Mosfet Amplifier Designs
A few alternative mosfet amplifier designs are included here, all of which are incomplete, and all of which I have rejected or abandoned for some reason. They could probably be made to work reasonably well with a little additional development. A few more alternatives may be added eventually.
Some of the same advantages of the MJR7 circuit could be included in a more conventional direct coupled circuit, with dual jfet input stage and a more symmetrical driver stage. The following diagram is just a first thought, it may have a few problems and could no doubt be improved. Most of the distortion will still originate in the output stage, and the result should be about the same as the MJR7 if the same feedback loop gain is still possible, which may not be so if the extra stage adds stability problems. Supply rejection will probably be worse, as will the low frequency damping factor, so there is really no advantage compared to the MJR7, and we have had to use twice as many transistors, plus a few more needed for some form of speaker protection. Even adding supply line fuses may be a problem, if we use one in each supply line then if just one fuse blows the output could swing up to the opposite supply rail, and some modification to prevent this may be needed. Not recommended!
We could go further and make a completely symmetrical version. There are some wrong ways to make symmetrical amplifier circuits, and a number of common problems to be avoided. The next circuit is just one of many possibilities, and although it could probably be made to work reasonably well there is not one single improvement compared to the MJR7. The distortion will be higher, the supply rejection will be poorer, and we still need speaker protection and some safe way to add fuses.
The emitter resistors for the two input transistors need to be chosen or adjusted to set the currents shown as 5mA. The lack of symmetry between npn and pnp types means these resistors will probably not have equal values. As usual the symmetry is an illusion, present in the circuit diagram but not in reality. The bias stabilisation transistors are a npn plus pnp pair and the original 100p plus 470R compensation is split into two 47p plus 1k to make the apparent symmetry complete, but again none of this is of any real benefit. The 3k3 resistors used to limit the input stage current at clipping are a really bad idea, and some alternative method is needed. Again, not recommended!