Im thinking about this again and again.
There is clearly a direct correlation between case temps, fan noise and air circulation.
Firstly, what is the general concensus about good case temp for a quiet PC? Perhaps 10c above ambient?
Flowing on from that is the relation of fan noise and air circulation inside the case. Naturally the faster you replace the air inside the case the cooler it becomes but also the noisier it gets.
So im looking for some thoughts (i realise there is no definitive ruling) but what do you think?
Furthermore i am keen to try to achieve a very slight positive air pressure in the case, to assist with fresh air being pushed into the case whilst minimising dust that is currently sucked through every crevice/gap in my HTPC case. I only have 2 exhausts fans, the 14cm psu fan and a 60mm useless case fan.
case temp vs fan noise v air circulation
Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee
-
sciberpunkt
- Posts: 14
- Joined: Tue Oct 24, 2006 5:42 pm
I still find myself tweaking something every now and again on my six month old system. Spending time researching those causal relationships between each fan and different temperatures and experimenting with different configurations should be considered a Zen practice. 
I've found that around 15C above ambient is more realistic (in my desktop-style case at least). I can run Folding@Home while defragmenting the hard drive for a couple of hours and stay under 40C. Venting the heat from the CPU out of the case bought me a few degrees. Replacing the 3.5" hard drive with a 2.5" one bought me a couple more. One 80mm fan spinning at 1000rpm dropped my hard drive temp by 10C. It's the small refinements that make the difference.
I've found that around 15C above ambient is more realistic (in my desktop-style case at least). I can run Folding@Home while defragmenting the hard drive for a couple of hours and stay under 40C. Venting the heat from the CPU out of the case bought me a few degrees. Replacing the 3.5" hard drive with a 2.5" one bought me a couple more. One 80mm fan spinning at 1000rpm dropped my hard drive temp by 10C. It's the small refinements that make the difference.
Re: case temp vs fan noise v air circulation
I think the right spot is where you are safely cooling your components, but no more than that.
Your CPU can get quite hot before switching to thermal protection. Assuming you've done all you can by using one of the best large CPU coolers and a fan that pushes as much air as possible at a given db level, I wouldn't worry too much about spinning the fan up too high to protect your CPU. CPUs are solid state devices that will last much longer than you'll ever need, and then some... overclocking is popular because CPUs can be pushed to the brink and still last for 5+ years, which is likely beyond their useful life for most of us. That's not to say you should run your CPU really hot just for the heck of it, but damage to the CPU is actually quite hard to do (again, assuming you're using proper gear), so if its a matter of running say the CPU @ 45C/40db or 50C/34db, I'd choose the latter every time, all else being equal.
Of course, in the real-world, not everything is equal, and so you do have to be careful that running the CPU hotter doesn't significantly put your PSU, video, MB or HDs under more stress. That's where the whole zen thing comes in... I think the right balance is different for every situation, since no two computers have exactly the same components, case or cooling apparatus. You have to experiment with the best practices and find the right combination for you.
From a perspective of ensuring adequate cooling, the only real concern I have is hard drives -- that's where all your data is stored and its a real bitch to lose it all because you got lazy and haven't backed up in a couple months. I've learned over the years that to increase hard drive reliability, you want to do two things: Keep them thermally stable and don't spin them up and down more often than you have to. That's why I always keep them cooled to between 30-40C with a fan and rarely ever spin them down (turn off the computer.) So far, I've had good luck when following these principles. I know that HD manufacturers claim higher tolerances, but I find that if your drives are constantly spinning up and down (like 2-3x every day), it quickly takes its toll on the moving parts and the risk of major catastrophe does increase, despite all those MTBF assurances. Every time your computer is off and you turn it on for 15 minutes or longer, your HDs internal temp goes from ambient to something like 25 degrees hotter than ambient (assuming no fan) -- and back down to ambient after you shut it off. The motor also has to overcome the significant inertia of the platters on startup. By adding a slow moving fan (as you have), you reduce the temperature shock by 10-15 celcius on startup. By avoiding the HD spinning down in the first place, you're more or less eliminating the yo-yo temp shocks and avoiding having to constantly deal with startup inertia. I've found that these are keys to drive longevity.
Aside from that, the only other thing I would offer is that for me, I have no qualms about running my computer a little louder when its chugging away at something processor intensive (like a game, or rendering, or mp3 conversion). It's when I'm nearby but not actively using it, or when I'm sitting there typing a document into a word processor that I appreciate near silence... The simplest solution that provides such flexibility is thermally controlled fan controllers. They allow you to get the best of both worlds: on demand power when you need it and near-silence the rest of the time.
Your CPU can get quite hot before switching to thermal protection. Assuming you've done all you can by using one of the best large CPU coolers and a fan that pushes as much air as possible at a given db level, I wouldn't worry too much about spinning the fan up too high to protect your CPU. CPUs are solid state devices that will last much longer than you'll ever need, and then some... overclocking is popular because CPUs can be pushed to the brink and still last for 5+ years, which is likely beyond their useful life for most of us. That's not to say you should run your CPU really hot just for the heck of it, but damage to the CPU is actually quite hard to do (again, assuming you're using proper gear), so if its a matter of running say the CPU @ 45C/40db or 50C/34db, I'd choose the latter every time, all else being equal.
Of course, in the real-world, not everything is equal, and so you do have to be careful that running the CPU hotter doesn't significantly put your PSU, video, MB or HDs under more stress. That's where the whole zen thing comes in... I think the right balance is different for every situation, since no two computers have exactly the same components, case or cooling apparatus. You have to experiment with the best practices and find the right combination for you.
From a perspective of ensuring adequate cooling, the only real concern I have is hard drives -- that's where all your data is stored and its a real bitch to lose it all because you got lazy and haven't backed up in a couple months. I've learned over the years that to increase hard drive reliability, you want to do two things: Keep them thermally stable and don't spin them up and down more often than you have to. That's why I always keep them cooled to between 30-40C with a fan and rarely ever spin them down (turn off the computer.) So far, I've had good luck when following these principles. I know that HD manufacturers claim higher tolerances, but I find that if your drives are constantly spinning up and down (like 2-3x every day), it quickly takes its toll on the moving parts and the risk of major catastrophe does increase, despite all those MTBF assurances. Every time your computer is off and you turn it on for 15 minutes or longer, your HDs internal temp goes from ambient to something like 25 degrees hotter than ambient (assuming no fan) -- and back down to ambient after you shut it off. The motor also has to overcome the significant inertia of the platters on startup. By adding a slow moving fan (as you have), you reduce the temperature shock by 10-15 celcius on startup. By avoiding the HD spinning down in the first place, you're more or less eliminating the yo-yo temp shocks and avoiding having to constantly deal with startup inertia. I've found that these are keys to drive longevity.
Aside from that, the only other thing I would offer is that for me, I have no qualms about running my computer a little louder when its chugging away at something processor intensive (like a game, or rendering, or mp3 conversion). It's when I'm nearby but not actively using it, or when I'm sitting there typing a document into a word processor that I appreciate near silence... The simplest solution that provides such flexibility is thermally controlled fan controllers. They allow you to get the best of both worlds: on demand power when you need it and near-silence the rest of the time.
-
Bluefront
- *Lifetime Patron*
- Posts: 5316
- Joined: Sat Jan 18, 2003 2:19 pm
- Location: St Louis (county) Missouri USA
I agree 100% with gourdo_1 about fan controllers. The perfect situation is a fan that slows down to almost inaudible when the computer is idling, and speeds up to achieve the desired temperature when the ambient is higher, or the computer is being stressed, with a few hundred rpms left in reserve.
This is not easy to achieve....it takes the right controller, the right fan, and the correct placement of an external sensor for the controller. The controller must have the right temperature curve, or must have an adjustment capability to achieve the right balance between airflow and temperature. You can spend a bunch of time and money to achieve this. I have yet to find a built-in fan controller that can be tweaked to perfection.....and I have not found a controller that relies on software to do this.
My solution.....A quiet fan that will run slowly but can speed up to 1600-1800 rpms (Scythe S-Flex 1600rpms) for the CPU, a fan controller with a temp curve about 28C-42C (NoiseMagic NTM3 with an extended sensor wire), and a CPU heatsink where you can mount a sensor easily in a variety of places to achieve the perfect temperature curve (Ninja).
This same controller can be used for every fan in your setup....one controller per fan. Adds some complication, but will give you a perfect balance after some tweaking....
This is not easy to achieve....it takes the right controller, the right fan, and the correct placement of an external sensor for the controller. The controller must have the right temperature curve, or must have an adjustment capability to achieve the right balance between airflow and temperature. You can spend a bunch of time and money to achieve this. I have yet to find a built-in fan controller that can be tweaked to perfection.....and I have not found a controller that relies on software to do this.
My solution.....A quiet fan that will run slowly but can speed up to 1600-1800 rpms (Scythe S-Flex 1600rpms) for the CPU, a fan controller with a temp curve about 28C-42C (NoiseMagic NTM3 with an extended sensor wire), and a CPU heatsink where you can mount a sensor easily in a variety of places to achieve the perfect temperature curve (Ninja).
This same controller can be used for every fan in your setup....one controller per fan. Adds some complication, but will give you a perfect balance after some tweaking....
-
spookmineer
- Patron of SPCR
- Posts: 749
- Joined: Sat Nov 11, 2006 6:02 pm
Re: case temp vs fan noise v air circulation
Having a lot of airflow will make it more turbulent (noise) but it will also just push air through the case that will never come into direct contact with a hot part (not that this is needed in a normal situation, because the temperature will distribute among the rest of the air, but the temperature coefficient of air is bad).GLO wrote:Naturally the faster you replace the air inside the case the cooler it becomes but also the noisier it gets.
A high airflow is very inefficient for this reason: a lot of the air will not even be warmed up a bit. Cooling wil be very good, but at the cost of a lot of noise and "spilled air": air that doesn't warm up.
An efficient airflow is what to aim for (imho). Slow enough to not be too turbulent (therefor quiet) and slow enough to "soak up" the heat from hot parts, but still being able to cool all the parts sufficiently.
From my (very short) experience, there is a fine balance of keeping parts cool, noise down and getting an efficient airflow.
I have a log stating all the readings I have so far but some parts are coming in, too soon to be definitive.
My system isn't quiet by a long shot, but I have removed 3 of the 5 case fans with the same temperature results.
I've been using a fan control to check temps and to regulate 1 of the 2 rear case fans (the other is not suited to control).
So far, my readings are (all @ idle):
Code: Select all
1x 80 fan @ 2000 1x 80 fan @ 2000 1x 80 fan @ off 1x 80 fan @ off
1x 80 fan @ 3100 1x 80 fan @ 1400 1x 80 fan @ 3100 1x 80 fan @ 2000
CPU fan @ 1600 CPU fan @ 1600 CPU fan @ 1600 CPU fan @ 1600
CPU: 53 °C CPU: 55 °C CPU: 57 °C CPU: 60 °C
GPU: 34 °C GPU: 35 °C GPU: 35 °C GPU: 37 °C
HDD1: 25 °C HDD1: 26 °C HDD1: 26 °C HDD1: 26 °C
HDD2: 26 °C HDD2: 32 °C HDD2: 32 °C HDD2: 34 °C
air in: 21,3 °C air in: 21,6 °C air in: 21,5 °C air in: 21,3 °C
air case: 25,2 °C air case: 25,2 °C air case: 29,1 °C air case: 29,1 °C
CFM case: 72 CFM case: 50 CFM case: 40 CFM case: 26 The intake air sensor is placed in the front fan grill, the case temp is placed to the upper part on the metal support for the PSU (the sensor doesn't touch the metal, so it really measures the air temp).
Below 50 CFM, my case temp starts to rise quite a bit.
The CPU temp rises gradually with decreasing airflow.
The noisiest part is the CPU fan (better heatsink coming up).
I suspect (and read from other users here) an amount of airflow lower then 50 CFM is well possible with good results.
I don't dare to let my case temp rise more, as the CPU is hot.
With other settings (CPU fan @ 950, case fans 1x @ 2000 1x @ 1500, CFM 50) it is really much more quiet, but running 3DMark03, it shut the program down when CPU temp reached 69 °C (it's ok, it's an old Barton 3200+). I don't have intake or case temps of that.
For me, this is a continuing process, I just started tweaking stuff and I'm not even halfway of what I want.
The fan control is very usefull because it provides readings I can draw conclusions from (as: with the heatsink I've got now, about 50 CFM through the case is the least I have to have).
Like many people have said, try things and monitor what happens, it's the only way for any specific case.[/quote][/code]
I'd say the perfect situation is a fan spinning at 600 RPM and providing enough cooling even at 30C ambientBluefront wrote:The perfect situation is a fan that slows down to almost inaudible when the computer is idling, and speeds up to achieve the desired temperature when the ambient is higher, or the computer is being stressed, with a few hundred rpms left in reserve.
. BTW, it's possible and probably not much harder to implement than a smart fan controller.