[SOLVED] 3-pin fans with Deepcool Fan Hub FH-04?

[dirk101]

Is there any way I can tell if Deepcool's FH-04 fan hub can control speed for 3-pin fans? Or does anyone here have experience with this component?

I have a Gigabyte Z590 UD AC motherboard, and am looking to connect 3 x be quiet Shadow Wings 2 120mm to a single fan header. The specs say that the motherboard sys fan header can deliver 2A current, and the Shadow Wings 2 specs say that a 120mm fan needs 0.12A. If my research is correct, three fans on one header should work.

I'm limited in terms of components available for the split fan connection, and my only viable option seems to be Deepcool's FH-04. My problem is that the product page claims "compatible with 3-pin fans", but I can't figure out if this actually means that the fans will have variable speeds, or if there's a risk that they will run at 100%, since it's a PWM hub. For other PWM hubs I've found, it turned out in reviews that they only run 3-pin fans at 100% due to how the limited circuitry works in the hub, but I haven't found anything similar for the FH-04 yet.

 

[Paperdoc]

Oddly enough, the answer is yes. And in my opinion, that is because this unit is wrongly labelled a "Hub", apparently because it looks like a closed box with ports inside holes.

To me, this is a SPLITTER, which is exactly what you need. A Splitter merely connects all its fans in parallel to the pins of the mobo header. All of the power for the fans comes solely from the header. As you have outlined, the header is certainly able to provide the power required by those three fans. So you MUST configure the SYS_FAN header you use properly - see your mobo manual, p. 25. Set its Fan Speed Control to "Normal", its Fan Control Use Temperature Input to the mobo sensor, not the one inside the CPU chip, and its Fan Control Mode to "Voltage". To control the speed of 3-pin fans, the header must vary the VOLTAGE supplied to the fans, not using the new PWM Mode.

To me a true HUB is a different device that has an extra connection arm that gets power directly from the PSU. It then sends that fixed 12 VDC power source to all its fans and also relays to them the PWM control signal from the mobo header. But that requires that the fan be of the new 4-pin PWM design, which you do not have. Hubs operating this way can NOT vary the VOLTAGE to their fans, and that's why they cannot control 3-pin ones.

 

[dirk101]

To me, this is a SPLITTER

I had the same thought, since it seemed to use a single cable to take all of its electrical information from the motherboard. But I'm largely ignorant about electrical engineering, and I figured maybe the circuitry in the device could somehow pull a contant 12V and always expect a PWM signal to pass on to the fan ports, no matter what fan type I plugged into it. I hadn't considered forcing the fan header to use voltage control, which would then force the "hub" to distribute that same variable voltage to its fan ports.

On a related note, just to be extra sure, do you know if a PWM fan can be controlled by voltage if I set the fan header to this? Or does the PWM chip in the fan not allow this somehow?

Fan Control Use Temperature Input to the mobo sensor, not the one inside the CPU chip

Why would it be better to set fan speeds relative to the motherboard sensor and not the CPU's? I've only read recommendations of setting fans according to CPU temps, and from my own testing and monitoring all 6 motherboard sensors, I've agreed so far. Under load in Cinebench, with all fans at max speed, the CPU goes up to a stable 80 but my two motherboard sensors stay below 40. Settings my fan speed according to one of these two sensors seems like it would risk running the fans too low to deliver enough cool air to the CPU cooler.

 

[Paperdoc]

Part of the backwards compatibility features designed into the new PWM fans is that they CAN be controlled by the older Voltage Control Mode. If the fan receives NO PWM signal, then its special chip cannot modify current flow, and the motor does whatever it can with the variable Voltage supplied to it. This is not as good as supplying the proper PWM signal set to that motor, but it does work.

Fan speed control is not really that. It really is TEMPERATURE control - that is, the target is a temperature at a sensor, and the variable that is manipulated to achieve that is the speed of a fan supplying the cooling air. For the Motherboard and its many components, the maker has set a temp sensor in a position they deem is the best representation of overall mobo component cooling needs, and pre-programmed the target temperature for that into the automatic default system for SYS_FAN or CHA_FAN headers. But many do also offer the user an option to change which sensor is used, without changing any other parameter of the control system. Often also there is another option to change the target temps and resulting fan speeds, so you can custom-configure a particular header if you have a special situation and some idea of how to create that custom setup.

Separately and similarly, the cooling system for the CPU chip does the same thing using the temp sensor already built into the chip itself, and the right temp target for that CPU. So that system controls the speed of the fan cooling the CPU chip, and does not care about anything else.

The design of these two parallel systems assumes that the air supply for CPU cooling will be whatever is inside the case, and expects that will be adequate for CPU cooling within the range of air flow the CPU cooler fan can generate. You are correct in suggesting that, if you misinform the SYS_FAN header that controls case fans about the temperature on the mobo (by giving it the temp inside the CPU instead), it will greatly speed up the case vent fans and over-cool the mobo and case. In turn, this will allow the CPU cooler to slow down its speed, because it is easier to cool the CPU if the air in the case is cooler. BUT that does NOT mean the CPU actually will be running cooler in most conditions. The CPU cooler speed will be adjusted to achieve the SAME CPU temp internally, no matter what the air supply temp is. The result will be very little impact on actual CPU temperature, and lower-than necessary temps on the mobo. The exception to this might occur at VERY high workloads if the case fan system really is not adequate. In that case, it is possible that, even with the mobo heated up and the case fans running fairly fast, they may not be a full speed. But if the CPU cooler already IS trying to run full speed and that's not good enough, then possibly raising the speed of the case fans MIGHT reduce the CPU temps a little.

As a general rule, I prefer to have each system - CPU cooling and mobo cooling - operate as designed, and using the relevant sensors in each case. What you propose might result in slightly better cooling of the CPU in some conditions, and likely WILL result in over-cooling the case, but you may decide that is a good plan.

 

[dirk101]

The CPU cooler speed will be adjusted to achieve the SAME CPU temp internally, no matter what the air supply temp is

I may be misunderstanding you, but you seem to imply that , for example, under a medium CPU load, the system will have a predefined temperature at which to keep the CPU (i.e. medium CPU load = keep it at 60 C), and then the CPU cooler will adjust its fan speed to keep the CPU at 60 C. If influences outside of the CPU cooler add extra cooling (case fans blowing on the cooler's heatsink or bringing cooler air to the CPU cooler fan), then the CPU cooler fan will slow down to allow the CPU to remain at 60 C with the added cooling taken into account. But doesn't this go against the idea of having a fan speed curve relative to CPU temps? Meaning that, if that CPU at 60 C gets additional cooling and drops to 55 C, then the CPU cooler fan speed will then appropriately drop to the lower speed designated for 55 C, and you get an overall lower CPU temp, the CPU cooler doesn't keep the CPU at an arbitrary value.


But if I understand your overall post correctly, you're suggesting to let the CPU cooler worry about CPU temp and the case fans worry about motherboard temps, which is an interesting perspective and it makes sense. But if this is the case, then why does every PC parts reviewer and their grandmother benchmark case fans cooling against CPU (and GPU) temps? If the case fans make a negligible difference in CPU cooling, wouldn't they benchmark case cooling against motherboard temps, if motherboard cooling is their intended purpose?

 

[Paperdoc]

You have understood my post correctly. It will get a little more complicated below. But I agree that reviewers who track CASE ventilation systems by CPU internal temperatures are getting it wrong! That temp is how to evaluate the CPU cooling system only! Now, a similar argument CAN be made about the temp of the GPU chip on a video card, for exactly the same reasons - IF such a card is supplied with more case air at a lower temp, then that card may be able to keep its GPU chip cooler. But the fact is that the actual GPU chip temp is best to use to evaluate the video card's own GPU cooler, and very few people ever try to CHANGE that to a better GPU cooler! So GPU temp indirectly is a reflection of case temp. and IF you want lower GPU temps without replacing the video card's fan system, the case temp is the only tool you have, poor as it may be!

The WAY (Control strategy) that an automatic system relates fan speed to sensor temperature can come in a couple of main classes of control. The simplest to is called Feed-Forward, because the information measured is used to predict what the manipulated variable should be, with no consideration or real result. s, for example, a "fan curve" in this system would be set up to say: if temp = 25C, set fan to 30%; if temp = 60C, set fan to 70%; if temp =85C or more, set fan to 100%. End of story, until a new measured value is different.

A Feed-Back system takes the result as measured AFTER the control action, and uses that to check and correct what the system is doing to achieve its target. The "result info" flows backwards compared to the flow of "product". So a similar example a Feedback system would have a target temp and set an initial fan speed. Then constantly it would compare that target to the actual measurement and decide whether to increase or decrease the fan speed, based on the DEVIATION of the actual measurement from the Target. This just keeps going on as temps change. Looks like: target is 60, fan at 40%; Actual is 68, increase speed to 44%; actual is now 67, keep speed unchanged; actual dropped to 57, reduce speed to 37%; etc. etc. This system really does try hard to keep the measured value on target. This is the most common system in industrial control systems, and LOTS of things. Think about how YOU (the controller) control the speed of a car as you drive - by checking the speedometer and altering your gas pedal pressure.

Years ago it certainly appeared that fan controls were using Feedback. More recently it looks like they have switched to Feed-Forward the way that "fan curves" are used. But NOTE the similarity: BOTH systems will change the fan speed setting according to the measured temperature. So both will reduce the fan speed if the temp drops for whatever reason. The difference is only in whether the system predicts a required speed (Feed-Forward), or tries to correct a failure to match a target (Feedback).

 

[dirk101]

Ok, now I understand. If you actually want your CPU to always stay at a desired temp, you would use a feedback system. If instead you want to keep the CPU at as low a temperature as possible depending on, say, fan noise preferences, you use feed-forward control through a fan curve. Thank you for sharing your knowledge. One last question, does any modern BIOS let you specify what type of control to use?

 

[Paperdoc]

No, you get no choice - it's in the software coded in the BIOS. From the appearance, I'd say most now are using Feed-forward. They all come with default settings including pre-defined "fan curves" of what speed to run a fan for what measured temp. MANY also allow you the option to specify your own "fan curve" with four or five points along the curve. This is useful if you believe you know better than the default settings the maker suggests. It is also useful IF you are setting up a custom cooling control loop for a paricular area that you have aimed a dedicated fan at - then you can choose a temp sensor IN that component IF there is one, and make your own "curve" for that one area.

As I said, in practice BOTH systems will speed up or slow down a fan according to the sensor temperature you tell it to use. The difference is that a Feedback system will also alter its speed to get as close as possbile to the target temp. A Feed-Forward system will just say: a measured temp of 60 C means I set the speed to 70% (or something like that). A different measured temp gets a different speed setting. If you want deliberately to run your unit (say, the CPU) cooler than the mobo maker suggested, you just set your own curve to run the fan faster at each temperature. Many also have options to use a fixed fan speed independent of temperature, and some who want max cooling always will use this to set full speed or "Turbo".