Here the situation is different at turn-on. The 1µF capacitor must be charged up mostly from through the diode and the 800K resistor. While charging up, the positive input of the op-amplifier will be more negative than the negative input and thus the op-amplifier's output will swing negatively to the end of its potential. As the tube heats up and the capacitor charges up the op-amplifier's output will move to the correct value to maintain the desired idle current. But once the capacitor's charge exceeds the voltage across the 200K resistor, the diode will stop conducting and drop out of the circuit's operation.

This idea has not been tested and is offered as only a thought experiment. Last month we saw how an LM337 makes an excellent current source. This month, we'll look in to using it in place of the voltage reference and op-amp in the previous circuit, as it contains a precise, low noise voltage reference and an op-amp. The first mental hurdle is realizing that the output of the regulator can be grounded and its input can be used as the output. The adjustment pin senses the deviation from its internal voltage reference and controls the amount of current flow from the input pin to the output pin to compensate.  By placing a resistance in series with the input pin the variation in current will define a variation voltage, a voltage that can be used to bias the tube's grid. The big problem is how to prevent the music from throwing the bias off. The solution is to shunt the AC through a capacitor, which would also serve to stabilize the new output of the this auto-bias circuit. n