Uggh, this is suppose to be an analog thread, but I just can't resist countering Rauls outrageous tube amplifier comments. I have designed, built and modified a number of amplifiers, both tube and solid state. So my comments are based on actual listening experience.
Tubes and transistors are both devices that produce gain using similar concepts but with quite different characteristics. Tubes have good linearity (low distortion) but low gain. Transistors on the other hand have very poor linearity (high distortion) and very high gain.
To get around the distortion problems most tube amplifiers and all transistor amplifiers use feedback to eliminate distortion. Here I use the term distortion in a very broad sense, anything that deviates from the original signal including frequency response, noise along, harmonic and IM distortion. More feedback means lower distortion, including the "equalizer" effect that Raul mentioned. Since transistors have very high gain and it is easy to produce lots of feedback and therefore very low distortion, ruler flat frquency response, high damping factor, vanishingly small harmonic distortion, blah, blah. Tube amps on the other hand have much more limited gain and can only produce limited feedback. The result is considerably worse distortion measurments, in spite of the fact that they are lower distortion devices to start with.
But there is a significant hitch to this story. It turns out that feedback has a clearly audible negative effect on sound. To date I don't know of anyone that is able to measure this affect. It does not show up in any of the standard distortion measures. This is part of why there is so much controversy about this topic. I have a lot of first hand listening experience regarding feedback and I assert that this is both real and significant.
So like so many things audio it all boils down to compromise. Feedback brings both good and bad to the table and the optimum balance is both system and taste dependent.
In a given context and a set of personal preferences there will be an optimal balance between feedback and distortion. That optimum balance will not be the same in a different system or with different tastes. So it's never as simple as choice between feedback or no feedback or for that matter tubes or transistors.
The goal for good amplifier design is the lowest possible distortion with the smallest amount of feedback. That means that the amplifier needs to have the lowest possible open loop distortion. Open loop distortion is the level of distortion before feedback is applied. A well designed amplifier implemented with high quality components will require less feedback to arrive at a distortion goal and will therefore sound better.
So if you look at amplifiers from the perspective of levels of feedback the high feedback end of the spectrum is the exclusive domain of trasistor amplifiers. On the opposite end of the spectrum is single ended triode amplifiers with no feedback at all. Each have their strengths and weaknesses.
Personally I find that nirvana is a zero feedback amplifier. But the no feedback path is not for the faint at heart. Everything has to be perfect or you will hear it. Component quality become hugely important. Speaker choices are very limited and there are major speaker/amp synergy issues. Feedback extracts a price but it does make things a lot easier.
Chris
