The Manely 300B mono blocks I reviewed had a 10 position user adjustable negative feedback dial.
I spent time going through every one. I preferred no negative feedback. It sounded the most natural and organic to me.
Here's some peculiar things about feedback.
In a tube amp, its usually applied to the cathode of the input tube. The audio is coming in the grid of the tube, so the two are mixed in the tube, which is
not linear.
IMO this is a poor practice but everyone does it that way, mostly out of tradition. IIRC, Norman Crowhurst wrote about this about 60 years ago but didn't propose a solution.
So the tube distorts both the feedback signal and the input signal. Higher ordered harmonics are generated, along with some intermodulations (which can be generated any time a non-linearity is present).
There is a way around that. If you were to invert the phase of the feedback (which can be done at the output transformer), you can then mix the audio with the feedback using a simple resistive divider network at the input of the grid. Resistors are a lot more linear than tubes are, so the feedback signal can be mixed in a far more linear environment.
The result is the feedback is then more accurately able to do its job, with less distortion introduced by its application.
There are tube products that do it this way although its rare in amplifiers. The Leak Point One preamp is a good example.
This happens to also be the way opamps have feedback applied to them. You can do it that way in a class D amplifier as well.
Here's another pesky issue: if the gain and bandwidth of the amplifier circuit is insufficient
and feedback is used, what will happen is the distortion vs frequency will show a rise at some frequency were the feedback is no longer supported by the gain and bandwidth. This can cause the distortion at higher frequencies to be unmasked and so the amp can have a brighter and harsher presentation. This was a common problem with solid state amps well into the 1990s.
Class D allows for a lot of gain and bandwidth (known as Gain Bandwidth Product). For example in our class D the GBP is about 20MHz, and since we 'only' have about 37dB of feedback, the distortion does not rise anywhere in the audio band. That is a big reason why the amp isn't bright. Most self oscillating class D amps have this property.