Simple class A amplifier how it works

[Tony Stewart]

Here I show 1mV @ 10kHz input how crossover distortion is avoided, https://tinyurl.com/y2zbh33e

 

[audioguru]

I do not think the high current opamp is fast enough to avoid 10kHz crossover distortion.
The Falstad simulation shows crossover distortion on the input of the circuit, not on the output.

 

[rjenkinsgb]

Simple scaleable class AB

Two points:
By definition that is class B, not AB; it has no standing current and both transistors are off passing through zero volts, not both on for a small range around zero as with class AB.

And: the simulator you are using is simply not accurate.
The same circuit simulated in LTSpice, using a similar high voltage, high speed opamp (LTC6090) with 50mA output capability.

Green trace input, blue trace opamp out, red trace load voltage:

Full screen grab:


ltspice file attached.

 

[Nigel Goodwin]

I'm sure you all can back up your opinions. I know I can https://tinyurl.com/yxmqb3gm Show me the excess power dissipation and crossover distortion with 100k=Aol and 33=Acl

Except you're playing computer games, not doing electronics, and simulators aren't the real world.

Power dissipation shouldn't be a problem, because it's pure class-B - but crossover distortion will be for that very same reason.

 

[rjenkinsgb]

Just out of curiosity, I've modified the circuit in to a very crude Class AB bootstrapped amp.
It's then got no visible distortion at all; you can see in the transistor current waveforms that both conduct when the output is near 0V, so there is a smooth changeover between the two.

And the sim file:

 

[Tony Stewart]

I agree all simulations are as imperfect as their assumptions. The Falstad Sim does not modulate hFE or GBW and these change with power levels. Although THD on fundamentals above 5kHz are almost undetectable by humans. I admit oversight on not calling it Class B and rJenksin circuit will certain perform class AB and eliminates the risetime demands to drive class B. However it is flawed in that cannot produce more output current than hFE Ib or about 10 Watts in this example

Yet the crossover demands can be met easily @ 1kHz and no one can hear any distortion of 10kHz. So it is an amusing simple solution for some, even in the real-world but not an optimum solution.

 

[Nigel Goodwin]

I agree all simulations are as imperfect as their assumptions. The Falstad Sim does not modulate hFE or GBW and these change with power levels. Although THD on fundamentals above 5kHz are almost undetectable by humans. I admit oversight on not calling it Class B and rJenksin circuit will certain perform class AB and eliminates the risetime demands to drive class B. However it is flawed in that cannot produce more output current than hFE Ib or about 10 Watts in this example

Yet the crossover demands can be met easily @ 1kHz and no one can hear any distortion of 10kHz. So it is an amusing simple solution for some, even in the real-world but not an optimum solution.

Considering how easily cross over distortion can be heard (it's VERY audible, and the main reason for the poor performance of very early transistor amps), I'm dubious it wouldn't be audible.

 

[Tony Stewart]

Some data is need ed to support opinions of what is audible.
Crossover error is reduced by 2MHz GBW of Op Amp and gain is 100k or 100dB minimum, and thus inaudible @ 1kHz is far << 1%
Yet significant if distorted at 10kHz producing 20kHz and up is inaudible.

 

[Tony Stewart]

The real flaw in this design is the limit of Veb of 6V so cutoff protection is needed so it would fail for high voltage

 

[rjenkinsgb]

The real flaw in this design is the limit of Veb of 6V

????
The emitters are tied together and can never be even one volt different from the bases in your version, or a couple of volts or so in the biassed version; I do not understand the comment.

 

[Tony Stewart]

Oops , Looking at wrong circuit, please ignore that..

Did you see why your AB circuit drives from the pullup resistor will be current-starved due to reduced base current near peak voltage, which is why traditional circuits use alternate configurations that are more complex.

Yet for some applications, my simple design with high feedback gain works not bad.

 

[rjenkinsgb]

Did you see why your AB circuit drives from the pullup resistor will be current-starved due to reduced base current near peak voltage,

No; that's the reason for the bootstrap capacitors. They cancel out the current reduction that occurs without them.

The caps cause the resistor junctions to track roughly constant voltages above and below the base drive voltage, so giving an near constant current effect at all but very low frequencies.

 

[Nigel Goodwin]

Oops , Looking at wrong circuit, please ignore that..

Did you see why your AB circuit drives from the pullup resistor will be current-starved due to reduced base current near peak voltage, which is why traditional circuits use alternate configurations that are more complex.

As rjenkinsgb has already pointed out, his circuit is bootstrapped - a VERY, VERY common configuration in many (most?) amplifiers, the alternative would be a constant current source. It's basically amplifier design 101.

I was quite amused by the fact that both halves are bootstrapped, which isn't somehting you commonly see.

 

[Tony Stewart]

Unfortunately even if the caps were 1mF with 1K and less than 20V drop or < 20mA base current and hFE=100, the emitters cannot produce more than 2A or 32W peak into 8 Ohms. Using Sil+ Sch Diodes DC output stage current may be excessive (0.5V/0.47=1A).

I think the direct drive to Class B is more effective.

 

[Tony Stewart]

Here's my simulation of your AB circuit with an input switch https://tinyurl.com/yxqtqwvk and sliders for Vin and f

 

[rjenkinsgb]

I did say it was a "very crude" design, it was just a mod to show the change in crossover distortion & class AB current changeover between the transistors.

Neither version is practical for serious use at any real power level; much higher output stage or device gain would be needed, plus temperature compensation etc.

 

[Tony Stewart]

PS... there is a bit of an "easter egg" in the schematic.... something that probably would let out the "magic smoke" in a real build of this design... anybody who can tell me what it is gets a "Wile E. Coyote Sooooper Genius" award....

The (Vled + Vbe )/0.1 Ohm flaming high bias current.

 

[unclejed613]

I do not think the high current opamp is fast enough to avoid 10kHz crossover distortion.

part of the problem with the causes of crossover distortion is that any op amp has a limited slew rate, so the transition between feeding base current to one transistor to feeding base current into the other transistor will take a finite amount of time, and that transition will still provide a small amount of time where neither output device is conducting... it gets worse if the positive and negative-going slew rates of the op amp aren't the same.... which happens with a LOT of op amps

 

[unclejed613]

The (Vled + Vbe )/0.1 Ohm flaming high bias current.

as mentioned in a previous post... Q7 limits the current to a safe level... the use of global feedback keeps the output node at 1/2 Vcc, and so the current source isn't outputting ALL of the current it's been set to deliver... i also already covered what i thought might be a problem with Q4 and Q5, and why i had originally thought they might be the "easter egg" i was referring to.... Wile E. Coyote remains in his cave working on a new Acme contraption...