amplifier.302190/
Here, I even made two lm3886 running in classA for 8 ohms.
As the added impedance act in current mode, maybe the voltage controler with keep the distortion very low all doubling the power at 4 ohms, each amp loading half of it.
I will see if such extra lm3886 can make the amp run in +/-35v with 4 ohms.

[Kokoriantz]

This is a negative 8 ohms. If 8 ohms load is used, the amp gets unloaded so no thermal feedback. If running without load, the amp will go hot. To allow sweet tweeter sound, I limited the frequency to 1khz to get loaded when used with 8 ohms. It can also be switched on in full range mode, only when 4 ohms load implied.
The scope shows 50hz 30v applied on 4 ohms to be about 8A and the generator current and the -8 ohms is about 4A each.

[temperance]

When claims are being challenged, you just change the subject and trow some nonsense into the thread.

[Kokoriantz]

Simulate it on LT one of the buffers with inverting and non inverting inputs and outputs.

[temperance]

Why should I waste time on those ridiculous claims?

Reply #81 shows to NON inverting buffers. The NPN version distorts the top of sine wave while the PNP version distorts to bottom of the sine wave. I don't need LT spice to see that.
Reply #83, the Marantz shows a switch at the input stage where you change the phase. Note that phase of the harmonics with respect to the fundamental does not change by flipping the switch. Flipping the speaker polarity as you stated earlier brings you back where you started.

You still didn't answer the question about the phase of the harmonics and instead of answering that question you just change the subject. The same goes for the LM3886 error correction. Why don't you show a working proto type and some measurements? 

Reply #83: again some meaningless simulation.

Maybe someone else is interested in you nonsense.

[Kokoriantz]

You have a non linear amplifier that amplifies more going negativly than going positively. You want it to reverse it to amplify more positively and less negatively. If you inverse the input polarity, the positive going becomes negative, applied to the amp, it will amplify more and the negative becoming positive it will amplify less. You invert it again to get the right phase but now the more amplified is positive and less is the negative.

[Kokoriantz]

Only for your eyes.
Read the phase of the 2nd harmonic. One is about 0 the other about -180.

[temperance]

Only for your eyes.

My eyes spotted something for your brain:

or you don't know how this simulator works or you are cheating. Just look at the amplitude of the fundamental in the FFT. Do you see what is wrong? Even I can see it while I don't know this simulator.

The sampling start time must have something to do with it. Change it to at least ten times the RC time of the output coupling capacitors and show us the result.

[Kokoriantz]

The fundamental in both is about 1v same phase.
Do your own simulations.

[temperance]

Try this. No phase difference to be seen between vout1 and vout2.

Harmonic   Frequency    Fourier    Normalized    Phase     Normalized
 Number      [Hz]      Component    Component   [degree]   Phase [deg]
    1       1.000e+03    9.362e-01    1.000e+00      89.91°       0.00°
    2       2.000e+03    3.212e-02    3.431e-02     179.76°      89.85°
    3       3.000e+03    1.920e-02    2.051e-02      89.77°      -0.14°
    4       4.000e+03    1.190e-02    1.272e-02      -0.27°     -90.18°
    5       5.000e+03    7.235e-03    7.728e-03     -90.35°    -180.27°
    6       6.000e+03    4.216e-03    4.504e-03     179.51°      89.59°
    7       7.000e+03    2.378e-03    2.540e-03      89.22°      -0.69°
    8       8.000e+03    1.064e-03    1.137e-03      -0.33°     -90.24°
    9       9.000e+03    2.220e-04    2.371e-04     -86.45°    -176.36°
Partial Harmonic Distortion: 4.298026%
Total Harmonic Distortion:   4.298576%

Fourier components of V(out2)
N-Period=1
DC component:-0.023504

Harmonic   Frequency    Fourier    Normalized    Phase     Normalized
 Number      [Hz]      Component    Component   [degree]   Phase [deg]
    1       1.000e+03    9.368e-01    1.000e+00      89.91°       0.00°
    2       2.000e+03    3.153e-02    3.365e-02      -0.27°     -90.18°
    3       3.000e+03    1.874e-02    2.001e-02      89.77°      -0.14°
    4       4.000e+03    1.164e-02    1.242e-02     179.84°      89.92°
    5       5.000e+03    7.110e-03    7.590e-03     -90.34°    -180.25°
    6       6.000e+03    4.155e-03    4.436e-03      -0.70°     -90.62°
    7       7.000e+03    2.342e-03    2.500e-03      89.18°      -0.73°
    8       8.000e+03    1.066e-03    1.138e-03    -179.87°    -269.78°
    9       9.000e+03    2.665e-04    2.845e-04     -86.94°    -176.86°
Partial Harmonic Distortion: 4.209819%
Total Harmonic Distortion:   4.210347%

[Kokoriantz]

Use eyeglasses. 2nd harmonics are 180° shifted.

[temperance]

I didn't see that. But anyhow, does that make all the difference? As an example I increased the distortion on purpose. See the attached image. They sound the same.

[Kokoriantz]

Here how the negative Z works.
The -Z is adjusted to be -8 ohms, it totally unloads the amplifier to reduce dramatically the distortion. With 4 ohms, it will double the power with the performance of 8 ohms.
The Thiel and zobel can be used for the stability of the -Z.
If you find this circuit non sense as #88, please, don't spit in the soup.