[dgojanov]

I am experimenting to drive dc motor from pwm source which is pic) and
driving mosfet tranzistor.

At longer dutty cycles motor is behaving as expected, high speed and high torque.
But at shorter duty cycles motor looses significantly torque.

Is there any explanation for this and any suggestion how to solve this problem

Thank you very much.

Darko

[doorknob]

At what frequency are you driving the PWM?

If it is at too low a frequency, perhaps increasing the frequency would help.

[dgojanov]

Thanks for suggestion

pwm is at 4 kHz

Will try to increase it

[doorknob]

4 kHz may be fast enough. I was worried that you might be using a substantially slower frequency.

Also, what is the inductance of your motor? If the inductance is too high, then maybe the current can not build up fast enough at the PWM frequency that you are using (for short pulses), in which case the cure may require either going to a higher motor voltage or maybe even a slower frequency. Do you have an oscilloscope that you can use to monitor the current waveform (measuring the voltage across a small-value series resistor) to the motor?

(still, I would expect torque to decrease at lower duty cycles, because the average current would be less than at higher duty cycles)

[dgojanov]

Thanks for your comments

I tried even to run at 20kHz and 30Khz but there is no change.

Otherwise this motor is behaving nicely under mach and gecko320x.

it holds torque evenly from 0 to max speed.

Voltage accross motor and mosfet is 36V.

I put series resistor of 1 ohm.

One good ting happened at igher frequences and this is that MOSFET is not any more heating up.

[H500]

You need to add some sort of feedback. If it works with a G320, then I assume your motor has an encoder. Make use of it.

[mike_Kilroy]

hmmmm.... if your control can use a torque amplifier then you dont need any encoder or speed feedback, but perhaps you should consider going the next step and add a current loop instead of just open loop.... put the 1 phm res between motor and common and bring its voltage bac to pic and make a PI loop with it to then command not a pwm % cycle but rather a current. pretty simple to add to your pic program; examples should abound on line. I believe I have seen many on the Pic prod site we have used (parallax).

4khz is plenty fast for what you are doing, and of course higher pwm freq will have no effect on what you see.

then know WHY it acts 'funny.' you are simply pwming output VOLTAGE not current..... know that armature voltage =Kb*rpm +IR drop, where IR drop makes current and thus torque (T=Kt*I). So put your pwm generated voltage, current, & speed into a spreadsheet and you will see in black and white (sorry, no color) why it 'acts funny' at low pwm %

[dgojanov]

Thanks to all of you who gave me advice.

Darko