ozapaydin wrote:
In the examples I've seen, a constant current is delivered to the load connected to the circuit after Vout-ADJ (with resistor).
Bruce's approach works the same way except it's puts the load on the other side of the source. You described above, the situation in the data sheet illustrated below:
Attachment:
Current Regulator.png
But this circuit is essentially unchanged, from the load's perspective, if the load is moved to the input side of the circuit and the adjustment pin and the reference resistor are tied to ground. The only limitation is that whatever is on the input must have enough potential above ground to make the regulator work. This would be the regulator dropout voltage plus the adjustment voltage. In the case of an LM317 this is a few volts at these current levels.
A better diagram to help understand is this one:
Attachment:
LM317 Current Source.png
Here there is no ground reference. The only requirement is a minimum voltage differential between the input and the output. This is the circuit that Bruce puts in the tail of the output pair (without the variable resistor). The tube characteristics set the eventual cathode voltage based on the current drawn by the regulator. But at startup and before current starts to flow, the cathode voltage is high enough for the regulator to begin functioning. Once it does, the current rapidly approaches steady state and the cathode voltage settles where it does based on the tube characteristics.
It is actually not uncommon to form a current source in this manner. The fact that Bruce applies it to tube amps is just another application of the constant current source. The same thing can be accomplished with a simple NPN transistor, but the biasing becomes a little more complicated.
Does this make the function a little more clear.