Audio amp base *current* bias

[Circlotron]

Normally an audio output stage has a temperature compensated DC Vbe applied to the bases of the BJT output transistors. As they heat up the applied Vbe is lowered to maintain more or less the same quiescent collector current. What would happen if instead we applied a small but constant *current* to the bases and let the the Vbe consequently look after itself. If this would work there would presumably be no need for thermal contact to the output transistors. Surely someone must have tried this.

[SuzyC]

Results:      Small Constant Current======>XoverDistortion   No Thermal Contact======>Thermal Runaway=====>Smoke

[srb1954]

Normally an audio output stage has a temperature compensated DC Vbe applied to the bases of the BJT output transistors. As they heat up the applied Vbe is lowered to maintain more or less the same quiescent collector current. What would happen if instead we applied a small but constant *current* to the bases and let the the Vbe consequently look after itself. If this would work there would presumably be no need for thermal contact to the output transistors. Surely someone must have tried this.

Transistor current gain H_FE tends to increase with temperature so, even if you supply a constant base current, the collector and emitter currents will still increase with temperature. The consequent increase in power dissipation and junction temperature effectively creates a positive feedback loop and it could lead to a thermal runaway situation.

[Kokoriantz]

This type of biasing does exist in class AB as class A with JLH1969. The increase of bias due to Hfe is positive feedback with a gain less than unity so it stabilizes itself. This type does not require emitter resistors.
Search on DIYAUDIO website for examples.