T&L Publications

Is there a "rule-of-thumb" for picking the PWM frequency relative to the motor's time constant?<p>It seems to me, that if PWM frequency was much too high, only high percentage duty cycle pulses would result in useful power. Low percentage duty cycle pulses would be gone before any significant current flow. (ie- 1% to 50% = 0 oz-in., 55% to 100% = 1 oz-in. to 50 oz-in..) The range where there is control over torque or RPM is reduced.<p>If the PWM frequency was much too low, motion would be jerky, and probably hard on the gears.<p>I also suspect that the PWM period should not be hamonically related to the mechanical time constant of the motor. Else resonent vibration shakes the motor apart, or at least an unpleasant noise.<p>Gut feeling is that PWM frequency should be set so that a 5% to 10% duty cycle pulse is about equal to the time constant.<p>Anybody have any facts to share?<p>Thanx

Think of it this way. For a DC motor to run at full speed you feed it DC.
If you want it to run half speed, feed it half DC and half off. And so on.
The frequency comes from the repetition.
Motor
Read Input
Write pulse related to input
GOTO Motor

No Harold, it's not that simple. Motors are inductive. They also have a mechanical time constant.

Hi, Dale. While there are a number of other factors that you could take into consideration, generally you just say that a frequency which is too low will result in mechanical resonances/excess noise, and a frequency which is too high will result in too much lost power from the switching transistors/MOSFETs being in the linear region. A good rule of thumb for smaller DC motors is 1 KHz to 5 KHz. If possible, sweep the PWM switching frequency, and observe the results for the motor you've chosen (listen for excess hum/buzz, check your switching transistors for excess heat/scope 'em out).<p>Good luck
Chris

I suggest something in the range of 40Khz to 1 MHz. If you pulse at 5 Khz, youre gonna hear it. It can get annoying fast.

The GE servos I have seen run at 400hz. It would seem that the inertia of the system would be a big factor. YOu would want to go as low as possible without causing vibration.

What i have been finding in experimenting with dc motors in the 100-200 watt range switching speeds of ~ 10-25 khz seem to be the most efficient but I believe that is dependant on the amount of commutations the motor makes. As the frequency increases, so does the gate current needed to charge the gate cap. If you have too little gate current, the mosfet spends too much time in its linear mode causing it to overheat. I have also found that the reverse diode's switching speed seems to make a differance.good luck
tj