Diabolical Artificer Wrote:BTW,how did you workout the phase shift?
Well, the input cathode follower presents a low impedance at its cathode. So the upper end of the 22k resistor connecting to cathode is effectively grounded.
At low frequencies, within the Baxandall network, the 680pF capacitors and the 0.01μF capacitor can be considered open-circuit.
So, the 12AU7 cathode (pin 8) couples via 0.22μF to a resistance of 144kΩ (22k + 100k pot'r + 22k). This is a single-pole roll-off, so it will give phase shift varying from zero at high frequencies to 90° at low frequencies.
From the 144kΩ resistance, there is a signal tapped off to the 12AX7 grid. But, this goes through another LF roll-off because the 0.1μF capacitor and the 1MΩ resistor form another high-pass filter. The same happens again.
The 12AX7 inverts (so 180° phase shift) and the 12AU7 cathode follower just passes this in faithfully.
There is not enough phase shift to cause oscillation because the gain around the whole loop is zero as the overall phase shift becomes 360°. But it takes only a few degrees of extra shift (such as cathode bypass roll-off) and the thing CAN become unstable because the aforementioned phase shift needs perhaps only to be 355°, at which point the loop gain will definitely not be zero!
So, it might be better to dump the 0.22μF capacitor from the 12AU7 pin 8 (just have a direct connection) and put it to give DC blocking from the INPUT cathode follower instead.
