PWM question

I have a dc motor (110v 0.1 HP) running nicely with controllable speed using the PWM output, optoisolator, MOSFET control element. However, it is annoyingly noisy because of the 490 HZ pulse repetition rate of the PWM. Somewhere on one of the forums I think I read that one or two of the PWM pins on the Arduino UNO run at double this rate and if this is so I will try that to see what it does to the noise.

Can anyone tell me if this is true and if so which pins? The Reference, at least where I read does not mention this though it does mention pins 5 and 6 as being different.
Any wisdom would be appreciated.

Otherwise the project is advancing nicely with few surprises--once I get a chance to think about it for a day or two.

The defaults for the Uno and other ATmega328 Arduinos are:

5,6 : 976.5Hz (timer0)
9,10: 490.2Hz (timer1)
11,3: 490.2Hz (timer2)

But you can reprogram the timers to run much faster. However reprogramming timer0 means
millis(), delay() and micros() will stop working properly.

BTW this might be useful:
http://www.gammon.com.au/forum/?id=11504
or even my little page:
http://sphinx.mythic-beasts.com/~markt/ATmega-timers.html

Thanks for the quick reply. I'll have to dig into the reprogramming approack

Could you post your circuit?

Because I think alternatively, something like this could work with the existing PWM frequency:

The value(s) may need to be tailored to your circuit.

Without the capacitor, you may need to use 10-20KHz PWM to lower noise levels and your motor drive transistor may not be capable of this or it may overheat.

dlloyd:
Could you post your circuit?

Because I think alternatively, something like this could work with the existing PWM frequency:
http://ww1.microchip.com/downloads/en/AppNotes/00771b.pdf
The value(s) may need to be tailored to your circuit.

Without the capacitor, you may need to use 10-20KHz PWM to lower noise levels and your motor drive transistor may not be capable of this or it may overheat.

That is not applicable to this situation, we are talking 75W, any attempt to slow down the switching
will generate lots and lots of heat in the MOSFET.

Correct ... I implied using a transistor (as in diagram). If using a transistor, it would need to be sized correctly with suitable safe operating area and would require heat sink. I guess a MOSFET is starting to look better in this regard.

No, you don't do this for a big motor, that's not the same kind of PWM. The current in the motor
is pretty constant across the PWM pulses, for the fan the current falls to zero during the PWM off,
according to that AN. Big motor has lots of inductance.

Even for only 1/10 HP (75W) DC motor? Also, I was suggesting an alternative that keeps the PWM at default rate (relatively slow 490Hz). The diagram shows a back EMF diode that should send the extra energy to the power supply. Wouldn't the power supply filter absorb the back EMF energy? I would think overall it would act like an RL or RLC filter, thereby lowering and smoothing the DC voltage seen by the motor and reducing audible noise. (Just wondering).

I'm very surprised that this simple question has generated so much discussion. I really have to get recalibrated concerning motor size--I considered 0.1HP as quite small; my work experience was with HP's in the 100's or in many cases several thousand with 3 phase SCR control.

Doubling the PWM frequency has worked very well. The noise level has gone down significantly and the frequency is less annoying. Also, something I did not anticipate is that the wave shape has gone to something which makes sense-- a hump with modulation slots where the rectified sine wave is above the CEMF and simply flat dc between humps. When you think about it 490 Hz gives a shade over 2ms per pulse and 1/2 wave of 60 Hz is 8.3ms so only 4 PWM pulses per hump at best and once the speed is up less than that. The inference here is that the higher the PWM frequency the better.

re the MOSFET, the object of the game is to work it in either the high current saturated region or in the high voltage no current region to keep the MOSFET power dissipation low. Halfway between gives high power dissipation so a hot MOSFET. I found the bleed resistor gate to source affects this--too high and turn off time stretches out longer than I was happy with.

MarkT--I'll get the elementary posted ASAP when I figure out how to do it; haven't posted one before.

Actually this is just one corner of a much bigger project.

First try at inserting an elementary

motor control.pdf (26.8 KB)

R1 bursts into flame, possibly exploding.

Try again...

R1 is no problem. It’s about 20W or so and gets hot but nowhere near smoking. Probably wouldn’t burn your finger.

Do the arithmetic. Vdc is about 110 average, resistance is 1500 ohms.

Watts dissipated is 110 X 110 / 1500 = 8.06 W

Wasting 20W to power a MOSFET gate? Getting rid of 20W is expensive too.

Capacitively coupled supply is the way to this.

This is descending into silliness.

Just because I use a resistor which I estimate is rated at 20 W does not mean 20 W is wasted. In actual fact the voltage dividing bridge uses 7.1853 Watts assuming 110vdc.

The present job is proof of concept and prototype construction, not final polishing which will come when I have a working system. The 10 mike capacitor is probably overkill too and I will look into it eventually. And I also found that driving the MOSFET gate takes more beef than you would think if you want to get fully across the active region into the low dissipation regions in all cases.

The original posting was to find out the double speed PWM pins which you told me and I got better results than I expected--- thanks

Why is suggesting a capacitively coupled supply for the MOSFET gate silly?

You are right its about 8W since you don’t have a smoothing cap on the DC bus (I would have
thought that could cause motor noise issues at 100 or 120Hz?), but 8W to get rid of is an unnecessary
pain - worth a couple of extra components to get easy thermal management I would have thought.

Besides all that it looks like the mosfet gate is discharging through a 15k resistor. That will limit your PWM frequency quite a bit.

What I meant by silliness is that my original question was to find the fast PWM pins, not a request to redesign the system based on armchair engineering.
I did not look into capacitive coupling but I may in the future.
I did do a lot of engineering and experimenting with the bridge resistors and the gate discharge resistor and found that with higher resistors the gate takes longer to discharge which will increase the mosfet dissipation.

If you take the IRF740 data sheet, draw it out to linear scale, put in the max power dissipation hyperbola then put in the load line for the motor armature resistance, 9.5 ohms, you will find that, at very low speed, you need all the gate drive you can get to get up to the Rds line and practically all the transition, cutoff to saturation is to the right of the hyperbola so through the high dissipation region. As the speed increases, the load line moves toward the origin so meets the Rds line at much lower current.
Therefore we need strong gate drive and very fast transition to keep the time in the high dissipation region to a minimum at low motor speed.

For the present, the design is working nicely and stays as is while I go on with several other blocks of the project.

If a gate driver like this was connected between the 4N25 and IRF740, then perhaps much higher PWM frequencies could be utilized (10KHz+). This should resolve any issues with the very low CTR of the 4N25.

Sorry i should have clarified, the gate resistor of 15k is too high. Its ok it you want a relatively low pwm speed (100 -300 hz or so) but once you get over that you will be in the linear region for a significant time. Losses will be high.

See how long it takes a 63nC gate to discharge through a 15k resistor. No experimentation is necessary. If your gate drive is 10 volts for example you are discharging that gate with 0.66 ma of current. This takes 95 uS to discharge the gate. If it takes 50 us (a guess, the 4n25 doesn't allow much current either) to charge and 95 to discharge (145 total) this should be about 1/20th of your pwm period or less to keep switching losses low.

That means you should not go under 2600 us seconds pwm period or 348 Hz. 100Hz to 300 is fine but if you want ultrasonic frequencies (get rid of the whine) you will need to drive the gate much faster than that. A lower resistor value will help but proper gate drive is the answer.

Edited: moved decimal place as per Marks correction below.