'Tis for a guitar amp. (I have never designed, nor built a power amp. In the words of Jeremy Clarkson: how hard can it be?).
Specifications:
Power rails would be either +/-35V (ish) or single-supply 70V (ish), which should allow for a 40W-ish clean power (taking the MOSFET drop into account).
My idea:
- use a single-ended input stage employing current feedback (as in the following examples: https://sound-au.com/project217.htm https://sound-au.com/project12.htm either with an output cap, DC adjustment or better yet, a DC servo)
- use a push-pull cascode VAS with dynamic gate bias as in Slone's OptiMOS amp for its soft clipping characteristics (right there in the middle)
- with such a VAS, possibly capable of beefier current than the usual, do without MOSFET drivers.
MOSFETs would probably be the Exicon devices (rated 125W).
Could it work? Replicating the OptiMOS design for a guitar amp is extravagantly overkill, and I don't feel I need all the symmetry or complexity. Also, 2nd harmonic distortion is not a big issue. If it's rich in low-order distortion, that's actually the design goal.
No, I don't want to go with a soft clipper (whatever it may be) before the power amp, because the response would be quite different, and most of all, it would be too easy 🤣 . The clipping should be generated internally by the high output impedance power amp with all its interactions with the speaker.
Specifications:
- at least 20W@8ohm, <3%
- soft knee clipping
- high output impedance (or mixed mode feedback, possibly)
- use of lateral MOSFETs for ease of biasing ('tis my first amp, after all)
- no tertiary power supply for the input stage and VAS (for simplicity and to actually keep power levels low)
Power rails would be either +/-35V (ish) or single-supply 70V (ish), which should allow for a 40W-ish clean power (taking the MOSFET drop into account).
My idea:
- use a single-ended input stage employing current feedback (as in the following examples: https://sound-au.com/project217.htm https://sound-au.com/project12.htm either with an output cap, DC adjustment or better yet, a DC servo)
- use a push-pull cascode VAS with dynamic gate bias as in Slone's OptiMOS amp for its soft clipping characteristics (right there in the middle)
- with such a VAS, possibly capable of beefier current than the usual, do without MOSFET drivers.
MOSFETs would probably be the Exicon devices (rated 125W).
Could it work? Replicating the OptiMOS design for a guitar amp is extravagantly overkill, and I don't feel I need all the symmetry or complexity. Also, 2nd harmonic distortion is not a big issue. If it's rich in low-order distortion, that's actually the design goal.
No, I don't want to go with a soft clipper (whatever it may be) before the power amp, because the response would be quite different, and most of all, it would be too easy 🤣 . The clipping should be generated internally by the high output impedance power amp with all its interactions with the speaker.
In fact, I wouldn't replicate the OptiMOS design, extremely overkill for guitar. I'd like just to lift the VAS design and choice of lateral MOSFETs for the output devices. The first ESP designs employ current feedback, on the other hand.
How do you make sure that the opposite current sources Q5-Q6 vs Q9-Q10 are 'in balance'? They're not connected in any way, so if one diverts by odd chance (tempco?) the (nfb) amp has a lot to do to remain it's stability. Being a part of the first stage, on the very input-feedback differential, a runaway is very possible. Any simu's, not only dc/ac/bode but also start/stop sequences. From there noise, distortion and more lively parameters.
That's part of the OptiMOS design that I wouldn't replicate anyway (and apparently the design works well).