Quiet PCs, Ducting Air from Case Fan to Heatsink?
Milo_Mindbender asks: "While listening to the whine of my heatsink fan I was wondering. It seems like a good way to get quiet cooling for the CPU would be to mount a fan in the back of my case and run a duct of some kind (folded sheet metal or some kind of hose) from the back of the fan directly onto the top of a fan-less CPU heatsink. You should be able to get the same amount of airflow with a large slow (quiet) case fan as you do with a little noisy cpu fan...and the air being blown onto the heatsink would be cooler as well. This seems like a fairly obvious idea so I'm wondering if there's some reason why it wouldn't work, or if anyone has tried it and could tell us how it turned out." Yeah, but what about the heat in the rest of the system? Depending on the size of your enclosure (and what's in it), you may or may not need more than one fan. Has anyone tried something like this and can comment on how well it worked?
Ducts - The Cheap Cooling Solution
Didn't work too well until there were 5 fans in the side of the case,
link
I'm glad people are actually starting to care about noise reduction in PCs. People tend to get a little defensive when the new fan they bought keeps the people in the next apartment awake, just so they can hold onto the few extra MHz they were able to squeeze out.
More on topic, back in my tech days I remember seeing a setup something like this in an IBM case. I don't think the processor had as much need for cooling as the current bleeding edge of speed (it was maybe a Celeron 300 or so) but the heatsink on the CPU was cooled by the fan in the power supply.
I'm no expert but it seems like ducts could work just fine especially if you had a fan the size of the side of your case. Maybe I'll strap a house fan on my box and just set it to low, I hope the magnets don't erase my hard drive.
You'd have two big problems.
1) Turbulence in the ducting would reduce your effeciency
2) Cooling the ducting itself
Your best bet would be to get a larger, 80mm heatsink and use a larger, quiter fan right on top of that. They are a little heavier and larger, so your mother board must be able to support it.
I have two dell PCs, the first is 4 years old, the second 2 years. They both use a plactic duct that is attached to a fan in the back and covers the CPU entirely.
The four year old is still really quiet, the other one is starting to make more noise. But that's because of a buggy fan on the video card...
-> maybe we can apply the same strategy there?.
Its kind of old, a FIC Neptune mini-NLX system. If you are at all familiar with the NLX chassis layout, the CPU is placed in the upper right corner of the motherboard, directly in front of the front case fan. There is a small duct focusing the fan on the CPU, and as there isn't even a place to plug a CPU-mounted fan in, you have to use a passive cooler and hope the case fan is sufficient. It was designed for Pentium II and first-generation celeron systems, apparently the motherboard can run coppermines though...so far, with a slow celeron, its been fine with no CPU fan.
"(Man) tries to live his own life as if he were telling a story. But you have to choose: live or tell." --Sartre
I think the main problem with your plan is the amount of airflow required to properly cool a CPU with just a heatsync. Unless it is a Celeron300A or C3 or something of that nature, if you use a just a heatsync you'll need an 80mm fan running at least 2000 RPM. At that speed, the noise the air makes is quite noticeable, even if you choose to get one of the ultra-quiet brands.
I finally decided to just get a new heatsync/cpu fan combo for around $30 from QuietPC and I have to say that not only is it almost silent, it keeps my system cooler than the stock Intel CPU fan. The PSU is another source of noise, and I upgraded my 300W Antec to a 370W TTGI-350SS for around $40. TTGI isn't as well known a name as Enermax or Zalman, but I've found my PSU to be just as noiselss as advertised.
I don't want to discourage you, but I don't really consider heatsync/case fan combos viable for a silent PC with an Athlon XP or P4. It might be expensive to experiment, but then again having such a quiet machine is worth it. I take great pleasure in surprising my friends by turning my monitor on the show them that my computer is already on.
The future isn't what it used to be.
I have seen that some Compaq machines use this method. I haven't seen them have any problems with heat, or noise. Your idea has been thought of and put into effect long before you thought of it. :) Good going. (This happens to me a lot!)
> A high speed x86 cpu wastes 90% of its power on operations who's result is thrown away.
This figure sounds pretty fishy. I know there's a lot of speculative execution going on in a superscalar chip, but 90% sounds way too high. Where'd you get this?
A couple years ago I bought an ATX case that came with a fan like you're talking about. I can't remember if the case is an Antec case or if I'm just imagining things. The fan's duct hooks up to the rectengular grill on the back panel of the case. The duct leads to a larger 5v fan that sucks air in from the rear and pushes it right around the processor. The box right now has my K6-2 in there and it running so I'm not terribly interested in shutting it off.
The K6 has a fan and a small heat sink that I've never turned off with the duct on so I don't know how well it would cool the heat sink. If you figure out it moves about the same volume of air at the same speed as the smaller sink fan you might want to give the idea a shot or see if you can find a fan like mine. One problem I envision is the slower fan not getting enough air over hot spots on the heat sink to keep it cool enough when the processor is running full tilt doing something. I'm also not to sure processors like the P4 or AthlonXP (the Palomino core at least) without a fan right on the heat sink is a good idea. The heat sink fans speed up as the heat increases where a regular 5v case fan is not going to.
I'm a loner Dottie, a Rebel.
My first computer was a Gateway Destination PII 233
What made the enclosure unique was that it was a horizontal case with no CPU fan. Instead of having a fan on the processor, they used a Huge-Ass heat sink and a shroud that would redirect air from the Power Supply fan. When I say huge, I mean this heatsink was about 5 x 3 x 3 inches in size! When I switched that motherboard & Processor to a vertical case, I could not reuse the fan shroud/ducting so for a while I actually ran it without any fan! Yeah, it was probably risky, but apparently there was enough natural convection heat transfer over the huge heatsink to avoid problems.
It's a shame that nobody makes a power supply with special air ducting for processors and heatsinks. It would be sweet if somebody did that and added a tempature based fan controller.
Actually, Windows 9x/ME does not use HALTs. It literally NOOPs but leaves the processor on. This is why 'software cooling solutions' exist - they simply correct this flaw, running just one notch above the idle task and issuing nonstop HALTs.
Windows NT/2k/XP is, thankfully, not so stupid.
Many high-performance heatsinks have incredibly powerful, incredibly noisy 60mm high-RPM fans. The best and the loudest are made by Delta. 54 dBa is not unheard of for their top-of-the-line models. That's way above anything I can stand, however.
Ducts and passive cooling are options, but they are not exactly optimal. A better solution is to use a larger-but-slower fan with equivalent airflow. To push the exact same airflow as a 60mm fan, an 80mm fan can spin *much* slower, thus producing lower noise. And with more space to work with, the fanblade tips can be shielded a little better on the outside rim, lowering noise dramatically. (It's the tips that make the most noise.)
But please don't use a fan adaptor on a 60mm heatsink. You need something designed to accomodate 80mm fans. For AMD socket chips, the Alpha PAL8492 is *wonderful*. Put some Arctic Silver and a lowspeed fan on that baby, and get better cooling than almost any noisemaker I've ever come across. I'm sure there are similar alternatives for Intel CPUs.
But there are some heat sinks that do use the case fan. They are large fan-like things, and if you like, you can get air ducts with them.
There's no problem with reducing or re-arranging the cooling for a CPU. It is not rocket science.
Perhaps it is irrelvent these days, but I have a 133MHz Pentium box that just does web-browsing, e-mail, and seti@home. A couple of years ago, the bearings on its fan got noisy. So I dug around in The Drawer for a couple of minutes looking for socket 7 heatsinks. It's been running with the largest chunk of black aluminum I had handy, without a fan, for years. It is/was also the most stable Win98 machine I've ever seen, going for months at a time between reboots. It's doing just as well now with Win2k.
I'll never have to bother with that CPU cooler again. The only critical moving parts now are the (solitary) PSU fan and hard disk.
Compared to the roar of the ethernet switch, firewall box, and gaming rig across the room it's essentially silent.
That all said, some people here seem to think there's some magic to having a high-speed whiney piss-ant fan directly on the heatsink.
Sure, a modern CPU is likely to be less-than-tolerant of running without active cooling. They make more heat, and have less surface area to spread and conduct that heat away from the core.
But, the heatsink doesn't care. It just wants air passing over/through it, and nearly all of them want that airflow coming from the top.
All fans in this context have very plain, easy-to-understand airflow ratings, expressed in Cubic Feet per Minute. If your current fan moves say, 40CFM, and keeps things sufficiently cool, a larger, quieter replacement and duct which is also capable of moving 40CFM will work just as well.
Yes, the duct impede airflow somewhat. Your CPU may run a degree or two warmer because of it. This, for all purposes, does not matter.
Just make sure it's coupled reasonably well to the heatsink, as it does take a very small amount of pressure to force cool air down between heatsink fins. You'll gain an ounce of efficiency in this way over simply directing air toward the heatsink.
FWIW, Alpha heatsinks are widely available without fans at all. They tend to have high-end qualities and price, but it's probably a drop in the bucket compared to the rest of this project.
Materials for the duct can be almost anything. I'd be inclined, personally, to use 1/16" ABS plastic, but only because I've got a bunch of it sitting around and it's a nice shade of purple.
If you feel like over-engineering things, try to radius (curve) the bends in the duct and promote laminar flow. Or just do a hard 90 degree bend -- at these velocities, it really doesn't much matter.
Other construction materials might include pipe. The plumbing section of a good hardware store will have for sale a plethora of nice white PVC fittings, which have the added benefit of promoting laminar airflow out-of-the-box. It shouldn't be too difficult to come up with some combination which can be bolted to the fan, have a nice, smooth reduction in size, a bend, and then fasten securely to the heatsink.
Or, for a rustic look, one could enlist the services of a someone who makes ducts for a living. A tin smith should be able to bend and cut something like this together in a few moments time, and he'll likely do it cheaply just because it's an interesting project that does not involve a furnace (Athlon XP notwithstanding). Having flanges built-in to drive screws through, into the fan and heatsink, would be secure and trivial to implement. (If I didn't have the ABS handy, I'd investigate this route first.)
Good luck!
Kid-proof tablet..
The clock is not gated and uses 80% of the CPU power. Taking a superscalar chip to use about half of its executed results it produces and stages left empty.
Also the use of NOPs in wait loops, Super pipelines systems having a monster mispredicted branch penalty, Speculatively executing both logical and arithmetic operations and muxing the result.
These are all standard practises that we will have to get rid of before we can make very fast chips without huge heat sinking systems.
Mouse powered Chips, Open source Processors and Lego
HP has done this for ages as well. I have a few 3 year old Kayak XA and XU workstations (P2/300 to 500) that have exactly this. When you take the case apart, you would notice a molded plastic duct that runs from two fans through the processor's heat sink, and back out through the back of the case.
Making any changes to the hardware required taking all of that crap out, but it did do an excellent job at keeping the components cool.
You can accomplish anything you set your mind to. The impossible just takes a little longer.
Even when the compressor is not running, those pipes are still cold due to thermal inertia. As soon as they get warm again, the compressor turns on again.
So the air coming out will not change too much in temperature - The compressor is simply cycled to maintain coil temperature at just above freezing.
retrorocket.o not found, launch anyway?
It's simply being moved.
Standard fan designs are VERY inefficient at even a small backpressure. Which is why you can't make a "poor man's supercharger" for cars using regular 'ole electric fans.
I remember reading the specs on a 48V 500CFM (approx.) fan - Flow dropped to 0 at less than 0.3-0.4 psi of backpressure.
retrorocket.o not found, launch anyway?
Results: I get an extra 5 degrees Celsius trimmed off of my CPU heat, and I get to run my HSFs in "quiet mode." Suddenly, my poor XPs can run a bit faster, and life is good.
Not to mention the benefit of funnelling the water which is dripping from the sides of the case down into a cup. :-)
Seriously though I hope you are either in a dry climate or you insulate the case from the outside world lest your poor computer drown.
Brand name computer maker (Dell, IBM, Compaq, etc.) do that on many model of business class PC.
You may not know, but there is a whole culture that had developped around cooling and case modification. People do air duct all the time out of cardboard, soft metal, acrylic, etc. There are many other option : mounting a larger fan right on your HS with an adapter, throttling your fan down (7volting, rheostat, voltage regulator [my favorite], PWM), using a quieter fan (Panaflo L1A are popular), etc. Check out the Case and Cooling Fetish forum of Arstechnica. 7 volts is another site I like very much.
:wq
I'm currently planing of moving my computer to the next room, shut the door, and have a quiet PC.
Idea was basicly to use a 5 m monitor cable extender and a usb hub for keyboard and mouse. Sound is no problem. Anyone has experience with such a setup ? is it a good idea ?
My mom has a first Gen P3 Gateway. Its cooled like that. Sadly, every time I take it apart, the duct falls out, and I have to tape it back in. Just make sure to use one heck of a heatsink, and maybe stack 2 fans??
The case my homebrew machine in is has a duct to bring air from the back to the top of the CPU's heatsink. My old athlon system ran cooler with the duct than with a heatsink fan. I ran it for several days each way, and using lmsensors kept a log of the temperature. Not only did the ducting make the case quieter because of one less fan, but it also kept the average and highest temperatures down by several degrees celsius.
Since then I have put a dual athlon board in that case, so the ducting had to go, because it only would have cooled one cpu, and even then it didn't clear the big heatsinks that came with the my new athlons. I found that cooling has been the biggest issue effecting stability in the dual athlon. In the machines original configuration it would lock up under high load, so I rearranged some stuff to bring the max temperature down to about 52C and 56C for each processor, and it has run at full load for weeks at a time with no problem.
I tried adding a front fan to bring air in, and that actually increased the average and max temperature in the case. I am not sure if that was due to increased turbulence or blowing hot air from the drives onto the cpus. Either way, it is important to remember that more fans doesn't automatically equal more cooling.
'from the back of the fan directly onto the top of a fan-less heatsink'
The air coming out of your case fan is carrying heat, not a real good idea.
Consider this instead.
A tube with an external intake.
In the section of the tube that passes over the heatsink and cpu, put slots or maybe a mesh section.
Carry the tube on up to the case fan so that air is drawn through the tube, not pushed into it.
The fan is going to have to have a substantial pull to be useful.