P channel mosfet amp

My design philosophy, when it comes to MOSFETs, is use N-Channel or P-Channel, one or the other, but not both in the same circuit. "Complementary" N-Channel/P-Channel pairs are a good deal less "complementary" than are NPN/PNP pairs.

I have used IRFP240/9240 in many designs and never had any problems.
They sound great.

A lot of the distortion introduced in MOSFET amps is killed by global feedback.

I have also designed quality class d amps and they dont sound any better than my class AB mosfet amps.
 
A friend of mine was low on n channel mosfets so he asked me if P channel mosfets could be used for one of my amp circuits, so this is what i came up with:

http://i.imgur.com/41nSt.jpg

I built it and it works just fine, it even simulates slightly lower THD than the lateral amp.

The IRFP9240 has a lower Gm (about half) than the "complement":rolleyes: usually employed around here, IRFP240. Yet it's still higher than the Gm of the lateral fet. This means there is less Vgs vs Id change for the Hexfet leading to a higher input Z and better phase margin. This may have something to do with the simulated distortion measurements. However if you do not intend to use them with class B bias, and I suggest if lower distortion is important that you don't, you will still need a couple hundred mA bias and thermal management. I suspect a more optimal bias for the IRFP9240 would be ~250mA. Even in quasi configuration you would still have the Gm droop problem at very low currents causing the gain of the output stage to drop in that part of the operating region leading to static crossover distortion.:yuck: Better to bias above this region. At least with quasi you don't have to worry about searching for compatable N-ch & P-ch devices of equal Gm to use as complements.:whazzat:

What I don't understand is why someone would underbias the output stage so that the non-linear effects of high frequency components from the resulting crossover distortion is fed back via a realatively slow global feedback loop to the inverting input of the IPS and integrated with the audio signal causing all kinds of nasty distortions when just a little bit more bias and idle heat dissapation from the outputs can add so much more linearity to the open loop transfer function?
For such an increase in performance adequate bias is a small sacrifice indeed.:headshot:

Want to use only 20-40mA? Choose a device with a higher transconductance such as a BJT.:)


PS: The gate resistors in your circuit appear quite large. Also the driver stage is quite undersized. I would think at least 20 - 30mA bias for the driver stage.:scratch2: