Using CPU Fan Header For PWM Fan Hub

[devocean]

Recently, I was debating which specific motherboard I should purchase for a budget build with 4x ARCTIC F12 PWM fans (2 dual ball bearing, 2 fluid dynamic). I decided that a Silverstone CPF04 (1-to-8 PWM fan hub, SATA powered w/ motherboard 4-pin connector for the speed control signal) would be most effective.
However, after reading the manual for the motherboard I had chosen (B150M Mortar), I realized that the system fan headers are not "true" 4-pin PWM headers, as the fourth pin is labeled as "No Connection". To circumvent this, I instead chose this motherboard:

https://us.msi.com/Motherboard/B150-PC-MATE.html

The B150 PC Mate has the same system fan headers, but also has an additional CPU fan header. The two CPU fan headers are, in fact, "true" 4-pin PWM headers.

So, my questions:
1. Can I plug the stock CPU cooler from my processor (i3-6100) into the CPU_FAN1 header, then the CPF04 fan hub connector (which will support 4x 120mm fans) into the CPU_FAN2 header?

1a. Assuming this is possible, will the fan speeds of the CPU cooler and the case fans attached to the fan hub be separate, or the same?

2. Would this work simply in one CPU fan header (assuming it is a "true" 4-pin PWM header), avoiding the need to purchase a motherboard with two CPU fan headers?

 

[Paperdoc]

You are reading and understanding the mobo fan header properties correctly.

1. Yes.

1a. They will be the same - sort of. The "trick" is that the fan control systems do NOT actually control to speed targets. They actually are attempting to achieve control of the TEMPERATURE measured at one sensor. In the case of the CPU_FAN mobo header, the sensor involved is one built into the CPU chip itself. Its signal is sent out on one CPU pin for the mobo to use. So what the CPU_FAN header does is send out a control signal (on the PWM line, Pin #4) to manipulate the fan speed to whatever it takes to deliver enough air flow to keep the TEMPERATURE inside the CPU chip on target. When you use a PWM fan hub like that Silverstone unit, all the fans plugged into it receive the SAME PWM signal. But what speed they actually run at depends on the specific fan design.

If you were using the PC-MATE mobo with two CPU_FAN headers as you propose, then both the actual CPU stock cooler on CPU_FAN1 and the Silverstone hub (and its fans) on CPU_FAN2 would receive the same PWM signals. All those fans would be automatically controlled by the mobo's CPU fan control system, according to the cooling needs of the CPU chip. For case ventilation fans this is not exactly ideal. (The mobo SYS_FAN headers intended for this use are guided by a different temperature sensor built into the mobo by its maker.) BUT there is a good relationship between system workload (and cooling needs) as determined by the CPU internal temperature, and the cooling needs of the entire mobo doing much the same workload. So using the CPU temperature sensor to guide overall case cooling is a reasonable, if not ideal, procedure.

I think you realize this already, but just to remind: when you use a 4-pin PWM fan hub, ALL the fans MUST be of the 4-pin design. (A PWM control system cannot control the speed of a 3-pin fan.)

2. Yes also. You can use the Mortar mobo with only one CPU_FAN header. Plug the Silverstone hub into this header. Now make SURE that you plug the actual CPU fan into the one hub port that is marked off separately. A mobo fan header can only accept the speed signal from ONE fan, so the hub forwards back to the CPU_FAN header ONLY the speed signal from that one marked-off port. All the other fans on the hub have their speed signals ignored, and you will not be able to "see" those speeds. What is vital here is that the mobo puts special effort into checking the proper performance of the fan cooling the CPU, to be SURE it is working OK and reduce the chance of CPU overheating. So it is VITAL that the actual CPU cooling fan be connected to the only hub port that will send its speed to the mobo.

You have one other alternative. The Mortar mobo has two SYS_FAN headers, but they are really only 3-pin headers in Voltage Control Mode, with a useless 4th pin. So you CAN control 3-pin fans from them. Now, you can use 3-pin Y-splitters to connect TWO fans (but not more) to EACH of those SYS_FAN headers, and thus connect four fans in total to the headers that ARE controlled specifically by the temperature of the mobo, rather than of the CPU chip. That would meet your needs as long as you do not plan to go to more than 4 case ventilation fans. To do this normally you would buy four 120mm 3-pin fans instead of 4-pin units, because they are a little cheaper (maybe). But there's even a further option. The design of 4-pin fans is such that, even when they are plugged into a 3-pin header operating in Voltage Control Mode (and yes, this can be done mechanically and electrically by design) they WILL have their speed controlled by such a header. Technically, 4-pin fans under PWM Mode are slightly better than when they are under Voltage Control Mode, but the difference is small. So you actually could use either 3-pin or 4-pin fans with two Y-splitters from the SYS_FAN headers and achieve full automatic case fan ventilation under mobo control, with that control based on the mobo temperature sensor rather than the CPU chip temperature.

 

[devocean]

If you were using the PC-MATE mobo with two CPU_FAN headers as you propose, then both the actual CPU stock cooler on CPU_FAN1 and the Silverstone hub (and its fans) on CPU_FAN2 would receive the same PWM signals. All those fans would be automatically controlled by the mobo's CPU fan control system, according to the cooling needs of the CPU chip. For case ventilation fans this is not exactly ideal. (The mobo SYS_FAN headers intended for this use are guided by a different temperature sensor built into the mobo by its maker.) BUT there is a good relationship between system workload (and cooling needs) as determined by the CPU internal temperature, and the cooling needs of the entire mobo doing much the same workload. So using the CPU temperature sensor to guide overall case cooling is a reasonable, if not ideal, procedure.

Thanks for your explanation.
Just to be clear: will the fan hub (when plugged into CPU_FAN2) and the CPU cooler fan share the same RPM, or is the control based on percentages? For example, say the max RPM of the CPU cooler is 1000 and my case fans have a max RPM of 1350. If the CPU cooler is running at 500 (50% of 1000) RPM, will my case fans run at that speed as well, or run at 675 RPM (50% of 1350)?

Does the same go for if I plugged the CPU cooler into the fan hub's signal slot on a motherboard with only one CPU header?

 

[Paperdoc]

Two different fans sharing the same control signal will operate at different speeds commonly. Your example of percentage of max speed is a good way to view that. This will happen for either of the two mobos you are considering. In saying that, I am making an assumption I believe reasonable. I assume for the PC-MATE mobo with two CPU_FAN headers that the mobo's automatic control systems for BOTH of those headers uses exactly the same logic and settings and temperature signal (from the CPU's internal sensor), and hence will send out exactly the same control signals.

The fact that the fan speeds are not the same is not important at all. An automatic fan speed control system by a mobo is NOT that. Although that's what we say, it really is a TEMPERATURE control system. For the CPU_FAN header, for example, the system constantly compares the actual TEMPERATURE measured inside the CPU chip to the target for that CPU type and then manipulates the signal to the fan on that port to make it blow just the right amount of air over the CPU cooler heatsink to get that internal temperature on target. The system does not care what the fan speed is. It cares only about what the measured internal temperature is, and the fan speed is manipulated to whatever it takes. Although the fan speed can be displayed for you, the temperature control system never even uses that information at all.

In your cases, OP, you plan to use the automatic CPU cooling system to guide the cooling fans for the case, too. So those case fans will get exactly the same signal as the CPU cooling fan. The automatic system will ensure that the CPU internal temperature is under control. It will not know or care about the mobo temperature, or about any fan speed. It is VERY likely that this will produce adequate cooling of the mobo components, even though those temperatures are not being monitored or used in this scheme, because of the correlation between CPU workload and mobo component workload. Alternatively as I said before, you could use splitters to connect the case fans to SYS_FAN headers which would make case ventilation more directly related to measured mobo temperature.

Thank you for Best Solution.

 

[devocean]

For the CPU_FAN header, for example, the system constantly compares the actual TEMPERATURE measured inside the CPU chip to the target for that CPU type and then manipulates the signal to the fan on that port to make it blow just the right amount of air over the CPU cooler heatsink to get that internal temperature on target. The system does not care what the fan speed is.


In your cases, OP, you plan to use the automatic CPU cooling system to guide the cooling fans for the case, too. So those case fans will get exactly the same signal as the CPU cooling fan. The automatic system will ensure that the CPU internal temperature is under control. It will not know or care about the mobo temperature, or about any fan speed. It is VERY likely that this will produce adequate cooling of the mobo components, even though those temperatures are not being monitored or used in this scheme, because of the correlation between CPU workload and mobo component workload.

Okay, these should be my last questions: though I feel as though they are stupid ones.
This is the motherboard I will plan on using:
https://us.msi.com/Motherboard/B150M-ECO.html#hero-overview
It's cheaper than the B150M Mortar. I'm thinking of going with one CPU PWM header, and run the cooler and case fans off of it.
These are the fans, 4-pin PWM, 2 of each:
http://www.amazon.com/ARCTIC-F12-PWM-Rev-Dynamic/dp/B00H3T1KBE
http://www.amazon.com/ARCTIC-F12-PWM-PST-CO/dp/B007YLUC0Q

1. When sharing the fan hub with the CPU cooler (which will be in the primary/signal slot), will the case fans all receive a relay control signal? Essentially, what I want to know is whether the PWM fans that are connected to slots 2/3/4/5 on the fan hub will run at full speed, or vary (according to whatever the CPU thinks is necessary to lower internal temperature).

2. On page 13 of the motherboard manual, it says that you can change the corresponding fan speeds that the CPU runs at when it reaches different temperatures through the BIOS settings. I decided to mention this because I wanted to double-check if whether or not the CPU header control signal is based on percentile-logic or otherwise.
http://prntscr.com/be0npq (screenshot of page)
Also, here is a download link for the PDF manual for the B150M ECO:
http://download.msi.com/archive/mnu_exe/M7994v1.1.zip

3. If so, then could I change the CPU Level 1 to, say, 30 (Celsius, I hope) to a 25% CPU fan speed (and, thus the case fans aswell, hopefully), to allow my fans to run quietly during idle temperatures? What CPU Levels 1/2/3/4 (Celsius to fan speed ratios) do you recommend for my specific build?

 

[Paperdoc]

1. All fans on the Silverstone hub will receive the same PWM signal - the one from the CPU_FAN header the hub is plugged into. Thus ALL of those fans will speed up and slow down as the PWM signal is changed by the mobo's automatic control system based on the CPU's internal measured temperature. The only thing unique about the hub's marked Port #1 is that ONLY the fan on this port (which you will ensure is the CPU cooler) will have its speed reported to the mobo CPU_FAN header. Hence that is the only fan's speed you will be able to "see". But all of them will change speeds just like the one you can see.

2. When you start the system up, that screen on manual p. 13 will have default values already. I suggest you leave them alone until you are used to the way your system performs, then consider customizing them later. It is just a way to establish your own graph of PWM signal versus measured CPU temperature. The PWM signal values are basically "% On" values. PWM stands for Pulse Width Modulation. A PWM signal is somewhat like a "square wave", but differs in an important way. A Square Wave is either at full value or at zero value, and it spends exactly half its time in each state. A PWM signal also changes between full On and full Off, but the relative widths of the On and Off states are varied. We refer to the proportion of time spent in the On state as the "% On" value. Now, the "% On" value does not correspond EXACTLY to fan speed as a percent of maximum, but it is reasonably well related.

I suggest you find some information on the proper temperatures for the CPU chip you buy. Each design is a little different. You should be able to find a few typical values for various workloads, plus a specified maximum operating temperature. Once you have observed what your system does in a normal range of uses and how much noise the fans make, you can begin to adjust the settings in BIOS if needed.