A PC cpu fan in an 8 core AMD system is very noisy at full tilt, and is rat her excitable. CPU temp only need go up by an imperceptible amount and it r evs up (idling for a minute.) I want to quieten the thing somehow. I'm thin king the control system is presumably pwm, and a series doide or 2 should h elp, and probably not cause problems. Am I right?
I wish the control system could be better, I want quiet and don't mind a ho ttish cpu. The other option is just to put the more usual low power fan on and hope it doesnt overheat when pushed. But presumably it would, hence the very high speed fan. Its got to be quieter for the pc to be useful.
If it doesn't turn out to be a software issue, you might want to consider a different, *better* cooling system rather than a poorer one. In particular, water cooling can be very quiet and keep your CPU cooler. I haven't looked at them lately, but they aren't all that complex. The water pump can be very quiet. The water is circulated through a, by comparison, large radiator which is cooled by a large, slow fan that can also be fairly quiet.
I was going to once make a water cooling unit for my desktop. I had a heater core for the external cooler. I never quite got the cooling block made before I lost interest. Seems gluing Plexiglas to aluminum doesn't work well using drug store epoxy.
I have the opposite problem. My laptop has an i7 and the exhaust is so hot I can't leave my hand in front of it. I never hear the fan no matter what is happening with the processing power. Maybe I have a software problem, lol. I know there are a *lot* of issues with this Lenovo. I'll never have another one. I thought my Toshiba was a low quality product, but this thing makes the older Toshiba look like the gold standard.
Nope. At best, that's a band aid. Maybe spending some time finding the cause of the problem would be more productive.
First, get a decent thermal sensor monitor. This isn't great but since it's from AMD, I assume it's been tweaked to work with your unspecified AMD chip. I like:
The fan is normally controlled by the BIOS. Go the unto the web site of your unspecified motherboard and check for an updated BIOS.
The fan should not start up that quickly, so I would check if the heat sink is properly attached to the CPU. It is possible to assemble a heat sink so that it is not parallel to the CPU top surface. The idea is for the thermal goo to fill in the cracks in the metal surfaces. A thin layer is best and too much goo makes things worse. If you have a thermocouple probe with a tiny tip, measure the temperature of the CPU and the heat sink. If you have an FLIR thermal camera, even better.
I've spent considerable time and effort reducing the noise from my machines, with good results. When watching video on the computer, I don't need loud fans. The noise made by the CPU fan is mostly controlled by the fan speed and by any obstructions in the air path. Cooling is strictly CFM (cubit feet per minute) or air flow. Blade tip vortexes also have a big effect on noise. When you install a small fan on the CPU heat sink, it has to spin faster to move the same number of CFM. The result is that rapidly turning fans are noisy, while large diameter slowly turning fans are quiet. If you take a look at Dell desktops, you'll find that they have one BIG slowly turning fan on the front or back of the machine. It's also carefully positioned so that it cools the CPU first, making the CPU fan a secondary cooler. Some don't even have a CPU fan, like my Optiplex
960. The 120 mm fan is on the left side of the black fan shroud: There's no fan in the power supply. The rear of the case is also heavily perforated to minimize back pressure.
Normal operation barely makes the fan spin, which is fine as that makes it very quiet. The tiny fan on the video card makes more noise than the single big case fan, so I'll be switching to a no-fan video card fairly soon. I've looked with other PC brands and case designs, most of which make far more noise than if they used a big fan.
It's difficult for me to make comparative noise level measurements, so I use a cheap Radio Shack sound level meter. My meter and data are in my palatial office so I can't post numbers right now. Bug me if you want to compare.
Incidentally, if you want to check your work, try either Prime95 of the Intel Burnin programs: Running these programs will speed up the fan and probably make more noise, but then your machine should not be doing that very often. The only time mine ever speeds up the fan is when I'm rendering an image with Poser or Daz Studio.
If you think you have it working, find a lung cancer candidate and have them blow some cigarette smog into the fan intake and check the airflow. If there is stagnation anywhere, you might want to build a duct, plenum, or baffle to redirect some of the air. I have one machine with a 3/4" ID "smurf tube" pushing air to a rather warm 2nd video card.
Use: to take over the job that the BIOS scribblers mangled.
No. Put a LARGER fan on and run it at slow speed.
Yep. I think you understand how it works.
--
Jeff Liebermann jeffl@cruzio.com
150 Felker St #D http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann AE6KS 831-336-2558
On some Dell machines, the fan is inaudible when new, but gets progressively louder over the years. After 3-4 years, you have to replace the bearings in the fan to stay sane.
Just a thought, the fan may be going bad, and either new bearings or a whole new fan may be the solution. Also, check for dust blockages that require the fan to work harder.
I made a cooling block out of some sheet copper and copper tubing. Copper is easy to work with and can be soldered using a small propane torch. I ju st used a large jar for the external cooler. By large I mean a gallon or t wo. Did not use any fan on the external cooler, but the water never got wa rm. That was a good while ago. Since then CPU's generate more heat.
That must have been quite a while back then. Even 10 years ago CPUs were in the 100 Watt club. I remember one guy's computer was silent other than the hard drive. He piped the coolant through the wall to a
50 gal barrel in his garage which never really got warm. Even the pump was on the other side of the wall. Not sure how he got rid of the PSU fan.
I got an aquarium pump for my project for free. I had contacted the manufacturer for info. When I explained what I was doing they sent me one. lol It really is quiet and moves a fair hunk of water. Half inch tubing is rather thick walled and bulky though. It would be a PITA to pipe it inside a standard computer case. Maybe metal pipes inside and flex tubing on the outside.
Also, sometimes there are options in the BIOS setup that let you tweak the fan controller algorithm - possibly labeled something like "Quiet" (fan doesn't speed up until the CPU warms up quite a bit), "Performance", (fan speeds up as soon as the CPU temperature goes up just a little), and "Max" (fan runs at 100% all the time). Adjusting this setting, if you have it, may help.
Usually somewhere in the BIOS you can adjust the criteria for fan control and the temperature to speed mapping and gradient. When my office is cool in the morning I have actually have the CPU fan stop if the machine is idling. I can keep mine under 50C flat out and fairly quiet. It is all but silent with only one core running flat out.
Are you sure that it isn't that the bearings are failing or something is loose and vibrating. Check the BIOS settings you may be pleasantly surprised that there is an easy fix in there.
ather excitable. CPU temp only need go up by an imperceptible amount and it revs up (idling for a minute.) I want to quieten the thing somehow. I'm th inking the control system is presumably pwm, and a series doide or 2 should help, and probably not cause problems. Am I right?
hottish cpu. The other option is just to put the more usual low power fan o n and hope it doesnt overheat when pushed. But presumably it would, hence t he very high speed fan. Its got to be quieter for the pc to be useful.
OK, done a little testing to understand better the issue.
The CPU climbs to 52C rapidly when powered up. The base of the heatsink is about as warm as the cpu, so thermal contact is good, and the fan is kickin g out a good stream of very warm air. So it seems that a) the cpu is eating excessive power on idle b) the fan is unnecessarily fast & noisy
BIOS offers 3 settings: eco, standard and overclock. I changed it from stan dard to eco, but am noticing no difference. There's nothing else that I can see is relevant, the power options section I made little sense of and will look for the mobo manual.
The fan itself was noisy from new, afaics due purely to it running so fast. The noise is high-rpm type not bearing type. The behaviour and noise haven t changed any. I've refitted it to the heatsink on a rubber gasket & tied i t on with rubber bands, and that has certainly helped, but its still sounds too much like an airport.
When I moved it last week the HS was totally clogged, but its completely cl ear since then, so that's not part of the picture.
I've no interest in water cooling. I know it works well, but don't even beg in to have the time.
It originally ran win7, and now runs avlinux 6.0.1, with the same result no isewise. Av is debian based.
There's no room for a larger fan unless I build a cowl. It might need it, b ut I'd much prefer to just calm the fan down and let the cpu run hotter. An d if possible also reduce cpu power use.
Finally the psu fan is large & quiet, so its not contributing significantly to noise. The graphics card is fanless, I'm not fond of fanned cards. (I d id have a sort of one once, back when high performance 16/32M cards were go ing cheap because they overheated badly. I just added a fan.)
About a decade ago I was working on the PPC970MP (used in the PowerMac G5) we had a "codename that processor contest" for the Apple processors. I won it with the "Bright Stars" theme and the MP was named Antares. My boss' boss sent an email announcing the winner and saying that he hoped it wasn't a comment on its power dissipation (120W, IIRC). I did a quick back-o-the envelope calculation that said that the power density was ~1E9 that of the sun. He wasn't pleased. No sense of humor. ;-)
I've had problems with this case type in both Intel and AMD CPU's. Some are not quite flat. The large surface area tends to collect "lumps" which cause gaps. I know that you said that the CPU and heat sink temperatures were "close", but I suggest you tear it apart, clean off all the heat sink compound with alcohol, and start over. I've fixed a few overheating machines this way. In a few cases, I found some debris mixed in with the heat sink compound that was causing the problem. However, mostly it's just getting unwieldy heat sink assembly to properly mate with the CPU assembly.
Good. Then Linux is not doing something in the background to increase power consumption.
About 50C at idle is about right. Here's an FX-8350 system showing about 50C at idle that was fixed with new thermal goo:
The max safe core temp limit for 4/6 cores is 71C. For 8 cores is
61C. Maximum socket temperature is about 70C. However, after reading some of the overclockers pages, these limits are conservative and you can go higher (if you must).
30C temp rise above ambient is about right.
Yep. At about 70C it starts to throttle down.
Reminder: I consider the excessive heating and the noisy fan two separate problems.
Yikes. That's much too fast. At idle, you should be running about
2000 rpm. However, there's a chance that whatever program you're using to measure the CPU fan RPM is wrong. Try the BIOS numbers or a different program. If you have a strobe light, try using it.
I'll guess(tm) that your 70mm fan is that "low profile" variety and only 10mm thick. Digging through some random data sheets, The top fan is rated at 4.2 watts, 4200 rpm, 25cfm, and 36.7 dBA noise which I think this fan is fairly typical. By comparison, a 120mm x
25mm fan such as: produces 37cfm, 20dBA noise which is 17dB or 40 times as quiet, and spins at only 950 rpm. See the difference that diameter makes? If you want a quiet machine, get a bigger fan.
Any restrictions to air flow in back? Just curious. I don't think this would make any difference in noise levels because mini-tower cases already have so many air leaks that back pressure unlikely. If it had been a small (SFF or USFF), that would be a concern.
No experience with this board. I do mostly Intel based MB's. I don't see anything that might affect cooling.
--
Jeff Liebermann jeffl@cruzio.com
150 Felker St #D http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann AE6KS 831-336-2558
You might find this video interesting, which shows what happens when you block the air flow on an AMD 8350 CPU.
AMD FX-8350 Cooler noise
Also, note the size of the fan, which looks about 100mm.
--
Jeff Liebermann jeffl@cruzio.com
150 Felker St #D http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann AE6KS 831-336-2558
Keep in mind that these CFM numbers are for no restriction, no back pressure. Once the fan is added to a reducing cowling back pressure will go up significantly. But it won't hurt to try it if all the parts are handy. Fans are a lot cheaper than fancy heatsinks.
Its the orignial stock AMD fan, rated 12v 0.7A. And yes, it idles at 6k rpm :/ Sounds like an airport.
I removed solidified heatsink compound and replaced, but its no better. See ms a bigger fan or less likely extensive soundproofing are the only options . I dont have anything that powerful in stock. I did briefly try underclock ing it by 25% but noticed no difference.
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.