So I tried my take on this project. Analog Lab Class B Amplifier in the Lessons in Electric Circuits textbook.

I used this schematic:

It was a total flop. I believe 8Ω x ß of 10 is 80Ω which way too low for U1 to drive. Makes me wonder if the original circuit worked. The TL082 may have more drive with ±12V. To get my version to work I'll need super gain transistors such as Darlington or Sziklia pair.

 

What are the white triangles next to R5 and R7?

 

What are the white triangles next to R5 and R7?

That is circuit ground. Not really ground, but the 0V reference point for the circuit.

 

AKA virtual ground.Its drive is whatever the second op amp can drive (That was not a complete schematic.)

 

Was it a flop as in you didn't get much power out of it, or did it not make sound at all?

If it's any consolation, Alpine had some excellent class A/B automotive amplifiers back in the day and they very frequently used a triple-darlington setup. Phoenix Gold also made stellar amplifiers and they used triple darlington on their M44 and M50 models. I'm sure there were others, but point being there must be a good reason these highly respected manufacturers went the triple darlington route. This makes me think the project you linked is just intended to make some sound, but not with any real power.

 

I believe 8Ω x ß of 10 is 80Ω which way too low for U1 to drive.

A ß of 10 is for full saturation of the transistor as a switch.
For linear operation the AC ß of the transistor is used (typically in the range of 75-300).

 

Even with a gain of 100 the loading would be 800Ω still kinda low and it could be much lower. Like I said, I built this and it didn't work. The transistors don't have to be in the feedback loop. For now I'll focus on the transistor Sziklia pair. End goal is to fully drive the output with out distortion. At 12V this would have been around 10.5V P-P. withe the Sziklia pair it would be around 12-1.2 = 10.8V P-P max. I'm using 8Ω resistors for now and watching the output with an oscope.

 

Post your actual circuit with part values.

 

I did except for the speakers which is a large wattage resistor. My big interest for this project was crossover distortion, at lower value sin waves which I did see at at. I was kind of surprised.

 

Even with a gain of 100 the loading would be 800Ω still kinda low and it could be much lower. Like I said, I built this and it didn't work. The transistors don't have to be in the feedback loop. For now I'll focus on the transistor Sziklia pair. End goal is to fully drive the output with out distortion. At 12V this would have been around 10.5V P-P. withe the Sziklia pair it would be around 12-1.2 = 10.8V P-P max. I'm using 8Ω resistors for now and watching the output with an oscope.

It's more about current being under the current limit than load impedance.
10.5V p/p is 5.25V peak, which is 650mA into 8Ω, at which point you should have a typical gain of about 90, so 7.2mA from the op-amp, which is well within the capability of a TL081.
What does the output waveform look like?

 

Sziklia pairs can be confusing to work with so I label The equivalent transistor EBC. My latest brain fart:

 

Below is the LTspice sim of your circuit modified:

• I didn't understand the output configuration, and couldn't get it to work, so I used the common two diode bias to prevent crossover distortion
• I didn't have models for the parts you used, so substituted for those.
• I adjusted R5 and R7 to give a gain for the maximum output with a 500mV RMS signal.

The output is shown just below clipping with nearly 0.9W power to the load.

 

So I tried my take on this project. Analog Lab Class B Amplifier in the Lessons in Electric Circuits textbook.

I used this schematic:
View attachment 322254
It was a total flop. I believe 8Ω x ß of 10 is 80Ω which way too low for U1 to drive. Makes me wonder if the original circuit worked. The TL082 may have more drive with ±12V. To get my version to work I'll need super gain transistors such as Darlington or Sziklia pair.

The amplifier has two issues I see at first glance:
1) U1b inverting input should be biased at 1/2Vcc and not at GND as it is.

2) R3 must be much lower than 10k ohms to supply gate drive to Q2. The downside is this current subtracts from the drive current from U1b to drive the gate of Q1.

Issue 1 can be taken care of by running 18k ohms from U1b pin 5 to Vcc. Issue 2 is a problem as mentioned above and this limits the power this amplifier can produce.

Given proper drive from the Op amp the TL082 can drive 10 mA and the TIP42 current gain is 50 at 500 mA. The circuit should be able to drive 500 mA peak into an 8 ohm speaker and this is 1W RMS. T


TIP41 datasheet https://www.onsemi.com/pdf/datasheet/tip41c-d.pdf

TL082 datasheet https://www.ti.com/lit/ds/symlink/tl082-n.pdf

 

Which amp are you talking about? Hint use the quote feature, all of your points are addressed in my schematic.

 

Crutschow's approach is in the right direction, but I think the uncontrolled gain of TWO transistors (kinda beta-squared) for each polarity is a bit high to get the crossover distortion down where you want it (and I like that LTSpice will compute that for you, even if it doesn't always get it right). Both of you might want to look up the term "Vbe multiplier" to replace the two diodes as well, that way you can play with the fractional # of diodes you want to get the crossover distortion down...and you DO know that network needs to be thermally coupled to the output devices right?

 

With a 12V supply and a TL081 opamp, you will be lucky to get 8V p-p or 4V peak. Then the additional loss of 0.7V peak from a Sziklai pair, the output to an 8 ohm speaker is only about 3.3V peak which is only 0.68W if the transistors have not less than "typical" specs.

 

Crutschow's approach is in the right direction, but I think the uncontrolled gain of TWO transistors (kinda beta-squared) for each polarity is a bit high to get the crossover distortion down where you want it (and I like that LTSpice will compute that for you, even if it doesn't always get it right).

On the contrary, the more gain the more feedback the less distortion; but the open loop voltage gain depends on the transconductance of the output transistors, which is Ic/26mV, not Hfe. Hfe only allows you to calculate the amount of current the op-amp has to supply.
If the open loop gain is high enough and extends to high enough frequencies then the circuit can compensate for under-bias.

Both of you might want to look up the term "Vbe multiplier" to replace the two diodes as well, that way you can play with the fractional # of diodes you want to get the crossover distortion down...and you DO know that network needs to be thermally coupled to the output devices right?

I suspect there are a few senior members of the forum contributing to this thread who don't take kindly to being accused of not knowing about Vbe multipliers and thermal coupling, and perhaps you should have mentioned it in less confrontational language.

 

you DO know that network needs to be thermally coupled to the output devices right?

Yes

but I think the uncontrolled gain of TWO transistors (kinda beta-squared) for each polarity is a bit high to get the crossover distortion down where you want it

And YOU apparently do not know that the current gain of the transistors has little to do with crossover distortion.