Ducted Zalman 7000CU w/Countercurrent Flow Cooling

Do-It-Yourself Systems
Viewing page 4 of 4 pages. Previous 1 2 3 4

MEASUREMENTS

Here are some temperatures (degree Centigrade over ambient) at full load while running Folding@Home:

Fans voltage
6/5 V
12V
PSU air intake
6°C
-
PSU air exhaust*
9°C
8°C
CPU air exhaust
16°C
11°C
Graphics card heat sink
16°C
14°C
CPU temp (according to MBM)
26°C
20°C

The temps shown are °C over ambient temperature.
* These readings are suspect as it is very difficult to identify the hottest point in the exhaust airflow. I believe the temperatures should be higher.

ADDITIONAL NOISE / THERMAL ANALYSIS
notes by Mike Chin, Editor

After the above info data was presented, I suggested a way for Han to "measure" the noise of the PC. I measured several Panaflo 80L1A fans in my lab, at several different voltages, and came up with this averaged noise data:

Panaflo 80L1A
(avg. of 4 fans @ 1M, 15 dBA ambient)
Voltage
SPL (at 1 meter)
12V
24 dBA
10.8V
22 dBA
8.8V
19 dBA
6.0V
at or below ambient; nomally inaudible

Since Han had several Panaflo 80L1A fans, I suggested he compare the sound from his PC against one of these fans. My thinking was that his samples of these fans are very close to mine, and the Panaflo is likely the most widely known and used fan among SPCR readers. So using it as a reference point would help readers get a very good idea of how much noise my PC makes. Here are Han's findings, again using folding@ home as the load:

Settings
Fan #
Fan
CPU
CPU Exhaust
PSU Exhaust
GPU
As loud as Panaflo
Normal
2
6V
26°C
17°C
9°C
16°C
@10V
Max cool
2
10.7V
19°C
11°C
8°C
14°C
@12V
Match 2-fan voltage
1
6V
32°C
20°C
17°C
16°C
@10V
Match 2-fan normal perf.
1
8.8V
26°C
17°C
10°C
16°C
@10V
Max cool
1
10.8V
23°C
14C
10°C
17°C
@11.5V?

NOTE: All temperatures are °C over ambient.
Ambient temperature was 22°C during the 2-fan tests and 23°C during the 1 fan testing.

Han's findings suggest the following:

1. At his normal configuration with two CPU fans in push-pull at 6V, the noise level is around 21 dBA/1m. This is very quiet, likely close to inaudible under a desk in a normal noise ambient environment. Han says, however, that the PC's noise signature has more low frequency sound than the Panaflo fan by itself, which is only natural given the three hard drives.

2. With the two fans set for maximum speed (via the Zalman Fanmate) for possible cooling during hot summer days, CPU cooling is improved by 5°C and the noise level goes up to around 24 dBA/1m. This is very good cooling and still pretty quiet performance. I suspect the duct itself acts to absorb some of the fan noise.

3. With the outside fan removed (but the duct still in place) and just the single internal Panaflo at 6V, the noise was not changed from 2 fans at 6V. This is probably because the three hard drives are loud enough to make ~21 dBA/1m the default noise floor of the system

4. For the single fan to match the performance of the 2-fans at 6V, it had to be driven to 8.8V in order to match the thermal performance of his normal configuration. But the noise level is still at just 21 dBA/1m due to the contribution of the three hard drives.

5. With the single internal Panaflo CPU fan at the max 10.8V, CPU cooling improved by only 3°C over the 8.8V level. The noise level was judged to be in between the Panaflo fan at 10.8V and at 12V; hence the 11.5V guesstimate. This puts it at ~23 dBA.

 

THERMAL IMPACT OF CPU VOLTAGE

Undervolting the CPU is a topic that has been much discussed here, in articles as well as in the forums. My system has been undervolted almost right from the start, but I was curious to see what the differences were, so I took some readings after setting the Vcore in the BIOS back up to the default.

I was stupified at the results. CPU-diode temp went up 9°, and CPU-exhaust temp went up 6.5° The effects of undervolting are much underestimated by most members of SPCR. Undervolting must be the most effective way to reduce the need for "lots of air", at least for this CPU. [Editor's Note: The power goes up/down by the factor of (new voltage ÷ old voltage) squared --in this case, (1.3 ÷ 1.475) squared = 0.777. In other words, the CPU power or heat drops 22.3%.]

Vcore & Temperatures
Vcore
1.3V
1.475V (default)
Ambient
22.5°C
22.5°C
CPU
49°C
58°C
CPU exhaust
41°C
47.5°C
PSU exhaust
33°C
33°C
GFX heatsink
41°C
40°C
Winbond chip
41°C
41°C

CONCLUSIONS

I believe this duct system allows the CPU to be cooled with the least amount air, the benefits of which are:

* Less problems getting enough air into the case.
* Slow revving fans means low noise.
* When airflow through the case is lower, it is reasonable to expect less dust to enter the case.
* The interior and the PSU in particular run cooler because most of the CPU heat is evacuated directly from the case.

Although my old rig was regarded as ‘pretty quiet’, this one is even more so, even with three hard disks. At night, I can’t hear it at seven feet, while I can hear it at five feet. At night, sitting at the console I can hear it, but only when I pay attention. What I hear are the three suspended Samsung disks. There is no whine, only a faint presence of mid-frequency noise.

Have fun if you decide to try a similar counter-current flow cooling project!

NOTES on Project Materials

The foam material used for the duct is called Plastazote, and it is amazingly versatile. It is a closed-cell, cross-linked polyethylene. My employer uses it as an inlay in shoes, and for lining splints, but it is also used for many other things. It's a firm foam, a little bit lighter and stiffer than the thick foam mouse-pads we all know.
You can draw your design on it, cut it with a knife, and use contact cement to glue it all together.

The foam block was packaging material. Easy to cut, tough to tear. I would guess it is made of recycled coffee-cups. (Editor's Note: Seems like some kind of closed cell foam.)

The elbow tubing used for the chimney comes from Noiseblocker, and is specifically made for this purpose. It is plastic with walls about 1/16" thick, and around US$2.50

* * *

Discuss this article in the SPCR Forums.



Previous 1 2 3 4

Do-It-Yourself Systems - Article Index
Help support this site, buy from one of our affiliate retailers!
Search: