There are three main types of amplifying device – the transistor, the vacuum tube and the FET (field-effect transistor).
Vacuum tubes are wonderful devices. They are very close to linear, which means that the amplified signal is pretty much identical to the input, except bigger. The main problem with tubes is that they require a lot of voltage to work, which leads to design difficulties, power supply problems etc. There is another problem – it is impossible to design a truly symmetrical tube amplifier. However, since asymmetrical amplifiers produce the pleasant even-order distortion we want, that's not such a bad thing after all.
FETs are also very close to linear. However, they are inclined to be a little noisier, which limits their use.
Transistors (bipolar transistors, to give their full name) resolve the problems of tubes and of FETs, and they are also wonderfully gainy – you can put a signal into a transistor and get a much bigger signal out. However, they are not linear. They are as far from linear as a ten dollar bill with Simon Cowell's face on it.
Fortunately there is a remedy. That is to take a portion of the output signal, invert its phase so it is now an upside-down replica of itself, and feed it back to the input. Magically, this feedback technique reduces distortion massively. It does this at the expense of gain, but bipolar transistors have much more than they need anyway.
So, using lots of negative feedback, a transistor preamplifier can be designed that has very low distortion.
But… what happens when the feedback runs out? When the signal clips, or reaches the maximum possible voltage, then in that moment negative feedback vanishes and the amplifier becomes unstable. It could take a millisecond or two to stabilize, and even this would be audible. It could take longer.
Tube and FET preamps don't need anywhere near as much negative feedback so they are likely to be much more stable under clipping. The problem is that designers tend to think that their preamps will always be used optimally, and they don't consider what will happen in the real world.