Thermally controlling AC fans

The alternative to direct air cooling

Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee

Post Reply
pdf27
*Lifetime Patron*
Posts: 264
Joined: Wed Nov 12, 2003 12:59 pm
Contact:

Thermally controlling AC fans

Post by pdf27 » Mon Nov 08, 2004 1:24 pm

I'm toying with the idea of designing a thermal control circuit for one of the Sunon 120x120x38mm 240v AC fans. This would have what would effectively be a variable resistor in series with the fan, and the resistor value will be controlled based on the temperature returned by a thermocouple placed somewhere in the water system, probably in the reservoir.
Firstly, does anyone think this is a bad idea (and if so, why?)
Secondly, does anyone have any suggestions as to what temperatures to vary the resistances over? The current plan is to switch the fan on at 35 deg C water temperature, and to full power at 50 deg C.

ferdb
Posts: 124
Joined: Fri Sep 10, 2004 1:13 pm
Location: austin, TX

Post by ferdb » Tue Nov 09, 2004 5:05 pm

No it's not a bad idea at all. You need to vary the resistance from 0 at the fastest to about 1k - 2k at the slowest. 35C and 50C seem like good starting setpoints. The easiest think to do would be to use one of the 1K 3watt pots from allied to set the low speed rpm (stock # 522-0049), and something like a pump relay to short out the pot at some temperature point and run the fan full speed. It's not proportional but it will protect your system from extremes. You could use a Selco Thermostat switch (stock #557-0026) which closes at 130F and opens at 110F and mount it on the resevoir, if you don't mind the temperature ranges it switches at.

Making a proportional temperature control is a little more complicated especially if you don't want the thermistor to be on the 120V side of the system (and thus connected to your resevoir etc.)

pdf27
*Lifetime Patron*
Posts: 264
Joined: Wed Nov 12, 2003 12:59 pm
Contact:

Post by pdf27 » Wed Nov 10, 2004 5:19 am

I'm UK rather than US, so the components (and voltages) will be slightly different. The plan is to have about 4 or 5 solid state relays shorting out a series of resistors giving me a series of different fan speeds. The control circuit isn't all that complicated, and I'm fairly happy with the logic circuits. The idea is for a thermocouple to run through an op-amp to all the logic circuits in paralell. These will trip at various speeds, and so give me some measure of proportional control. Right now I'm just looking to see if there are any obvious flaws in the concept, rather than in the detailed design (which should be easy enough to test).

hvengel
Posts: 205
Joined: Sat Oct 19, 2002 12:06 am
Location: Concord, Ca

Post by hvengel » Wed Nov 10, 2004 8:41 pm

Have you had a look at these AC speed controls? http://www.controlres.com/catalog.htm#1

pdf27
*Lifetime Patron*
Posts: 264
Joined: Wed Nov 12, 2003 12:59 pm
Contact:

Post by pdf27 » Thu Nov 11, 2004 5:09 am

Had a quick look. They look to be far bigger and more complicated than what I'm planning, can't go to a low enough voltage (I want to be able to have the fan off below a certain water temperature) and look like they'll probably be quite expensive.

hvengel
Posts: 205
Joined: Sat Oct 19, 2002 12:06 am
Location: Concord, Ca

Post by hvengel » Thu Nov 11, 2004 3:46 pm

These limit the min. fan voltage to 30% and this can be set for 30%, 40%, 50% or 60% of supply voltage. This does not seem like a major limitation to me since 30% would be like running a 12V fan at 3.6 volts. In addition the web site says to contact customer support if you need some other setting for the idle speed. In fixed speed mode you can go as low as 27%. There is a switch that allows you to decide if the fan will run or not below the low temp set point. So turning off the fan is not an issue. You can also select the delta between the low and high temp set point of 4C or 10C and the high temp set point can be 30C, 35C, 40C or 45C. You can also set the loop response time of the unit. There are many options for sensors including 11 temp (one is for liquid), air pressure, humidity and flow sensors. You can connect as many as 3 temp sensors and it will use the highest reading. You can also use an external DC voltage (0V to 10V) to regulate the speed. There are complete specs in PDF format on the web site and the one I looked at was 9 pages long. I was looking at the Nimbus unit. Yes this unit is over kill and may be more than you need but it looks to me like it will do about anything you might need. The real question is how much does it cost and is this too much. I did not check on cost so I don´t know what these cost and it is not listed on the web site.

ferdb
Posts: 124
Joined: Fri Sep 10, 2004 1:13 pm
Location: austin, TX

Post by ferdb » Thu Nov 11, 2004 8:26 pm

pdf27 wrote:I'm UK rather than US, so the components (and voltages) will be slightly different. The plan is to have about 4 or 5 solid state relays shorting out a series of resistors giving me a series of different fan speeds. The control circuit isn't all that complicated, and I'm fairly happy with the logic circuits. The idea is for a thermocouple to run through an op-amp to all the logic circuits in paralell. These will trip at various speeds, and so give me some measure of proportional control. Right now I'm just looking to see if there are any obvious flaws in the concept, rather than in the detailed design (which should be easy enough to test).
You might try switching in different sized capacitors instead of resistors. This is how they do those 3 speed wall mounted ceiling fan controllers. That way you won't have any significant power dissipation to deal with. You'll need to experiment to find the values that work well for you, they'll be in the 0.2 - 10uf range. Note that at some range of values the voltage across the fan will actually go above the line voltage because the inductance of the fan and the capacitor are resonating. Don't forget to include a snubber circuit so you don't kill your circuit from the inductive kick when you turn the fan off.
There might be helpful info here
http://www.fairchildsemi.com/apnotes/an_bydoc_d3.html
http://www.fairchildsemi.com/collateral ... er_ssr.pdf
a number of usefull parts in there for your project. Good luck

Lwood
Posts: 25
Joined: Tue Nov 11, 2003 12:17 pm
Location: Maryland, USA

Post by Lwood » Fri Nov 12, 2004 5:55 am

To play Devil's advocate....

I don't think I would go with a setup like this. I haven't watercooled, so you can take what I say with a grain of salt. With watercooling, there shouldn't be that much variance in temperature, assuming that changes in ambient temperature aren't great. In that situation, the temp/fan speed should settle and remain fairly stable over the long term. Experimentation with a manual fan controller should allow you to find that sweet spot. Using a temperature feedback system you run the risk of a system cycle, where the fans keep ramping up and down, which could be loud and annoying.

pdf27
*Lifetime Patron*
Posts: 264
Joined: Wed Nov 12, 2003 12:59 pm
Contact:

Post by pdf27 » Fri Nov 12, 2004 7:13 am

Lwood wrote:To play Devil's advocate....
I don't think I would go with a setup like this. I haven't watercooled, so you can take what I say with a grain of salt.
Neither have I, so this is entirely theoretical.
Lwood wrote:With watercooling, there shouldn't be that much variance in temperature, assuming that changes in ambient temperature aren't great.
Why not? The water temperature won't vary that much due to the much higher thermal capacity of water compared to air, but the amount of heat dumped by the radiator to atmosphere will vary with the heat dumped by the CPU to the water system. There will be a time lag, but the principle remains the same.
Lwood wrote:Using a temperature feedback system you run the risk of a system cycle, where the fans keep ramping up and down, which could be loud and annoying.
There won't be any major ramping up and down, simply due to the high thermal capacity of water. However, if I simply run the thermocouple directly to the circuitry then it could cycle between two fan states in some circumstances. Ideally I would like to have some kind of set/reset circuit so that the fan system only checks the thermocouple every 10 minutes or so and changes the fan speed based on that. The logic circuitry could be problematic, depending on if I can find a clock signal from somewhere.

ferdb
Posts: 124
Joined: Fri Sep 10, 2004 1:13 pm
Location: austin, TX

Post by ferdb » Fri Nov 12, 2004 8:51 am

the amount of heat going in the system can vary a lot from idle to full bore computing or gaming. Temperatures can swing as much as 15-20C going from idle to full load if you have fans set low on a WC system. The limitation is trying to dump all that heat into the air going through the system and if there is a lot of heat and not much air the temperatures go way up.

Post Reply