very "textbook" single ended Class A amplifier i designed

Simply build the thing and test the mentioned step by step. It takes a bit of effort, but it is audio engineering, diy, because these steps have different influences on the sound - which cannot be determined by peek-measurement methods;-)
 
You need to write down your own specification first.

I assume you want the amp to be a simple Class A like you have been experimenting with.

What is the amp to be used with? Speaker or headphones?

If it is to drive a speaker how much power do you want? That will determine the supply voltage.

If a speaker would you consider Class AB which runs much cooler?
 
One more resistor and one TO-92. Problem with the LM317 is the dropout voltage. It needs that to stay in constant current, too. With a 15 volt supply you can ill afford to lose another two volts. Losing one vbe plus Rds(on)*Idq is a little more palatable. And if the FETs are already on hand and the 317 isn’t, it a no brainer which to use,
 
May I suggest a circuit like:

1728849276738.png

Source: https://www.diyaudio.com/community/threads/hybrid-zen-amplifier-lm317-efficient-and-simple.402477/
 
That’s not it. You’ve added a mild amount of negative feedback, that’s all. Boot strap goes to the collector load of Q1, and you need to direct couple into the gate. Once you do that, adding NFB from output to the base of Q1 might be a good idea, to bring the gain back down and linearize it.
1728851821461.png

is that what you meant by "direct couple into the gate"? it seems to have the reverse efect
 
  • Like
Reactions: U102324
i first one to simulate a thing that works the best and then make it in real life.

It's a trap. How do these different types of distortion compare in real life?:

1) A 'raw' BC546 gain stage with no constant current source, with high voltage swings so it has lots of modulation, connected to a MOSFET follower.

2) With the BC546 connected to the lower MOSFET (common source), so its collector voltage modulation is 20-30x less, but/and the MOSFET is now a gain stage. Feedback could be used to reduce the total gain, or

3) connect the BC546 as a follower. Something also has to be done with the upper MOSFET. It could be reconfigured as a CCS and it's probably easier to use a P-channel for that.

More ideas:
Could a PTC thermistor be used in a resistor divider network controlling the gate voltage of the MOSFET CCS? So if the MOSFET gets hot, reduced gate voltage compensates the reduced on-resistance of the IRF.
 
is that what you meant by "direct couple into the gate"? it seems to have the reverse efect
No, what I meant was what you did in #34. Then split R8, and boot strap back to the output. That will give you the largest possible output of anything you’ve tried so far, and all off a single 15 volt supply. To get the most gate drive possible, make the upper half a diode (1N4148). That could however, add audible artifacts that two resistors won’t, and a maximum drain current of say 4A (2X IdQ) you only need about 3 or 4 volts above the source to fully turn the FET on. At higher currents you might want more, but not here.
 
No, what I meant was what you did in #34. Then split R8, and boot strap back to the output. That will give you the largest possible output of anything you’ve tried so far, and all off a single 15 volt supply. To get the most gate drive possible, make the upper half a diode (1N4148). That could however, add audible artifacts that two resistors won’t, and a maximum drain current of say 4A (2X IdQ) you only need about 3 or 4 volts above the source to fully turn the FET on. At higher currents you might want more, but not here.
sorry for being mean/gen but it almost works
1728862764483.png