quick Q about diy pc fan speed controller.
Mar 6, 2007 at 6:05 PM Thread Starter Post #1 of 19
drkazndragon

drkazndragon

Head-Fier
Joined
Apr 6, 2006
Posts
66
Likes
0
what type of pot (potentiometer) should i use? found some 100k linear pots at frys but the only thing that one does is turn the fan on and off... the fans are 12 vdc btw. any help would greatly be appreciated, thanks.
 
Mar 6, 2007 at 11:44 PM Post #2 of 19
Are you using the pot on it's own or with a small transistor/pwm circuit ?

I believe using a pot on it's own with a fan generally isn't all that good for them as pots aren't meant for high current. For that sort of purpose without some circuitry you'd use a rheostat.

Pot choice well, use linear, the resistance depends on how you're using it, how much adjustment you want etc.
 
Mar 7, 2007 at 12:05 AM Post #3 of 19
im going to use the pot by itself. the 80mm fan im using has a max rpm of maybe 5k. so i want adjustments somewhere around 35%~75% rpm speed. i want my pc to be as quiet as possible, i even underclock the cpu so it wont heat up too much. on 24/7 ^_^
 
Mar 8, 2007 at 12:17 AM Post #4 of 19
Quote:

Originally Posted by splaz /img/forum/go_quote.gif
pots aren't meant for high current. For that sort of purpose without some circuitry you'd use a rheostat.


Go rheo, definitely. I started into a project like that last year. I stopped when I nearly killed my expensive power supply (and that is with everything hooked up correctly...oy!) It just didnt like the connections, I guess. Beware!
 
Mar 8, 2007 at 2:00 AM Post #7 of 19
Reasons why your 100K pot won't work is that you need to drop a few volts, 5 to maybe 7 off the 12. I have a 120MM fan here, quiet-ish and it takes .08 Amps at 12 volts. One would need to guess or better, measure what actual current yours would require at lower voltages and lower speeds.

To calculate a voltage-dropping resistor's value, one needs to know the current, and the voltage drop at that current. V = IR and R=V/I.

I've guestimated a few numbers below to give you an idea what ohm values you would have needed. Clearly, a 100,000 Ohm pot won't have the resolution to change speeds. A 1k or 2K might. A variance of 130 ohms out of 100000 is a 0.1% change!
Power in watts = Current squared times resistance, and is right at or beyond the max for some larger pots. so your pot would survive only a short while.
R = 5v / .05A, R = 100 ohms, P = 0.25 watts
R = 6v / .04A, R = 150 ohms, P = 0.24 watts
R = 7v / .03A, R = 230 ohms, P = 0.20 watts

The top one would be fastest, dropping 12 volts down to 7. As you'll read through SilentPC, you'll note common experience shows many fans will run at 5v, but fail to start turning on their own at so low a voltage.

I suggest you use technique other than a pot as other head-fiers have remarked!
 
Mar 8, 2007 at 8:27 PM Post #9 of 19
Pots don't allow high currents - at least not those you'll find in brick and mortar stores. Rheostats are really expensive.

I'd use the LM317 route, but a trade off is the double diode voltage drop (~1.2V), so 12V is never possible. Making such a simple $3 circuit is easy and reliable.
smily_headphones1.gif
 
Mar 9, 2007 at 12:35 AM Post #12 of 19
lm317, simplest and easiest way to have a linearly variable voltage control of the fans with very cheap parts, all you really need is a lm317 chip, a resistor from the out to adj pins and a pot from adj to gnd and put a small heatsink on the lm317 and that is it

Tangent's 'tread' circuit board does the same thing with just a whole lot of extra unnecessary parts
 
Mar 9, 2007 at 1:17 AM Post #13 of 19
When I was more into overclocking, I experimented with any number of devices. In the end, my interest in overclocking waned, and I now rely on good airflow and low power fans to get the job done.

That being said, I did build a fanbus with some of those round pots from Digikey and it worked just fine. The key was to use the fan specs and ohms law to calculate the right value.
 
Mar 9, 2007 at 1:18 AM Post #14 of 19
One big question in the design that you'll want to ask is how efficient you want the controller to be. You might be better off asking people at SPCR for this; after all, quoting DaKi][er, "Audio electronics- Where we strive for inefficiency" =P I'm going to do a general overview anyway.

There are two general types of fan controllers you could use: switching voltage regulators and linear voltage regulators. Linear ones are easy to implement and very low noise (which is why we use them in audio here all the time). However, they are by nature very inefficient. Switching regulators are electrically noisy, and usually quite a bit more complex to implement, especially if you're working towards low noise (which luckily isn't very important inside a computer, which is already very electrically noisy), but they are considerably more efficient. There's a PWM controller often used for motor and fan controls, which is kind of like a switching regulator. Problem with this (you'll probably read this on SPCR) is that it may make your fan a bit noisier because of its method of operation. There are also switching DC regulators, which would probably reduce this (mechanical) noise significantly--might want to think about using one of these, although I personally haven't seen any fan controllers which use it

you'll want to verify this info on spcr, I haven't really looked into this much after finishing my new computer a couple of months ago
 
Mar 11, 2007 at 1:22 AM Post #15 of 19
As already mentioned, a POT-like device with enough current capacity would typically be called a rheostat.

The best mating of fans to any controller is to use fans with a max (@ 12V) RPM spec that is just a little more than you might ever need, not a 5K RPM fan you will always run at less than 1/2 speed (which is what I expect, 2500 RPM is still a little audible and a reason to move up to a larger fan or heatsink if a nearly inaudible system is the goal). The reason for this is that below a certain threshold the fan motor torque causes a pulsation, you can actually run one at a slowest absolute value if picking the slower spec'd part in same fan family. Ideally you want the resultant controlled speed of the fan to be as high a % of the original speed as possible, to arrive at any particular RPM result. It just takes some experience or trial and error if you have a new type of case or parts to see how temps respond to fan speed changes.

Generally with the fan you have proposed you'd need a resistance around 68Ohm (+- a few dozen ohm). With a slower default speed fan, particularly those very large fans which hardly need to spin at all to move enough air in an underclocked PC, you might need rheostat values up to about 150Ohm. Thus for best control you want to select a rheostat valuejust above your desired upper value.

The thing about rheostats is that while they're easy to implement, they also get quite expensive. Like with power resistors you don't really want to run one at it's max wattage rather derating it so it doesn't run hot. Last thing you want is to add complications, significant failure points to a vital cooling subsystem in a computer, it should stay as reliable as reasonably possible.

For best fan control you do not want a switching or linear voltage regulator unless you are only using very cheap fans which have terrible motor torque, undersized magnet. By "best" I reallymean smooth operation at low RPM, not best range of adjustments on most of the RPM range till you get to a lower thrshold, which can be a wash depending on exactly what is compared and contrasted among the different methods of control.

Otherwise the slower the fan turns, the more the fan starts to stall and lurch faster again (X times a revolution, 4 with some) with only a voltage control, making it even louder that it could have been but with lower airflow too. To keep the fan turning smoother at low speed the result is better with current control (limitation). A rheostat will do this, or a simple series resistor, or even an LM317 in current, not voltage, regulation mode.

Personally I don't want or need knobs sticking out on a PC so I often use one rheostat rigged with a couple fan adapters (4 pin molex socket and 3 pin mini 0.1" types) to dial-in the resistance any particular fan-PC-environment combination will need, then having that value a cheap and unobtrusive 1-2W resistor can be chosen for permanent installation instead of the rheostat.
 
You must log in or register to reply here.

Users who are viewing this thread

Total: 2 (members: 0, guests: 2)
Back
Top