A tube amp, whether Marshall or not, works something like this -

When you turn the power on and the standby switch is on, power starts to flow through the power supply, but is only diverted to the heaters in the tubes. This lets them warm up a little before they have to run 100%, and helps cushion them a little, lets them live longer. After 60-90 seconds, you flip the standby switch off and power starts to run through the whole amp.

Once this happens, everything is online - plate voltages are at working level, the tubes are heated up and conducting, etc.

Tubes are generally run in such a way that they are relatively linear at low signal strength, but become progressively less linear the harder they’re pushed. What this means to you is that at low gain and low volumes, tubes will amplify what you put in pretty accurately (the same as transistors, incidentally). So, not much distortion at low volumes and low gain.

Once you start adding gain and turning up the volume, though, you push the tubes harder, and tubes start to become less accurate - ie, they start adding distortion. This mostly takes the form of lower order harmonics (2nd and 3rd), which tends to sound musical… and louder, as well, especially due to the high quantities of 2nd harmonic distortion.

Some other stuff starts happening, too - when transistors are overloaded they clip hard - they chop off the top of the sine wave - while tubes tend to compress the sine wave, which means a smoother, more musical feel to the overall sound.

( the technical explanation is that when you chop the top off of a sine wave you create a square wave, which means you’ve just added tons of odd-order harmonics to your signal. compression still adds odd-order harmonics, but far, far fewer than hard clipping )

So… you’re always using the tubes, just that they start working a little different when you push ‘em harder.

Saul