Cool mosfet can be a little tougher depends.
Circuit is adaptable to mosfets no prob.
It will basically include the bootstrap and current sources as suggested as well.
Circuit is adaptable to mosfets no prob.
It will basically include the bootstrap and current sources as suggested as well.
Why not try a 'push pull' output stage? like so:
You can increase efficiency as well as using only N channel mosfets using this approach 👍
You can increase efficiency as well as using only N channel mosfets using this approach 👍
you reorganised the current source elements and made the thing push pull, how much power at idle does it draw?
Before jumping on the "your hearing must be deceiving you" bandwagon, are you absolutely certain that all of the distortion has improved when HD goes down?
The trouble with push-pull output stages is that they're only balanced @ 0V out, and unbalanced at every other voltage.
A quick look with a 2-tone thought experiment, shows that the harmonics of a high frequency riding on top of a low, modulator frequency, will undergo amplitude modulation, as one half of the amplifier more or less cancels out the distortion of the other half, depending on the phase of the modulator. Who cares if the long-term DC offset is in the uV range, when the dominant source of modulation are typical ±2V bass tones?
If even harmonics H2, H4, etc., are perfectly cancelled in the middle, this is a worst-case scenario because they reappear dynamically. Whenever real-world bass frequencies are closest to 90° or 270° phase, they push higher frequencies off-centre, so modulation is maximised.
I have a working hypothesis that slight 'detuning' by using simpler, "less perfect" circuits helps reduce the dB level of amplitude modulation, and that this is subjectively more important, even if HD is increased. IMO, the endless fuss about H2 vs H3 actually boils down to how well or badly the evens (H2 etc.) are cancelled out.
In a single-ended design, there is of course going to be modulation, but there's next-to-no chance of glitches or "s-bend" type reversals in the transfer function. Unless the earlier stages try to be too clever with a push-pull VAS or balanced LTPs (that lose their fragile balance when the common-mode voltage follows the music) or something like that.
Maybe I'm arrogant, but to me this seems like a big insight, that could save someone from going down a rabbit-hole of overly-complicated design, only to be disappointed in the horrible state of humanity when other people add a 0.1% THD tube frontend to their 0.001%THD masterpiece.
very late reply but i realised i could use the 3rd transistor the way they use it in the jlh1969 and have some kind of an feedback loop to reduce distortion.
R6 is probably way to high.
I also saw a lot of people using bjt current sources and fet drivers.
It is to high, I suspect the high frequency performance and square wave response is poor with such a high value.
