I'm currently designing a power amplifier with a complementary feedback pair output stage. I have found that I get much lower crossover distortion if I connect the emitters of the driver transistors directly to the amplifier output (as shown in the circuit on the left in the attached PDF), rather than the more conventional approach of connecting them to the emitters of the output transistors (as shown in the circuit on the right).
Can anyone think of any reason why connecting the driver transistor emitters directly to the output might be a bad idea? Obviously it means that the output transistor emitter resistors are now inside the local NFB loop, but I can't see why this might be a problem.
Can anyone think of any reason why connecting the driver transistor emitters directly to the output might be a bad idea? Obviously it means that the output transistor emitter resistors are now inside the local NFB loop, but I can't see why this might be a problem.
You may have been using the wrong quiescent current for the proper CFP output. If you believe in 'first watt Class A' (i.e. high quiescent current) then CFP is not for you.
More importantly; in the picture on the right, R2 and R3 provide thermal stability.
In the picture on the left, R2 and R3 don't do anything useful and the amp will probably suffer from thermal runaway (i.e. the output transistors get burned out).
In the picture on the left, R2 and R3 don't do anything useful and the amp will probably suffer from thermal runaway (i.e. the output transistors get burned out).
Godfrey,
without driver emitter resistors, would the drivers over-bias on rising temperatures and they would "burn out" sooner, rather than the outputs?
I think that is why with the CFP, Vbe Multiplier does not monitor the output device temperature and instead reads driver temperatures and pre-drivers, if fitted.
without driver emitter resistors, would the drivers over-bias on rising temperatures and they would "burn out" sooner, rather than the outputs?
I think that is why with the CFP, Vbe Multiplier does not monitor the output device temperature and instead reads driver temperatures and pre-drivers, if fitted.
The drivers can be biased high to get the first or few Watts into ClassA operation.
Is it Roender (and others) that uses drivers biased @ ~100mA
There are other means in the full design, which I have not included in the diagram for simplicity's sake, that prevent thermal runaway and maintain constant quiescent voltage across the emitter resistors.
I should add that I want to keep the output stage firmly in class B to avoid gm doubling effects.
Please explain.
ClassB for an audio amplifier, surely not. (unless your explanation turns out to be Self's definition).
If the drivers have no emitter resistors then how do you
My biasing circuit monitors temperature in both the driver and the output transistors, combining them such that the quiescent voltage across the emitter resistors remains within a few tens of microvolts regardless of device temperature. I'm therefore not too concerned about thermal issues.
As for the Class B versus Class AB debate, I'll avoid entering that here. Suffice to say that class B is what this particular design requires.
As for the Class B versus Class AB debate, I'll avoid entering that here. Suffice to say that class B is what this particular design requires.
I guess if he says it works, then it works. Asking us to identify possible problems when we can't see the circuit seems a bit of a dead end though.
Apologies for the confusion, AndrewT - I am referring to R2 and R3, which are, of course, Q2 and Q4's collector resistors (I'm too used to talking about emitter resistors in emitter-follower designs).
As I say, "class B versus class AB" is another subject altogether and one for another thread. My question is solely to do with where the driver emitters are connected.
As I say, "class B versus class AB" is another subject altogether and one for another thread. My question is solely to do with where the driver emitters are connected.
With the drivers connected directly to the load as shown in the ckt on the left, isn't it just another case of 'current dumping' by the output pair ?
If the drivers are conducting and the OP is off then as the current rises the OP can turn on and dump current in the load. Exactly how many lower power chip amps are made to handle larger currents than their standard rating.
The quiescent current at which the driver and op operate is very important.
If they are all in classB mode then the driver in the circuit on the left should turn on very fast and cause the OP also to turn on fast. In the circuit on the right it appears that this probably wouldn't happen the same way and also generate more crossover distortion.
If the drivers are conducting and the OP is off then as the current rises the OP can turn on and dump current in the load. Exactly how many lower power chip amps are made to handle larger currents than their standard rating.
The quiescent current at which the driver and op operate is very important.
If they are all in classB mode then the driver in the circuit on the left should turn on very fast and cause the OP also to turn on fast. In the circuit on the right it appears that this probably wouldn't happen the same way and also generate more crossover distortion.
- Status
- Not open for further replies.