Good, you got it to work, BUT:

Hi,

Yes this design is far from optimal, i wonder how perfect he is expected to get it.

For another example, one cap (C3) and two resistors (the two you talked about R6 and R8) can be completely eliminated by connecting the base of Q3 directly to the collector of Q2, and biasing Q2 accordingly. There is no really good reason for capacitively coupling Q2 to Q3 that's only needed when the DC bias of the previous stage will overwhelm the input and/or output of the next stage. Q3 is an emitter follower so the DC bias difference is very minimal, on the order of about 0.7 volts.

 

Yes this design is far from optimal, i wonder how perfect he is expected to get it.

Hi Al,
It worked perfectly well to the specification as set out by his class tutor, not someone else's idea what the specification should be.

As we both know, when you deliberately exceed the limits of the specification of any project, at some point it will fail.

The expression Good, you got it to work, BUT: is in my honest opinion a pathic way to disparage the work of a student.

Eric

 

The student should have been taught to bias a transistor with calculations and/or a simulation, not randomly.
I reduced the overloading of R6 and R8 on Q2, reduced the resulting extra gain and re-biased Q2:
EDIT: Note that the capacitance of C3 must now be reduced to keep the 20Hz cutoff.

 

Hello again,

Here is what i meant by "you don't have to capacitively couple Q2 to Q3".
This works better also.

 

When the PNP output transistor is DC-coupled to Q2, the PNP emitter resistor forms a voltage divider with the load, reducing the maximum undistorted output level:

 

When the PNP output transistor is DC-coupled to Q2, the PNP emitter resistor forms a voltage divider with the load, reducing the maximum undistorted output level:

Hello there AG,

Here is the complete circuit I was working with.
I should point out that the choice of R5 was merely to show how a better bias point for that output would work better when direct coupled with Q3. It may have been necessary to change R7 as well or even any stage before that. So the circuit was not meant to win any awards just show how the biasing would work when using direct coupling. I'm sure you are aware that direct coupling can be used when there is no chance the DC bias point will wander too much and be able to bias the following stage adequately.

I should also add your circuit is interesting too as well as your notes and I'm sure with your experience you could improve upon this circuit quite a bit.