Need some help to smooth a PWM signal, to stop resonance in motor.

[apples]

I bought a kit and made a pwm motor speed controller for running a 12 v car fan blower motor. Trouble is the pulses are causing humming or resonance in the motor itself.

So what are my options in getting a smooth clean (straight line on my scope) voltage to drive the fan? The speed controller is rated at 20 amps and the fan motor pulls at least 10 amps because it blew the fuse in the back of my multimeter

I have a collection of resistors, trasistors, capacitors, diodes, FETS etc (just getting started in electronics so bought a bunch of stuff).

Posted a video on Vimeo showing the problem I have and also a screen shot from my oscilloscope.

Any help is appreciated.

Peter

 

[(*steve*)]

Try changing the frequency to one that the motor doesn't hum to.

 

[Electrobrains]

Try changing the frequency to one that the motor doesn't hum to.

Definitely a good idea!
I have made several similar projects and always put the frequency in the not audible region (25-30kHz). It has worked very well.

I first suggest you put a (power) diode directly over the terminals of the fan (cathode on the plus side!), to catch those spikes that are seen on the oscilloscope (and are running along the whole cable length). That will reduce EMC/EMI problems and might even remove your noise totally (it will keep the current constantly flowing inside an inductive motor).
To make it even better, connect a small capacitor (for instance 10nF) or RC block in parallel to the diode.

In case you cannot move the frequency and the other things don't improve the situation, you could add a low frequency LC filter, but that would cost a bit and be bulky.

 

[Harald Kapp]

Smoothing out the PWM would definitely thwart the idea of using PWM in the first place: reduce losses.
You could go directly for a linear regulator in that case.
I think *steve*'s and electrobrain's tips are a good start to isolate the source of the hum and to finally reduce it.

 

[apples]

1. There is a trim pot on the board that you can adjust. The motor hums at all the adjustable frequencies. At the moment it is the quietest position I can find which is fully wound in one direction.

2. Diodes. What to use? Some of my bigger diodes are:

1N5408
http://www.diodes.com/datasheets/ds28007.pdf

1N5822
http://www.futurlec.com/Diodes/1N5822.shtml

But as the motor surges at startup at over 10 amps maybe I can not use these ones? Well the +10 amp surge was when I was testing before I had the PWM controller. I just hooked it direct tot he 12 battery. Maybe now with PWM there will not be such a high start up current?

The diodes seem to have heaps of voltage handling capacity the 1N5408 is rated at 700V RMS. But its current rating is only 3.0A (average rectified current output).

But has a non-repetitive peak forward surge current 8.3ms single half sine-wave superimposed of rated load of 200 amps. Gee that would be nice to know what that means in english.

Electrobrains,

I first suggest you put a (power) diode directly over the terminals of the fan (cathode on the plus side!), to catch those spikes that are seen on the oscilloscope (and are running along the whole cable length). That will reduce EMC/EMI problems and might even remove your noise totally (it will keep the current constantly flowing inside an inductive motor).


Is this what the diode will do? see picture.

 

[Harald Kapp]

You put the diode across the motor terminals in reverse polarity, meaning cathode to motor-plus, anode to motor-miinus.
The diode will not see the motro current (because it is inparallel to the motor, not in series). The diode will see the voltage across the motor, which is only 12 V (give or take some) when the motor is driven by the PWM controller. During the "off" state of the PWM signal the motor will act as a generator, converting the mechanical energy from its spin to electrical energy. The result is a negative voltage across the motor terminals. In this situation the diode will become conducting (because its cathode is at the moor-plus terminal where now the voltaeg is negative). A conducting diode has a voltage drop of ~0.7V, the current that flows will depend on the characteristics of the fan when acting as generator. You will have to measure this, you won't find it in the datasheet of the fan. You could try using the 1N5408 you mentioned.

However, I doubt that this will eliminate the noise. Give it a try and see for yourself what happens.

By the way: have you compared the noise with the noise comming from the fan when attached directly to the battery? From the video it sounds to me like the noise not comming from the PWM but from the fan itself due to the friction of the air.

 

[Electrobrains]

I think Harald gives a good explanation.

I would suggest using 1N5822, because it's faster and has lower voltage drop (Schottky).

I don't think any surge current would kill your diode. If there would be a problem, it would be over heating through the general high power (10A?), that would have to be carried for an unknown time each cycle (most critical when the motor is slowing down - acting as a generator).
But it's easy: Check the temperature of that diode for a while when you start testing and quickly switch off if it's getting too hot.

You could also increase the roughness of the diode, by putting two of them in parallel (twist them together in the same direction). Due to the inner resistance, it will distribute the power a bit.

When you measure the signal, I suppose you will have a nicer square wave now. If not, then connect a high value resistor (eg. 1k) to the output and down to ground.

I include a drawing with the configuration of the motor having common plus connection. I suppose this is your case too, isn't it?

 

[(*steve*)]

The diode will not see the motro current (because it is inparallel to the motor, not in series).

Well......

Because the motor is an inductive load, the diode will see the full current that was flowing through the motor as the voltage driving the motor switches to zero.

The voltage generated by the freewheeling motor is less of an issue.

 

[Harald Kapp]

Right, I didn't express myself clearly. What I meant was the forward current during PWM high. Of course, the motor, being an inductor cannot have an infinite di/dt when the PWM signal goes low, therefore the current is forced through the diode during that time.

As to "the voltage is less of an issue": if the voltage on the motor's terminals is not limited, it will rise to a very high value since the inductor will generate as much voltage as is needed to keep the current flowing, You sure know that this is the reason why relays usually have a backward diode across the coil contacts.
But with the diode in place the voltage is limited to < 1V and then it is no longer an issue,

I wonder if the PWM controller doesn't have an internal protection diode exactly for this reason? So the PWM controller's transistors would be protected from such a reverse voltage without additional circuitry.

 

[Electrobrains]

..I wonder if the PWM controller doesn't have an internal protection diode exactly for this reason? So the PWM controller's transistors would be protected from such a reverse voltage without additional circuitry.

It sure has a protection. See the oscilloscope photo. The high energy spike is clamped down at the supply voltage level (or a few volts above).
In fact, the protection of the output transistor is always at hand. Even the body diode of a modern power MOSFET has a quite well defined break down voltage, that can be used for clamping (just be aware of risk of over heating).

Why I believe the noise level could be reduced, is not that clamping is made, but where it's made. The spikes would be caught at the source (motor), instead of loading the cable and system with high energy, high frequency pulses, that give many negative effects.

What is really strange with this circuit, is the PWM frequency. If I read correctly, the frequency would be as low as 130Hz!

 

[CDRIVE]

Apples, I watched your video and the Diamond plate caught my eye. Did you fabricate that housing?

 

[apples]

Yep I grabbed some scrap 3mm checker plate and welded up the housing.

I tried a didoe a few weeks ago and it smoked. What about this one here:
http://www.jaycar.com.au/productView.asp?ID=ZR1039

It is 100V @ 20amps.

I am going to get one to try as I am going into town now, but it says 10A per leg. I think that my fan will surge over 10 amps when it starts.

 

[CDRIVE]

<snip>
What is really strange with this circuit, is the PWM frequency. If I read correctly, the frequency would be as low as 130Hz!

You never answered this question from Electrobrains.

 

[apples]

From the document that came with the kit it says: "This provides a frequency range of adjustment between about 120Hz and 1.2Khz

Are you trying to say that above about 20KHz (actually that other person said 25-30KHz) it should not be audible to the human ear?

Can I measure that frequency with my scope some how? RIGOL DS1052E