As design engineers, we all have different rules of thumb design crib sheets. If we want to design a servo-controlled power amplifier we engineer for maximum control and fast step rate. On the other hand, if we want to build a syrupy-sweet sounding amp we’d use a soft-focus input like a JFET or a vacuum tube. (If you’re a design engineer and want to test this just listen to the difference between a 5532 and a TL082 op-amp).
Each amplifier design is specific to the task at hand because no amplifier is perfect. When we focus our efforts in one area like high damping, low distortion, or high wattage, we do so at the expense of another.
I’ve mentioned this before but it likely bears repeating. Engineering is the art of minimizing compromise to achieve the design criteria. We gain here by losing a little over there.
The broader question posited in today’s post’s is so system dependent it’s almost difficult to answer with any clarity, though we can offer a few observations. Musically pleasing amps generally have the following attributes: voltage-centric devices at their inputs (FETS or tubes), stable open loop performance (they don’t require feedback to keep amplifying), relatively high bias at their output stages.
There are hundreds of designs in the world of high-end audio. Some are more musical than others.
If you want to delve a bit deeper I put together a video on the subject you can watch here.