My setup is DQ77KB / Core I3-3225 / 16GB / MSata SSD / Windows 8.
My question relates to how the fan control settings in BIOS change the behaviour of the CPU fan. Under most condition the CPU temperature in my system hovers around 32C with the CPU fan almost always spinning at ~1100RPM. I'd like to determine if under low load conditions I can get the CPU fan to turn off completely in order to reduce/eliminate fan noise. In the BIOS I've tried different settings for Control temperature, Over Temperature, All-on Temperature, Min Duty Cycle, Max Duty Cycle but nothing seems to change the spin rate of the CPU fan. For example if I set the following in the BIOS:
Control Temperature: 55C
Over Temperature: 65C
All-On Temperature: 75C
Min Duty Cycle: 0%
Max Duty Cycle: 50%
With a CPU temperature of 32C shouldn't these settings cause the fan to NOT rotate at all since Min Duty Cycle is set at 0%? and only start spinning when a Control Temperature of 55C is reached? or am I mis-interpreting the meanings of these settings?
Minimum Duty Cycle sets the minimum for the fan to run but still within the fan's RPM range of operation. In different words, it sets the lower speed bar and may actually force the fan to work faster at a higher value set. The RPMs will never be zero because the fans always have minimum and maximum RPMs.
The Maximum Duty Cycle establishes the top limit, and therefore it can lower the fan's RPMs when running faster. This is normally not recommended if the processor is going to face heavy loads.
I would leave those settings as default and set the other temperature values (Control, Over, All-On…) to a high value, (55C - 65C is probably ok) because the fan will speed up once the processor reaches such values.
Let me try to do a slightly better job of explaining what Joe is saying. The Intel boxed fans are Type A 4-Pin fans. That is, there is some minimum duty cycle that these fans will respond to. If the duty cycle falls below this minimum, the fan operates as if it was set to this minimum. For more information about this, you can read the 4-Pin fan specification (available here: http://formfactors.org/developer/specs/4_Wire_PWM_Spec.pdf http://formfactors.org/developer/specs/4_Wire_PWM_Spec.pdf). In general, this minimum is somewhere in the 20%-30% duty cycle range.
Warning: the following information is specific to 7 Series boards; it does not apply to 8 Series boards and it only partially applies to 6 Series boards.
The most efficient settings for the processor fan are typically the default settings. The fan speed control subsystem in the BIOS reads the "Tcontrol" temperature from the processor and establishes the fan speed control curve based upon it. The "Tcontrol" temperature is established during the validation of the individual processor's silicon. By definition, this is the temperature above which, in the absence of other information, the processor fan must operate at 100% duty cycle. The fan speed control curve implements a transition from the minimum duty cycle to the maximum duty cycle over a 15 degree (Celsius) range. The top of this range is anchored, by default, at the processor's "Tcontrol" temperature. The "shape" of the fan speed control curve between the two temperature limits is dependent upon the setting of the Responsiveness parameter. This is best explained by a diagram:
Ok, so far, seems simple enough, right? Here's where we throw a wrench in the works...
Each fan speed controller is capable of responding to the hotter of two temperature inputs. By this I mean that the higher of the duty cycle requirements of the two temperature sensors associated with it is going to be applied. In the configuration parameters for each fan, there are two parameters, Primary Temperature Sensor and Secondary Temperature Sensor which specify which two temperature sensors are associated with (and responded to by) this fan. You can choose to specify "None" for the Secondary Temperature Sensor parameter if you only want one temperature associated, however.
Here's the other confusing factor. Due to hardware limitations, the above information on fan speed control curve "shape" applies only to the primary temperature sensor. For the secondary temperature sensor, the fan speed control curve has a completely different "shape". This is best explained with another diagram:
I hope this gives you a reasonable introduction to what is going on.
Insofar as the processor fan is concerned, stopping the fan is not a very good idea. Proper thermal response is based upon a consistent application of cooling, not something that becomes hit or miss (consider: some fans can be slow to react to changes in duty cycle, especially with off to on transitions thrown in). Secondly, for best acoustics, the best Control Temperature is the processor's "Tcontrol" temperature. Some folks say - and I don't necessarily disagree - that lowering the Control temperature to, say, 5 degrees below the "Tcontrol" value can be a good thing if you want to establish a thermal buffer that will absolutely ensure that you see the best possible results in Turbo mode. Lowering it any more is typically unnecessary, however...