PWM frequency

[KarbonHed]

Here's a question that has just flat out stumped me.

I'm designing a winder and for the PWM controller I have a basic circuit figured out, but... just as I'm going for the breadboard to test it I realise that I have no idea what FREQUENCY the PWM sould be oscillating at.

So my question is how do you determine what frequency any given motor should be driven at?

Or am I heading for an afternoon of swapping caps and resistors until I find a happy selection of values that work?

[Jazz P Bass]

Exactly what type motor do you have?
And what is the controller circuit topology?
A triac circuit will work with an AC motor very well.
http://www.freescale.com/files/micro...ote/AN3471.pdf

[KarbonHed]

I must really not be understanding this.

Here's the basic idea I'll be working from. Picture 1 is the circuit, 2 dual opamps one as an oscillator, who's frequency is what I'm after. The other the comparator that does the PWM and a buffer for the PWM pot. That's probably not needed but there's a spare opamp stage so what the hey.

The second image is how I currently think it works. Your triangle wave goes into the comparator. Depending on how much voltage is going into the other input you get a variable pulse width. More pulse width and the longer power is applied to the motor so the greater the speed. Less voltage shorter PW etc.

If the carrier wave is too slow the motor is just going to lurch between pulses. So there's obviously a lower cut off where it's basically unusable. But what is the upper frequency? More importantly what's a happy medium?

Or am I looking at this totally the wrong way here?

[tedmich]

I believe most PWM circuits between below kHz work AOK for DC motors, I've built them mainly for LED dimming circuits and there you just need to exceed 100Hz so your eye sees it as continuous. The lower frequency controllers, hell using a 555, run just fine at a couple hundred Hz while the multi kHz PWM make heat and are more expensive and less usable at the low speeds typically used in winders. If you are designing a BIG motor speed control you use expensive kHz closed loop PWM (even PID based) controllers. But I should think a 200Hz PWM should work fine on a winder. I have never heard of a choppy PWM maybe if you are using it at <10% of the motors typical RPM?

[David King]

I wouldn't go overboard on the PW switching frequency for a DC motor. Most of them seem to switch at around 4kHz and often give you the option of tuning it higher by an octave or two to keep things simple. Ideally you want a frequency that doesn't resonate in the motor coils too loudly. If you can set the freq with a pot then just fine tune it for the speed range you will be running the motor at for the quietest operation. I'd settle for 2X or 4X the top RPM of your motor.

[marku52]

What you are aiming for in a pwm frequency is to have the motor inductance filter most of the current ripple out of the pwm motor current. since motor torque is proportional to current, current ripple = torque ripple. Do you know the motor L and R? The current will decay during the off time according to an L/R time constant, just like a capacitor will charge according to an RC time constant. You could calculate a pwm freq which would incur 10% current ripple at your minimum pwm, for example

Or, to do it empirically you need an AC current probe to measure the decay of motor current during the of time. Or just guess. Start with 5Khz, and see if the motor runs smoothly at your lowest speed (which would incur the maximum off time). If you run the freq too fast, you incur more switching losses in the mosfet (it will get HOT), too slow and the motor current will be too ripply, which will cause torque and velocity ripple.

Also, since you are running at 5V, you MUST use a "logic level mosfet". This type can be fully switched on with only 5V on the gate. regular mosfets require 10V of gate drive, and may not get fully turned on with only 5V. Again, this will make the mosfet run HOT, and limit your max motor torque.