How have you come to this conclusion? How have you measured the voltage? How have you measured the speed?

Firstly the pwm - you say the voltage at the output of the L293D is not linear. What happens when you replace the motor with a resistor? This will prove/disprove the non-linearity here. Using a normal multimeter to measure the voltage will probably give strange results due to the pulsing.

Secondly, the motor. Is the motor speed vs voltage linear when you use a variable DC supply?


At a guess, due to your low switch frequency of 100hz, that may be causing the issue - usually pwm is done at supersonic frequencies at >10khz. The inductance of you motor may be the culprit. PWM is pretty basic - at 50% you're supplying half the available energy. 25% a quarter etc.

Do some testing and that will lead you to the answer.