Biasing output transistors with constant base CURRENT

[Circlotron]

So we all know that you can't bias your output transistors with a constant voltage because as they heat up the Vbe drops and the transistors turn on harder and pull more current. Not stable with temperature changes. What though if we apply a small but constant CURRENT to the bases of the output devices? Changes in Vbe with temperature would then have no effect. Does the current gain vary much with temperature though? Maybe that's not the right word – we are not dealing with changes in collector current vs changes in base current – just a fixed base current vs changes in die temperature. Any opinions?

[sofaspud]

BJT's are current-controlled devices, i.e. there is a fixed relationship between Ib and Ic.

[DF96]

BJT collector current very accurately follows the exponential of base-emitter voltage, over many decades of current (but with a temperature-dependent scaling). BJT collector current is very roughly proportional to base current; the ratio varies being generally lower at very high and very low currents.

[Bonsai]

I think voltage biasing is the best solution! better to rather spend some efort on improving the temperature stability, or on developing some form of error correction circuit to counteract distortion due to temperature induced bias errors.

[godfrey]

Originally Posted by Circlotron

Does the current gain vary much with temperature though?

Yes. That's normally shown on the datasheet. e.g. See the pic below. Current bias of the output stage is still workable though. For example the 1969 JLH class A amp uses it.

[sofaspud]

Current bias of the output stage is still workable though.

Constant base current bias?

[Circlotron]

Originally Posted by godfrey

For example the 1969 JLH class A amp uses it.

I was primarily interested in class B application, but thanks for the graphs.

[Elvee]

The Circlophone falls more or less into that category.

[Circlotron]

So it does! Hey, that's pretty good.