Simple Mos amp

[emanuel_hr]

Hello everyone!
Lately I got curious about the sound of simple amplifier topologies, so i decided to try one. First I wanted to build a GrandMos, but i couldn't find the VAS mosfets 2sk216/2sj 77 so I decided to modify a little the schematic to suit my needs. Since i got so absorbed by the modifications, i included another buffer stage between the vas and the input. Also the feedback was modified a la Nelson Pass, it simply gives better simulated performace and now the output bias current and offset settings are independent. The compensation was redirected in order to eliminate the RHP zero. The output devices are laterals from Exicon with around 200 mA bias per pair(400mA in total).
The simulator says it is stable with 2uF cap load parallel with 8 ohm. The phase margin it gives me it is totally absurd(and I don't know why).



Regarding simulated performance, is quite good(for such a simple amp):
THD20khz at 45Vpk output with 8ohm paralel with 1uF is 0.006%



Same conditins at 1khz gives 0.001%



With only resistive load it gives THD20k 0.0048% and THD1k 0.001%.
The interesting fact is that at 1khz it gives mainly third harmonic but at 20k second harmonic is greater(with third close by). With capacitive load the higher order harmonics get bigger(5th and 7th) as you can see in the picture above. The slew rate is very high, around 700V/us, at least that's what the simulator says.

Any advice or comments are welcome.

[AndrewT]

Hi,
does the front end, Q1 to Q6 inclusive need a well regulated quiet power supply? or is the PSRR adequate without regulation?

Is that nested feedback? R7, R8 GnfB but of the CFB style (is it equivalent to current feedback?)
and R11, R12 forming the inner FB loop?

Why the LEDs on the drivers?

The phase margin it gives me it is totally absurd

what do you mean, excessive or miniscule or ...?

[emanuel_hr]

The front end would certainly benefit from regulated supply lines. I have tested with a 100 hz 1Vpk superimposed sinewave on both +and - rails and at the output i get 3mVpk so this would give aprox. 50db PSRR(or 56dB if you consider only one power supply line). At 20k I get 4.44 mVpk, thus 47dB.
The global feedback is similar to the F5, some call it low impedance feedback, others current feedback. The leds are used to drop some voltage without being forced to use excessive degeneration.

About the phase margin, it seems way too low for the stability predicted in capacitive loads(witch is very good). I use Microcap V9 for simulation but not for a long time. If someone knows how to simulate correctly loop gain and pahse margin with this software, I would appreciate very much.