Question Methods for slowing down two fans ?

[dev_cyberpunk]

I have a Thermltake Tower 500 case and want to slow down the two rear fans.

The two ways I can accomplish this is connecting the power in series and use the same PWM or use a flip flop on the PWM to half the speed by dividing the PWM TTL by two.

Also, should I lift the RPM lead or double the rpm signal?

Which way should I go, or is there another way of doing that ?

 

[Lutfij]

If your motherboard has support to manage each fan header, I'd get a PWM fan splitter/y-cable and then have those 2 fans off of one header, then set a custom profile in BIOS for that one header. This is assuming that the fans you're referring to are PWM.

Make and model of your motherboard? Make and model of your fans?

 

[Misgar]

When I have enough motherboard fan headers, I attach each fan to a separate header and control their speeds individually from the BIOS fan curve settings.

Alternatively, I'll use a a fan splitter if I don't have enough fan headers, or a fan hub.

On old systems where manual control is sufficient, I'll fit a front panel 3.5 or 5.25" controller, or use PCI brackets in spare slots. No need to dig out a CD4027B and a scrap of Vero board.

 

[faalin]

What motherboard do you have, most of the time you can set a fan profile in the BIOS, from the motherboard software, or a free program like speedfan.

If you want something hardware you could go with this LINK

 

[dev_cyberpunk]

Make and model of your motherboard? Make and model of your fans?

supermicro X10SLH-N6-ST031 motherboard
Iceberg Thermal IceGale Xtra 120mm

The fan options in IPMI only allow me to set the low rpm and profile to auto speed, heavy i/o speed profile and full on.

 

[dev_cyberpunk]

Another option I just thought of is using a series resistor on the 12V like the Noctua speed reducers.

Problem with integrating someone's speed controller is they might not handle the current draw. But of course I know how to build that seven different ways too and probably have the parts on hand to accomplish this.

Just trying to figure this out on my little workstation I want to build for myself with one of the newer style cases I never get to build in. Which is a challenge to get air cooling to work well in it.

 

[Paperdoc]

There is a family of similar mobos with varied features, so getting a manual for THAT particular mobo us hard. BUT the manual I saw seems typical of the entire family as far as fans go. It shows four case vent fan headers labelled FAn1 to Fan4, plus one CPU cooler header labelled FANA. Its says each of those headers will feed to a 4-pin PWM fan, and the power limit from EACH is up to 2.5 A max current. The fans you talk about are spec'd at 0.54 A max per fan (higher than many because of high airflow and pressure specs). Still, with that high current limit per header you would have NO problem connecting two such fans to a single header using a simple SPLITTER like this

https://www.amazon.com/EKWB-EK-Cable-Y-Splitter-2-Fan-10cm/dp/B00K2F8PFA/ref=sr_1_2?crid=18F0EJRQF0MTR&dib=eyJ2IjoiMSJ9.rc8jIuBZBRMcmb63IEEZWZp8V6bT1e6XuDBkSEWj1LGwpTKt-MNa-kA0fpIDrvRyHwWswCh4tLJW2bm4l4EYlwRPV88jMDj1ZnC9dDxrNCXRc_NzT2BC_Z5qc0poEiiIl041cxvxzbjVnuYN1Mk1-Nq-IvqFLzN5-k_YVj5zn9MB4QpjbLCZzSp36tUje-jQxYyAGL_-BTcQ6xSANHhHffidBzdrz2KdFEu1eu_YY0Q.uHY4j1oYQj3bnD118fYcc-BPabmoZyLKI2lPBtJTF48&dib_tag=se&keywords=fan+splitter+4+pin&qid=1738694959&sprefix=fan+splitter,aps,125&sr=8-2&th=1

or something similar. NOTE that a Splitter is this simple - one input connector to plug into a mono header and two (or more) output connectors for fans. It has NO additional third cable to connect to the PSU directly.

Also note that such a device an return to the host header the speed signal of only ONE of its fans (the one linked has that output labelled with a tag). This does NOT impact ability of the header to control speed of ALL fans connected to it. But IF you really want to have every fan's speed displayed somewhere, you must connect each fan to its own header, then configure each header the same way so they all perform the same way.

Once you have your fans connected to headers in this manner, you can use the configuration details of the header(s) to set how their speeds are controlled automatically to keep up with changing heat load as your mobo's workload changes. That includes means of specifying low speeds for minimal temperature readings. You do not have to rig up custom fan modifications!

 

[dev_cyberpunk]

There is a family of similar mobos with varied features, so getting a manual for THAT particular mobo us hard.

It wasn't a big deal to get the manual and CD that came with the system. I have a Supermicro account and that is the real only way to download anything from them. Even though they don't make it easy to navigate in their customer portal. Because the buttons you would think you would need to click on do nothing, but the text links works.

Its just that with the default software and even in its ipmi web there isn't a way to customise the fan options. That is why I was looking for a quick and easy solution.
But I'm not 100% convinced that the back fans are going to be effective there since I am using hi flow fans and could channel the air through the back side of the motherboard.

But it would be nice to figure out how to set these fans individually if possible. I know Linux has an IPMI fan utility supermicro created for the motherboards, so later today I will try programming them from a usb boot of ubuntu.

 

[Misgar]

Looking over an image of your SuperMicro server class board I can see most of my suggestions won't work. I was assuming a typical desktop mobo. Is this the layout?

I have a similar Supermicro + Xeon board in an old rig, but I don't use IPMI fans. I'll check my cooling solution later on today and get back to you.

The fan options in IPMI only allow me to set the low rpm and profile to auto speed, heavy i/o speed profile and full on.

Is there any reason why you can't use one of the IPMI options? I'm not particularly bothered by the Delta fans in my HP servers at startup. The high pitched banshee wail reduces to a "pleasant" loud hum after the OS loads. Not to everyone's taste though I admit.

Another option I just thought of is using a series resistor on the 12V like the Noctua speed reducers.

Check the Wattage of the resistor(s) if you choose this route. If you drop 6V and the current reduces to 200mA, that's still 1.2W dissipated in the resistor. Don't use an eighth or quarter Watt resistor, but I suspect you already know what you're doing if you were considering flip flops.

I'd still prefer something a little more sophisticated than a resistor, e.g. a variable control.

 

[Misgar]

I've just checked my SuperMicro X10SRA board. It has 4 fan headers. I needed 6 fans, so I used a couple of splitter cables (no resistors).

 

[dev_cyberpunk]

Looking over an image of your SuperMicro server class board I can see most of my suggestions won't work. I was assuming a typical desktop mobo. Is this the layout?

Its a X10SLH board. 5 fan headers FAN2-4 and FAN A are the case and FAN1 is the optional one for the heatsink which I verified that when I took it out of its 1U case.

Its an ebedded server board with six 10Gb ethernet interfaces that I can either use them in system or assign them a SATA port for iSCSI.

I built one up in a raidmax case with this stock passive heat sink and with automatic setting it quietly serves the home network with fans running at 620 rpm. With this heat sink in the Thermaltake case, the fans scream at 2700 rpm. With a stock socket 1150 Intel heatsink and fan its 1500 rpm.

I can force it to a point manually in Linux, but I think I need a better heatsink than the 90W heatsink on the 3.5Ghz 80W processor. The router system has a 40W 2.7Ghz CPU. So I ordered a 190W heatsink and I will find out if that will set the fans lower in auto without having to manually force them in Linux.

 

[dev_cyberpunk]

I found some info on the fans and how to set them manually, even though I prefer to have the fan controller ramp them automatically with temperature.

The fans in normal mode were cycling up and down, which I need to troubleshoot why its doing that. Because either the fan parameters are not correct or some other error is going on and its not being displayed anywhere but ramping the fan speed.

This board does not behave like the one I bought off of ebay as a bare motherboard. I removed it from an assembled unit so there might be some configuration with the hot swap power supplies on the hardware level that I probably need to disable somehow. Its just trying to figure this out without having to pay a Supermicro tech to fix it. Another symptom it has is it doesn't stay shutdown even with the watchdog disabled in Bios as well as the jumper.

 

[dev_cyberpunk]

Is there any reason why you can't use one of the IPMI options? I'm not particularly bothered by the Delta fans in my HP servers at startup. The high pitched banshee wail reduces to a "pleasant" loud hum after the OS loads. Not to everyone's taste though I admit.


Check the Wattage of the resistor(s) if you choose this route. If you drop 6V and the current reduces to 200mA, that's still 1.2W dissipated in the resistor. Don't use an eighth or quarter Watt resistor, but I suspect you already know what you're doing if you were considering flip flops.

I'd still prefer something a little more sophisticated than a resistor, e.g. a variable control.

The automatic mode cycles the fan speed up and down , which could be something else wrong, but at the meantime, I hear the two back fans being more noisy so I would like to reduce their speed proportionally, and use the auto mode so the fans will speed up if it gets hot instead of operating at one speed.

Full on and heavy io are the only two settings that the fans don't speed up and slow down.

I am going to have to find my heat sink junk from the Noctua heat sink I have on another unit, as it did come with a speed reducer and I didn't use it. I think they used a 2W resistor. The only thing is you loose relative power within the fan using this method, but a lot cheaper to come up with than modifying the PWM TTL signal with logic chips.

 

[dev_cyberpunk]

The issue I was told by a hardware engineer is the fan controllers on the x10 were designed for fans that run 1000 to 50,000 rpm. So, lower speed fans have to be set manually but there should be a way to make something that would scale the rpm so the controller thinks its a high speed fan.

The ramping of the fan speeds is a particular hardware error, but this error is introduced because I'm using slower high volume fans and the computer thinks that they are faulty.

 

[Paperdoc]

MAYBE. I do not know how that mobo does things. On all common computer mobos the fan header speed control system does NOT pay any attention at all to the speed reported by the fan. It sole target is the TEMPERATURE of the item to be cooled, and it sends to its fan a signal that really is a "% of full speed" command. That MAY be a Voltage between 5 and 12 VDC for an older 3-pin fan whose speed can be controlled only by Voltage variation. For more recent 4-pin fans the Voltage supplied is always 12 VDC, but the header sends out on Pin #4 the PWM signal. This is used by a special chip in the fan that modifies current flow from the 12 VDC supply through the windings to alter speed. Whichever the speed control signal system, it is based entirely on comparing the current temperature at a relevant sensor to the programmed target.

There is one other second important function on all common mobo fan headers that could be involved here. It DOES monitor that speed signal for NO signal (or sometimes, less than a min target). If the speed is considered too low or NOT there, the BIOS's first step is to raise the speed command signal to the fan to run at full speed to ensure it re-starts. If that produces an acceptable speed signal the header then tells the fan to return to the proper speed. If that speed is acceptable, too, end of story for now. But if it has become too slow again, the cycle will repeat and the fan speed will cycle up and down constantly. On the other hand if the fan does NOT re-start, that will cause the BIOS to pop up on the user screen a warning of fan failure so corrective action can be taken.

Considering that this mobo is normally in a server rack, such failure monitoring may not be part of its BIOS. Or it may be, BUT the action taken when failure is detected might NOT be a warning window on a screen nobody is watching! It might be to log the error, or to send out an error code on a port somewhere that a server rack monitoring system routes to a system management console.

This could get tricky to alter, even if you could find a screen to adjust fan header parameters. The signal to control fan speed may or may not need changing. BUT if the FAILURE detection system needs to have its minimum speed setting altered for the fans you are using, few mobos have any way to change that.

 

[dev_cyberpunk]

MAYBE. I do not know how that mobo does things. On all common computer mobos the fan header speed control system does NOT pay any attention at all to the speed reported by the fan. It sole target is the TEMPERATURE of the item to be cooled, and it sends to its fan a signal that really is a "% of full speed" command. That MAY be a Voltage between 5 and 12 VDC for an older 3-pin fan whose speed can be controlled only by Voltage variation. For more recent 4-pin fans the Voltage supplied is always 12 VDC, but the header sends out on Pin #4 the PWM signal. This is used by a special chip in the fan that modifies current flow from the 12 VDC supply through the windings to alter speed. Whichever the speed control signal system, it is based entirely on comparing the current temperature at a relevant sensor to the programmed target.

There is one other second important function on all common mobo fan headers that could be involved here. It DOES monitor that speed signal for NO signal (or sometimes, less than a min target). If the speed is considered too low or NOT there, the BIOS's first step is to raise the speed command signal to the fan to run at full speed to ensure it re-starts. If that produces an acceptable speed signal the header then tells the fan to return to the proper speed. If that speed is acceptable, too, end of story for now. But if it has become too slow again, the cycle will repeat and the fan speed will cycle up and down constantly. On the other hand if the fan does NOT re-start, that will cause the BIOS to pop up on the user screen a warning of fan failure so corrective action can be taken.

Considering that this mobo is normally in a server rack, such failure monitoring may not be part of its BIOS. Or it may be, BUT the action taken when failure is detected might NOT be a warning window on a screen nobody is watching! It might be to log the error, or to send out an error code on a port somewhere that a server rack monitoring system routes to a system management console.

This could get tricky to alter, even if you could find a screen to adjust fan header parameters. The signal to control fan speed may or may not need changing. BUT if the FAILURE detection system needs to have its minimum speed setting altered for the fans you are using, few mobos have any way to change that.

I think its because its running the power supply monitoring for a redundant power supply because the service is turned on in the ipmi side, but no way to disable it. But I restored setting from a machine that has the same motherboad, however it was purchased as the motherboard instead of it being pulled out of a 1U working unit. I guess I could troubleshoot to see if this is true, just have to find a way to temporarily put the power supply from the 1U into the Thermaltake Tower 500 the board is sitting in.

The microprocessor that runs the ipmi runs the fans and don't have a lot of user controls and anything custom has to be passed to it via command line to change settings.

The x10 rpm signal is going straight into the micro controller, so since its not conditioning the signal like a normal motherboard does, the lower speeds are too noisy on the rpm signal to accurately register the correct rpm. Which is a known flaw in these motherboards. The x12 and x13 series don't do this (or at least the ones I have).