Scythe Kamaboko CPU HSF

Cooling
Viewing page 3 of 3 pages. Previous 1 2 3

INSTALLATION

The installation hardware package is complete, allowing for use with socket 478 P4s, socket 775 and all K8 variants. The clips on the HS engage the standard socket 478 mounting bracket. For both 775 and K8, additional hardware allows the same clips to be used. It does require motherboard removal with 775 and K8.

Installation was done on our usual socket 478 motherboard. Figuring out how to mount the Kamaboko was a simple matter. The downward blowing fan meant that it was good enough just to slip the heatsink into the stock retention module, the orientation of the heatsink doesn't matter because it is symmetrical.

Actually clipping the HS in place was another matter. The clips require a great deal of pressure. I had to borrow an extra set of hands (Mike's) to hold the heatsink steady while I used both my hands to put muster enough pressure to engage the clips into place. This dgree of pressure seems unnecessary, and places the motherboard under an undue amount of stress.

TESTING

Test Platform

  • Intel P4-2.8A The Thermal Design Power of this P4-2.8 (533 MHz bus) is 68.4 or 69.7W depending on the version. As the CPU is a demo model without normal markings, it's not clear which version it is, so we'll round the number off to ~69W. The Maximum Power, as calculated by CPUHeat & CPUMSR, is 79W.
  • AOpen AX4GE Max motherboard - Intel 845GE Chipset; built-in VGA. The on-die CPU thermal diode monitoring system reads 2°C too high, so all readings are compensated up by this amount.
  • OCZ DDRAM PC-3700, 512 MB
  • Seagate Barracuda IV 40G 1-platter drive (in Smart Drive from Silicon Acoustics)
  • Seasonic Super Tornado 300 (Rev. A1)
  • Arctic Silver Ceramique Thermal Compound
  • Two-level plywood platform with foam damping feet. Motherboard on top; most other components below. Eases heatsink changes and setup.

Measurement & Analysis Tools

  • CPUBurn processor stress software
  • Motherboard Monitor 5 software to show CPU temperature
  • T.H.E. KP-6M reference omni microphone
  • M-Audio Tampa mic preamp
  • M-Audio Firewire 410 external digital sound interface
  • B&K 2203 sound level meter
  • SPCR lab custom-built multi-channel variable DC power supply for fan testing
  • Zalman Fan Mate 2 to control fan voltage on the motherboard
  • Digital display anemometer to measure fan airflow

The ambient temperature during testing was 21°C. Ambient noise level for testing and recording was 16 dBA/1m. Tests were performed with the fan running at 12V, 7V and 5V.

RESULTS

Scythe Kamaboko
Voltage
CFM
RPM
Load Temp
°C Rise
°C/W MP
°C/W TDP
dBA/1m & MP3 files
12
26
2000
47°C
26
0.33
0.38
7
16
1300
55°C
34
0.43
0.49
5
11
n/a
58°C
37
0.47
0.54

Voltage: Fan voltage
CFM: Measured airflow rate of fan on heatsink, in cubic feet per minute
RPM: Fan speed; not measurable at 5V
Load Temp: CPU temperature after 20 minutes running CPUBurn
°C Rise: Temperature rise above ambient at load
°C/W MP / TDP: Temperature rise per Watt, based on 79W Maximum Power or 69W Thermal Design Power rating of CPU
dBA/1m: SPL measured with high accuracy B & K SLM at 1 meter
MP3 file: High resolution recording of fan on HS at 3" distance. Note that it is preferable to save these sound files to your hard drive, and then play them.

12V: Our impression of the fan at stock voltage was fairly good. Noise is dominated by a low frequency motor hum. There is also a small amount of turbulence noise, although this is less of an annoyance than the focused frequency of the motor noise. Overall volume is not loud by normal standards, but it is only borderline quiet by SPCR standards.

Cooling performance is about as can be expected from this design: Adequate, but nothing to brag about. There may not be a lot of headroom for cooling hotter processors, especially if the heatsink is simply dropped into an existing case with no additional attention to airflow.

7V: The drop to 7V contributes to a dramatic drop in noise level. The stock fan at this level could certainly be considered quiet. Turbulence noise has disappeared almost entirely, and the motor noise is reduced to a low growl. Inside a case, this noise would quickly sink into the background, and may well be below ambient noise level if the other components in the case are not also silenced. A very low level buzz is now audible from within a foot or two, but this should be inaudible in a real setup.

The downside is that cooling performance is now marginal. Our test 2.8 GHz Northwood processor is probably about the warmest CPU that could be properly cooled inside a case without an additional source of airflow. We do not believe that a Prescott core P4 could be cooled at this level without triggering throttling under load. On the other hand, not many Athlon 64s draw more power than our test processor, so the Kamaboko could be well suited to cooling an AMD based system.

5V: Reducing the fan to 5V puts the finishing touches on the improvement in noise that was seen at 7V. In most environments, the ambient noise level will be higher than the noise produced by the Kamaboko. A subjective listening test revealed no turbulence noise and only the faintest trace of motor hum in the lowest audible frequency range. The buzz that became audible at 7V is now the main source of noise, but, as mentioned, it is inaudible further than a couple of feet away. This fan would be at home in a truly silent system, and would be below ambient noise under almost all circumstances.

The performance hit to achieve silence with this fan is not terribly large. A 3°C rise from the 7V level is probably acceptable under most circumstances. However, as at 7V, the range of systems that can be cooled at this level is restricted. It would probably be worth running the fan at 5V rather than 7V, since most systems that can be cooled at 7V can also be cooled at 5V. Our guess is that the difference cooling effectiveness represents about a 200 MHz difference ¬ó about one processor model step. The difference in noise between 5V and 7V is quite dramatic, since in many cases it will be the difference between a quiet system and a silent one.

Recordings of Comparable HSF:

MP3: Arctic Cooling Super Silent 4 Ultra TC MP3, 22 dBA/1m

MP3: Panaflo 80L - 7V - 17 dBA/1m -- on most any heatsink

MP3: Coolermaster Hyper 48 - 9V - 21 dBA/1m

HOW TO LISTEN & COMPARE

These recordings were made with a high resolution studio quality digital recording system. The microphone was 3" from the edge of the fan frame at a 45° angle, facing the intake side of the fan to avoid direct wind noise. The ambient noise during all recordings was 18 dBA or lower. It is best to download the sound files to your computer before listening.

To set the volume to a realistic level (similar to the original), try playing this Nexus 92mm case fan @ 5V (17 dBA/1m) recording and set the volume so that it is barely audible. Then don't reset the volume and play the other sound files. Of course, all tone controls and other effects should be turned off or set to neutral. For more details on how to calibrate your sound system playback level to get the most valid listening comparison, please see the yellow text box entitled Listen to the Fans on page four of the article SPCR's Test / Sound Lab: A Short Tour.

CONCLUSIONS

Aside from being marketed as inexpensive, the Kamaboko is also sold as a quiet heatsink. The Kamaboko lives up to its billing. With the help of a simple voltage controller, it is quite suitable for a silent PC.

One aspect of the Kamaboko's design that we did not test was its performance with external airflow. It is possible that the elevated position of the fan allows for a significant improvement in performance if a secondary fan is used to suck airflow through the channel underneath the fan. This could be a case exhaust fan or a power supply with a bottom mounted fan depending on the the orientation of the Kamaboko on your particular motherboard. Keep in mind that this is just conjecture until somebody decides to test it out and post the results on the forums (hint hint).

The disadvantage of the Kamaboko is its cooling performance. It is possible that the Kamaboko could handle a high-heat Prescott processor with the fan at full speed in a case with good airflow; there's no way it could do that with reduced fan speed. The suitability of the Kamaboko is highly dependent on the system in which it is used: It could be wonderfully quiet or horribly inefficient depending on how much heat it is asked to sink.

Realistically, if you want to run the fan quietly, we would not recommend using this HSF with a CPU any hotter than our test platform's P4-2.8 Northwood, which is hardly burning edge by Intel standards. It will do fine with most recent Athlon 64s or Semprons, which run as cool or cooler than the P4-2.8 Northwood. The bottom line is that it is a relatively small, not particularly effective heatsink mated to a very quiet 92mm fan. We expect the Kamaboko Z to be a better performer (but that model is equipped with a different 92mm fan, one with a variable speed control, which could be very different acoustically).

Pros

* Quiet fan can be made silent by undervolting
* Modest price
* Broad CPU / socket compatibility
* Modest weight

Cons

* Modest cooling performance
* Non-Socket 478 systems require elaborate mounting procedure
* Extremely tight mounting clips

Finally, here is a photo of the Kamaboko Z that helps explain the curiously large gap between the fan and the top of the cooling fins:

Much thanks to Scythe USA for the Kamaboko sample. (We would have preferred the Z, though.)

* * *

Comment on this article in our Forums.



Previous 1 2 3

Cooling - Article Index
Help support this site, buy from one of our affiliate retailers!
Search: