About US$0.01 per watt in single-ended configuration, about the price of one TIP41.

 

When I lookup ICs at Digikey then they say that many ST Micro amplifier ICs are obsolete. Unisonic Tech and Coetec in the Orient make many of the ST Micro amplifier ICs but I don't know who sells them and who designs and who copies them.

Here are my comments on Mike's bridged amplifier:

 

You speaker is very inefficient at 86dB. You can get similar speakers that are 96dB. They would require 1/3 the power to produce the same sound level, and therefore would run 3 times as long on the same battery.

 

You speaker is very inefficient at 86dB. You can get similar speakers that are 96dB. They would require 1/3 the power to produce the same sound level, and therefore would run 3 times as long on the same battery.

Correction. . . a tenth, not a third. It would require a third of the voltage.
A 96dB@1W@1m speaker might be hard to find these days. Speakers have become less efficient over the recent years: a combination of specmanship and materials science.
New materials allow the voice coil to run hotter without failing. That gives a higher power rating. That then allows the use of smaller (cheaper) magnet, the sensitivity falls but the maximum sound output stays the same.

 

Yes. I have mainly 2N3904, 2N3905, 2N2222, TIP41, TIP42, TIP31 and TIP32 transistors as active components.

If that's what you've got then this will knock the spots of any IC amplifier design. SPICE gives the distortion at -87dB (0.004%) and a flat frequency response to 100kHz

 

386 ohm resistance is too low for my circuit? Well at least thats the sum of resistances entering the transistor bases unless I'm missing something

 

386 ohm resistance is too low for my circuit? Well at least thats the sum of resistances entering the transistor bases unless I'm missing something

Input impedance is set by R15 = 22k

 

I'm trying to figure out how the 56 ohm resistors would melt the transistors. TIP41 datasheet states maximum continuous base current is 2A. If I were to pretend the 330 ohm resistor is a short, then doing math, I get 14.4/56=257mA, I guess its because doing further math leads to 3.7W. I think they would want to fry in this case only if good heatsinks aren't attached.

 

I'm trying to figure out how the 56 ohm resistors would melt the transistors. TIP41 datasheet states maximum continuous base current is 2A. If I were to pretend the 330 ohm resistor is a short, then doing math, I get 14.4/56=257mA, I guess its because doing further math leads to 3.7W. I think they would want to fry in this case only if good heatsinks aren't attached.

It's down to the difference between the Vbe voltages and the diode forward voltages. If the diode voltages are too high, then the transistors have more Vbe which leads to more collector current and therefore more heating.
If you decrease the resistors in the bias chain (56Ω, 330Ω) the current in the diodes increases, and the voltage across them increases.
When the transistors start to get warm the collector current for a certain value of Vbe increases, leading to more warming, and lower Vbe, until a thermal runaway situation develops and the transistors fail.
It is common practice to include a small value of resistance in the transistor emitters to prevent this.

 

I said the little 2N3904 driver transistors are so overloaded that they will melt.

 

I may be able to get away with PN2222 as they are meant to handle higher current.

 

Maybe its my speaker that is causing the TIP41 transistors to be hottish.
I modified my circuit a bit so the pull-up resistors audioguru referred to have been doubled in value and I now used a 6V supply and when I did tests, the audio is clear but the TIP41 transistor was getting warm.

Im betting everything comes down to the speaker because its only 8 ohms and doing math on voltages 6 and up gives me wattages greater than 2. Maybe my bet is wrong...

 

Maybe its my speaker that is causing the TIP41 transistors to be hottish.
I modified my circuit a bit so the pull-up resistors audioguru referred to have been doubled in value and I now used a 6V supply and when I did tests, the audio is clear but the TIP41 transistor was getting warm.

Im betting everything comes down to the speaker because its only 8 ohms and doing math on voltages 6 and up gives me wattages greater than 2. Maybe my bet is wrong...

NO - its the bias - @DickCappels mentioned it as long ago as post #19

 

I noticed something else. I was playing with more values in spice and when I made the bootstrap capacitors almost non-existant (setting them from 1000uF to 1pF each) I was able to achieve higher amplitude.

 

I noticed something else. I was playing with more values in spice and when I made the bootstrap capacitors almost non-existant (setting them from 1000uF to 1pF each) I was able to achieve higher amplitude.

You will almost certainly achieve lower output without the bootstrap capacitors. By removing them, you prevent the drive voltage to the positive output transistor going above the supply voltage. If you see the opposite, it must be due to an effect somewhere else in the circuit.

 

Its probably because I'm using the 8 ohm resistor tied to the outputs to simulate a speaker and probably that 8 ohm resistor is almost pulling the outputs together. Is it a bad idea to simulate this 8 ohm resistor as a speaker in spice? I couldn't find a speaker symbol in spice.

 

Its probably because I'm using the 8 ohm resistor tied to the outputs to simulate a speaker and probably that 8 ohm resistor is almost pulling the outputs together. Is it a bad idea to simulate this 8 ohm resistor as a speaker in spice? I couldn't find a speaker symbol in spice.

A real speaker is approximately 6Ω in series with 600uH (neglecting the bass resonance, which rarely makes any difference to the performance of an amplifier)

 

I wonder if I should factor in resonance. My output waveform changed a bit after going back to the bootstrap arrangement and replacing the 8 ohm resistor with a 6 ohm resistor and 600uH inductor in series. I didn't specify any other parameters for the resistor or inductor like max wattage etc but that probably wouldn't matter with spice or would it?

 

Not usually much point. It’s not easy to do, especially with a bass-reflex cabinet which has two resonance peaks, and the output impedance of a decent amplifier is pretty much zero at the frequencies involved.

 

No, LTspice will happily pass a hundred amps through a 50 ma diode.

You can get carried away optimizing performance with LTspice. Indeed, that can be done on the workbench too. The key question is whether a change is needed to improve (in this case subjective) performance. There comes a point at which it is more efficient to make it and try it out, lest the circuit be over-designed.