variable speed control for 12 volt fan

[Sojourner]

The bottom line about heat from diode....design heat load of whatever 8 or 15 or whatever amps will be at a given factory spec. limit. However if you go over ampere rating....in time it will smoke to puff. The lower ampere draws thorough a given ampere rated diode the cooler it will be.

FWIW

Sojourn for Christ, Jerry

[boogiethecat]

John,
Motors work fine with "high frequency" switching, at least when you're talking about 4.5 kilohertz.   Maybe a few hundred Khz or Mhz would be too high for a motor but under 50khz it's a no brainer.  Your control will work fine.
EMI really isn't an issue with this kind of circuit.  The ONLY down side to using low frequency PWM (anything in the audio range) for me is that if you still have good ears, you may get driven nuts by the whine the motor makes at the frequency the PWM is operating at.  400hz is tolerable but for me up in the 2-6khz range sounds like a really annoying whistle.  Above 15khz for most of us old farts won't even be heard.

Tim's circuit, yes it needs a "catch" diode inversely parallel to the motor or as Rick pointed out.  More importantly, it's only half the circuit and needs a PWM generator... trivial for us elektronikers but elusive for someone who isn't in to electronics to fill in the blanks. As an example of how a mosfet is wired it's almost correct (catch diode missing), but it's useless as a "real" schematic- it needs a lot more parts to function.

Diodes and a rotary switch?  It'll work but electronically speaking it's EXACTLY the same as a rheostat, just scads more complicated to build and gives you steps instead of a smooth transition between settings.  If you're even considering this, I'd suggest that you just get a proper rheostat (although I would never use either myself).  As Rick properly pointed out, It'll waste exactly as much heat at a given motor speed as will a rheostat, as the diodes are basically functioning as resistors in this application.
  I do have a place I love to use diodes with motors... with my drinking water pump.  I found that my sureflo pump went too fast for my drinking water spigot, and so it pulsed on and off quite annoyingly.  Putting two hefty diodes in series with the motor slowed it down just enough that the spigot was now faster than the motor could pump and the pulsing went away.  But that's the only kind of application I'd consider using diodes to slow down motors for...otherwise too many parts to do the job poorly that a decent PWM would do better.

Stan, I'm surprised at your comments "The other drawback to any solidstate method is you cannot get full speed."  These days as I'm sure you're aware, mosfet "on resistance" in the "Switched" mode (ie how much resistance is measured across a mosfet when fully turned on) is down in the milli-ohms. It's totally common to see less drop across a fully on mosfet than the average wire hooking to it.  Full speed with solid state circuitry is an everyday occurance in cars, battery powered tools, and basically everywhere.  The way you used your mosfet was probably in the "follower" mode (a MOSfet with a pot in to vary the input signal) and in that case you'll never get fully "on", instead you'll only get your power supply voltage minus what it takes to turn on the mosfet, usually 2-3 volts.  But this isn't a good way to do motor speed control, because now you're using the mosfet in "linear" mode and just like the switched diode thing, your mosfet is simply acting like a variable resistor and wasting exactly the same energy as would a simple rheostat.


Anyway, the bottom line is that the best way to do this is to use PWM.  For what it's worth, that's what is used in every battery operated variable speed tool... because it's simple, efficient, cheap, small, and works the best.  Maybe a suggestion would be just rob the PWM circuit out of a 14 volt drill, and there you are....

Cheers

[Rick Brown]

Here is the PWM circuit I use to drive a 24 Volt 16 Amp blower motor -> http://home.att.net/~intermountainac/PWM_1.jpg

As one never knows what evil lurks beyond his circuit board the optical isolators (U18 etc) are a good bet here.  The signals C10 etc. may come from a source as simple as a 555 timer.  Note that the electronics side is a 3.3 Volt system so your resistor values may differ.  The 4 pin connectors attached to the transistors rotate CCW when mating with the isolator connectors.  The output is configured so that the electronics may control both high and low side drivers.  The IRF4905L transistors are P channel MOSFETS drawn as presented by a TO-220 package (a fabrication guide).  The 249K resistors are for keeping the transistors off when you remove the electronic control circuit.  Because we really switch the power to the load rather than the ground to a hot load this is a high side switch.  The lower right circuit shows the 'catch' diode (drawn as a TO-220 package) added for an inductive load.  Notice the diode is directly across a pair of wires that go directly to the load.  Local grounding as is common on our vehicles is poor practice for PWM circuits.  You should bring both the power and ground wires for your load directly to the diode. 

How fast to run it?  The transistor only dissipates power when it is switching (to a first approximation).  Measured switching time for the circuit above is about 4 micro seconds.  If, for simplicity, one assumes half of the full current and voltage exist during the switching time then the power dissipated by the transistor during that interval is (16/2 Amps * 24/2 Volts) * 4 micro seconds = 384 micro Watts.  This happens twice per Hertz so if you switch at 1000 Hz the power lost is about .75 Watts.  That is independent of the power delivered to the load.  Switch at 10,000 Hz and you need to worry about heating issues.  I run at 960 Hz and can hear that tone when low power is being supplied to the load, but for a load such as a pump or fan to be useful it's going to make some noise also which quickly swamps out the PWM noise.  In summary: Crank up he frequency until you can't hear it, but keep your finger on the transistors.
-RickBrown in Reno

[gus]

This has been a very interesting dialog on modern electronics to an old (vacumn tubes) Navy electronics type, but I'm not sure the question was ever answered?

I would be tempted to go with the simple but effective rheostat from NAPA!!

[JohnEd]

Boggie,

I have a PWM "kit" that i bought that has a switch in it numbered 1H46SB.  The description says that if you want to drive a device that takes more power than a small fan you have to add a heat sync.  I can get those but I am wondering what the max current would be with a proper sync.  Thanks

Rick,

I need a pwm that will handle 16 amps at 12 volts and the one you showed would do that....I guess.  I can't figure out the schem you provided.  i never heard of a optical isolator.  I am not clear on any of those drawings.  I can't tell what gets connected to what.  There are connections "C3 thru C8" and what goes there?  There is also a C3 thru C8 that appears to be terminal boards....connected to what?  Do the ckts on the left interconnect with the ckts on the right?  How?  I am really in the dark here and I was a tech for 30 years and that was tubes and transistors but i think I should understand something....and I don't.  Where is the component that varies the switching frequency?  I understood that caution you gave about keeping your finger on the transistor while increasing the freq.  Good advice!

Thanks and I hope you can give the guidance at the level of detail, read that as really dumbed down, I need to build one of these and implement it.

John

[Rick Brown]

Here are some rheostat calculations.  Keep in mind that Watts = Volts * Amps and Volts = Amps * Ohms.  Assume a 24V, 16A load is resistive and you want to cut the power to it by a factor of 2 (24V*16A=384W & 384/2=192 Watts) with a rheostat.  The resistance of the load must be 1.5 Ohm (24V/16A=1.5 Ohm).  At half power through the load (from the basic equation P=I^2*R) the current is 11.3A (sqrt(192/1.5)=11.3) and the voltage across the load is 11.3A * 1.5 Ohm = 17V and the voltage over the rheostat is 24-17=7V.  The value of a series dropping resistor (or rheostat) must be 7V/11.3A=.6 Ohm and the power is 0.6*11.3*11.3 = 79Watts.  Big rheostat, but who cares as long as the engine is running and you have some place to mount it?

PWM becomes useful when you run from the battery.  I calculated in another post above a PWM circuit to control the same load would dissipate about 0.75W.

For the circuit: The optical isolator is a needless complication for a simple one-device controller.  Optical isolators consist of a light emitting diode (LED) and a photo transistor (PT) housed in a light tight package.  When you pass current through the LED it shines on the PT and turns it on to conduct current on the load side.  Thus the load is controlled by light and a big spark coming from the load side will not disrupt the control electronics.  I'm using a micro-controller for control of the PWM among other things and so use optical isolation to protect the control electronics.

Notice the vertical rectangle with four connector pins represented below the text "IRF4905L".  That is a connector at the end of three 22 gauge wires which lead to the transistor mounted on a heat sink.  It plugs into the electronics board such that the top wire connects to the junction of the opto-iso pin 8 and the 1.00K resistor.  The unused pin comes into play when the circuit is a low side driver.

Sorry about the dual "C8 etc" notation.  C8 on the right is a connector number screened on the board.  C8 on the left is the name of a signal from another sheet.  Poor documentation practice: My excuse is I didn't think anybody else would ever look at it.  The larger circles with nearby "16ga" notation represent terminals on a barrier block interfacing to the outside world.

In my case a computer determines the switching frequency.  In a more practical, single-device, controller a NE555 timer like device would be the controller.  I see you can run a NE555 at up to 16 Volts so I suspect you could build a 12 Volt controller using only the NE555, it's timing components and the IRF4905L.

You can google up "NE555" or any of these device numbers and get the manufacturer's data sheets which tell how to use the product.

The TO-220 package is well suited for mounting to things metal which act as a heat sink.  Digikey part #345-1000 is thermally conducting, electrically isolating pads and part #4724K is mounting hardware.
-RickBrown in Reno, NV