Ouch. Maybe it is worth to check how much will be shipping from the Audiophonics?price after shipping from UK for small quantity (~$46/pair!!)
https://www.audiophonics.fr/en/composants-electronique-transistors-puces-c-512.html
Just a remainder, “select grade” laterals are not P to N part matched at all and there is no reason not to order cheaper non-binned ones.
The binning is for paralleling them I believe. Matching n to p means adding some gate-source capacitance to match those up so the drivers see a symmetrical load, although I'm not sure its worth it as many circuits have the gates shorted to each other at AC anyway, and the internal capacitances are very non-linear anyway (its a MOS capacitor basically).
I think double-die devices are already bin-matched so that's going to be cheaper than matched single dies.
I think double-die devices are already bin-matched so that's going to be cheaper than matched single dies.
Member @mislaaav nudged me to check the TMC compensation with this amplifier, as he is working on his own version of similar design, that has a TMC compensation.
I’ve checked two pole (TPC), transitional Miller (TMC) and three pole which is a combination of TPC+TMC. Those compensation methods bring, in real circuit, some 6 – 12 db lower distortion (20 dB in simulation). For now, TMC seems the best choice as it provides the same unconditional stability as a standard Miller compensation.
Result are just a little better numbers, nothing more. Sound is, of course, exactly the same. It should be, as we are taking about distortion that is, to start with, one millionth part of the output signal.
I took the chance to rearrange parts like with @cibo build. With only 300 mm deep case, using two separate ground loop breaker circuits instead of one dual, enables better PS placement.
I’ve checked two pole (TPC), transitional Miller (TMC) and three pole which is a combination of TPC+TMC. Those compensation methods bring, in real circuit, some 6 – 12 db lower distortion (20 dB in simulation). For now, TMC seems the best choice as it provides the same unconditional stability as a standard Miller compensation.
Result are just a little better numbers, nothing more. Sound is, of course, exactly the same. It should be, as we are taking about distortion that is, to start with, one millionth part of the output signal.
I took the chance to rearrange parts like with @cibo build. With only 300 mm deep case, using two separate ground loop breaker circuits instead of one dual, enables better PS placement.