Firstly, with a permanent magnet DC motor, yes the speed is fairly proportional to the voltage, but just like transistors in an audio circuit, there is a band where this linear(ish) relationship exists. if you put 1V on a 12V motor it will probably not turn, as the power developed at that voltage will not be enough to overcome the inertia of the rotor or load. As a ball park figure, you will probably not get any useful controllable motion below about 1/3 of the rated motor voltage. if that.
But this very much depends on the physics of the motor and its load.
And......this is for simple open loop speed control. close the loop with feedback and you can make a PM DC motor turn at 1RPM. But that is a whole different kettle of fish.
Now. PWM.... Lets put this in simple terms, and make analogies with audio. because all the same principles apply.
Think of class D amplifiers. These are just PWM controllers just like for motors. Ill get back to that.
The REASON for PWM: Just like in audio it is all about efficiency. When a transistor is ON or OFF - acting like a switch - it is efficient. it looses virtually no power as heat. all power goes to the load (ON) or no power goes to the load (OFF).
If you simply use a potentiometer, or a transistor in its linear range to vary the voltage to a load (in an analogue or class A style), the surplus power from the supply is wasted as heat from that analogue amplifying (transistor) or attenuating (Pot) device. Think class A amplifiers. They dissipate lots of heat and are inefficient. But they are the most analogue (look up the real meaning of this word) type of amplifier, and hence to most people have sweet tone.
Now, if you have a square wave at a high frequency, and vary the ON / OFF time ratio, you change the average amount of power getting through, but because the device is switching on and off it is wasting no energy as heat. This is PWM...... PULSE (square wave) WIDTH (how wide the pulse is) MODULATION (changing the width)
The critical thing here is the "response time" of the thing you are controlling. A motor or an incandescent lamp has a very slow response time, with a PWM frequency of 20kHz, switching on and off power 20000 times per second to the filament in the lamp, it can not heat up and cool down that quickly, so it just sees the average and settles at a brightness proportional to the average amount of power the controller is letting through. Just like with the motor, it will settle at a speed proportional to the average amount of power the controller is letting through. To drive this concept home... think what would happen if we reduce the PWM frequency to 1Hz? The lamp and the motor CAN react to a signal switching on and off once per second, so you would see it as a flashing light, and how long it is on for each flash is the length of the pulse width.
If you get a simple PWM controller and you have an oscilloscope - have a look at the voltage at your motor terminals. it will be a square wave. (but probably messy due to real world factors, back EMF from the motor, etc.)
So, to bring it back to audio to sum up... Class D amplifiers are just PWM controllers, but the PWM switching frequency is so fast that the average output of the rapidly ON / OFF switching device, is is seen and averaged out by the speaker as a analogue voltage (and there is lots of added filtering and tricks to help) But in the end, there is no analogue amplification in a class D amplifier. just a switch, efficiently switching on and off really really really quickly, and as a result it is very efficient (no heat sinks required as not much heat is produced) and batteries last longer (that is why your phone and MP3 player will have a class D amplifier)
I am sure all you audio guys know this already.......but it is the same as motor control with PWM!
Cheers,
T