Transistor distortion and reducer circuit

[audioguru2]

I found this info about the horrible distortion of a transistor without negative feedback.
I have seen the distortion many times on my 'scope but didn't measure how much.
Here the distortion is shown to exceed 15% and shows a way to reduce it without adding negative feedback which would also reduce the gain.  ;D

View attachment 36818

 

[Alun]

How is the diode introducing negative feedback?  ???

 

[Alun]

Either way C2 or C3 are connected to ground when the wave is both posative and negative I can't see how the diode does anything, if anything it looks like it should introduce more distortion.

 

[audioguru2]

Hi Alun,
Isn't that a neat way to cancel a transistor's distortion by making it more linear?  ;D
The transistor's gain decreases when it conducts less current because the impedance of its base-emitter junction increases. The diode in the circuit conducts more current when the transistor conducts less, and since it is in series with the emitter, the gain of the transistor increases almost as high as it is when conducting more current.

This circuit actually cancels distortion, while ordinary negative feedback only reduces distortion.
The entire article is here:
http://www.edn.com/article/CA450599.html?spacedesc=designIdeas

 

[Alun]

I see, but how do op-amps and other transistor amplifiers avoid this distortion?

Is it anything to do the the constant current sources in the current mirrors?

Or is it just brute force by using a very high gain and lots of negative feedback?

 

[audioguru2]

Hi Alun,
In an opamp the changing base-emitter impedance is minimized by using an extremely high impedance current source instead of a collector resistor so that the change of impedance of the base-emitter is negligible. A series emitter resistor is also used to minimize the change. Multiple stages like this are used and of course then the gain is very high which allows for an enormous amount of negative feedback to reduce the remaining distortion down to nearly nothing.

Oh yeah, don't forget that the differential inputs and totem-pole outputs also cancel a single transistor's non-linearity.  ;D   

 

[Kevin Weddle]

Audioguru, that diode reduces distortion by means of it's variable resistance counteracting the transfer curve of the transistor. Transistor distortion is very small and is a result of how the beta changes from let's say 150 155 160 which is good to 152 155 158 which is less good because it is not predictable. Remember that as long as you can account for the original values of the signal, your system is adequate. Just as long as you can predict how the signal is affected. There are so many things that you cannot see with the scope it is a surprise that you see distortion. I really doubt your observation on this one.

 

[Alun]

Hi Alun,
In an opamp the changing base-emitter impedance is minimized by using an extremely high impedance current source instead of a collector resistor so that the change of impedance of the base-emitter is negligible. A series emitter resistor is also used to minimize the change. Multiple stages like this are used and of course then the gain is very high which allows for an enormous amount of negative feedback to reduce the remaining distortion down to nearly nothing.

Oh yeah, don't forget that the differential inputs and totem-pole outputs also cancel a single transistor's non-linearity.  ;D   

Is this why FET amplifiers are less noisy than BJT op-amps?
Is it because FETs are voltage amplifiers not current amplifiers and they don't suffer from this non-linearity?

 

[audioguru2]

Transistor distortion is very small and is a result of how the beta changes

Kevin,
The graphs show a transistor circuit with nearly half its positive-going waveform missing. It is nowhere near full output but its distortion is an extremely high amount of 15%. Distortion might be 50% at full output. Nowadays, an amplifier with only 0.5% distortion at full output is considered to be poor.

The beta change of a transistor with changes in its collector current have nothing to do with the change of the transistor's transconductance, in fact it is the opposite and cancels some of the distortion caused by transconductance change and is included in the distortion measurement.

There are so many things that you cannot see with the scope it is a surprise that you see distortion. I really doubt your observation on this one.

The very high distortion is obvious on the graphs and I have seen it many times on my 'scope.
Of course, this is a display of a transistor's transconductance, which is with an extremely low-impedance voltage input. In practice, a transistor is usually partially driven by a current source which could be the collector resistor of a preceding transistor stage. Since it is current-driven through a resistor, its voltage gain is severely reduced.

 

[audioguru2]

Is this why FET amplifiers are less noisy than BJT op-amps?

Hi Alun,
A FET-input opamp is less noisy than a BJT-input opamp usually only when they have a very high source impedance. With a low source impedance their noise is about the same and is determined by their design.

Is it because FETs are voltage amplifiers not current amplifiers and they don't suffer from this non-linearity?

The transconductance of FETs is called "Square Law" because their voltage gain nearly quadruples with each doubling of input voltage. At medium to high output swings it results in very high distortion.
On the graph attached, the input voltage changes in 0.25V steps but the output current shows low gain with low input voltage and much higher gain with higher input voltages.  View attachment 36879