• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

Cheap Alternative to PWM?

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

wormwood

Member
Joined
Sep 7, 2002
Location
Tucson, AZ
I think I found a cheap and simple alternative to PWM, but first let me explain the problems I've had with PWM. 7V'ing fans was out of the question, as they were still too loud, any lower and the suckers wouldn't always start.

I've done a number of PWM fan controllers, and they all exhibit one common problem. Certain fans (like Delta) create a humming sound depending on frequency. Even at frequencies that should be inaudiable (proven on my scope) harmonics come in to play. On most fans, it's not a problem.

The second issue is that they all have ICs, meaning they are overly complex for the average person. Granted, they are simple ICs (555's and the like) but a lot of people freak on anything over 3 pins.

The problem with simple voltage reduction is that a fans start-up voltage is a bit higher than the minimum it will run at, i.e. setting a fan to 6 volts it will still run, but it may not start.

The advantage of PWM is obvious: An retrigerable timer connected to the fans feedback allows you to pump VCC to the fan if it isn't running, meaning it can be set to a lower speed safely.

Last night, I figured out a REALLY simple solution. Power On Reset, just like we put into digital ciruits. No chips or anything.

Basicly, the whole circuit is just a TIP-120 (could use FET), a pot, and a large cap (I used a 220uf). A thermister could be used in addition.

The cap is attatched between the base and VCC (12V). Center of the pot is attatched to the base. The pot is also connected to VCC and ground. The collector is attatched to VCC, and the emitter goes to the positive lead on the fan.

The net result is the fan gets VCC for a couple of seconds, and then drops down to the set voltage.

Why is this so neat? Say you have a thermistor. When you first fire up your computer, it's relatively cold, and thus the minimum speed would be the fan start-up voltage vs. the slowest speed the fan can run at. By adding the RC circuit to the base (power on reset), the low could be the minimum running voltage, making the system much quieter.

In case your wondering, yes, the pot is doing double duty. Not only is it setting the voltage of the fan, but it's also acting as the "R" in the RC circuit.

Despite chips like 555's and 741's being dirt cheap, they aren't free. The entire circuit I described can be built for a few bucks.

$0.65 TIP-120
$0.85 10K miniature potentiometer
$0.30 220uf electrolitic (radial)

These are all qty 1 prices from a local supplier here in Arizona. They can be had cheaper elsewhere, especially in higher quantities.

On a side note, I have a couple delta 120mm fans running off it with no heat sink on the TIP-120. For more than that, a heat sink might be a good idea, but they can be had for pennies.

It's kind of neat when these Delta 120mm's kick in for a couple seoncds, and then drop down to barely moving. They are still pushing a moderate amount of air, but you have to put your ear right next to them to hear them (watch your hair if you don't have a fan guard!!!). Even at 7V, these fans could be heard quite easily, so this is a very noticable improvement.

Anyway, if this is all just common knowledge, I appologize. It's the first time I personally have ever used (or seen) a Power On Reset in an anolog circuit, though.
 
I've been looking into a way to control the speed of my fans, and the article on procooling forums really convinced me that pwm is the way to go(for me anyway). I will agree that the circuit is much more complicated than a rheostat or voltage regulator, and has taken quite a bit of searching on my part to understand what is going on and pick out the necessary components to order. You did bring up an interesting point, something that I have been wondering about, the noise the fan might make. Have you tried different capacitor sizes to try and mininize that vibration? What frequency was the 555 timer running at? I expect my parts to arrive sometime this week and I'm going to set it up on a solderless breadboard first and do some experimenting. Once I make any necessary adjustments I'll solder everything up and install it. That's a nice idea to bypass the transistor with a capacitor, it definately makes the circuit more versatile.
 
You can add the cap to just about anything, including a PWM circuit or one of those pre-made fan bus systems. It allows you to run at a lower speed (or voltage in this case, less duty cycle in PWM) because the fan is always able to start-up.

Trust me on this, I've built several variations of that PWM circuit (it isn't the orriginal source), and unmodified, the circuit I described earlier is far superior (and costs about 1/4).

The first problem with the cirucit isn't so obvious until you hook up a decent scope. The fan is basicly a large inductor. When the voltage is cut (741 acting as a comparitor) all the residual energy (up to -100V) is pumped into the diode. This is not only hard on the components, it is also hard on your power supply. The solution is simple: A small RC circuit is put in prior to the 2N2222 to round off the edges at about 1V prior to high and low. This way, the residual feedback never goes out of range, and the excess energy is used in physical motion. The effect is the square wave has rounded edges.

The second problem with the circuit is near fatal to the concept. At really low frequencies, you get thumping. But at really high frequenies, the voltage isn't on long enough to start the fan and thus can't work any better than a rheostat set to the average of the voltage/duty cycle. The reason is simple: It's an inductor. It takes a bit of time for the voltage to reach full value.

With the exception of the Deltas, I've had great results at about 800Hz-1KHz. I've tried nearly everything from 3Hz to 50KHz on many diferent types of circuits. None of them where able to do any better than the simple circuit I described, even with all the extras (adding Power on Reset, RC to handle feed back, and even a seperate stage to watch the pulses comeing back from the fan to force VCC if the fan wasn't running). The point is, it took 2 555s (1 for the freq generator, one as a pulse watcher), a 741, 5 caps (3 of them large), 2 pots (1 to vary frequency, one ot vary duty cycle), a diode, a FET, and a 2N2222, all to achieve the same results (ultra quiet fans) as the 3 parts solution I described earlier.

Prior to a few days ago, no one was more of an advocate of PWM than me. 7 Volting a fan just isn't quiet enough, and up until a few days ago, I thought PWM was the only way to achieve less noise than 7V. I was wrong, and the solution was so simple, I just couldn't believe it.

Really, you owe it to yourself to give it a try. The 3 parts I mentioned can work fine on a PWM circuit (one of my PWM circuits had a TIP-120 vs. a FET, I wanted minimum cost and parts count). The 220uf cap isn't listed on their circuit, but trust me, you want it between VCC and ground, or else your 555 can become unstable (the fan does way higher than rated current draw on pulse on, and that sag can trigger one of the op-amps in the 555). Of course, the 10K pot is also in their circuit.

Regardless of what you do, here is a link you might find helpfull.
http://www.cadsoftusa.com/
They have a freeware package for schematic and board layout that is quite nice. The difference between freeware and what you pay for is parts count and board size, but both are large enough that no mater how complex a PWM circuit you design, it'll handle it. It even has a autorouter that's pretty OK.
 
Excellent, thanks for the insight. From what you describe you wound up with a really elaborate pwm circuit.;) I'll definately try adding a large cap to my pwm when I get the parts. I did a google search on rc circuits and came up with this site: First Order RC Circuit. It demonstrates nicely the smoothing effect the circuit has on a sudden on and off current. Basically this take the on then off output of the FET and rounds off the edges of the wave, correct? I was also thinking of putting a cap across the +12 and ground leading into the circuit to hopefully absorb and fluctuations cased by the pulsing. Hehe, now I can't wait to set it up!:p Thanks for your post.
 
Back