Some of the MJR7 amplifier component values are critical for dependable stability. In some cases a combination of several capacitor and resistor values is important. Sticking to the specified values there should be no problem, but if, for example, we wanted to double the bandwidth from 60kHz to 120kHz we would perhaps try reducing the two 470p capacitors in the input circuit to 220p, but this could increase the feedback loop unity gain frequency beyond 10MHz potentially causing instability. To avoid this we also need to reduce the 10p across the 200k feedback resistor to 4p7. That is necessary anyway, because it contributes to the 60kHz bandwidth.
If instead we want to reduce gain, we could start by reducing the 200k feedback resistor to 100k, but then we also need to increase the parallel capacitor from 10p to 22p. That is still not ideal, it should help to also reduce the 470R in series with the 100p shunt compensation capacitor down to 220R.
If we want to increase the gain we can increase the 200k feedback resistor, but to keep the original 60kHz -3dB frequency we need to reduce the 10p parallel capacitor by the same factor, so to increase the gain from the orignal x18 to x27 we need 300k in parallel with 6p8.
After any change in the 200k feedback resistor remember to readjust the output stage operating voltage, i.e. the voltage of the mosfet sources should be set to about half the supply voltage using the 4k7 preset.
I have not tried these modifications in a real circuit, these are just my conclusions from simulations, which are not always entirely dependable, so I strongly advise sticking to the specified components and layout, it already works well.
There is less difficulty using dual-die mosfets for higher power, in this case simulations suggest it will make very little difference to stability provided we increase total mosfet quiescent current to at least 150mA.
As I have mentioned before, one fairly important component is the BC560C (pnp) in the input stage, here high gain (500+) and high fT (200MHz+) are needed. Also low base-spreading resistance (rbb') is good for both input stage transistors. This is not always specified, and some supposedly 'low noise' transistors such as the KSC1845, 2N5089 and MPSA18 appear to have quite high values which can combine with input capacitance Cbe to add significant phase shift potentially reducing stability margins. The specified 2SC2240BL input NPN should be used if available, but if not then the surface-mount 2SC2713-BL is similar, and if all else fails the BC550C is also probably good enough.
I have not tried different transistors, only the 2SC2240BL and BC560C, and previously the also obsolete 2SC2547E (npn) and 2SA1085E (pnp). The Toshiba data sheet for the 2SC2240 actually specifies 'low 1/f noise', which is unusual, I never saw that mentioned in any other transistor data, and it appears to be true from my own measurements compared to the supposedly lower noise 2SC2547E. The 2SC2240 noise starts to rise below 10Hz but the 2SC2547 starts an octave higher at 20Hz. As expected this low level noise under 20Hz was not found to be audible.
The amplifier can be used at higher supply voltages if higher power is needed, and I tested it myself at 94V without finding any problems, other than the driver stage transistors getting a bit hot and needing heatsinks for long term reliability. The current source KSA1381 has the highest current of the medium power devices, so this at least needs some help. At high supply and low impedance loads the mosfets may also be at risk, and then my advice would be to use the dual-die types, which have higher specified current limit, 16A instead of 8A for the single device. The power rating also is doubled from 125W to 250W, but then for higher continuous power output a better heatsink is still needed.
Profusion can now supply the Exicon lateral mosfets in selected bands based on the value of Id at Vgs = 0.5V. The lowest (red) band is from 105 to 125mA and the highest (white) from 185 to 205mA. They will not guarantee to supply any requested band, but for a quantity of the same type all will be the same range. If unmatched samples now include those rejected as outside the matched groups then the chances of getting a device with high Vgs outside the range of adjustment could be increased, so paying a little extra for matching may be worthwhile.
Another popular modification is simply to use a different type or value input capacitor. Anyone who seriously believes they can hear differences between capacitor types can of course use something with audiophile approval, but note that the MJR7 distortion measurements include the effect of the cheap polyester type I used, so it doesn't appear to be a problem. I would be surprised if it added more than a nanovolts third harmonic at 1kHz at the output. I did a calculation of capacitor distortion appearing at the amplifier output, based on the capacitor tests by Bateman, where the worst case polyester capacitor produced -90dB 3rd harmonic with 4Vrms across it. For a 1V input at 1kHz there is only 6mV across the input capacitor of the MJR7, and making reasonable assumptions about distortion level versus voltage I estimate a contribution to amplifier output distortion around a tenth of a nanovolt. The minimum audible level from my own speakers at the 3kHz 3rd harmonic frequency I found to be 300uV, so the capacitor distortion reaching the output is then theoretically 130dB below audibility. My advice remains 'small physical size and low leakage', (but not high-k ceramics, which can have fairly high distortion), and maybe increased value to 4u7.