Question Controlling case fans

[Conradfwebster]

The case I want has 1 x 140mm front fan and 1 x 120mm rear fan supplied. Both non-PWM. May add 1 x Silent Wings 3 PWM 120mm fan to front and one to base. Using Asus Rog Strix B550-E can I connect 3 x 120mm fans to Chassis Fan Headers 1,2 and 3, then the 140mm fan to CPU-OPT to control all via FanExpert 4?

 

[Paperdoc]

You can do that as you say, but there's a better way. Why? The CPU_FAN, CPU_OPT and AIO_PUMP fan headers all use the temperature sensor inside the CPU chip for their guide on what fan speed to run. The three CHA_FAN headers use, instead, a sensor on the mobo, and that is the best guide for case ventilation. So for optimum fan control, you need a way to connect four fans to those three headers. The simplest device for that is Splitters.

The simplest of these devices has one input "arm" with a female connector with four holes to plug into a mobo header, and two output "arms" each with male (with pins) where you can plug in a fan. Example:

Amazon.com: JBtek All Black Sleeved PWM Fan Splitter Cable 1 to 2 Converter, 2 Pack: Electronics

Buy JBtek All Black Sleeved PWM Fan Splitter Cable 1 to 2 Converter, 2 Pack: Computers & Accessories - Amazon.com ✓ FREE DELIVERY possible on eligible purchases

www.amazon.com

That's a 2-pack of 2-output Splitters, and you CAN use this 4-pin model for your 3-pin fans. When you plug in a 3-pin fan, it simply does not connect to Pin #4.

Do not get a HUB. That is a different type of device and often the two things are mis-labelled. A HUB has input and output "arms" or may look like a circuit board or closed box. What is distinctive is that a HUB also has an extra "arm" that must plug into a power output from the PSU to get power for its fans. You do not need one of these, and they only work with 4-pin fans.

You need to know three things when choosing a Splitter.

1. A Splitter gets all the power for its fans from the host header, and hence us limited to that header's max output. The limit is 1.0 A total current load for all fans on one header. The fans you are considering all use less than 0.3 A max each, so connecting two fans to one header is perfectly OK.

2. The method of controlling the speed for 3-pin fans is different from how to control 4-pin ones. So you should NOT mix 3- and 4-pn fans on the same header. In your case, use two CHA_FAN headers and one Splitter on each of them, with two fans per header. Pit the two 3-pin fans on one header together, and the two 4-pin fans on another. See your mobo's BIOS Manual here

https://dlcdnets.asus.com/pub/ASUS/mb/SocketAM4/ROG_STRIX_B550-E_GAMING/E17679_ROG_STRIX_B550-E_GAMING_WIFI_BIOS_EM_WEB.pdf

on p.48 for options to set for EACH of the CHA_FAN headers you use. Set them as:
Chassis Fan Control to DC for the 3-pin fans.r PW< for the 4-pin fans.
Chassis Fan Source to Motherboard to use that temperature sensor.
Chassis fan Profile to Standard for automatic control based on temperature.

After you have the headers configured, use Esc to get to the Monitor Menu, then choose Exit at upper right to get to the Exit Menu (p.28). Choose Saver Changes and Exit to save your settings and reboot.

3. Any mobo header can deal with the speed signal sent back to it from ONE fan only. So the Splitters will ensure the host header only gets one fan's signal. If you look closely at the two outputs of each header, you will see that one has all four pins, and one is missing Pin #3. The speed signal of the fan plugged into the output with the missing Pin #3 is simply never reported - it is ignored. This has NO impact on ability to control speed. It merely means that, when you go looking for fan speed readings, you will see on each of those headers the speed of only one of its two fans.

 

[Conradfwebster]

Thank you for that comprehensive reply. I suppose I could have the two PWM fans attached to one CHA_FAN header and the supplied 2 fans at front and rear attached to headers 2 and 3 in order to use all the headers. I thought DC mode meant they were at full speed all the time but now I assume they must be controllable similar to PWM?

 

[Paperdoc]

Yes, that arrangement is fine, too.

There is confusion, leading to misinformation, about control of the two fan designs. BOTH can be controller by a mobo header - it depends only on whether that header is configured to do things properly.

The older 3-pin fan design is controlled ONLY by changing the VOLTAGE supplied to it on Pin #2. (Pin #1 is Ground or negative, and Pin #3 is the speed pulse signal sent back to the header.) It ranges between +12 VDC for full speed to about 5 VDC, the minimum to keep the fan running without stalling.

The newer 4-pin PWM design does it differently. (Note that this is NOT the same way the PWM speed control has been done in larger DC motors.) The computer version of this is that the motor always receives the full +12 VDC supply from Pin #2, but also receives the new PWM signal from Pin #4. Inside the motor case there's a small chip that uses that signal to modify the flow of current from the fixed +12 VDC supply line through the motor windings to change its speed. This has modest technical advantages. Part of the backwards compatibility features of this design is that, if such a fan is attached to a 3-pin fan header, its speed still will be controlled. In that scenario it does not receive any PWM signal and so its chip cannot modify current flow. But the SOURCE power from Pin #2 is VARYING, so the fan speed IS controlled by the header.

If you plug a 3-pin fan into a 4-pin header that is using the newer PWM Mode of electrical control signals, that fan receives the fixed full +12 VDC power supply from Pin #2, and never receives the PWM signal from Pin #4, but it could not use that anyway - it has no chip. So that fan runs full speed all the time.

So, if you match the fan design to the method or MODE that the header uses to send out its electrical signals, either fan type is controlled by the header. If you MIS-match, there are two possibilities. A 4-pin fan on a 3-pin header still WILL have its speed controlled. A 3-pin fan on a 4-pin PMW header will always run full speed. Those details often get mixed up leading to mis-statements. In the early days of introduction of 4-pin fans, various designs of mobo fan headers were used, some even mis-labeled, and many incapable of dealing with both fan types. That contributed to the misunderstandings.

Modern fan headers on mobos are almost always of the 4-pin type. But that does NOT tell you what electrical signals it sends out. It is now VERY common that each fan header has a configuration option available in BIOS Setup called its MODE. The choices are Voltage Control Mode (aka DC Mode), PWM Mode, and Auto. Auto is supposed to test the connected fan at every start-up (by using PWM Mode to try to change its speed and observing the speed signal returned to the header), and make it own adjustment to use the correct Mode for that fan. If you want to avoid any possibility of getting the setting wrong with this process (and yes, that can happen in some situations) you fix the header to use only the setting you are sure is always appropriate for that header

The result is that EITHER fan type CAN have its speed controlled automatically by a mobo header IF it is configured to use the appropriate MODE. Note that this option is separate from another setting of the header's PROFILE, which is how it decides what speed that fan should run, whether that is a fixed setting or some strategy based in a measured temperature of the item to be cooled. Once the speed choice is made, then the MODE setting establishes how the electrical signals should be sent out to the fan to achieve that speed.

 

[BonzaiDuck]

Thank you for that comprehensive reply. I suppose I could have the two PWM fans attached to one CHA_FAN header and the supplied 2 fans at front and rear attached to headers 2 and 3 in order to use all the headers. I thought DC mode meant they were at full speed all the time but now I assume they must be controllable similar to PWM?

I use the $12 Swiftech splitters. They "look like" hubs, but they're not. They're driven by the PWM wire from the motherboard, but powered directly from the case. Or maybe I misunderstood what PaperDoc was saying, but these work the way he notes as common to "splitters".

ASUS board BIOS's allow running the fans as either voltage-controlled or PWM, and of course you'd run the splitter off a header configured for PWM control. To use the 3-pin fans, you'd connect directly to the motherboard header, configure it for voltage-control, and if there's more than one fan, you'd connect them in parallel and use the tach-sensor wire from only one of the fans. It's intuitive: if you're going to run a string of voltage-controlled fans, it's more convenient that you use all fans of the same kind, so the RPM numbers are representative. I suppose the same thing holds true for PWM-control, though.

Even so, you can always use spare fan-headers for connecting the tach-wire only -- with either type of control.

 

[Stuka87]

Yeah, I would use a fan "hub" if you want multiple fans controlled by a single fan header. That way they are all speed controlled based off CPU/case temps. But the hub supplies the 12V power. As depending on fan draw, you could draw too much current from a single header if you are powering multiple fans off of it.

 

[Paperdoc]

What you describe of the Swiftech Splitters does make them actual SPLITTERS in my view.

My way to label these two devices is this. A SPLITTER has only two types of "arms" (connections): it has ONE input "arm" with a female (with holes) connector to plug into the mobo header. It has two or more output "arms" (or ports) with male connectors (with pins) to plug in your fans. It has no other connections. It may appear to be a collection of cable "arms", or a small circuit board with headers, or a box with openings or "ports" for the fan plugs. Its only source of power for the fans is from the host header, so the total load connected must conform to the limit of the header (usually 1.0 A max total current).

A HUB has ONE input arm (female connector) to plug into a mobo header, and two or more outputs (male with pins) to plug in your fans. But it also has a THIRD connection "arm" that must plug into a SATA or 4-pin Molex power output from the PSU. ALL of the power for the fans comes from the PSU, and none of that power is drawn from the mobo header. This avoids the power limit of the mobo host header. However, it is sometimes forgotten that there still is a limit, although higher. A SATA power output from the PSU normally is limited to 4.5 A max load (due to limits of the contacts used in the connector); a 4-pin MOLEX PSU output can supply more because it has heavier contacts. Normally neither of these is an issue because it takes a LOT of modern fans to reach 4.5 A max current draw. The cable to the header deals only with two signals: it gets from Pin #4 the PWM signal and distributes that to all its fans (this does not load the header significantly), and it returns to the header on Pin #3 the speed signal from ONE of its fans. In almost all Hub designs, it merely distributes the +12 VDC and Ground power lines and the PWM control signal line to all its fans. Thus the Hub can ONLY operate as a PWM Mode device, and it can NOT control the speed of a 3-pin fan. A few Hubs with a different design can do that last job. Very like Splitters, Hubs may appear to be groups of cable "arms", a circuit board with headers, or a closed box with ports. In my view, the best way to distinguish between SPLITTERS and HUBS is that THIRD "ARM" that a Hub has to connect to the PSU.

A mobo header can deal with the speed signal returned to it from only ONE fan (it is a train of 5 VDC pulses, two per revolution, generated in the fan), which it counts to determine speed. The Splitter or Hub thus will ONLY return to the host header the speed signal from ONE of its connected fans; all the other are simply ignored. In a design that looks like cable "arms", the simplest way to accomplish this is that only ONE of the male outputs contains all four pins, while all the others are missing Pin #3 so that output makes no connection to the fan's speed signal. In other designs, the only output that does return its fan's speed signal is marked in some manner - typically by highlighting lines or labels like CPU FAN. That label is a bit misleading, because unless you ARE using this unit to connect your CPU fan to the CPU_FAN header, that Hub output should be used only for one of the case fans you are using, but it MUST have some fan plugged into it. Note that NOT receiving the speed signals from all fans does NOT impact ability to control fan speeds - the system does not use the speed info for speed control. However, it DOES impact the header's second important function: fan FAILURE DETECTION. All headers monitor the speed signal received and will issue a prompt warning if there is no speed signal (or in some cases, if the speed falls below a minimum setting). Some headers (especially the CPU_FAN header) may take more drastic action if a fan fails. When you use a Splitter or a Hub to connect several fans to one header, that failure monitoring cannot be done for ALL of the fans. Hence, YOU should check all your fans from time to time to ensure they all still are working.

 

[Paperdoc]

Stuka87 said,

Yeah, I would use a fan "hub" if you want multiple fans controlled by a single fan header. That way they are all speed controlled based off CPU/case temps. But the hub supplies the 12V power. As depending on fan draw, you could draw too much current from a single header if you are powering multiple fans off of it.

If you have concerns about max load, this is correct. As it happens in this thread, OP proposes using at most 2 common fans per header, and that does NOT come close to the limit for the headers. So Splitters are sufficient. Moreover, one pair of fans are 3-pin design, and most Hubs cannot control the speed of that fan type.

 

[Conradfwebster]

Now I am considering either just one extra 140mm PWM front case fan or that plus a 120mm base intake fan. I have read that a base fan will introduce turbulent airflow rather than linear, which is supposed to assist cooling. However, another writer said that a base fan only assisted cooling by 2 degrees max. Will leave psu and gpu with zero fan mode and cpu dark wing 4 fan also on auto. Any thoughts would help.

 

[Paperdoc]

The role of of turbulence in heat removal is confusing because it really is complex. If you consider the flow of an air stream over a smooth hot surface for heat removal, super-smooth air flow results in very little air flow perpendicular to the surface. So the small amount of air right in contact with the surface absorbs heat, but continues on to the next area still in contact with that surface. Because it is now hotter than before, it absorbs heat from the new area more slowly. Overall, this reduces the total heat removal rate. On the other hand, if there is some perpendicular air flow component - THIS is what we call "turbulence", meaning air flow in multiple directions - then the heated air moves away from the surface and is replaced with cooler air, thus improving heat removal.

All of that is at a "micro-scale" - examining closely the details VERY close to the surface. At the "macro-scale" - details of air flow throughout the entire case - it is not so clear. In that view you do want relatively smooth flow of the major air currents just to be sure there is LOTS of net air flow though the case and over the major components. Larger objects that cause air flow to veer away or reverse can cause so much turbulence that it actually decreases overall flow over and through the case, so that the average air temperaure is higher than you could achieve with less turbulence. That can reduce heat removal.

For an example more people might be familiar with, consider the little devices called "spoilers" mounted on the rear of many cars. The look like little "wings" across the roof line or trunk lid, at a slight angle from horizontal. They cause the air flowing over the car surface and off the rear of the car to be modestly turbulent instead of absolutely smooth. So why are they there? It turns out that without one of those you get very little turbulence in the air leaving the rear of the car and coming back together behind the car, and this causes an actual low-pressure vacuum area behind the car that "sucks" it backwards, resulting in a force that tries to pull the car backwards and slowing it down a little bit. The modest turbulence created by this air flow "spoiler" breaks up the air flow patterns in that area of re-filling the empty space behind the car and reducing the vacuum. In high-performance cars for racing, this small improvement in effective use of the engine's power gives a few extra mph in speed and an important advantage. The idea has been applied to many consumer cars partly for "racing" appearance, and partly for actual modest improvement in gasoline consumption. This is a more commonly-seen real-world application of how to use modest turbulence to gain an advantage. But note that nobody places a big vertical panel up from a car trunk lid to make huge turbulence!

Inside your computer case, having an air flow come up from the bottom to mix with a main flow coming in horizontally at the front can produce modest turbulence in the main case space. (This, of course, is an indirect effect - the main reason for that bottom air inflow is to cool the bottom chamber where the PSU is.) As long as that bottom inflow is much smaller than the main inflow from the front, you won't generate massive turbulence to impede overall air flow.

Now, to the matter of how to assess the impact of these air flow decisions. You have read comments that the impact is only very few degrees on some temperature measurement. While that is true, that is a poor way to asses the impact. The basic fact is the all automatic computer case (and CPU) cooling systems change their fan speeds to reach a set target for the TEMPERATURE read by a sensor. In the case of the general temperature inside your case, the sensor used is mounted in the mobo by its maker, and the system changes the case fan speeds to keep that measured temperature on target. If you increase the fan capacity (either with more fans, or new fans with higher air flow capacity), what happens is simply that the automatic system slows the fans down to get back to that temperture target. So your first observation is that the TEMPERATURE has not changed, or only by a small amount that hardly matters. But the real changes that DO matter are twofold. First, the fans now each are running slower, stretching their lifetime, using marginally less power, and making less noise. And secondly, when you get to very high workloads and high heat generation, the fans can speed up and very likely provide all the heat removal you need without ever getting to their maximum speed - the do NOT limit your workload under any likely conditions. For some systems this is a real improvement, although it certainly is possible to over-do it and use more fans that you ever will need.

So, OP, go ahead and add those two fans you suggest. I am sure it will NOT produce enough turbulence to cause problems, and it may actually improve things. Worst case is you have marginally too much cooling capacity and the fans all run slower and quieter. That is not any problem at all!

 

[BonzaiDuck]

What you describe of the Swiftech Splitters does make them actual SPLITTERS in my view.

My way to label these two devices is this. A SPLITTER has only two types of "arms" (connections): it has ONE input "arm" with a female (with holes) connector to plug into the mobo header. It has two or more output "arms" (or ports) with male connectors (with pins) to plug in your fans. It has no other connections. It may appear to be a collection of cable "arms", or a small circuit board with headers, or a box with openings or "ports" for the fan plugs. Its only source of power for the fans is from the host header, so the total load connected must conform to the limit of the header (usually 1.0 A max total current).

A HUB has ONE input arm (female connector) to plug into a mobo header, and two or more outputs (male with pins) to plug in your fans. But it also has a THIRD connection "arm" that must plug into a SATA or 4-pin Molex power output from the PSU. ALL of the power for the fans comes from the PSU, and none of that power is drawn from the mobo header. This avoids the power limit of the mobo host header. However, it is sometimes forgotten that there still is a limit, although higher. A SATA power output from the PSU normally is limited to 4.5 A max load (due to limits of the contacts used in the connector); a 4-pin MOLEX PSU output can supply more because it has heavier contacts. Normally neither of these is an issue because it takes a LOT of modern fans to reach 4.5 A max current draw. The cable to the header deals only with two signals: it gets from Pin #4 the PWM signal and distributes that to all its fans (this does not load the header significantly), and it returns to the header on Pin #3 the speed signal from ONE of its fans. In almost all Hub designs, it merely distributes the +12 VDC and Ground power lines and the PWM control signal line to all its fans. Thus the Hub can ONLY operate as a PWM Mode device, and it can NOT control the speed of a 3-pin fan. A few Hubs with a different design can do that last job. Very like Splitters, Hubs may appear to be groups of cable "arms", a circuit board with headers, or a closed box with ports. In my view, the best way to distinguish between SPLITTERS and HUBS is that THIRD "ARM" that a Hub has to connect to the PSU.

A mobo header can deal with the speed signal returned to it from only ONE fan (it is a train of 5 VDC pulses, two per revolution, generated in the fan), which it counts to determine speed. The Splitter or Hub thus will ONLY return to the host header the speed signal from ONE of its connected fans; all the other are simply ignored. In a design that looks like cable "arms", the simplest way to accomplish this is that only ONE of the male outputs contains all four pins, while all the others are missing Pin #3 so that output makes no connection to the fan's speed signal. In other designs, the only output that does return its fan's speed signal is marked in some manner - typically by highlighting lines or labels like CPU FAN. That label is a bit misleading, because unless you ARE using this unit to connect your CPU fan to the CPU_FAN header, that Hub output should be used only for one of the case fans you are using, but it MUST have some fan plugged into it. Note that NOT receiving the speed signals from all fans does NOT impact ability to control fan speeds - the system does not use the speed info for speed control. However, it DOES impact the header's second important function: fan FAILURE DETECTION. All headers monitor the speed signal received and will issue a prompt warning if there is no speed signal (or in some cases, if the speed falls below a minimum setting). Some headers (especially the CPU_FAN header) may take more drastic action if a fan fails. When you use a Splitter or a Hub to connect several fans to one header, that failure monitoring cannot be done for ALL of the fans. Hence, YOU should check all your fans from time to time to ensure they all still are working.

Apparently the Swiftech device corresponds to the second type or "hub" you described. There's the two-conductor plug to the mobo-header for PWM and tach-wire. All the fans are driven from the PSU (I may have said "case" before but my mistake).

You also give me useful information about the PSU's power and the type of plug used. It wouldn't matter for a lot of fans: some of your ~100+CFM fans still only draw a half an Amp. You could string three 0.3A fans together and the draw would yet only be below 1.0A.

But I had this . . . idea . . . a curiosity -- for using a DC fan with a 3.0A power draw -- A San-Ace that can put out something like 230CFM (and put on your airport-runway ear-mufflers). Of course, using a Swiftech device just for that fan would still be practical, since the power must absolutely come from the PSU. The mobo probably wouldn't last long providing that much power to a fan-port.

 

[Conradfwebster]

I now understand the various options, very useful explanations.

 

[Paperdoc]

BonzaiDuck, what you suggest is possible. As you say, the heavy power requirement of such a fan MIGHT be met from a standard Hub. There really are three other questions for that plan:
(a) does it have a "standard" computer case fan connector on the end of its cable, or would you have to custom-wire it?
(b) is it really designed to operate as a PWM-type computer case fan? The way that PWM technology is implemented for computer case fans is NOT how that is used industrially for PWM control of larger DC motors. In the older industrial applications, the heavy-load DC power supply for the DC motor applies the PWM control wave signal to the DC power source, and what gets delivered to the motor is already modulated. In small computer case fans, the 12VDC power arriving at the motor is NOT modulated. The system also delivers the PWM control signal on a separate line to that motor case, where a small modulation chip applies it to the power line to control flow of current from that source through the motor windings. So the motor in a computer can includes components a normal industrial DC motor does not have, and uses the inputs differently. You need to know which way that proposed San-Ace motor is designed and wired.
(c) check the rating specs of the Hub you propose to use. Many have a spec for the max current (or maybe expressed in Watts) for each individual output port. that depends on how heavy the wiring or printed circuit traces are inside the Hub, and may well be less than the max TOTAL current the Hub can pull from its source.

 

[BonzaiDuck]

BonzaiDuck, what you suggest is possible. As you say, the heavy power requirement of such a fan MIGHT be met from a standard Hub. There really are three other questions for that plan:
(a) does it have a "standard" computer case fan connector on the end of its cable, or would you have to custom-wire it?
(b) is it really designed to operate as a PWM-type computer case fan? The way that PWM technology is implemented for computer case fans is NOT how that is used industrially for PWM control of larger DC motors. In the older industrial applications, the heavy-load DC power supply for the DC motor applies the PWM control wave signal to the DC power source, and what gets delivered to the motor is already modulated. In small computer case fans, the 12VDC power arriving at the motor is NOT modulated. The system also delivers the PWM control signal on a separate line to that motor case, where a small modulation chip applies it to the power line to control flow of current from that source through the motor windings. So the motor in a computer can includes components a normal industrial DC motor does not have, and uses the inputs differently. You need to know which way that proposed San-Ace motor is designed and wired.
(c) check the rating specs of the Hub you propose to use. Many have a spec for the max current (or maybe expressed in Watts) for each individual output port. that depends on how heavy the wiring or printed circuit traces are inside the Hub, and may well be less than the max TOTAL current the Hub can pull from its source.

So glad I checked back on this thread as soon as I did.

Here's the link for the fan I purchased, and there are customer reviews which might be indicative of . . . something . . Sanyo / San-Ace 120mm x 38mm 12V/3A [PWM] 4-pin

Oh -- I see there was only 1 review in which the buyer said "Great!" No help there. But scroll down to the seller's large-typeface blurb where he mentions "gaming computers, bitcoin servers, etc."

Based on a caution you already gave, I checked what amperage draw this portends for my Seasonic Titanium 750W PSU. Regular SATA hard disks pull as much as 30W; this beast will pull 36W. Somewhere else, I had an indication that the SATA PSU ports would provide up to 54W. [Or was it Amps? I'm gettin' old . . . ] Overall, though, a 750W PSU used for a system which was more than adequately powered by a 650W Seasonic unit should have plenty of power.

What do you think? I have a couple other options for this exhaust fan -- each maybe double the price of the San-Ace. But they don't put out 230CFM at maximum speed.

The Swiftech fan hub doesn't provide any info about power limitations. I can continue searching, but haven't found anything yet, and that includes their PDF available from Performance PCs.

 

[Paperdoc]

Your PSU capacity is not really an issue. The fan will draw power from the
12 V rail which has a limit of about 50 A, so the 3 A for this fan is not a huge increase in load. The real issue is how to get that power to the fan. As I said, it cannot be done from a normal mobo fan header - it would need the use of a suitable fan Hub that draws power directly from the PSU. That Hub need two characteristics. One is power source, and even though a SATA output is limited to 4.5 A, that's enough; a 4-pin Molex source can do even more. But the internals of the Hub need to handle 3 A per output port, and many do not - their internal wiring is too light. I found this unit

https://www.amazon.com/2-Pack-Pin-Molex-Splitter-Cable/dp/B07F354TP3/ref=sr_1_99?crid=1QCC8BO8CNJOC&dchild=1&keywords=pwm%2Bfan%2Bsplitter&qid=1615907305&sprefix=PWM%2Bfan%2Bsplitter%2Caps%2C201&sr=8-99&th=1

I looks like a collection of cable "arms", but it IS a Hub. It draws all fan power from the PSU via a 4-pin Molex output, and gets the PWM signal from a mobo header. It also returns to the header the speed of the fan you plug into its only output with all FOUR pins. So you plug your heavy-duty fan into that one output, and leave the others unused. The unit specifies it is constructed with 24 AWG wires. I looked that up, and for in-chassis wiring applications, that gauge of wire can handle up to 3.5 A. So this unit CAN handle the max current load of that fan.

I note this fan is thicker than most: 38 mm thick, vs the more common 25 mm. And of course with its high top speed and air flow, it creates much more noise that others.

Just another note. Because this fan can deliver MUCH more air flow than most, if you use it with the normal settings of fan speed vs mobo temperature sensor reading in the default automatic fan speed control system from a SYS_FAN header, that system may well try to run this fan at minimum speed becasue that's all the heat it needs to remove. My impression is that you intend to over-cool your system deliberately. So you will need to change how its speed is controlled. You could set your own custom "fan curve" to set the lower fan speeds above that minimum for the lower temperature range. Alternatively, you could just set the fan speed to a fixed full speed (or almost full speed), but then it might never go to max speed when your workload gets really high.

 

[BonzaiDuck]

Your PSU capacity is not really an issue. The fan will draw power from the
12 V rail which has a limit of about 50 A, so the 3 A for this fan is not a huge increase in load. The real issue is how to get that power to the fan. As I said, it cannot be done from a normal mobo fan header - it would need the use of a suitable fan Hub that draws power directly from the PSU. That Hub need two characteristics. One is power source, and even though a SATA output is limited to 4.5 A, that's enough; a 4-pin Molex source can do even more. But the internals of the Hub need to handle 3 A per output port, and many do not - their internal wiring is too light. I found this unit

https://www.amazon.com/2-Pack-Pin-Molex-Splitter-Cable/dp/B07F354TP3/ref=sr_1_99?crid=1QCC8BO8CNJOC&dchild=1&keywords=pwm%2Bfan%2Bsplitter&qid=1615907305&sprefix=PWM%2Bfan%2Bsplitter%2Caps%2C201&sr=8-99&th=1

I looks like a collection of cable "arms", but it IS a Hub. It draws all fan power from the PSU via a 4-pin Molex output, and gets the PWM signal from a mobo header. It also returns to the header the speed of the fan you plug into its only output with all FOUR pins. So you plug your heavy-duty fan into that one output, and leave the others unused. The unit specifies it is constructed with 24 AWG wires. I looked that up, and for in-chassis wiring applications, that gauge of wire can handle up to 3.5 A. So this unit CAN handle the max current load of that fan.

I note this fan is thicker than most: 38 mm thick, vs the more common 25 mm. And of course with its high top speed and air flow, it creates much more noise that others.

Just another note. Because this fan can deliver MUCH more air flow than most, if you use it with the normal settings of fan speed vs mobo temperature sensor reading in the default automatic fan speed control system from a SYS_FAN header, that system may well try to run this fan at minimum speed becasue that's all the heat it needs to remove. My impression is that you intend to over-cool your system deliberately. So you will need to change how its speed is controlled. You could set your own custom "fan curve" to set the lower fan speeds above that minimum for the lower temperature range. Alternatively, you could just set the fan speed to a fixed full speed (or almost full speed), but then it might never go to max speed when your workload gets really high.

I didn't anticipate any problem running this off the ASUS Sabertooth Z170S fan ports. They're configurable as either "DC" voltage-controlled or PWM in an extensive BIOS Menu. I've got CPU_FAN, CPU_OPT_FAN, PUMP, CHA-FAN1, --2, -- 3 and ---4, and about three ASST_FAN[x] ports. They all have manually-definable fan curves with "MIN RPM" settings, a maximum, mid-range and minimum temperature/PWM% pairing.

With the existing rig, there is only a pair of 3-pin DC fans in parallel (about 0.6A total) connected as voltage-controlled with 1 tach-wire for one fan only. Everything else runs off a Swiftech hub/splitter connected to the CPU_OPT_FAN port and powered by the PSU, with a CPU exhaust fan @ 0.5A connected to the CPU_FAN port as PWM.

The project I have in mind will do something similar, but I won't use any twin 3-Pin arrangements. They'll all be PWM fans connected to the Swiftech hubs (I'll want two of them). The 120x38mm fan in question will probably use the device you recommended for connection to CPU or CPU_OPT fan ports.

I think I should be able to set the San_ACE to run at minimum 2,000 to 3,000 RPM and maybe top out at 5,000 if CPU temperature rises to whatever I set it to --- 60C, 65C -- whatever.

I'll have to modify my TR accordion duct to accommodate the extra 13mm fan width. I've got two of those ducts in their factory boxes and they were cheap, so . . . no big loss if I'm not happy with it. But basically, one would just slice the duct between its ribs and over-lap the two pieces, making it stubbier and able to fit the 38mm fan without smashing it together between the fan and cooler.

It's an experiment! No big loss, for a $10 San-Ace fan. I have an appointment in July to see my ENT doctor for another hearing test. They haven't recommended hearing aids, yet. I just feel confident that if anybody can get rid of noise that isn't just air-turbulence, I can do it.