Experimentally determining the coldest I can get with air

[Blain]

Experimentally determining the coldest I can get with air

Standard water or air cooling will never be any lower than ambient temperature.

 

[Syzygies]

Some time ago I used a Scythe Kama Bay in three 5.25" bays of my Antec P182 case, and rigged a wind tunnel back to my push-pull True. I saw a nearly 5 C difference overclocking at full load. This made me think.

Fast forward through Megahelams and Noctua NH-D14 coolers, I build in Mountain Mods cases, eight or nine case fans each. The cases cost several hundred dollars more than many people find reasonable, but they're like renting an unfurnished apartment; every other case on the market is a furnished apartment, and it's junk furniture that's usually in my way. My newest build is in their Plateau: A horizontal motherboard tray that better handles the 1 Kg weight of my cooler, and slides out for easy work, like repairing an older American car. That's the real appeal of their cases: How do you really feel about going back into the case, once it's crowded with gear?

Getting the inside of a closed case down to room temperature is just a matter of scale. It takes an absurd number of fans, and most people won't bother or want to pay for the fans. But this approach is easier than building wind tunnels in already crowded cases.

If you really want to see the limits of air, test your system open bench with a commercial-grade house fan blowing on it. If you can't replicate these results inside your case, then your case is holding you back.

 

[Emulex]

no actually you think of your case as the vessel and the ducting you make is the fluid dynamics work (or use smoke - not the ganja variety it is too sticky) and model airflow your way.

you know its a fun subject to learn and quite honestly there is a reason every workstation/server uses passive cooling. heck even my mac pro has no fans on anything. because passive cooling with proper ducting using simulation and sensors is far superior (esp with single failure points of a fan).

 

[JamesWatt1]

7. Controlling for ambient temperature
One problem I ran into was controlling for the ambient temperature. A suggestion I was given was to report CPU core temperature – ambient temperature instead of just CPU core temperature. For example, if CPU was 70 degrees C and ambient was 20 degrees C, I would report 50 degrees C. One concern I had was that I’m not sure that CPU temperature varies linearly with ambient temperature. Can anyone shed light on this?

Here are some highlights from previous experiments measured this new way.

49.2 C (67 C – 17.8 C): first duct attempt
47.5 C (65 C – 17.5 C): case front and sides removed
46.4 C (62 C – 15.6 C): Rosewill RFA-120-BL heat-sink rear exhaust, Coolermaster A12025 case rear exhaust.
45.6 C (67 C – 21.4 C): Coolermaster FA12025 heat-sink front intake, Coolermaster A12025 case rear exhaust.

8. Removing the rear fan grill
One suggestion I got was removing the rear fan grill (http://postimage.org/image/br30n09w). I removed it and replaced it with another fan guard that would let me keep my fingers but have less of an impact on air flow (http://postimage.org/image/br6bpp9g). I’ll take the 1 degree C cooler thank you very much.

45.6 C (67 C – 21.4 C): rear fan only with old grill
44.4 C (66 C – 21.6 C): rear fan only with new grill

46.3 C (65 C – 18.7 C): front and rear fan with old grill
45.4 C (67 C – 21.6 C): front and rear fan with new grill

9. Increasing negative pressure
Deimos3428 had some suggestions for increasing negative pressure. I tried them with the Coolermaster FA12025 front intake on the heat-sink and a Coolermaster A12025 rear exhaust on the case.

44.4 C (66 C – 21.6 C): Control
43.7 C (64 C – 20.3 C): Covering the top exhaust hole with two pieces of paper & Coolermaster A12025 case side exhaust
42.4 C (63 C – 20.6 C): Covering the top exhaust hole with two pieces of paper

These results are puzzling. Deimos, as you suggested, covering the top cooled the CPU down (two degrees is a nice result). However, covering all the non-fan vents as I did in my varying case enclosure experiment proved disastrous (54 C = 72 C – 18C was the hottest above ambient I got in all my testing). Your suggestion of adding a side fan to exhaust seemed natural given the success of having a rear exhaust fan only. But it happened to reduce cooling performance. The relationship between pressure and temperature in PC cooling is complicated. Any ideas for teasing it out?

 

[JamesWatt1]

10. Improving the duct
There were a couple of issues with the last version of the duct.

The rear exhaust fan was blocked. This was because the duct hose wasn’t flexible enough to get from the back of the heat-sink all the way to the exhaust hole. I replaced the duct with more flexible duct.

Not enough air was being drawn in. There was no front intake fan. I added one. The front of the duct was blocked by the 5 ¼” bay grill. I removed it (http://postimage.org/image/brhwh4pw/).

With those fixes, here’s what I got. Both measurements were taken with the front and rear grills removed.

44.9 C (66 C-21.1 C): Silverstone FM121 front intake & rear exhaust
38.2 C (59 C-20.8 C): Duct with Silverstone FM121 front intake & rear exhaust (http://postimage.org/image/br9mse90/)

11. Cooling the northbridge
One concern that was raised with the duct was that the north-bridge would run too hot. To remedy the issue, I tried putting a Rosewill RFA-120-BL on top of the case. Both trials increased the cpu – ambient so much that I decided against the top fan. All trials were done with the duct with Silverstone FM121 front intake & rear exhaust.

38.2 C (59 C-20.8 C): Control
41.2 C (62 C -21.8 C): Top fan set to exhaust
44.3 C (65 C -20.7 C): Top fan set to intake

Future actions
· I am going to pick up a vacuum and a pressure gauge to try to see what the deal is with pressure and temperature.
· I am going to buy some fans with the same CFM, but different static pressures and see how they perform.
· I will try adding some more space between the heat-sink and fan.
· I will try cooling the air before it enters the duct.

 

[deimos3428]

9. Increasing negative pressure
Deimos3428 had some suggestions for increasing negative pressure. I tried them with the Coolermaster FA12025 front intake on the heat-sink and a Coolermaster A12025 rear exhaust on the case.

44.4 C (66 C – 21.6 C): Control
43.7 C (64 C – 20.3 C): Covering the top exhaust hole with two pieces of paper & Coolermaster A12025 case side exhaust
42.4 C (63 C – 20.6 C): Covering the top exhaust hole with two pieces of paper

These results are puzzling. Deimos, as you suggested, covering the top cooled the CPU down (two degrees is a nice result). However, covering all the non-fan vents as I did in my varying case enclosure experiment proved disastrous (54 C = 72 C – 18C was the hottest above ambient I got in all my testing). Your suggestion of adding a side fan to exhaust seemed natural given the success of having a rear exhaust fan only. But it happened to reduce cooling performance. The relationship between pressure and temperature in PC cooling is complicated. Any ideas for teasing it out?

Makes sense to me.

Your exhaust fan forces air out, creating negative pressure in the case. Air rushes in to equalize the pressure. Too many holes and the incoming air doesn't move quickly; too few holes and air barely moves at all. You're trying to find the right intake surface area to maximize the volume of air flowing through your case.

As for the side exhaust, that was for spot cooling. In your case, it looks like it's disturbing the main airflow too much; probably not worth doing. That said, you might consider experimenting with the speed of the two fans and see if you can find a combination that works.