In the Douglas Self Graphs (Small Signal Audio Designs, I'm referring particularly to figure 3.3, 3.6, 3.8, 3.11, but i believe also in any other books), the THD is a percentage based on a value of V input level.

Of course i understand the different amount of distorsion based on increasing Rloads but I can’t get why, in general, the distorsion is depending from the incoming Vrms level.

This is a very basic question but your hint will help me to unstuck.

Follows an example graph.
If you can tell me what I must double check in order to understand the concept, I will be happy.

 

Note to anyone else reading this: to save you having to find a copy of Small Signal Audio Design, the TS is referring to a simple NPN emitter follower with a load resistor.

It is because the current through a transistor changes, and that requires a change in Vbe.
Current is exponentially related to Vbe, so the function is non-linear.

 

Note to anyone else reading this: to save you having to find a copy of Small Signal Audio Design, the TS is referring to a simple NPN emitter follower with a load resistor.

It is because the current through a transistor changes, and that requires a change in Vbe.
Current is exponentially related to Vbe, so the function is non-linear.

That explanation is correct. That is the reason for using negative feedback, because, at least in theory, it cancels out the distortion caused by non-linear amplification.

 

I can’t get why, in general, the distorsion is depending from the incoming Vrms level.

Transistors have a non-linear change in current versus input voltage.
At small signal levels you are operating over a small portion of the non-linearity which generates only a small distortion.
The larger the signal, the more the non-linearity has an effect on the perceived distortion.

Below is the LTspice simulation of a simple one BJT amplifier for two input voltages with no negative feedback:
Note the noticeable distortion on the top side of the larger yellow output waveform.

As seen in the table below, the total harmonic distortion for the small signal is <2% whereas the large signal distortion is nearly 10% (largely due to the increase in the 2nd harmonic).

 

Distortion comes from one or both of two sources, either non-linearity or else clipping, which is a special case of non-linearity.
Adequate amounts of negative feedback can reduce non-linear distortion quite a bit, and selecting components that are more linear also helps. Selecting the operating condition for most linear operation also helps.

 

Using a second non-linear component in such a way that the non-linearities cancel out is another technique (see "long tailed pair")
Although clipping is another form of non-linearity, negative feedback will make it worse, not better.

 

Using a second non-linear component in such a way that the non-linearities cancel out is another technique (see "long tailed pair")
Although clipping is another form of non-linearity, negative feedback will make it worse, not better.

When negative feedback is applied over an entire amplifier it works rather well. Usually that long-tailpair is found in just the inverter section of an amplifier. And clipping can be restricted to operation areas outside the amplifier's intended range.
The reason distortion increases with amplitude is that the effect of any non-linear action increases in a non-linear manner as the amplitude increases. That should be obvious by the definition of non-linearity. The only exception I can think of is an amplifier operating in the "class C" mode, which is well into cutoff and often into saturation, and this relies on resonant circuits to restore a sine wave output.

 

I was totally neglecting that a change in voltage doesn't linearly correspond to a change in current.
I need to triple check some fundamentals.
Negative feedback as an expedient is described later in the quoted book of Douglas Self.
Your help is always precious and I'm thankful for the time you spend to help people.
J