Question "Passive" radiator

[liamwhalley20]

I am just wondering if this will work, I am planning on buying 4 arctic f12 pwm fans as a push pull configuration on my radiator, these fans only start up when the cpu is under load, I already have 1 of these fans as an exhaust in my case, so my question is with these fans not spinning at low load could this cause problems with the aio? The fans on there right now are way too loud even with fan speed on 'silent' but they just cause a racket really and I would like to make it alot quieter as I dont like my pc sounding like it is going to take off every 2 minutes

 

[Lutfij]

What AIO are you using right now? You might want to mention what your heatware is as well. IMHO, if you're looking at the off the shelf AIO's with thin radiator profiles, you're not going to get by with quiet fans regardless of static pressure since the fins per inch on AIO radiators(apart from those made by EK, Swiftech and Alphacool) are too tight and require high CFM and static pressured fans.

 

[liamwhalley20]

Deepcool maelstrom 240t
I5 4670k @ 4.5 Ghz 1.201 Vcore

 

[tennis2]

Your mobo is responsible for fan speeds (or lack thereof).
There's no harm in turning radiator fans off below a certain CPU temp threshold. When your CPU exceeds that threshold, the mobo will simply turn the fans on and cool down the fluid quicker.
One thing to consider is running the fans at a low speed minimum (instead of off) that's below the audible range in your room. This will provide greater cooling potential than passive and may prevent the fans having to ramp up higher than necessary to "catch up" if you're going passive. This would create a less variable fan noise profile, which is generally less noticeable to users than having fans continually changing speeds.

 

[liamwhalley20]

Yes, I have tried all of the different fan profile settings in the bios but it's either fans on loud all the time or they are at low speeds but then speed up every couple of minutes when the cpu is under load theres no inbetween where they can just stay quiet all of the time I have tried some coolermaster ones on there and they seemed to work alot quieter but those fans are on their way out as the bearing is making a noise in them

 

[tennis2]

What temps are you targeting?

How are you connecting the fans to the mobo? If it were me, I'd have the two pushers on a splitter from the CPU fan header (since, especially from this era, the CPU fan header has the best fan speed control), and the two pulls at a constant speed on a splitter from another fan header.

Have you set the header that your AIO pump is connected to to 100%? Or are you varying the pump speed/voltage?

How many case fans pushing air into the case? How many out?

I have a 3570K @4.2GHz 1.2V and my 120mm AIO with push/pull rarely spins my fans up. During gaming sessions I think I'm typically around 60C if I recall. (I need to remeasure this because I think I need them spinning faster to get my GPU waste heat out better). I have 2x120mm fans as intake and my AIO is the only exhaust.

Just doesn't seem like a 240mm AIO with 4 fans should even be breaking a sweat with your setup.

 

[liamwhalley20]

Temperatures aren't an issue at all, I hit 71°c on prime95 26.6 stress test but in games i only hit around 55, the pump is on 100% yes, and the fans for the radiator are on the fan hub deepcool provided, I have 2 intake (radiator fans) and 3 exhaust around the case, it is just the noise created by the radiator fans is a bit much really especially when trying to watch something

 

[tennis2]

If the fans are hooked up in pairs as I described (and depending on your mobo's fan controls) you could consider turning the pullers off and keeping the pushers at minimum rpm for movie watching and other tasks that are unlikely to raise your temps much. But hey, if full passive mode works well, that's obviously the quietest.

More than one way to skin a cat (what an odd saying...)

 

[Paperdoc]

You should realize that the fans on your radiator are SUPPOSED to change their speeds as your workload changes. Also realize that forcing them to run slow means you are NOT getting the cooling of your CPU chip that it needs at higher workloads.

That said, I realize that some fans are noisier than others for similar air flow. And it really is air FLOW that is required. But especially on a radiator, the ability to deliver that flow against a higher backpressure (air flow restriction by narrow air passages) is important, so you need to look also at the max backpressure spec for any fan you consider.

IF you have your radiator fans connected (via the small Splitter unit supplied by Deepcool) to the CPU_FAN header, then that header is how those fans are controlled. Remember that fan speed, noise and air flow (thus cooling) all go together. The normal automatic fan control system used by the CPU_FAN header actually is a TEMPERATURE control system. That is, it has a temperature target to meet as measured by a sensor built into the CPU chip itself. It will change the fan speed to whatever it takes to keep that internal temperature on target. The fan speed is the manipulated variable, not the ultimate target. Normally the CPU_FAN header can ONLY use that internal sensor to guide it, but IF you have those fans plugged into a different header, then check its configuration settings. Some fan headers allow you to specify which temperature sensor the header is using, and a different sensor on the motherboard is always an option. IF your rad fans are connected this way, make sure their header is configured to use the sensor inside the CPU chip.

Most mobo fan headers also have an option beyond the "normal" automatic one. This is often called "Manual" or "Custom". It allows you to specify a different "fan curve" from the pre-programmed one. That is, there's a graph of what speed the fan should run for each temperature in the operating range, and usually you can specify four or five points along the curve. This can be used in two ways. One is to make your own curve and force the fan to run slower and quieter than "normal" at all times, BUT that means you are also forcing your CPU to run hot all the time. The other is more subtle and may be useful to you. Often a fan is running just on the edge of switching from one speed to another, and almost any change in workload causes the fan speed to jump. You can set your own curve to change the temperature at which the fan speed jumps so that it happens less frequently and is less annoying. To do that you do not need to make the fan stay too slow at very high temperatures.

With that background that applies no matter what fans you have, back to your plan. I think you are a bit confused on a couple of points. First, most fans do NOT stop at low temperatures and workloads. Automatic fan speed control systems deliberately avoid that by never sending the fan signals to run so slow that they stall - they always run at a minimum speed, or faster when needed. That is, unless you configure the fan control system to defeat this. Now, for users who are not using mobo automatic fan control systems, Arctic does supply a line of CASE VENTILATION fans that DO have their own temperature sensors built in and CAN stop at low temps. For example, their F12 TC model has this feature - note the "TC" suffix in the name. But that is NOT the fan you specified.

Secondly, you do NOT want a fan designed for general case ventilation, because they are designed for max air flow against very little backpressure. See this page from Arctic

ARCTIC fan guide - Which fan for which use case?

With the P-Series and F-Series fans, ARCTIC has fans for both high static pressure and high airflow. Which fan is best for me?

www.arctic.ac

Your plan was to buy an "F" series fan optimized for max air flow against low backpressure. For use on a radiator, you want fans from the Arctic P series like this

https://www.arctic.ac/worldwide_en/p12-pwm.html

 

[liamwhalley20]

I understand what you're saying but I am not confused at all I am just curious, as I have stated I already own 1 of the arctic f12 pwm and it stops and starts when cpu is under load, arctic themselves say that the f12 pwm and tc and silent all tap into the pwm signal from the cpu header regardless of it being connected to the cpu header, and deepcool clearly shows in the manual to plug the fan hub into the system fan header and pump into cpu header. Now, temperatures are all fine they're not the issue at all, it just the noise. As it is getting annoying. I have tried manually setting fan speed in bios but that just does the same as any of the other profiles in the bios either quiet and fans speed up every 2 minutes or they're just loud constantly. Now the first is the lesser of the two evils but I would like to eliminate this all together as I built this to be quiet and right now that isn't the case.

 

[tennis2]

No fan can take a signal from a header it's not connected to.

 

[Karadjgne]

Case fans should never stop, that's just a recipe for disaster.
I have an eco setting on my gpu. The target temp is @ 65°C. Below 65°C the fans are off. That's great, if you don't game. Gaming puts the gpu temp from 32°C to +65°C in a matter of minutes. So then the fans kick on at 70+% until the gpu goes to @ 60°C, then off. And on. And off... 0rpm to almost 2000rpm, then back to 0rpm in seconds, repeated every 10-15 seconds. Whomever thought that was a wonderful idea should have been throat punched. Twice. Hard. By an angry Sumo.

Needless to say, that eco setting (fortunately for the gpu) is disabled. The fans spin at a constant 20ish% at idle and will ramp up to a constant 45ish% gaming, and the gpu barely ever gets beyond 60°C.

Having any kind of eco crud compromising cooling by shutting down fans and airflow, just allows temps to get out of control until something steps in to fix that issue, and will do so with gusto.

Case Fan: Noctua - NF-P12 redux-1300 PWM 54.32 CFM 120 mm Fan ($13.90 @ Amazon)

Just 2x of those, no need for push/pull, won't really do anything. Don't use the other fans as anything but a Frisbee. Set 100% duty cycle for 80°C, set minimum duty cycle for 45°C. That'll kill the vast majority of Windows caused ramp-ups at idle and fan speeds will almost never get over @ 70%

 

[Solandri]

Not quite sure what you're asking. But a radiator simply transfers heat from the radiator to the air. A passive radiator works due to the heated air rising, forcing cooler room air to enter the bottom of the radiator. The hotter the radiator gets, the hotter it heats the air, and the more tendency there is for the air to rise out of the radiator. If this natural convection is sufficient to cool the radiator when you're at idle, then the fans will never kick on (at idle). The fans are only necessary when the heat load from your PC exceeds the capability of this natural air convection to remove heat from the radiator.

There are actually several natural heat transfer mechanism which ramp up in response to increasing temperature even if you don't have a fan. There's natural convection described above. Radiation (the heat you feel on your face when you stand next to a fire) increases with increasing temperature. And the rate of direct heat transfer (from radiator to the air) is proportional to the temperature differential. So assuming your air temp remains constant, the hotter the radiator gets, the faster its rate of heat transfer to the air. All this happens without any temperature sensors or active monitoring. The last one is why an increase in heat generation rate (first derivative of temperature) results in a proportional increase in temperature, rather than a constantly increasing temperature like you'd expect from the math.

 

[tennis2]

I have an eco setting on my gpu. The target temp is @ 65°C. Below 65°C the fans are off. That's great, if you don't game. Gaming puts the gpu temp from 32°C to +65°C in a matter of minutes. So then the fans kick on at 70+% until the gpu goes to @ 60°C, then off. And on. And off... 0rpm to almost 2000rpm, then back to 0rpm in seconds, repeated every 10-15 seconds.

Wow, that's sucky. I've never had that problem with my card. I believe my fanless limit is somewhere around 55-60C though. Perhaps that's the difference. My GPU fans' speed transitions aren't terribly rapid though either.

 

[liamwhalley20]

As I said my objective is to literally reduce noise from the radiator fans I have installed, so any fan that will be quiet and also provide enough cooling for the radiator, I thought the f12's would suffice but clearly not

 

[liamwhalley20]

And my gpu doesnt have a fanless limit, the fans are on all of the time from start up, but they do not create that much noise at the temps stay around 50 on 60% fan speed

 

[Paperdoc]

The Arctic F12 PWM fan can deliver 53 CFM at its max speed of 1350 RPM. The F12 TC with the same max speed and air flow is a THREE-pin fan so its speed can NOT be controlled by a header using the newer PWM Mode of control. However, it has its own temperature sensor that reduces fan speed at low temps, so its speed is related to temperature even though the mobo is not doing that work. The F12 Silent is also a 3-pin fan whose speed can be controlled by a mobo header ONLY if that header is using the older Voltage Control Mode (aka DC Mode) - PWM headers cannot control this fan. And it is "silent" because its max speed is only 800 RPM, delivering a reduced 37 CFM air flow. All of these fans CAN be connected to a 4-pin header, but if that header is using PWM Mode for control then the first one will be speed-controlled, the second one will do its own control even though the header cannot, and the third will have NO speed control. NONE of these can be under mobo speed control unless they are connected to a mobo fan header, either directly or using a proper Splitter or Hub.

Ah, I missed a detail in how your fans are connected. Deepcool recommends that the PUMP be connected to the CPU_FAN header for a particular (good) reason. Assuming that header by default is using the newer PWM Mode for control, the pump will receive a constant +12 VDC power supply from Pin #2 and thus will run at full speed all the time as designed. But more importantly, that header does some extra-tight monitoring of the SPEED signal from whatever is plugged into it to check for FAILURE of that device. In an AIO cooling system, failure of the pump is much more dangerous than failure of one of the radiator fans, so it is always best to connect the pump to this header unless you have other comparable options. Then that leaves the need to control the speed of the FANS on the radiator, and still that needs to be done by using for the temperature information the sensor inside the CPU chip, and NOT the one built into the mobo. In your case, it appears your only choice is to use a SYS_FAN header. (You have not told us your mobo model, so we cannot check further details.) Many mobos have an option in the configuration of any SYS_FAN header to choose the correct temperature sensor, but the default one is usually the motherboard sensor. IF that is your case, you should change it. If your rad fans are being guided now by the motherboard temperature sensor, then their behaviour is NOT being guided properly.

On the other hand, if you cannot make that configuration change, there IS a way to get the rad fans under control by the CPU_FAN header that IS using that internal CPU chip sensor. So if you need to know how that is done, post back here for advice.

 

[liamwhalley20]

Well I have about 3 splitters lying around plus the hub that deepcool provided, so if I used the hub with the pump and fans linked into the cpu header instead of the sys fan header that could work? I dont have an option to choose which sensor it works off

 

[Paperdoc]

Yes, that's the idea, but there is an important detail for you. It is vital that the speed of the PUMP is returned to the CPU_FAN header. A fan header can only deal with the speed signal from ONE device - any more and it gets to confused and gives wrong answers and bad error messages. So the Splitter you have will connect the speed signal of only ONE of its devices back to the host header, and will ignore all the others. My guess is that, if you look closely inside the four sockets of that Splitter, #1 will have all four pins in it, and the other three will each be missing Pin #3. So be sure that the PUMP is plugged into the socket with all four pins, and the rad fans are plugged into other sockets.

This will put all of them - pump and two rad fans - under automatic speed control by the mobo based on the internal temperature of the CPU chip. (The pump speed actually will not change, and that is by design.) The PUMP will be monitored for failure and its speed can be displayed for you as the "CPU Fan Speed". You will never see the rad fan speeds, and they will NOT be monitored for failure. So YOU should check them from time to time to be sure they both are still operating correctly.

 

[liamwhalley20]

Yes, the hub is clearly marked anyway saying which is the first header, I shall do that soon and report back, yes I will be sure to check that they're working from time to time, I will eventually be buying a new system so I will be sure to get a motherboard with better fan control than the one I currently have XD

 

[Karadjgne]

While that sounds good in theory, I have to disagree with pump AND fans on the same header UNLESS there's an alterior firm of fan control. There's very few aio pumps actually designed for variable speed control. Nzxt is the only Asetek pump that comes to mind, and it's a hybrid built by Asetek in conjunction with Nzxt, not a purely Asetek supplied pump. Coolit pumps are not designed for variable speeds either, the reason being that they are already such low pressure/low volume pumps that any deviance from a constant 12v supply has detrimental affects on both ability and longevity of the motor.

Pump can be on any header, if a 3pin, it must have fan control either bios set for 100% duty cycle, disabled fan control, of is using mobo software or SpeedFan, set for 100% constant. PWM does not affect voltage, so is always at 12v constant.

Fans can be on any other header. Corsair directions specify pump on cpu_fan and fans on sys_fan. Only because if the cpu_fan header tach reads 0rpm, it'll shut down the pc or refuse to boot, showing a cpu_fan error. Personally, I'm not a big proponent of that idea, most motherboard bios is too simplistic to seperate sensors from headers, which leaves pump running 100% (cpu_fan is a dedicated pwm header, no matter what mode is used for the sys_fans) and most case fans at the mercy of whichever sensor it's tied to. I've seen sensors that always run hot because they are parked next to the Northbridge or Sata controller chipsets and sensors that rarely change because they sit right above the psu. SpeedFan, HWInfo, Asus suite all report me having one sensor at -125°C and another at +255°C, not good for fan control at all.

So for me, pump goes on cpu_aux, cpu_pump, or adjusted sys_fan or even direct to psu, and fans go to cpu_fan. The only exception is higher-end aios where fans are directly connected to the pump housing and controlled by software via usb/Sata, which leaves the pump/fans powered by Sata, controlled by software and pump tach on cpu_fan to avoid cpu_fan error.

Regardless of vendor directions, there really is no right or wrong way to hook up an aio. There's only what works and what doesn't, for you personally. I've seen plenty of ppl not use usb and software control (corsair link had more bugs than an anthill for many years), so put a splitter on cpu_fan with the fans, and pump went to cpu_aux or other. Same hookup as a basic aio, using bios settings. It's not wrong to do it that way, it's just not what the vendor envisioned in design, and it's not what they prefer since the chances of pump failure are higher than fan failure.

Of course I ran my aios 24/7, 360 days a year, for over 6 years, shutdown only to clean, and wouldn't you know the first and only thing to fail was a fan.

 

[tennis2]

I agree, my pump is on the dumbest fan header my mobo has cut it just needs a constant 12V/100% all the time.

Mobos from the Haswell and earlier era's didn't have great fan control, so like I said before, the CPU fan header is the best place to connect the fans that need to vary speed the most. It's a shame to waste that functionality on a pump that just needs a constant 100% duty cycle.

If your pump fails and temps go crazy, your CPU fan(s) will be running 100% all the time (which you may or may not notice, especially if you're only using 800rpm fans). If all else fails, modern CPUs just throttle back frequency as much as needed to keep temps in a safe range.

 

[Paperdoc]

My suggestion to connect the pump AND the two rad fans to the same CPU_FAN header via a Splitter is based on three points.

1. On many mobos, monitoring for failure of the "fan" (or whatever) plugged into the CPU_FAN header is more intense than for other headers, so that failure will trigger an obvious warning message plus (in many cases) completely shut down the system without even waiting for a high temperature to be detected by the CPU's internal temperature sensor. On such systems, often it will refuse to start up if the CPU_FAN device fails to show a speed signal immediately when power is first turned on by the On / Off button. As I said, my belief is that, with a AIO system the PUMP is the critical unit that needs this level of failure monitoring. If it fails, cooling of the CPU is severely reduced very quickly. On the other hand if ONE (or even BOTH) fans on the radiator fail, cooling is reduced but it will result in slow increase in CPU temperature until the other CPU protection system (high internal CPU temp detected) causes the sequence of reduced processor speed, followed by shut-down if temp still continues to rise. I believe this slower rise is less dangerous to the CPU than a rapid rise that is not corrected quickly. I recognize that some mobos have dedicated PUMP headers that probably (not clear in most manuals) are subject to careful failure monitoring, but in this thread OP has not told us what mobo is involved, so we cannot advise use of a header that may not exist.
2. The radiator fans need to have their speed governed by the temperature sensor inside the CPU chip, because in almost all AIO systems it is these speeds (NOT the pump speed) that control removal of heat from the CPU chip. Virtually all CPU_FAN headers can use ONLY that sensor's readings as a guide, so using that header to govern the rad fan speeds makes sense. Many mobos (but not all) allow you to control the speed of their attached fans from that same CPU internal sensor from a different header, such as CPU_OPT or some SYS_FAN or CHA_FAN headers (optional choices in their configurations). Again, in this thread we don't know whether OP has those choices available. If not, the only header that we can be sure uses that sensor for guidance is the CPU_FAN header.
3. I agree that this AIO system (like most) is designed to have the pump run at full speed all the time. A great many of those use a little "trick" based on the compatibility features designed into the 4-pin fan control system. In this newer system, the power supply to the fan on Pin #2 is always +12 VDC, unlike the older Voltage Control Mode which reduces that voltage to reduce fan speed. Then the header puts out on Pin #4 the PWM signal that the new fan design uses (via its own internal chip) to modify flow of current from that fixed voltage supply through the motor windings, thus changing motor speed. BUT if you connect to such a header an older-type 3-pin fan, it does not receive that PWM signal that it could not use anyway (it has no special chip), so all it "knows" is that its power supply is always +12 VDC, and it runs full speed all the time. For a fan, this means you get good (full) cooling, but no speed control, a reasonable default behaviour for this mis-matched configuration. Almost all AIO system pump motors are of exactly the same type as 3-pin fan motors, so they exhibit exactly this behaviour. They have connectors on their cable ends for a 3-pin system only. They do not pick up the PWM signal from PIn #4 and cannot use it, anyway. The result is that, IF the pump is connected to a 4-pin header that really is configured to use the PWM Mode to control its "fan" it WILL always run full speed as designed.

OP, on my last post I neglected to add one configuration detail. Your CPU_FAN header configuration options may well include choice of control MODE. IF there's a choice, it will usually include three items: PWM, Voltage (DC), or Auto. If you set it to Voltage then you MUST also remember to set it to full speed always. If you set it to Auto (often the default setting) there is a good possibility it will decide that the connected device (the pump) is a 3-pin older fan type and change itself to use that older Voltage Control Mode and that would let it run slower pump speeds. By far your best choice is to let the system use that quirk of design: set this option to PWM for sure, and your pump will run full speed always. Then the fact that the two other devices on that Splitter (the rad fans) are real 4-pin fans means they WILL receive the PWM signal and have their speeds controlled by that CPU_FAN header.