DC motor (pump) control problem

I'm working on a project where I control the regulation of the water level in tanks from the PC via a microcontroller STM32.

My problem is this: when I am sending a low duty cycle (less than 40%) at the pump, it does not rotate! This then causes a response delay of the system so that the controller sends 100% .. The control signal of the pump thus resembles a PWM signal! 0% or 100%!

Is this problem is the circuit of the power amplifier https://www.electro-tech-online.com/attachments/power-amplifier-jpg.83626/ (I use NPN transistors instead of IGBT, ie running attack and not voltage)? Or should I define a dead zone for the pump in the program?

Is the circuit of the power amplifier has a role in this problem?

Do you have any ideas for improvement? solutions?

My problem is this: when I am sending a low duty cycle (less than 40%) at the pump, it does not rotate!

Click to expand...

Then can't you use a higher duty cycle?

Edit: The opamp is not a rail-to-rail output type. Its max output will be a couple of volts less than the supply. There is also 2V or so dropped across the Darlington-connected output transistors. So your motor will see about 4V less than the supply. Not surprising it's sluggish.

Look at electric fans with a speed switch.
The switches are like this: OFF-- FAST-- SLOW
The fan will not work well with OFF--SLOW--FAST

When I start fans with a computer I go from off to full on for 1/3 second then drop back to slow. You need to drive the fan hard until it gets started then back down.

---edited----
Alec_t is right.