开云体育

Hans, I just analyzed the amplifier circuit.

Based on my back-of-envelope simulation, most of the open-loop voltage gain is in Q14. The overall gain is approximately 5.6k * (gm of Q14) * (gm of Q507) * load resistance. Now, there are two major open-loop poles. One is 5.6k and Ciss of AOD403, which is about 3nF, plus the Miller effect of Crss of Q507 (0.6nF times the gain of that stage). This comes to a bit below 10kHz. Another pole is the output resistance of the amp against 1uF. When Q507 is on "lightly", which I assume from about 1V voltage VDS, the pole from this is also around 10kHz.

That is, the open loop Bode plot has two poles close to each other, where the open loop gain is quite high. I think this is risky in terms of stability. I think one way to increase the stability is to move one of the either pole much lower to make it a dominant pole. This circuit is a power supply, so I think the most appropriate course is to increase the 1uF output capacitor to 10uF or larger. In fact, I would look into the largest capacitance that does not interfere with the envelope modulation.

This amp has a quite large open-loop gain and gain-bandwidth product. I think a dominant pole approach will leave more than plenty of bandwidth for narrowband communication.

This circuit is low frequency basic amplifier (voltage regulator) so it is well suited for SPICE simulation.