why will my amplifier design fail?

[kleinPE]

Hallo everyone,

yesterday i came up with an idea for a tiny 50-100 w amp, but i think it is too simple to be as good as my brain thinks. I know the transistors are probably not the best suited and the capacitor should be bigger , but i only used them as placeholders in a long and failed effort to simulate the distortion.



My first plan was to get arduino or raspberry pi with oscilloscope. measure the output and correct the input in software to get a perfekt output (kind of the electric equivalent to a servo-corrected subwoofer) but then i realized an OpAmp will let me do that in hardware.

in my theory the upper transistor will push the output Voltage up if it is below 40x input and the lower transistor will suck the outputvoltage down if it is above 40x input.

I know there is the possibility of both transistors working at the same time , but that should be preventable with some XOR/NAND logic and a trimmpot instead of R2 to make really sure both OpAmps will act with the same gain. But i am not an engineer, I am just 4 month into my apprenticeship as an electrician and dont even know the difference between different transistor types.

So any insight you can provide is very welcome.

[TimFox]

Why are R7 and R6 so high?  10 megohm?

[Kleinstein]

The circuit has a problem, as both OPs try to control the output votlage. Ideally there could be a values of R2 so that both would try to get the same result. Hower in real world there is an offset error at the OP and the 2 OPs are aiming for slightly different targets, even of only a minute difference.
This would by not a big problem if the upper OP wants the voltage to be lower and the lower OP wants the voltage to be higher:  the upper OP can only control the positive current and the low lower OP only the negative current. So in this case one Side would be off and one OP as active at a time.
There is still a smaller problem with the limited base to emiter voltage (max some 6 V for most transistors) so this may stress the transistor base quite a bit.

The more problemativ case is when the upper OP wants the voltage higher than the low one. Than both OP would fight each other, sending current from both sides. With the high gain of the OPs this could be up to the limits set by the OPs maximum ouput current.

On transients there is a chance that the second case may happen even of normally set for the first case. In principle the circuit is similar to a class B amplifier.

[kleinPE]

@TimFox:
I just thought : there is no signal going through them , so lets make them as high as possible without completely isolating that ground connection.
what value would you have chosen ? for me this is all learning by doing.

@Kleinstein:
Danke, i always thought the base of a transistor is almost isolated from C and E and just acts as a switch. thats why i put +-48V around the transistors.
Is there maybe another kind of semiconductor, that is more suitable ? it is no problem if that would mean huge heatsinks . this is going to be for a loudspeaker that will be taken on construction sites where an additional PC-fan wont matter as long as it does not burst in to flames. but there will also be raspberry pi as bluetoothreciever/ audioinput, so it is trivial to include a shutoff function for high temperatures.

[CaptDon]

Fails for a bunch of reasons. How do you control the idle current for the outputs to be AB1, AB2 or B? Second, as Tim mentioned 10 megaohms in series with your supplies??? How will the output transistors draw any current? Or did you mean 10 milliohms??? Not a likely choice either. Their was an amplifier circuit similar to your idea years ago. It used a now obsolete un-obtainable opamp that could handle something like +/- 50vdc on the supply rails, maybe more. It was in a little T-05 multilead can. They were sort of a 1 I.C. 2 transistor solution to a 40 watt audio amplifier.I wish I could remember the number LM540? ua540? It has been far to many years since I saw the circuit, but I do run across it in one of my design books from time to time and think "What a useful I.C., why did they stop making it?" Would have been useful for servo amps and all kinds of stuff.
B.T.W. is your supply for your I.C. +/- 48vdc?? Good luck with that.

[Kleinstein]

Audio amplifiers are really standard cicuits. There is no needs to invent a new cicuit for a long solved problem. If sound quality is no a high priority, one could use just 1 OP to drive both transistors. This would give a slightly poor sound quality from cross over (a bit like a LM358), but no problems with both sides fightting each other.

Normally just get an audio amplifier chip or a ready made amplifier. Most amplifier chips include protection measures that would complicate a slf made circuit quite a bit.
Other wise look up a classic discrete audio amplifier. The high voltage makes it a bit cumbersome to use OPs.

For some simulators 10 M would stand for 10 milli-Ohms and this would be no problem

[kleinPE]

@Kleinstein @CaptDon
Thanks a lot guys for enlightening me. 98% of my knowledge about electronics comes from watching EEVblog, BigClive, and a bit Mr.Carlson.
The 10M resistors and the ground connection next to the transistors is only there, because i could not figure out how to place a single supply with +48 on the top and -48 on the bottom in Qucs.

Soundquality is actually very important. we already have a speaker( a dumpsterdive ) that can go loud and if you dont know the tracks it sounds ok , but whenever i play sth. that i enjoyed at home i am dissapointed. on friday i found a nice suitable speaker kit but could not find an amplifier kit with 50-100 Watts, that i can trust. but after i wrote that sentence i checked again and find a lot.

I didn´t know the data sheets include reference designs and partslists for everything i need . so i thought an oldschool design with discreet transistors would be easiest.
but texas instruments has enough info that i will leave out the passive crossover and use an active one for almost the same price.

but thanks again for your input. without you i probably would have screwed around with my idea for another 2 or 3 days.