In theory that could work, but then 3 surface mount 1N4148 in series would do as well; there is no added value to use 3 diode-connected transistors compared to that
The thing is to have a heat transfer from heatsink (or much better: directly at the output transmitter) without a heat resistance path.
Soldering 3 devices together and then encapsulating them in a non-conducting way unto a sort of holder does mean a (significant) heat resistive path, which translates into a temperature sensing delay, which translates into a swinging bias voltage across the output devices.
Depending on the amp circuit and the way it is used (power demanded), this could result into a situation where the thermal runaway of the output transistors is faster than the bias temperature sensor (i.e. the 3 encapsulated devices) can compensate (which means....smoke). And when the power demand is over, the cool down response is delayed in the same fashion up to a point that not enough bias voltage may be supplied.
This is why a flat-faced bias transistor directly mounted under the mounting screw of one of the output device (of the associated channel) is the most proper and most common execution.
Also because only one PN junction needs to warm up / cool down, hence speeding up the compensation process with a factor equal to the bias spreader multiplication factor.
In case of 3 devices in series, in a thermal resistive path, the speed is reduced by the slowness of each PN junction to follow the temperature.
The thing is to have a heat transfer from heatsink (or much better: directly at the output transmitter) without a heat resistance path.
Soldering 3 devices together and then encapsulating them in a non-conducting way unto a sort of holder does mean a (significant) heat resistive path, which translates into a temperature sensing delay, which translates into a swinging bias voltage across the output devices.
Depending on the amp circuit and the way it is used (power demanded), this could result into a situation where the thermal runaway of the output transistors is faster than the bias temperature sensor (i.e. the 3 encapsulated devices) can compensate (which means....smoke). And when the power demand is over, the cool down response is delayed in the same fashion up to a point that not enough bias voltage may be supplied.
This is why a flat-faced bias transistor directly mounted under the mounting screw of one of the output device (of the associated channel) is the most proper and most common execution.
Also because only one PN junction needs to warm up / cool down, hence speeding up the compensation process with a factor equal to the bias spreader multiplication factor.
In case of 3 devices in series, in a thermal resistive path, the speed is reduced by the slowness of each PN junction to follow the temperature.
Last edited: