ideal lowest PWM frequency for fan

[Simon]

I need to drive a spal blower in an air-conditioning box with PWM, now my experience of PWM drive tells me that the fan motor will "sing" as the rotor oscillates back and forth. I don't want to go too high in frequency as that will just make my driving MOSFET hotter and limit the PWM resolution due to the limitations of the driving PIC. Is there a sweet spot or do i have to just get just above most peoples hearing range ?

I remember making a PWM drive with a 555 soldered to screws in a blackboard eraser to run the blower i ripped out of my car before scrapping it. I got an 80W fan to run off a 30W supply (up to a suitable speed so as to not overload the supply). I had a cap attached at one end only to I could push it into parallel to slow down the frequency, at a lower frequency the power transistor (that I think was a BJT) didn't get as hot and the fan speeded up ! but the regulation was not so fine tuned, with the higher frequency i could get lower RPM before cut off but it was certainly using more heat in the BJT.

[nctnico]

Did you try to use a diode to discharge the base of the transistor? It usually helps to make a BJT switchoff faster and thus reduce dissipation.

[Simon]

well in this case I'll be using a mosfet, will a mosfet behave better ? my main question is on the frequency and noise.

[SeanB]

How about using a lowish frequency and adding a LC filter on the output? Will help with EMC as well, no fast rise time pulses, and smooth DC to the motor.

[Simon]

so essentially creating a SMPS type output but not regulated with feedback. hm, what sort of size components will I need for a 12.5A motor ? space is a premium. will using variable DC be as good as PWM ?

[benemorius]

You nailed it - the audible frequency range is often the primary limiting factor in such cases.

Variable DC is not as good as PWM in at least one significant way. Torque is greatly reduced by reduced voltage, so you lose the benefit of being able to drive the motor to a lower minimum RPM as you could using pulsed high voltage. For instance, a 12V 5000 RPM fan might stop spinning completely at a reduced voltage of 5V with an rpm of 3000, making 3000 RPM its minimum speed. However, the same fan might be capable of going much lower to just a few hundred RPM if you pulse the drive current at its rated voltage of 12V. Thus the minimum RPM is reduced and the total range of operation becomes much greater.

Also, heat dissipation in the fet can be reduced by increasing the edge rate of the gate voltage during switching and therefore reducing the amount of time it spends conducting in its saturation region.

[Smokey]

The motor "sings" if you PWM in the audible frequencies, ie 20hz to around 20khz.  Most industrial servo amplifiers start at about 25khz.  That's about as low as you want to go to avoid the motor singing.  Except for being annoying and having courser current regulation, lower frequencies won't necessarily hurt the system.  I'm not sure what your circuit is, but you should check out IGBTs for switching amplifiers.  It should say right on the data sheet what the rated switching speed is.  There is usually a grade right around 25khz, and another grade around 100khz.  Another good spec to look at in the data sheet is the current rating at high temp, usually 110degC or so.  It's dangerous to design around the 25degC current spec for something like that.  It's going to get hot. 
You didn't say what voltage you want to run at, but for 12A continuous you are probably going to want to use PWM.  If you go linear you are going to have to have a huge heatsink.
A good place to start learning about servo amplifiers is the OSMC (open source motor controller) project.  They use parallel TO-220s in the air, which you could replace with one TO-247 screwed down to a backplate, but still a decent design.  Then once you get that you can start adding current feedback and make a real controller.
Careful with filtering the output of a PWM.  The components all need to be rated at the current level of the system, and 12A inductors are bigger than you think they would be at frequency.

[Simon]

I'm driving a 24V 12.5A fan, no I don't think DC will be an option. I'll do some tests just under 20KHz

[David_AVD]

Have you checked the input capacitance spec of the MOSFET you intend using?  Some can be really high and with a standard logic drive the MOSFET can spend a lot of time in that slope region.

[Simon]

I’m not familiar with the blower or what the application is I’m assuming automotive based on your other post. If you want to get the full performance out of the blower you have to deliver is rated current . If you don’t need a 12.5A blower save some money and buy a smaller one.

What the devil are you on ? haven't you ever heard of a variable speed fan in a vehicle, yes this is for a vehicle and there is a nice speed pot to go with this so I have to do something with it 

Thank you for your suggestion of a mosfet, I'll have 12V inside the control system anyway.

[Simon]

Have you checked the input capacitance spec of the MOSFET you intend using?  Some can be really high and with a standard logic drive the MOSFET can spend a lot of time in that slope region.

I was going to use the attached, or should i use a driver IC ?

[Simon]

I think I'll use a driver IC, i can get one for the same price as the descrete parts. I see that most mosfets have their characteristics defined for VGS of 10V, anything wrong with 12V if the max is 20V, is 10V just a standard used for most parts for comparison ?

[SeanB]

Use a driver IC, as it will do the level conversion from 3.3/5V as well which you still need with the complementary pair. It will probably be smaller footprint wise as well, though you probably will need a small copper area as heatsink.

Nothing wrong with 12V, you will get a slightly lower on resistance, though using a 15V rail will give the lowest, but with a slightly higher driver dissipation.

[Simon]

Well i don't want to over complicate matters, I'm quite happy with 12V as I have a 12v source in the circuit.

[Simon]

The driver IC I have and i expect all those I'd buy in 8 pin dip/soic is 2 channel, can I parallel them for more output?

[nctnico]

I think I'll use a driver IC, i can get one for the same price as the descrete parts. I see that most mosfets have their characteristics defined for VGS of 10V, anything wrong with 12V if the max is 20V, is 10V just a standard used for most parts for comparison ?

When designing for automotive the circuit should be able to withstand 60V at the power supply input. An NE555 should have enough output power to drive a MOSFET. The trick with MOSFETs is to choose one which can handle just enough current. And don't get blinded by the lowest RDSon. More current and lower RDSon mean a higher gate capacitance which slow a MOSFET down.

[Simon]

so an irf540n migyt not be that bad after all ? mind you i'm vering to a to-247 to ease assembly

[SeanB]

If you have it use it, only if you needed a extra rail you can tweak it then.

What are you doing? Retrofit a refrigeration controller?

[Simon]

I'm designing an air conditioning box ECU so I need to control fan speed and a water valve position based on temp settings and sensors