case fans speed

JanxSpirit · 23 Jul 2016 at 11:31

danlightbulb said:
So am I best leaving it all on auto in the BIOS then, and not bothering with fan profiles in windows?

It's up to you. I just set all fans from the BIOS to run as fast as they can almost silently when the PC is idle at the desktop. They soon start to ramp up when needed.

doyll · 23 Jul 2016 at 12:17

JanxSpirit said:
It does on at least some Gigabyte Z170 motherboards. The fan headers are clearly auto-switching between 3-pin and 4-pin driving modes depending on the fans attached to them. I've checked all available modes with both 3-pin and 4-pin fans, and they all work the same with both fan types.

There is no mention of this in the manual for the Z170-Gaming K3, but it's definitely doing it with no need for user intervention. Even the CPU_OPT header does it which came as a bit of a surprise as I expected this to be strictly 4-pin only.

How is this auto-switching done?
I've seen the Asus 4-pin headers with a microswitch on PWM side that PWM fan plug trips. Didn't know Gigabyte has an auto system

danlightbulb · 23 Jul 2016 at 13:16

JanxSpirit said:
It's up to you. I just set all fans from the BIOS to run as fast as they can almost silently when the PC is idle at the desktop. They soon start to ramp up when needed.

I have a silent option in the BIOS, but won't that raise temperatures?

What temperatures are acceptable at idle and at full load (best I have to simulate full load at the moment would be the intel stress test).

I have noticed that the cpu fan is more noisy than the case fans, but it spins faster too - up to 1500rpm. The case fans are very quiet up to around 1000rpm then a bit noisier above this, maxing out at around 1300 rpm.

I could choose to run the case fans at 900rpm all the way to 60 degrees then ramp it up above this, it would still be fairly quiet. Or I could let them go all the way down to 600rpm at idle temperatures (say up to 40 degrees then start ramping them?)

I don't know which is the best configuration?

JanxSpirit · 23 Jul 2016 at 13:41

doyll said:
How is this auto-switching done?
I've seen the Asus 4-pin headers with a microswitch on PWM side that PWM fan plug trips. Didn't know Gigabyte has an auto system

I guess it must just be detecting an open-circuit on the fourth pin. That would be very easy to do with appropriate hardware/software.

JanxSpirit · 23 Jul 2016 at 13:49

danlightbulb said:
I have a silent option in the BIOS, but won't that raise temperatures?

Maybe.

What temperatures are acceptable at idle and at full load (best I have to simulate full load at the moment would be the intel stress test).

I think you're worrying about temperatures way too much. Anywhere below 95degC when stress testing is fine for the CPU. Thermal throttling will kick in then, assuming you've not disabled it in the BIOS. It's never going to get anywhere near that hot under normal usage anyway.

Bear in mind that all of the components you're using are designed to run flat-out for at least the period of their warranty with stock cooling and no case fans. Any case fans at any speed are a massive advantage over no case fans at all, so chill and enjoy. :cool:

doyll · 23 Jul 2016 at 14:00

JanxSpirit said:
I guess it must just be detecting an open-circuit on the fourth pin. That would be very easy to do with appropriate hardware/software.

Or would be detecting a closed circuit?

At least according to what a fan engineer told me. He said the PWM circuit in the fan sends power to the controller, and controller uses that power to control the signal back. Not saying that is in fact true, but just that is what I was told.

puppetworx · 23 Jul 2016 at 14:07

danlightbulb said:
So next question is what should I be trying to achieve in setting up profiles

I'd make sure they all have a smooth ramp so you don't hear any abrupt speed changes. Also set the fan speed of the exhaust fans to lag behind the intake fans to create positive pressure (for less dust in your case).

If you have any overclocks make sure the fans are at full speed before their point of failure too.

Tealc · 23 Jul 2016 at 15:21

doyll said:
Or would be detecting a closed circuit?
At least according to what a fan engineer told me. He said the PWM circuit in the fan sends power to the controller, and controller uses that power to control the signal back. Not saying that is in fact true, but just that is what I was told.

Open or closed circuit has exactly the same result with PWM fans used in computers. An always on signal on the PWM pin drives the fans at 100% and so does an open circuit (or 0% PWM).

As far as I know, and I have never reverse engineered a fan so haven't confirmed it, the 5v PWM signal switches on and off Mosfet transistor, which in turn then switches on and off, like a relay, the main fan motor. The open closed circuit thing and minimum speeds and what not is controlled by the controller chip.

Maybe I will pop my Apache open one day.

danlightbulb · 23 Jul 2016 at 16:28

puppetworx said:
I'd make sure they all have a smooth ramp so you don't hear any abrupt speed changes. Also set the fan speed of the exhaust fans to lag behind the intake fans to create positive pressure (for less dust in your case).

If you have any overclocks make sure the fans are at full speed before their point of failure too.

Thanks.

Interestingly the case fans don't seem to make a lot of difference. I set a flat case fan profile so they are steady at 600rpm, and only set the CPU fan to ramp up with temperature.

Ambient air temp in room - 26C
System temp - 38C with case fans at 40%
CPU idle temp - 31C with its fan on around 45%

During a stress test:
System temp - 38C with case fans unchanged
CPU temp - 58-63C with its fan ramping to around 70% (1350rpm)

So the system temp was unchanged.

I then repeated the test with the case fans at a steady 80% (around 1100rpm - marginally noisier but certainly not loud by my definition), and got exactly the same temperature results!

Not sure what this means. Extra airflow from case fans not making any difference. Maybe things will change once I get a GPU in there.

Also a strange thing I find - how can my CPU temp at idle be less than the system temperature? There should be nowhere for the CPU heat to go if the system temp is higher, and the two should equalise?

Edit - the fan speeds aint making a great deal of difference at all, so I think I'm just going to make them fairly quiet until temps hit 60C then ramp up to full speed. My system seems either to be at 30C at idle or above 50C when I'm doing something. Nothing in between, so there isn't a great deal of point in spending time tweaking fans in the 30C to 50C range.