Hi, I need some advice on fan and sensor placements. I have a BeQuiet DarkBase Prof 900 Rev.2. It has three silentwings3 front fans, one silentwings3 back fan and one singlentwings3 top fan (near the back). I also have two Noctua CPU fans. My case came with a fan controller board with 8 PWM fans connectors but it may not be good to use it as the motherboard BIOS treat all the fans connected to that board as one. The ASUS Fan Extension Card comes with three PWM fan connectors and 3 sensors. What is the recommended way to connect the fans? Where should I install the three temperature sensors? Thanks.
Question Please advice on fan connections and sensior locations (ASUS Fan Extension Card)
- Thread starter anandtechreader
- Start date
I'll offer some general comments, most of them not specific to your situation.
Virtually all mobos have two "groups" of fan control systems built into them. One concentrates on cooling the CPU chip, and the other on the mobo generally. In each case, they really are TEMPERATURE control systems, NOT fan speed controls. That is, each has a temperature sensor in a particular important place, a target temperature to be achieved there, and a few "tuning parameters" associated with how it will do its job. It constantly monitors the temperature at the probe and adjusts the speed of the associated fan (plugged into its header) to achieve that temperature target. Thus it controls temperature, and does that by manipulating fan speed. It does not really care what that fan speed is. In fact, it does not even use the fan speed value in its control functions. But it also has a secondary function, and that is to monitor the fan speed signal for failure. "Failure" usually is a complete loss of speed signal, but it may be a speed which is lower than some limit value either pre-set or set by you in BIOS Setup configuration. For case fans a failure normally results in a warning message to you so you know there is a problem to fix. For the CPU very often the action is more aggressive because this expensive component could be permanently destroyed by overheating. The action will start with a warning to you, but may be followed very soon afterwards by shutting down your entire system quickly, without even waiting for the temperature sensor in the CPU to show a rapid temperature rise. Some such systems also can refuse to let you boot up if the CPU cooling system fails to start up.
There is a temperature sensor built into all CPU chips and feeding its value out on one chip pin to the mobo. The CPU_FAN header always uses this sensor to guide it. That same sensor signal normally is used for any additional headers in that group, such as a CPU_OPT header. Its signal also MAY be available to you to use at your option for other mobo fan headers in some systems. There is a second sensor built into the mobo by its maker at a spot (and with a target setting) that normally is used to guide the SYS_FAN or CHA_FAN headers. In the BIOS Setup configuration screens for these headers many let you choose whether to use the mobo sensor or the CPU chip's internal sensor for guidance, just in case you want to re-purpose a header for some function related to the CPU. There also are many mobos that have more than one sensor on the mobo - one "general" one, and one or more additional for specific components like the North or South Bridge chips, etc. These MAY be of use to you if you plan to arrange a particular fan to cool that particular area of the mobo, as opposed to just general case ventilation.
Almost no mobo has any sensor or control system for an added video card - there are a few exceptions for mobo makers who use a non-standard means of passing info from one of their own add-on video cards back to the mobo. In general, video card makers use their own on-card sensor, fans and control system to manage cooling of card components, and you can monitor and adjust those systems using a utility that comes with the video card on its CD. But generally video card cooling is NOT done by any mobo system.
Fans come in two main designs from the perspective of electrical connections and control method. (There are a couple other less common ones.) The older design uses THREE wires to connect to the mobo header and a female connector with three holes. It is about 8 mm (3/8") wide and has two ridges down one side. The mobo header (male) has three pins, and beside them is a plastic "tongue" sticking up - the ridges on the female connector fit around that so there's only one way to plug it in. The wires are Ground (Black) on Pin #1, +VDC (red) on Pin #2, and Speed (Yellow) on Pin #3.The power supplied on Pin #2 varies in order to control speed, from +12 VDC max down to about +5 VDC min. Any voltage less may cause the fan to stall and not re-start until the voltage is raised. The speed signal is a series of pulses (2 pulse per revolution) generated in the motor and sent back to the mobo for counting on Pin #3. It is normal practice that, on start-up, all the fans are sent signals for full speed operation to ensure that they start. A few seconds later as the POST process completes it starts reading actual temperatures and reduces fan speeds to whatever is needed then.
The newer design uses FOUR wires and pins on a slightly wider but very simular connector, with that fourth pin just beyond the ridges on the outside of the female connector. Mechanically, this connector is almost the same as the older 3-pin one, so you can always connect any 3- or 4-pin fan to any 3- or 4-pin mobo header and it will work within some limits. Electrically, the connections are similar - Ground and Speed lines are the same. Pin #2 now carries a constant +12 VDC, and Pin #4 carries the new PWM signal. Inside the motor there is a small chip that uses the PWM signal to modify the flow of current from the +12 VDC supply through the windings to achieve speed control. This PWM Mode of control has a few tehcnical advanteges over the older Voltage Control Mode (aka DC Mode).
If you connect a 3-pin fan to a 3-pin header, or 4-pin to 4-pin, it all works as expected. If you mis-match, it works with a couple of cautions. Plug a 3-pin fan into a 4-pin header and the fan gets a constant +12 VDC on Pin #2 and recevies no PWM signal from Pin #4, but it can't use that anyway because it does not have the required chip. So it runs full speed all the time. Plug a 4-pin fan into a 3-pin header and it receives no PWM signal because there is no Pin #4, but it does receive from Pin #2 a VARYING voltage supply, so it speed IS controlled by the header. It's just that this Mode does not use a few technical advantages of the PWM Mode of control. A 3-pin fan header can only exercise control by the Voltage Control Mode. On many mobos a 4-pin header MAY be configured in BIOS Setup to use either of the two control Modes, or even to automatically detect which fan type is connected and adjust its own Mode. But it is possible to "fake" this automatic feature. If the header has 4 pins but actually only uses Voltage Control Mode and sends out no PWM signal on Pin #4, the fan's speed WILL be controlled no matter which fan type is plugged in, and you may never know the details.
Almost all mobo fan headers can supply 12VDC power at up to 1.0 A current in total to all devices connected to that header. This limit applies to each header separately. There are some mobos that have one heavy-duty fan header able to provide up to 3.0 A. Often you have more fans that mobo headers, so you need a way to connect more than one fan to a header. There are two very different types of devices for this. Unfortunately many sellers confuse the labelling of them. I will use the Functions in their design, rather than their physical appearance, to differentiate. The simpler device is a SPLITTER. All it does is connect all the fans on its outputs in parallel to the input from the mobo header so that all the signals are shared. Thus all of its fans do the same thing. This also means that all the power for its fans must come from the header and is subject to the limit of 1.0 A max current. Now, the mobo circuits that count the speed pulse signal from the fan can do so with only ONE sequence of pulses fed to it. Two or more overlapping pulse trains cause terrible confision and false readings. So any Splitter (or Hub below) will only send back to its host mobo header the speed signal from ONE fan. It simply does not connect the speed signals from Pin #3 of the other fans to anything and ignores them, often simply be eliminating that Pin #3 on the output connector of the "other" fan(s). A SPLITTER has only two types of connectors. It has one "arm" ending in a female fan connector that plugs into a mobo header. It has two or more "arms" ending in male (with pins) connectors to plug in fans. It has no other "arm" types. Physically it may look like a collection of three or more cables or "arms", or it may appear to be a small printed circuit board with output headers and one cable to the mobo header. Note that, becasue of the arrangements of connectors, a 4-pin SPLITTER also works just fine in a 3-pin system.
The other device type here is the HUB. It gets power for all its fans directly from the PSU using a cable that plugs into either a 4-pin Molex or a SATA power output connector. It does NOT connect the mobo's lines from Pin #1 or Pin #2 to its fans because those power lines are being fed a constant +12 VDC from the PSU. It connects to the mobo Pin #3 the speed signal from only ONE of its fans, just as the Splitter does. It takes the PWM signal from the mobo's Pin #4 and simply shares that in a parallel connection to all its fans, but this does not overload the mobo header. Again, all of its fans do the same thing. This device avoids the power limit of the mobo header, but to do that it relies on the ability of each fan to use the PWM signal to control itself. Thus, a HUB cannot control the speed of any 3-pin fans, and it MUST have a PWM signal from a mobo header that is configured to use the new PWM Mode for control. (There are one or two Hubs that are different and actually do a conversion from PWM Mode to Voltage Control Mode so they can be used with either fan type, but they are not the cmmon design.) A HUB has the single Input "arm" and several output "arms" PLUS a third "arm" that ends in a wide power connector to plug into a PSU output, and that is how you can differentiate this device from a SPLITTER. Physcially, a Hub may be a collection of cable "arms", a printed circuit board, or a closed box with ports open along edges.
In general, I consider it a good idea to use the automatic fan control systems built into the mobo for your fans, bearing in mind that there are two different functional groups - CPU cooling and case cooling. In each case you have one or more headers to use, and several fans in each group. To connect more than one fan to a header you can use Splitters for smaller numbers of fans that meet the current limit of each header, or you can use Hubs. But IF your fans are 3-pin, then only a SPLITTER can be used to pass on to its fans the varying voltage from a header using the older Voltage Control Mode (aka DC Mode). However, if you have a mixture of fan types, you may be able to re-group them. Use Splitters with any 3--pin fans on their headers using the old Mode, and use either Splitters or Hubs for 4-pin fans on headers using the PWM Mode. In each case to start I recommend you let the header use its normal automatic fan control strategy, but ensure each header is using the correct temperature sensor if there is a choice. Many headers also offer you the chance to use a fan speed control "curve" of your own design, but I suggest you start with the pre-programmed setup until you can observe how well that works and decide whether you can improve on it.
Now, you have the ASUS Fan Extension CARD which many do not have. It adds two important features. One is that it adds more fan headers each with their own power capability and configuration settings, so you can do more control of more fans. That may mean you do not need any Splitter or Hub. The board also comes with temperature sensors you can mount on components of your choice. They can display the temperatures at thoose points, and they are available for use to guide any of your fan headers. Personally I would be somewhat reluctant to use them for automatic control, and here is why. First of all, it can be difficult to accomplish good theremal contact between a probe and a device, and also to ensure that good contact is maintained over time - it does not shake loose. But perhaps more importantly, how would you establish the "correct" temperature target for that measurement point in order to configure the automatic control for that header? Certainly you might do that over a longer time of experimentation, using criteria I don't really know but you might develop. Myself, I'd likely just use the probes to observe temperature i think interesting, without making control of any fan dependent on them.
Virtually all mobos have two "groups" of fan control systems built into them. One concentrates on cooling the CPU chip, and the other on the mobo generally. In each case, they really are TEMPERATURE control systems, NOT fan speed controls. That is, each has a temperature sensor in a particular important place, a target temperature to be achieved there, and a few "tuning parameters" associated with how it will do its job. It constantly monitors the temperature at the probe and adjusts the speed of the associated fan (plugged into its header) to achieve that temperature target. Thus it controls temperature, and does that by manipulating fan speed. It does not really care what that fan speed is. In fact, it does not even use the fan speed value in its control functions. But it also has a secondary function, and that is to monitor the fan speed signal for failure. "Failure" usually is a complete loss of speed signal, but it may be a speed which is lower than some limit value either pre-set or set by you in BIOS Setup configuration. For case fans a failure normally results in a warning message to you so you know there is a problem to fix. For the CPU very often the action is more aggressive because this expensive component could be permanently destroyed by overheating. The action will start with a warning to you, but may be followed very soon afterwards by shutting down your entire system quickly, without even waiting for the temperature sensor in the CPU to show a rapid temperature rise. Some such systems also can refuse to let you boot up if the CPU cooling system fails to start up.
There is a temperature sensor built into all CPU chips and feeding its value out on one chip pin to the mobo. The CPU_FAN header always uses this sensor to guide it. That same sensor signal normally is used for any additional headers in that group, such as a CPU_OPT header. Its signal also MAY be available to you to use at your option for other mobo fan headers in some systems. There is a second sensor built into the mobo by its maker at a spot (and with a target setting) that normally is used to guide the SYS_FAN or CHA_FAN headers. In the BIOS Setup configuration screens for these headers many let you choose whether to use the mobo sensor or the CPU chip's internal sensor for guidance, just in case you want to re-purpose a header for some function related to the CPU. There also are many mobos that have more than one sensor on the mobo - one "general" one, and one or more additional for specific components like the North or South Bridge chips, etc. These MAY be of use to you if you plan to arrange a particular fan to cool that particular area of the mobo, as opposed to just general case ventilation.
Almost no mobo has any sensor or control system for an added video card - there are a few exceptions for mobo makers who use a non-standard means of passing info from one of their own add-on video cards back to the mobo. In general, video card makers use their own on-card sensor, fans and control system to manage cooling of card components, and you can monitor and adjust those systems using a utility that comes with the video card on its CD. But generally video card cooling is NOT done by any mobo system.
Fans come in two main designs from the perspective of electrical connections and control method. (There are a couple other less common ones.) The older design uses THREE wires to connect to the mobo header and a female connector with three holes. It is about 8 mm (3/8") wide and has two ridges down one side. The mobo header (male) has three pins, and beside them is a plastic "tongue" sticking up - the ridges on the female connector fit around that so there's only one way to plug it in. The wires are Ground (Black) on Pin #1, +VDC (red) on Pin #2, and Speed (Yellow) on Pin #3.The power supplied on Pin #2 varies in order to control speed, from +12 VDC max down to about +5 VDC min. Any voltage less may cause the fan to stall and not re-start until the voltage is raised. The speed signal is a series of pulses (2 pulse per revolution) generated in the motor and sent back to the mobo for counting on Pin #3. It is normal practice that, on start-up, all the fans are sent signals for full speed operation to ensure that they start. A few seconds later as the POST process completes it starts reading actual temperatures and reduces fan speeds to whatever is needed then.
The newer design uses FOUR wires and pins on a slightly wider but very simular connector, with that fourth pin just beyond the ridges on the outside of the female connector. Mechanically, this connector is almost the same as the older 3-pin one, so you can always connect any 3- or 4-pin fan to any 3- or 4-pin mobo header and it will work within some limits. Electrically, the connections are similar - Ground and Speed lines are the same. Pin #2 now carries a constant +12 VDC, and Pin #4 carries the new PWM signal. Inside the motor there is a small chip that uses the PWM signal to modify the flow of current from the +12 VDC supply through the windings to achieve speed control. This PWM Mode of control has a few tehcnical advanteges over the older Voltage Control Mode (aka DC Mode).
If you connect a 3-pin fan to a 3-pin header, or 4-pin to 4-pin, it all works as expected. If you mis-match, it works with a couple of cautions. Plug a 3-pin fan into a 4-pin header and the fan gets a constant +12 VDC on Pin #2 and recevies no PWM signal from Pin #4, but it can't use that anyway because it does not have the required chip. So it runs full speed all the time. Plug a 4-pin fan into a 3-pin header and it receives no PWM signal because there is no Pin #4, but it does receive from Pin #2 a VARYING voltage supply, so it speed IS controlled by the header. It's just that this Mode does not use a few technical advantages of the PWM Mode of control. A 3-pin fan header can only exercise control by the Voltage Control Mode. On many mobos a 4-pin header MAY be configured in BIOS Setup to use either of the two control Modes, or even to automatically detect which fan type is connected and adjust its own Mode. But it is possible to "fake" this automatic feature. If the header has 4 pins but actually only uses Voltage Control Mode and sends out no PWM signal on Pin #4, the fan's speed WILL be controlled no matter which fan type is plugged in, and you may never know the details.
Almost all mobo fan headers can supply 12VDC power at up to 1.0 A current in total to all devices connected to that header. This limit applies to each header separately. There are some mobos that have one heavy-duty fan header able to provide up to 3.0 A. Often you have more fans that mobo headers, so you need a way to connect more than one fan to a header. There are two very different types of devices for this. Unfortunately many sellers confuse the labelling of them. I will use the Functions in their design, rather than their physical appearance, to differentiate. The simpler device is a SPLITTER. All it does is connect all the fans on its outputs in parallel to the input from the mobo header so that all the signals are shared. Thus all of its fans do the same thing. This also means that all the power for its fans must come from the header and is subject to the limit of 1.0 A max current. Now, the mobo circuits that count the speed pulse signal from the fan can do so with only ONE sequence of pulses fed to it. Two or more overlapping pulse trains cause terrible confision and false readings. So any Splitter (or Hub below) will only send back to its host mobo header the speed signal from ONE fan. It simply does not connect the speed signals from Pin #3 of the other fans to anything and ignores them, often simply be eliminating that Pin #3 on the output connector of the "other" fan(s). A SPLITTER has only two types of connectors. It has one "arm" ending in a female fan connector that plugs into a mobo header. It has two or more "arms" ending in male (with pins) connectors to plug in fans. It has no other "arm" types. Physically it may look like a collection of three or more cables or "arms", or it may appear to be a small printed circuit board with output headers and one cable to the mobo header. Note that, becasue of the arrangements of connectors, a 4-pin SPLITTER also works just fine in a 3-pin system.
The other device type here is the HUB. It gets power for all its fans directly from the PSU using a cable that plugs into either a 4-pin Molex or a SATA power output connector. It does NOT connect the mobo's lines from Pin #1 or Pin #2 to its fans because those power lines are being fed a constant +12 VDC from the PSU. It connects to the mobo Pin #3 the speed signal from only ONE of its fans, just as the Splitter does. It takes the PWM signal from the mobo's Pin #4 and simply shares that in a parallel connection to all its fans, but this does not overload the mobo header. Again, all of its fans do the same thing. This device avoids the power limit of the mobo header, but to do that it relies on the ability of each fan to use the PWM signal to control itself. Thus, a HUB cannot control the speed of any 3-pin fans, and it MUST have a PWM signal from a mobo header that is configured to use the new PWM Mode for control. (There are one or two Hubs that are different and actually do a conversion from PWM Mode to Voltage Control Mode so they can be used with either fan type, but they are not the cmmon design.) A HUB has the single Input "arm" and several output "arms" PLUS a third "arm" that ends in a wide power connector to plug into a PSU output, and that is how you can differentiate this device from a SPLITTER. Physcially, a Hub may be a collection of cable "arms", a printed circuit board, or a closed box with ports open along edges.
In general, I consider it a good idea to use the automatic fan control systems built into the mobo for your fans, bearing in mind that there are two different functional groups - CPU cooling and case cooling. In each case you have one or more headers to use, and several fans in each group. To connect more than one fan to a header you can use Splitters for smaller numbers of fans that meet the current limit of each header, or you can use Hubs. But IF your fans are 3-pin, then only a SPLITTER can be used to pass on to its fans the varying voltage from a header using the older Voltage Control Mode (aka DC Mode). However, if you have a mixture of fan types, you may be able to re-group them. Use Splitters with any 3--pin fans on their headers using the old Mode, and use either Splitters or Hubs for 4-pin fans on headers using the PWM Mode. In each case to start I recommend you let the header use its normal automatic fan control strategy, but ensure each header is using the correct temperature sensor if there is a choice. Many headers also offer you the chance to use a fan speed control "curve" of your own design, but I suggest you start with the pre-programmed setup until you can observe how well that works and decide whether you can improve on it.
Now, you have the ASUS Fan Extension CARD which many do not have. It adds two important features. One is that it adds more fan headers each with their own power capability and configuration settings, so you can do more control of more fans. That may mean you do not need any Splitter or Hub. The board also comes with temperature sensors you can mount on components of your choice. They can display the temperatures at thoose points, and they are available for use to guide any of your fan headers. Personally I would be somewhat reluctant to use them for automatic control, and here is why. First of all, it can be difficult to accomplish good theremal contact between a probe and a device, and also to ensure that good contact is maintained over time - it does not shake loose. But perhaps more importantly, how would you establish the "correct" temperature target for that measurement point in order to configure the automatic control for that header? Certainly you might do that over a longer time of experimentation, using criteria I don't really know but you might develop. Myself, I'd likely just use the probes to observe temperature i think interesting, without making control of any fan dependent on them.
Thanks for the info. All the fans I use (cpu-heatsink fans and case fans) have four pins. Attached is an illustration of the location of the five case fans.
On the motherbaord (see attachment), it has the following fan and pump connectors: (A) CPU_OPT, (B) CPU_FAN, (C) W_PUMP, (D) H_AMP, (E) CHA_FAN1, (F) CHA_FAN3, (G) EXT_FAN and (H) CHA_FAN2.
(B) CPU_FAN is connected to the main fan which came with Noctua NH-D15S. (A) CPU_OPT is connected to the aditional Noctua fan which I bought. Is it better to connect the top and back case fans to two of CHA_FAN1, CHA_FAN2 and CHA_FAN3 while connect the three front fans to the three PWM ports on the ASUS Fan EXtension Board or the 3 front fans to CPU_FAN1-3 and back as well as top fans to two PWM ports of the ASUS FAN Extension Board?
As for sensors, ASUS told me that I don't need to use them.
On the motherbaord (see attachment), it has the following fan and pump connectors: (A) CPU_OPT, (B) CPU_FAN, (C) W_PUMP, (D) H_AMP, (E) CHA_FAN1, (F) CHA_FAN3, (G) EXT_FAN and (H) CHA_FAN2.
(B) CPU_FAN is connected to the main fan which came with Noctua NH-D15S. (A) CPU_OPT is connected to the aditional Noctua fan which I bought. Is it better to connect the top and back case fans to two of CHA_FAN1, CHA_FAN2 and CHA_FAN3 while connect the three front fans to the three PWM ports on the ASUS Fan EXtension Board or the 3 front fans to CPU_FAN1-3 and back as well as top fans to two PWM ports of the ASUS FAN Extension Board?
As for sensors, ASUS told me that I don't need to use them.
Attachments
Last edited:
I doubt it would matter which way you do it. I am not clear whether the three ports of the Extension card can each be configured differently, or whether they all operate exactly the same. I think they can be separately controlled using Fan XPert or BIOS Setup, but I'm not sure. However, that really does not matter. All of those fans (except the two Noctuas) do one job - supply air flow through the case for general cooling. So I suggest that you configure all of them exactly the same, and they all will do the same thing.
Since you have connected both Noctua fans to separate headers (but I presume they are both mounted on the CPU cooler heatsink fins in a push-pull arrangement), just ensure that the CPU_FAN and CPU_OPT headers are configured the same as each other.
With this arrangement you are not creating a really complex fan system, but you are getting lots of cooling with all faqns under mobo-based automatic control, goof airflow balance for a small positive internal pressure, and readings of the speeds of every fan separately. That also means that every fan will be monitored for failure, since all of them have their speeds available in BIOS. And of course you do not use any Splitter or Hub.
Nice cooling design!
Since you have connected both Noctua fans to separate headers (but I presume they are both mounted on the CPU cooler heatsink fins in a push-pull arrangement), just ensure that the CPU_FAN and CPU_OPT headers are configured the same as each other.
With this arrangement you are not creating a really complex fan system, but you are getting lots of cooling with all faqns under mobo-based automatic control, goof airflow balance for a small positive internal pressure, and readings of the speeds of every fan separately. That also means that every fan will be monitored for failure, since all of them have their speeds available in BIOS. And of course you do not use any Splitter or Hub.
Nice cooling design!
Thanks. I shall keep in mind to configure the CPU_FAN and CPU_OPT the same way.
I tried the following configuration:
3 front case fans to ASUS Fan Extension Board. Back and Top case fans to CHA_FAN2 and CHA_FAN3. I went to the BIOS but it did not display CHA_FAN1. When I went to QFan Control, CHA_Fan 1, 2 and 3 were displayed along with others. However this time CPU_OPT was not displayed. In other words, I can independently control CHA_FAN 1, 2 and 3 but I cannot control CPU_OPT. I asked ASUS Tech support but they are still waiting for higher level tech support from HQ to reply. Could it be that there is a bug or limitation in the BIOS or QFan? Even the fans are configured automatically to be the same, BIOS reported that the case fans spin a bit differently. Perhaps since no two fans are made exactly the same and such minor differences (maybe <20 rpm) is acceptable in this case?
Given that the BeQuiet Dark Base Pro 900 Rev 2 comes with a PCB that can connect up to 8 PWM case fans (two set of 4 connections, one set on each side) and that these fans can be controlled by the motherboard via a PWM cable, I wonder if there is any merit in keeping the ASUS Fan Extension Board. From the test mentioned above, it looks like due to a bug/limitation, I can only independently control some, but not all, the fans. Since the 5 case fans are the same Silentwings3, does it matter whether the motherboard treats them as one lumped fan or two set of lumped fans? Any merit in being able to control each individual fan independently except for higher fault tolent (i.e. not all fans fail to together)? If I change some case fans to RGB fans, will it be better to be able to control individual fans independently?
Last edited:
It is odd that CHA_FAN1 was not shown in the BIOS Setup pages. But apparently it is shown in the QFan utility, and I'm guessing it seems to work correctly there. The fact that the speeds are all slightly different is normal. All fans are VERY slightly different even when they are the same model, so they actually do run at slightly different speeds (and changing sligfhtly) when given the same control signals. Remember, the control system for them does NOT try to achieve a particular speed. It only changes the signals sent to the fans to achieve a TEMPERATURE measurement at the relevant sensor. Exactly how the individual fans respond to those identical control signals has some variation.
That case's fan control board will not change that performance. What it does simply is to accept the PWM control signal from ONE mobo header and send it out to ALL of the fans plugged into its ports. The fact that the eight ports are in two groups is merely convenient "real estate" arrangement, and has no bearing on control and performance. Moreover, like all fan Hubs, that device will send back to its host mobo fan header the speed of ONE of its fans, and none of the others. BY using separate mobo fan headers for each fan (including those on the Extension Card), you get three advantages:
(a) You can see the speeds of every fan individually,
(b) Each fan is being monitored for failure separately,
(c) IF you ever wanted to, you could congifure each fan to do a different speed-versus-temperature curve.
I can see that none of those is especially important to you, but they are factors.
If you change any to RGB fans, you'll need to know a few things before proceeding. First, that would make NO difference to fan control and cooling (except, of course, if the new fans have different air flow performance from your current fans). On an RGB fan, there are TWO connection cables with separate functions. One is solely for the fan motor and looks exactly like a fan cable, and plugs into the same header. The other is a RGB lighting device cable the connects to an RGB header or Controller and provides power and control solely of the RGB device built into the fan chassis.
The earliest fan designs that included lighting devices are called LED Fans. Their LED's are simply connected in parallel to the motor power feed, so they light up when the motor gets power. So this fan type has only one cable for the fan motor, and no separate connection for the LED's. They also dim when the motor is running slow. They provide only one colour and can do no changing displays. These days there are two common and incompatible designs of RGB fans, plain RGB and a more complex system called Addressable RGB or ADDR RGB or ARGB. The "plain" variety uses a 4-pin connection system that provides a common +12 VDC supply and three separate Ground lines, one each for the Red, Green and Blue LED's along the light strip length. Along the strip, all the LED's of one colour do exactly the same thing at any one moment. The control system can mix the three colours to get many, and can change them. The ADDR RGB system uses THREE electrical connections, but the connector looks a lot like the 4-pin one, just missing one pin. These provide a common +5 VDC and Ground power supply, and a Control Line. Along the strip the LED's of three colours all are in Nodes with one LED of each colour plus a control Chip. Instuctions for the display are contained in addressed data packets sent down the Control Line and all of the chips in the strip listen to this line. Each chip responds only to the control packet addresed to it and activates only its group of three LEDs. Thus more complex displays can be done by the controller, like a sequence of colours chasing themselves along the strip.
To control and power any RGB device you need them connected to a Controller. The controller MUST be of the same type as the RGB devices - you cannot mix the two types in one circuit. (NOTE the big differences in voltage supply and control mechanism.) Some mobos come with no such controller systems, some come with one or two plain RGB headers with 4 pins (controller in the mobo, too), some come with one or two ADDR RGB headers, and some come with headers of both types on board. Those that have such headers also supply on their CD's of utilites a software tool for control of the display, and you use those resources to power and control your RGB devices. Most often the RGB header of either type has a limit of 3.0 A max current available to all the devices you plug into the mobo header (if you have two headers, that limit applies to each header separately). Each mobo maker has their own trade name for their system (ASUS Aura Sync is one), but you need to understand that the software tool works for EITHER type of mobo header. So the type of lighting device you buy is NOT determined by the name of the mobo maker's software, but by the type of HEADER included on the mobo. YOUR mobo is the ASUS WS Z390 Pro according to the page image you posted Feb 7. That mobo has ONE header of the plain RGB type (4-pin, 12 VDC power) located near the bottom front of the mobo, so choose RGB fans accordingly if you choose that route.
That case's fan control board will not change that performance. What it does simply is to accept the PWM control signal from ONE mobo header and send it out to ALL of the fans plugged into its ports. The fact that the eight ports are in two groups is merely convenient "real estate" arrangement, and has no bearing on control and performance. Moreover, like all fan Hubs, that device will send back to its host mobo fan header the speed of ONE of its fans, and none of the others. BY using separate mobo fan headers for each fan (including those on the Extension Card), you get three advantages:
(a) You can see the speeds of every fan individually,
(b) Each fan is being monitored for failure separately,
(c) IF you ever wanted to, you could congifure each fan to do a different speed-versus-temperature curve.
I can see that none of those is especially important to you, but they are factors.
If you change any to RGB fans, you'll need to know a few things before proceeding. First, that would make NO difference to fan control and cooling (except, of course, if the new fans have different air flow performance from your current fans). On an RGB fan, there are TWO connection cables with separate functions. One is solely for the fan motor and looks exactly like a fan cable, and plugs into the same header. The other is a RGB lighting device cable the connects to an RGB header or Controller and provides power and control solely of the RGB device built into the fan chassis.
The earliest fan designs that included lighting devices are called LED Fans. Their LED's are simply connected in parallel to the motor power feed, so they light up when the motor gets power. So this fan type has only one cable for the fan motor, and no separate connection for the LED's. They also dim when the motor is running slow. They provide only one colour and can do no changing displays. These days there are two common and incompatible designs of RGB fans, plain RGB and a more complex system called Addressable RGB or ADDR RGB or ARGB. The "plain" variety uses a 4-pin connection system that provides a common +12 VDC supply and three separate Ground lines, one each for the Red, Green and Blue LED's along the light strip length. Along the strip, all the LED's of one colour do exactly the same thing at any one moment. The control system can mix the three colours to get many, and can change them. The ADDR RGB system uses THREE electrical connections, but the connector looks a lot like the 4-pin one, just missing one pin. These provide a common +5 VDC and Ground power supply, and a Control Line. Along the strip the LED's of three colours all are in Nodes with one LED of each colour plus a control Chip. Instuctions for the display are contained in addressed data packets sent down the Control Line and all of the chips in the strip listen to this line. Each chip responds only to the control packet addresed to it and activates only its group of three LEDs. Thus more complex displays can be done by the controller, like a sequence of colours chasing themselves along the strip.
To control and power any RGB device you need them connected to a Controller. The controller MUST be of the same type as the RGB devices - you cannot mix the two types in one circuit. (NOTE the big differences in voltage supply and control mechanism.) Some mobos come with no such controller systems, some come with one or two plain RGB headers with 4 pins (controller in the mobo, too), some come with one or two ADDR RGB headers, and some come with headers of both types on board. Those that have such headers also supply on their CD's of utilites a software tool for control of the display, and you use those resources to power and control your RGB devices. Most often the RGB header of either type has a limit of 3.0 A max current available to all the devices you plug into the mobo header (if you have two headers, that limit applies to each header separately). Each mobo maker has their own trade name for their system (ASUS Aura Sync is one), but you need to understand that the software tool works for EITHER type of mobo header. So the type of lighting device you buy is NOT determined by the name of the mobo maker's software, but by the type of HEADER included on the mobo. YOUR mobo is the ASUS WS Z390 Pro according to the page image you posted Feb 7. That mobo has ONE header of the plain RGB type (4-pin, 12 VDC power) located near the bottom front of the mobo, so choose RGB fans accordingly if you choose that route.
Thank you for the informative reply. Yes, my motherboard is ASUS WS Z390 Pro. The case fans are just a few RPM different so it is normal.
Currently, my fans (all 4-pin PWM) are connected in the following way:
CPU_FAN slot of the motherboard is connected to Noctua Heatsink fan
CPU_OPT slot of the motherboard is connected to 2nd Noctua Heatsink fan
CHA_FAN2 slot of the motherboard is connected to the back case fan
CHA_FAN3 slot of the motherboard is connected to the top case fan
EXT FAN1 of ASUS Extension Board is connected to the top front case fan
EXT FAN2 of ASUS Extension Board is connected to the middle front case fan
EXT FAN3 of ASUS Extension Board is connected to the buttom front case fan
However, under the BIOS, it only lists the following. EXT FAN3 is not listed. Not sure if it is a bug or limitation of the BIOS. * means the fan symbol is spinning.
CPU FAN *
CHA1 FAN
CHA2 FAN *
CHA3 FAN *
HAMP FAN
CPU OPT FAN *
EXT FAN1 *
EXT FAN2 *
Under Q-Fan Control, it only lists the following. CPU_OPT is not listed and thus not user controllable. Again, not sure if it is a bug or limitation of Q-Fan Control.
CPU FAN
CHA1 FAN
CHA2 FAN
CHA3 FAN
HAMP FAN
EXT FAN1
EXT FAN2
EXT FAN3
W_PUMP+
If you were me, would you use the ASUS Extension Board or the BeQuiet's fan controller board?
Currently, my fans (all 4-pin PWM) are connected in the following way:
CPU_FAN slot of the motherboard is connected to Noctua Heatsink fan
CPU_OPT slot of the motherboard is connected to 2nd Noctua Heatsink fan
CHA_FAN2 slot of the motherboard is connected to the back case fan
CHA_FAN3 slot of the motherboard is connected to the top case fan
EXT FAN1 of ASUS Extension Board is connected to the top front case fan
EXT FAN2 of ASUS Extension Board is connected to the middle front case fan
EXT FAN3 of ASUS Extension Board is connected to the buttom front case fan
However, under the BIOS, it only lists the following. EXT FAN3 is not listed. Not sure if it is a bug or limitation of the BIOS. * means the fan symbol is spinning.
CPU FAN *
CHA1 FAN
CHA2 FAN *
CHA3 FAN *
HAMP FAN
CPU OPT FAN *
EXT FAN1 *
EXT FAN2 *
Under Q-Fan Control, it only lists the following. CPU_OPT is not listed and thus not user controllable. Again, not sure if it is a bug or limitation of Q-Fan Control.
CPU FAN
CHA1 FAN
CHA2 FAN
CHA3 FAN
HAMP FAN
EXT FAN1
EXT FAN2
EXT FAN3
W_PUMP+
If you were me, would you use the ASUS Extension Board or the BeQuiet's fan controller board?
Hi, it seems you managed to read RPM speeds with the Dark Base Pro 900 Rev.2. But Have you used the hub ? I just built my new PC with the Dark Base Pro 900 Rev.2 so, but I cannot read any RPM speeds apart for the CPU Fan. I pluged all (3) case fans to the hub (except the CPU fan). It works but bios does not show any speed for them. Normaly, according to the manual, if you set the selector to the extrem left, the system should be controlled by the mb - and show the speed of at least one fan, no (if I understood what you explained before)? Currently when I do so, the fan are at max speed. What should I do ? I have the Gigabyte Z390 Aorus Extrem. Thank you for your help. Best
tom56, you are correct. The manual says that, with that slider all the way left, the mobo fan header will control the speed of all the fans connected to this Hub. It does not say specifically that any of those fan's speeds will be shown. But there are a couple of things you MUST do to get that to work.
First, you must connect an included cable between the Hub's port "A" at upper right and a mobo 4-pin SYS_FAN header. This cable is required BOTH to supply the PWM control signal and to pass ONE fan's speed back to the header. Your mobo has four of these, three at bottom near the front, and one at top near the back edge. Use whichever is most convenient to reach the Hub. Next, once that is done, you must make sure that the particular mobo header you plugged that cable into is configured correctly. See mobo manual p. 44, and use the F2 key to switch to Classic View screen (p. 45) where you can select the Smart Fan 5 Settings screens - see p. 54-55. Start by selecting the particular SYS_FAN header you are using. Set Fan Speed Control to Normal, Fan Use Temperature to use the motherboard (not CPU) sensor, Fan/Pump Control Mode to PWM. When done, use Esc to go back to the main Classic screen, and from there use the Save and Exit tab to get there (p.66), then choose the Save and Exit option to be sure your new settings are saved. This should provide the required PWM signal from the mobo to the Hub.
Now, regarding speed. Any mobo header can deal with only ONE speed signal coming back to it from a fan. So any proper Hub will send back the speed signal from ONE of its connected fans and ignore all the others. For this purpose it is built to use only one of its output ports for that, so you must ensure that one of your case fans is plugged into the correct Hub output port. Unfortunately which of the eight that one is, is not shown in the manual. (I am assuming this Hub does this, although the manual does not say clearly that it will send back a speed signal from one of its fans.) This has NOTHING to do with the ability of the Hub to control the fan speeds - it is only important for ability to display one fan speed. So, try changing which of the eight Hub output ports you plug one of your fans into, until your BIOS can show you a fan speed on that mobo header you used. And remember, only the mobo header used to connect to the Hub will show you a speed, and it can only show one ONE fan's speed.
First, you must connect an included cable between the Hub's port "A" at upper right and a mobo 4-pin SYS_FAN header. This cable is required BOTH to supply the PWM control signal and to pass ONE fan's speed back to the header. Your mobo has four of these, three at bottom near the front, and one at top near the back edge. Use whichever is most convenient to reach the Hub. Next, once that is done, you must make sure that the particular mobo header you plugged that cable into is configured correctly. See mobo manual p. 44, and use the F2 key to switch to Classic View screen (p. 45) where you can select the Smart Fan 5 Settings screens - see p. 54-55. Start by selecting the particular SYS_FAN header you are using. Set Fan Speed Control to Normal, Fan Use Temperature to use the motherboard (not CPU) sensor, Fan/Pump Control Mode to PWM. When done, use Esc to go back to the main Classic screen, and from there use the Save and Exit tab to get there (p.66), then choose the Save and Exit option to be sure your new settings are saved. This should provide the required PWM signal from the mobo to the Hub.
Now, regarding speed. Any mobo header can deal with only ONE speed signal coming back to it from a fan. So any proper Hub will send back the speed signal from ONE of its connected fans and ignore all the others. For this purpose it is built to use only one of its output ports for that, so you must ensure that one of your case fans is plugged into the correct Hub output port. Unfortunately which of the eight that one is, is not shown in the manual. (I am assuming this Hub does this, although the manual does not say clearly that it will send back a speed signal from one of its fans.) This has NOTHING to do with the ability of the Hub to control the fan speeds - it is only important for ability to display one fan speed. So, try changing which of the eight Hub output ports you plug one of your fans into, until your BIOS can show you a fan speed on that mobo header you used. And remember, only the mobo header used to connect to the Hub will show you a speed, and it can only show one ONE fan's speed.
This (last) part of the previous answer goes to my specific question. If one uses a fan controller hub (I'm considering the Phanteks PWM-PH; http://www.phanteks.com/PH-PWHUB.html), will the fan controller report the fan(s) speed BACK to the motherboard? [I already have both CPU fans attached to the motherboard (SuperMicro H11DSi-NT, dual socket AMD EPYC), as well as reservoir fans and case fans running through a separate fan controller board for liquid cooling the RTX TITANs and the front of the case - they run as expected, cycling up and down as per a PWM response. They are NOT detected by the motherboard (BIOS) or via the the SuperDoctor5 SM GUI utility (that runs once the O/S is booted), which leads me to believe that their current status is NOT being reported back to the motherboard via the PWM fan header. I did not do the initial build.] I want to place two 70 mm PWM fans at the back to increase exhaust and one fan 120 mm PWM at the bottom to increase draw through into the case.
The issue that I'm running into is that if I plug the PVM fans directly into the motherboard fan headers they are being detected as "failing" as they are spinning below 500 RPM and the SuperMicro motherboard goes into an alarm phase, cycling up the fans to maximum (sounds like a wind tunnel), and then letting the fan speeds fall back to below 500 RPM, and repeating the process.
So, my questions is: How to use the PWM signal with a fan controller hub without having it report back to the motherboard?
I doubt you can... Fan pin 3 returns the RPM to the controller (and only one RPM). Pin 4 is for PWM and I'd guess the mobo controller needs the pin 3 feedback for functionality. But intuitively I guess there's a chance PWM might work without feedback, and/or might work on a device vs device basis.
If I was desperate I'd try cut the pin 3 wire and hope the mobo wont spout an error but will continue the temp/PWM profile. Or maybe find a signal generator (or hacked from a separate fan) for a generic value to spoof pin 3 out of warning mode and hope the mobo ignores the feedback values to run your desired PWM profile. Both are long odds though and I do not recommend either.
This mobo is one I've not seen before, and its manual does NOT include mention of some configuration features common on other mobos. Those include the ability to adjust the minimum speed of a fan that will trigger a fan failure warning and action, and a means to set up a custom "fan curve" - that is, to set several values of fan speed for measured temperatures. I also am accustomed to seeing at least two different temperature sensors on a mobo. One is the sensor already built into every CPU chip; since your system has two such chips, I would have expected to see two such CPU temp sensors available for use. These are used specifically to guide the cooler system (fan and heatsinlk, or other) for each respective CPU chip. Then I normally see at least one general motherboard temp sensor commonly used to guide the case ventilation fans that usually are connected to CHA_FAN or SYS_FAN headers.
Your mobo has eight fan headers, and all of them are specified as being ONLY used for PWM Mode control. Thus they are NOT well suited to use with older 3-pin fans. Of these, six are labelled with numbers 1 thought 6, and the other two are labelled A and B. I might GUESS that the A and B headers are for coolers on the two CPU chips (not clear which is which) and the other six for case ventilation fans. HOWEVER, that is just my GUESS, and there is no statement in the manual about what those eight headers are supposed to serve.
There are two major types of device used to connect several fans to a single header. The simpler is a SPLITTER which simply connects all its fans in parallel to the fan header power lines, and thus all fan power must come from that host header. The header itself normally is limited to a max of 1.0 A current to all its fans, so that limits how many fans one can use with a Splitter. (Typical fan currents max out about 0.10 to 0.25 A each.) A fan HUB, on the other hand, does it differently. It MUST have a PWM signal from the host header (via Pin #4) and must also have a connection to a Molex or SATA power output connector from the PSU. That power connection is the source of power for all the fans, thus NOT using any power from the host header and avoiding its limit. The HUB distributes power from the PSU to all its fans, and the PWM control signal from the mobo header to those fans. Thus it depends on having ONLY the newer 4-pin PWM fans attached, because only those fans can use the PWM signal to modify the fixed +12 VDC power supply they receive and control their own speed.(See the end.)
Since your mobo has ONLY PWM Mode fan headers, ALL of the fans in your system really should be of the new 4-pin type. (Any 3-pin fan connected to such a header will always run full speed). So if you do have only 4-pin fans, I suggest you do NOT want that Phanteks PWM Hub. It is one of a VERY few Hubs that operate differently. It takes the PWM signal from the mobo header and creates in its own circuits a set of six older-style 3-pin fan headers that use the older Voltage Control Mode. (In that Mode, fan speed is controlled by varying the voltage supplied on Pin #2.) This "conversion" feature is VERY useful if you have PWM fan headers but 3-pin fans. However, I hope you do NOT have that mis-match to deal with. So I suggest your better alternative is a different 4-pin fan Hub, and here are just a couple of examples. This one from Silverstone had 8 output ports
www.newegg.com
This Deepcool FH-10 model has 10 outputs
Do NOT get the Deepcool FH-04 device - it is just a Splitter.
I assume your plan is to connect several CASE vent fans to a single mobo header, and the actual CPU coolers are being done separately, which is best. Many of these Hubs tell you to connect its fan header cable to the mobo CPU_FAN header, but you do NOT need to do that. That advice is solely for use with some older mobos that used the required PWM Mode of control only on the CPU_FAN header, and you do not have that limit. But do note that all of them have ONE of the output ports identified (often with a "CPU" label) as the only port that will send its fan's speed back to the mobo.
You are hoping to avoid sending a fan speed signal back to the host mobo header to stop it from detecting a slow fan and trying to speed it up continually. I am hoping NOT to do that, and here's why. A mobo fan header has three functions. It provides power to the fan. It provides a means to control the speed of the fan. And lastly, it monitors the fan's speed signal for FAILURE. Yes, the speed CONTROL does not use the speed signal, but FAILURE detection does.) Your mobo's fan headers are detecting fan speed less than its lower limit and trying to correct that with some success. If they receive NO speed signal, they will likely become more aggressive in their actions with prominent warning messages and perhaps by shutting down your system for lack of cooling (even though the TEMPERATURES may be OK). Now, when you use a Hub and it sends back only one fan's speed signal to monitor, the mobo cannot check all the other fans for failure, but that's just the way these work. So it really is best to ensure that the Hub DOES send back the speed of one fan by plugging one into the special output port. But that still leaves you open to your current problem.
I suggest a few ways to fix this.
1. Look though the BIOS Setup screens to see what fan header configuration options you have, in case there are some that the manual does not tell you about. Look for a way to set the minimum acceptable speed (the "fan failure" speed limit) for the fan. Or, see if there is a place to specify that the system will never try to run the fan speed slower than some value.
2. If those are not found in BIOS Setup screens, look in the Superdoc 5 utility and see if such control options are offered there.
3. Last option is to try to "fool" the system. One common cause of this problem for others has been using large 140 mm fans - they typically offer much higher air flow rates but at lower rotational speeds, and this can trigger the low-speed problem you have experienced. On the other hand, most smaller fans run at faster speeds when given the SAME control signals as large fans. You are considering 70mm fans (although that is pretty small, but I assume you have reasons) and a 120 mm fan. If the new fans have a maximum speed spec higher than your current fans, then ensure that the FASTER fan design is the one you plug into the only Hub output that will send back a speed signal. That may result in having faster speed signals sent back, and stopping your problem.
There's a bit of confusion above about what fan signals are and do, so let me explain FYI. This is limited to the newer 4-pin PWM fan system - I'll ignore the older 3-pin fans for this. The signals on the mobo fan header are:
Pin #1: Ground
Pin #2: +12 VDC constant power supply
Pin #3: Speed signal - 2 pulses per revolution generated in the motor and sent back to the header for counting
Pin #4: PWM signal - 25KHz pulses used by a chip in the motor housing to modify current flow from the +12 VDC source through the windings, thus changing speed.
The control system really is a TEMPERATURE control system. That is, in each case the system looks at an actual temperature measured by a sensor and compares that to a target, then changes fan speed if the real temperature is off target. It will increase or decrease fan power signals to whatever it takes to get the right temperature, but it actually does not worry about the real SPEED of the fan. The speed signal is used almost exclusively to verify that the fan is working and not stalled.
Your mobo has eight fan headers, and all of them are specified as being ONLY used for PWM Mode control. Thus they are NOT well suited to use with older 3-pin fans. Of these, six are labelled with numbers 1 thought 6, and the other two are labelled A and B. I might GUESS that the A and B headers are for coolers on the two CPU chips (not clear which is which) and the other six for case ventilation fans. HOWEVER, that is just my GUESS, and there is no statement in the manual about what those eight headers are supposed to serve.
There are two major types of device used to connect several fans to a single header. The simpler is a SPLITTER which simply connects all its fans in parallel to the fan header power lines, and thus all fan power must come from that host header. The header itself normally is limited to a max of 1.0 A current to all its fans, so that limits how many fans one can use with a Splitter. (Typical fan currents max out about 0.10 to 0.25 A each.) A fan HUB, on the other hand, does it differently. It MUST have a PWM signal from the host header (via Pin #4) and must also have a connection to a Molex or SATA power output connector from the PSU. That power connection is the source of power for all the fans, thus NOT using any power from the host header and avoiding its limit. The HUB distributes power from the PSU to all its fans, and the PWM control signal from the mobo header to those fans. Thus it depends on having ONLY the newer 4-pin PWM fans attached, because only those fans can use the PWM signal to modify the fixed +12 VDC power supply they receive and control their own speed.(See the end.)
Since your mobo has ONLY PWM Mode fan headers, ALL of the fans in your system really should be of the new 4-pin type. (Any 3-pin fan connected to such a header will always run full speed). So if you do have only 4-pin fans, I suggest you do NOT want that Phanteks PWM Hub. It is one of a VERY few Hubs that operate differently. It takes the PWM signal from the mobo header and creates in its own circuits a set of six older-style 3-pin fan headers that use the older Voltage Control Mode. (In that Mode, fan speed is controlled by varying the voltage supplied on Pin #2.) This "conversion" feature is VERY useful if you have PWM fan headers but 3-pin fans. However, I hope you do NOT have that mis-match to deal with. So I suggest your better alternative is a different 4-pin fan Hub, and here are just a couple of examples. This one from Silverstone had 8 output ports
SilverStone PWM Fan Hub System Cables, Black (CPF04) - Newegg.com
Buy SilverStone PWM Fan Hub System Cables, Black (CPF04) with fast shipping and top-rated customer service. Once you know, you Newegg!
This Deepcool FH-10 model has 10 outputs
Do NOT get the Deepcool FH-04 device - it is just a Splitter.
I assume your plan is to connect several CASE vent fans to a single mobo header, and the actual CPU coolers are being done separately, which is best. Many of these Hubs tell you to connect its fan header cable to the mobo CPU_FAN header, but you do NOT need to do that. That advice is solely for use with some older mobos that used the required PWM Mode of control only on the CPU_FAN header, and you do not have that limit. But do note that all of them have ONE of the output ports identified (often with a "CPU" label) as the only port that will send its fan's speed back to the mobo.
You are hoping to avoid sending a fan speed signal back to the host mobo header to stop it from detecting a slow fan and trying to speed it up continually. I am hoping NOT to do that, and here's why. A mobo fan header has three functions. It provides power to the fan. It provides a means to control the speed of the fan. And lastly, it monitors the fan's speed signal for FAILURE. Yes, the speed CONTROL does not use the speed signal, but FAILURE detection does.) Your mobo's fan headers are detecting fan speed less than its lower limit and trying to correct that with some success. If they receive NO speed signal, they will likely become more aggressive in their actions with prominent warning messages and perhaps by shutting down your system for lack of cooling (even though the TEMPERATURES may be OK). Now, when you use a Hub and it sends back only one fan's speed signal to monitor, the mobo cannot check all the other fans for failure, but that's just the way these work. So it really is best to ensure that the Hub DOES send back the speed of one fan by plugging one into the special output port. But that still leaves you open to your current problem.
I suggest a few ways to fix this.
1. Look though the BIOS Setup screens to see what fan header configuration options you have, in case there are some that the manual does not tell you about. Look for a way to set the minimum acceptable speed (the "fan failure" speed limit) for the fan. Or, see if there is a place to specify that the system will never try to run the fan speed slower than some value.
2. If those are not found in BIOS Setup screens, look in the Superdoc 5 utility and see if such control options are offered there.
3. Last option is to try to "fool" the system. One common cause of this problem for others has been using large 140 mm fans - they typically offer much higher air flow rates but at lower rotational speeds, and this can trigger the low-speed problem you have experienced. On the other hand, most smaller fans run at faster speeds when given the SAME control signals as large fans. You are considering 70mm fans (although that is pretty small, but I assume you have reasons) and a 120 mm fan. If the new fans have a maximum speed spec higher than your current fans, then ensure that the FASTER fan design is the one you plug into the only Hub output that will send back a speed signal. That may result in having faster speed signals sent back, and stopping your problem.
There's a bit of confusion above about what fan signals are and do, so let me explain FYI. This is limited to the newer 4-pin PWM fan system - I'll ignore the older 3-pin fans for this. The signals on the mobo fan header are:
Pin #1: Ground
Pin #2: +12 VDC constant power supply
Pin #3: Speed signal - 2 pulses per revolution generated in the motor and sent back to the header for counting
Pin #4: PWM signal - 25KHz pulses used by a chip in the motor housing to modify current flow from the +12 VDC source through the windings, thus changing speed.
The control system really is a TEMPERATURE control system. That is, in each case the system looks at an actual temperature measured by a sensor and compares that to a target, then changes fan speed if the real temperature is off target. It will increase or decrease fan power signals to whatever it takes to get the right temperature, but it actually does not worry about the real SPEED of the fan. The speed signal is used almost exclusively to verify that the fan is working and not stalled.
Lots of info, thanks. I would have just started experimenting an saw what worked...
For my personal interest and perhaps backing up my previous suggestions I was hoping if you can clarify how the PWM signal on pin 4 works. Is it a relative signal (i.e. go faster or slower than what the feedback reads)? Or is it an absolute value (i.e. just follows some profile based on the temp sensor)? In which case how does the motherboard controller know the max speed (and therefore percentages) of the fan? You did state the control system is based on temperature (which is what I referred to as a "profile") but you said "it actually does not worry about the real SPEED of the fan" which seems impossible when PWM profiles are based on percentages and different fans have different native speeds..
Thanks for the feedback. Interestingly, when the machine was built, there was a Phanteks PWN Hub included with the parts (it's in a Phanteks Enthoo Primo case) and that's what was used for the reservoir fans, the front and rear fans and the side fans (all 120 mm PWM fans). I've attached a photo (sorry for the blurriness), where you can see that there's a separate power, a 4 pin PWM CPU fan header (receiving input) and then each of the six outgoing "arms" are all 3 pin connectors to 4 pin PWM fans. There's NO feedback going back to the motherboard - just the the two CPU fans (that are directly plugged into the motherboard) are registering speeds. HOWEVER, the case/reservoir fans are responding to the temperatures. I suspect that the PWM fan headers are not in fact PWM headers - this is also what the company that did the initial build suspected. So, I'm not sure what's going on...
---------------------------------------------------------------------------------------
EDIT: Maybe it's because there's no fan plugged into the white fan header? See **IMPORTANT NOTE** in the Phanteks PWM-PH manual? http://www.phanteks.com/assets/manuals/PH-PWHUB.pdf
Last edited:
I can confirm that the Phanteks PWM-PH can be used to PWM module PWM fans without returning fans speeds to the SuperMicro motherboard(s). Just DO NOT plug a PWM fan (or any other) into position FAN1 on the fan controller hub. This set-up assumes that you have the CPU fans plugged into the motherboard fan headers (directly; in other words, the mobo is getting it's fan info from the the CPU fans and dispensing it's PWM via the other fan headers). A greasy workaround, but it does work.
OK, clarification of how PWM works. Remember a Square Wave? In that, the signal is either fully on or fully off - nothing in between. Further, the "On" state takes up exactly 50% of the time, and "Off" also is 50%. The transition time from Off to On is VERY fast - as near to instantaneous as the designers can make it. A true Square Wave has a "Duty Cycle" of 50% - that is, it is in the "On" state 50% of the time.
A PWM control signal is similar, with the important change that the Duty Cycle can be anything from 0% to 100%. In the case of computer PWM signals, the PWM line signal (Pin #4) has a 5 VDC peak voltage and runs at around 20 to 25 kHz frequency. Inside the motor there are a few active elecrtonic components doing several jobs, and one of them is to use that PWM signal to control the flow of current from the fixed 12 VDC power supply line (Pin #2) though the motor windings. Thus what is applied to the windings switches rapidly from zero volts to 12 VDC and then back to zero in every cycle of the signal. So the windings get full 12 VDc power for only the "On" part of the PWM signal cycle and during that time they generate a torque to turn the motor; then there's a period of no torque when the PWM signal requires "off". So the power supplied to the motor has the same Duty Cycle as the PWM control signal provided from Pin #4. That is how the motor's speed is set.
NOTE that this signal does NOT specify a speed. It specifies a % of the time that the motor actually gets power to make it turn. The PWM signal and the control system that generates it does NOT care what the max speed of the motor is, or what the real current speed of the motor is. VERY roughly one expects that the actual speed at any time will be a percentage of its design max speed, where that percentage number is close to the Duty Cycle number of the PWM signal. Thus the SAME PWM signal can be supplied to ANY computer-type 4-pin PWM fan and it WILL control the fan's speed over its full useful range.
There are two major types of automatic control strategies: Feed Forward and Feedback. In a feed Forward system you need to have a thorough model of how the entire system and its components operate as certain parameters are changed by the control system, plus some idea of what is the final target you are trying to achieve. In a computer CPU cooing fan system, you can control a signal (say in the case of a 4-pin fan, the PWM signal), and ultimately your aim is to prevent the interior of the CPU chip from overheating, and that last is specified by an ideal operating TEMPERATURE and a lmiting max temp. But to control the cooling to achieve this you would need to know the details of what airflow the fan can generate at various signal settings, how that airflow is reduced by backpressure (resistance to air flow), the heat transfer rate that the heatsink can produce at various air flow values, and a few other things. ALL of these change with every fan dsign and every heatsink and every CPU chip, and some (air flow and heat transfer rate) change as dust collects in the system. A Feed Forward control strategy is just impractical for this.
Feedback control systems are much simpler and VERY widely used. They don't worry about all the "how to" details - they concentrate on the RESULT. For this CPU cooling situation the strategy is to set the target operating temperature and max temp of the CPU (specs the mobo maker gets from the CPU makers and puts into their BIOS chip), then measure the actual temperature inside the chip via a sensor the CPU maker included; its signal is sent out on one pin to the mobo. So the feedback control strategy simply needs to compare the current measured temperature to the target and decide whether the CPU needs more cooling to lower that temp, or less cooling.That is translated quickly into how much should the fan control signal should be raised or lowered. After a suitable small pause, repeat, and repeat, and continue.... NOTE that there is NO need to know the actual fan speed, or the actual air flow rate, or any of those details. The strategy only needs to manipulate the fan control signal to whatever it takes to get the measured temperature inside the CPU chip to the target.
Note also that the variable that is FED BACK to the control system is the RESULT, the measured TEMPERATURE, not the fan speed. This control system does NOT use the fan speed for its work. Although we often call it a fan speed control system, it is a TEMPERATURE control system, and its means of executing control is to manipulate the fan speed. The new output setting going to the fan (the "% On" or Duty Cycle Value of the PWM signal in this case) is an "absolute" value of a particular %, not just an "increase by 3%" kind of instruction.
Now to the use of the Phanteks PWM Hub in your system. You are using it because it was already included with your case. Bottom line: that is just fine - KEEP ON USING THAT.
Now for why I suggested another Hub if you had not already had that one. The newre 4-pin PWM fans were designed with significant "backwards compatible" features to ease the transition form the older 3-pin fans. For 3-pin fans, speed is controlled by varying the VOLTAGE supplied to it on Pin #2. For the 4-pin fans, that voltage is always fixed at +12 VDC, and the new PWM signal from Pin #4 modifies the motor power as I described above. So, what happens when you mis-match the fan type with the header control MODE? Of you plug a 3-pin fan into a 4-pin header using PWM Mode, the fan receives a full 12 VDC on Pin #2 at all times and does not get the PWM signal; but it has not special chip so it could not use the PWM signal, anyway! That fan always runs full speed, so you get good cooling, but no way to reduce the cooling and noise of the fan. If you plug a 4-pin fan into a 3-pin header that is using the older Voltage Control Mode (and a 4-pin header can be made to do this, too!) the fan receives from Pin #2 a VARYING voltage from 12 VDC (full speed) down to 5 VDC (any lower and it might stall), so its speed IS controlled by this signal. The fan gets NO PWM signal so its chip cannot modify that power supply going to the windings.
From a technical perspective, a PWM motor design works best when fed the intended PWM signals. This design allows lower start-up speeds and lower minimum run speeds than can be done when the older varying-Voltage type of signals are supplied. There are a few other advantages to this system, so IF there's a choice, it is slightly better to use the new PWM Mode of control with these 4-pin fans, but is is NOT required.
The Phanteks PWM Hub is different from the vast majority of Hubs because it does not merely distribute the PWM signal to its fans. It uses its own internal circuitry to convert the PWM signal set it gets from the host mobo header into VOLTAGE control Mode signals output on its six 3-pin ports. The power for all those ports comes from the PSU directly, and thus it does not overload the capacity of the host mobo port, just as other Hubs do not. Now, as I detailed above, either type of fan CAN have its speed controlled by the older Voltage Control Mode because the 4-pin fans were designed wit this back-up capability. So if you have only a 4-pin PWM Mode mobo fan header and only 3-pin fans, or a mixture of 3- and 4-pin fans, the Phanteks PWM Hub is one of only a VERY few that really can do the job. And in fact, it will do that also if all your fans are 4-pin. By the way, the version of the Hub you received with your case, OP, has no top cover on it. Another version widely sold for self-installation does have a cover, and it is the one that causes problems trying to plug 4-pin connectors into its ports - the TOP has the small openings that are a problem.
On that Hub, the White Port #1 is the ONLY one that can send back to the mobo host header its fan's speed. It is labelled for use with the actual CPU coolng fan for other reasons, but when you use it for only case ventilation fans, it is a good idea to connect one of them to this white port. Otherwise the mobo header will never get a fan speed and will send alarms out about failure of the case fan it supplies. Now, that is not a big issue. In fact, the fan speed signal IS used to detect fan FAILURE when a single fan is connected to a port, even though it is not used for speed control. But since the header cannot deal with more than one fan's speed, the Hub can only send back a speed from one fan. Thus FAILURE of all the other fans on that Hub can NOT be detected. So you cannot rely on automatic failure detection of all your fans.
A PWM control signal is similar, with the important change that the Duty Cycle can be anything from 0% to 100%. In the case of computer PWM signals, the PWM line signal (Pin #4) has a 5 VDC peak voltage and runs at around 20 to 25 kHz frequency. Inside the motor there are a few active elecrtonic components doing several jobs, and one of them is to use that PWM signal to control the flow of current from the fixed 12 VDC power supply line (Pin #2) though the motor windings. Thus what is applied to the windings switches rapidly from zero volts to 12 VDC and then back to zero in every cycle of the signal. So the windings get full 12 VDc power for only the "On" part of the PWM signal cycle and during that time they generate a torque to turn the motor; then there's a period of no torque when the PWM signal requires "off". So the power supplied to the motor has the same Duty Cycle as the PWM control signal provided from Pin #4. That is how the motor's speed is set.
NOTE that this signal does NOT specify a speed. It specifies a % of the time that the motor actually gets power to make it turn. The PWM signal and the control system that generates it does NOT care what the max speed of the motor is, or what the real current speed of the motor is. VERY roughly one expects that the actual speed at any time will be a percentage of its design max speed, where that percentage number is close to the Duty Cycle number of the PWM signal. Thus the SAME PWM signal can be supplied to ANY computer-type 4-pin PWM fan and it WILL control the fan's speed over its full useful range.
There are two major types of automatic control strategies: Feed Forward and Feedback. In a feed Forward system you need to have a thorough model of how the entire system and its components operate as certain parameters are changed by the control system, plus some idea of what is the final target you are trying to achieve. In a computer CPU cooing fan system, you can control a signal (say in the case of a 4-pin fan, the PWM signal), and ultimately your aim is to prevent the interior of the CPU chip from overheating, and that last is specified by an ideal operating TEMPERATURE and a lmiting max temp. But to control the cooling to achieve this you would need to know the details of what airflow the fan can generate at various signal settings, how that airflow is reduced by backpressure (resistance to air flow), the heat transfer rate that the heatsink can produce at various air flow values, and a few other things. ALL of these change with every fan dsign and every heatsink and every CPU chip, and some (air flow and heat transfer rate) change as dust collects in the system. A Feed Forward control strategy is just impractical for this.
Feedback control systems are much simpler and VERY widely used. They don't worry about all the "how to" details - they concentrate on the RESULT. For this CPU cooling situation the strategy is to set the target operating temperature and max temp of the CPU (specs the mobo maker gets from the CPU makers and puts into their BIOS chip), then measure the actual temperature inside the chip via a sensor the CPU maker included; its signal is sent out on one pin to the mobo. So the feedback control strategy simply needs to compare the current measured temperature to the target and decide whether the CPU needs more cooling to lower that temp, or less cooling.That is translated quickly into how much should the fan control signal should be raised or lowered. After a suitable small pause, repeat, and repeat, and continue.... NOTE that there is NO need to know the actual fan speed, or the actual air flow rate, or any of those details. The strategy only needs to manipulate the fan control signal to whatever it takes to get the measured temperature inside the CPU chip to the target.
Note also that the variable that is FED BACK to the control system is the RESULT, the measured TEMPERATURE, not the fan speed. This control system does NOT use the fan speed for its work. Although we often call it a fan speed control system, it is a TEMPERATURE control system, and its means of executing control is to manipulate the fan speed. The new output setting going to the fan (the "% On" or Duty Cycle Value of the PWM signal in this case) is an "absolute" value of a particular %, not just an "increase by 3%" kind of instruction.
Now to the use of the Phanteks PWM Hub in your system. You are using it because it was already included with your case. Bottom line: that is just fine - KEEP ON USING THAT.
Now for why I suggested another Hub if you had not already had that one. The newre 4-pin PWM fans were designed with significant "backwards compatible" features to ease the transition form the older 3-pin fans. For 3-pin fans, speed is controlled by varying the VOLTAGE supplied to it on Pin #2. For the 4-pin fans, that voltage is always fixed at +12 VDC, and the new PWM signal from Pin #4 modifies the motor power as I described above. So, what happens when you mis-match the fan type with the header control MODE? Of you plug a 3-pin fan into a 4-pin header using PWM Mode, the fan receives a full 12 VDC on Pin #2 at all times and does not get the PWM signal; but it has not special chip so it could not use the PWM signal, anyway! That fan always runs full speed, so you get good cooling, but no way to reduce the cooling and noise of the fan. If you plug a 4-pin fan into a 3-pin header that is using the older Voltage Control Mode (and a 4-pin header can be made to do this, too!) the fan receives from Pin #2 a VARYING voltage from 12 VDC (full speed) down to 5 VDC (any lower and it might stall), so its speed IS controlled by this signal. The fan gets NO PWM signal so its chip cannot modify that power supply going to the windings.
From a technical perspective, a PWM motor design works best when fed the intended PWM signals. This design allows lower start-up speeds and lower minimum run speeds than can be done when the older varying-Voltage type of signals are supplied. There are a few other advantages to this system, so IF there's a choice, it is slightly better to use the new PWM Mode of control with these 4-pin fans, but is is NOT required.
The Phanteks PWM Hub is different from the vast majority of Hubs because it does not merely distribute the PWM signal to its fans. It uses its own internal circuitry to convert the PWM signal set it gets from the host mobo header into VOLTAGE control Mode signals output on its six 3-pin ports. The power for all those ports comes from the PSU directly, and thus it does not overload the capacity of the host mobo port, just as other Hubs do not. Now, as I detailed above, either type of fan CAN have its speed controlled by the older Voltage Control Mode because the 4-pin fans were designed wit this back-up capability. So if you have only a 4-pin PWM Mode mobo fan header and only 3-pin fans, or a mixture of 3- and 4-pin fans, the Phanteks PWM Hub is one of only a VERY few that really can do the job. And in fact, it will do that also if all your fans are 4-pin. By the way, the version of the Hub you received with your case, OP, has no top cover on it. Another version widely sold for self-installation does have a cover, and it is the one that causes problems trying to plug 4-pin connectors into its ports - the TOP has the small openings that are a problem.
On that Hub, the White Port #1 is the ONLY one that can send back to the mobo host header its fan's speed. It is labelled for use with the actual CPU coolng fan for other reasons, but when you use it for only case ventilation fans, it is a good idea to connect one of them to this white port. Otherwise the mobo header will never get a fan speed and will send alarms out about failure of the case fan it supplies. Now, that is not a big issue. In fact, the fan speed signal IS used to detect fan FAILURE when a single fan is connected to a port, even though it is not used for speed control. But since the header cannot deal with more than one fan's speed, the Hub can only send back a speed from one fan. Thus FAILURE of all the other fans on that Hub can NOT be detected. So you cannot rely on automatic failure detection of all your fans.
Right, exactly! I guess my biggest concern was the "failure" alarm on SuperMicro MOBOs that incorrectly report fan failures for low speed (non-SM fans). (BTW: you can take the cover off the after market version; there's a little screw). But yes, you are correct, the big caveat is that if a fan fails, I won't know about it unless I visually confirm. Thanks for all the technical explanation.
TRENDING THREADS
-
Discussion Intel current and future Lakes & Rapids thread- Started by TheF34RChannel
- Replies: 23K
-
Discussion Zen 5 Speculation (EPYC Turin and Strix Point/Granite Ridge - Ryzen 9000)- Started by DisEnchantment
- Replies: 11K
-
Discussion Speculation: Zen 4 (EPYC 4 "Genoa", Ryzen 7000, etc.)- Started by Vattila
- Replies: 13K
-
Discussion Intel Meteor, Arrow, Lunar & Panther Lakes Discussion Threads
- Started by Tigerick
- Replies: 8K
-
Facebook
Twitter