|
Socket-A Heat Sink Comparison
Date Reviewed: July 3, 2001
Reviewed By: Jason Rabel
Companies: GlobalWinUSA,
VantecUSA, Thermosonic,
Thermaltake, PowerCooler
Thermsonic Thermoengine:
The Thermoengine is a radically different approach to making a
heat sink, it has a round core with fins on all four sides (as seen in the
pictures below), also the TIM (Thermal Interface Material) is very high quality,
to achieve the same results with regular thermal compound I had to lap the base
first since it had a grooved texture. The picture below shows the stock fan,
however I tested it with a Delta fan. There are a couple models of the
Thermoengine out on the market today, a hollow core and a solid core. Both have
shown identical results, but just for the record mine is one of the hollow core
earlier models.

(click on image for close-up)
Power Cooler PCH137:
The Power Cooler PCH137 is an all copper / thin fin design. The
clip is a little different than the traditional style, it doesn't really bend.
Instead there are two bars that go through the heat sink and that is what the
clip rests upon. Still, the design is solid and it's not that bad looking of a
heat sink. The PCH137 is also sold as the OCZ Gladiator which they put a 37CFM
fan on it, however to keep the testing consistent I used a standard 38CFM
Delta fan.

(click on image for close-up)
Testing:
Testing was done using a KT7A-RAID with a 1GHz Athlon (Courtesy of PCNut)
overclocked to 1.13GHz (8.5 x 133). I used a synthetic ester based thermal
compound, which I have found works just as good as AS but is much easier to
clean up.
Taking the temperature is a little different
than what we have done in the past. Instead of using the built in thermistor
that touches the bottom side of the CPU, I opted to also use a temp probe that
touches the side of the core on the top side of the CPU. This provides much more
accurate temperature readings since it is at the source of the heat. I am
including the under CPU temp readings also for people to see how they compare,
they are kind of interesting because some heat sinks caused some bizarre
numbers.

To heat the CPU up, I used Prime95 since it puts a consistent load on the
CPU. After several people griping about how Prime95 doesn't put an actual 100%
load on the CPU, I did a little research and emailing and discovered it only
puts about an 85% load on the CPU. So my solution to compensate is simple. First
I used a program called Radiate (Download Radiate.zip from our FTP, the original
file can be found somewhere at Overclockers.com) to calculate out the wattage of
an Athlon at 1.13GHz at 1.85v. This number turned out to be 67 watts. So if I
was to run Prime95 with that setting, we take 85% of that number and we get
almost 57 watts, which is no good. So, to compensate I upped the voltage in the
program till I got something that would be about 67 watts at 85%. This turned
out to be 2.05v, which at 100% would be 80.4 watts, but 85% of that number would
be 68.34 watts. Well, the closest I could get on my modded motherboard was 2.04v
so I figured I was pretty much right on the money. Yes I know this is a crude
way to do it, but it is much better than just ignoring the fact that Prime95
doesn't heat up the CPU all the way.
The heat sinks were tested twice to make sure the numbers were consistent. So
now that we got a smokin CPU, how do they measure up?
| Ambient Temperature: |
| 82.1F +/- .5F |
27.8C +/- .5C |
|
Heat Sink Model |
CPU Idle
Top |
CPU Load
Top |
CPU
Load Bottom |
| Vantec CCK-6035D |
90.9F |
32.7C |
120.4F |
49.1C |
120.2F |
49C |
| Thermaltake Dragon Orb 3 |
90.9F |
32.7C |
126.5F |
52.5C |
125.6F |
52C |
| GlobalWin CAK38 |
91.4F |
33.0C |
121.1F |
49.5C |
120.2F |
49C |
| GlobalWin WBK38 |
91.8F |
33.2C |
122.2F |
50.1C |
125.6F |
52C |
| Thermosonic Thermoengine* |
90.32 |
32.4C |
122.2F |
50.1C |
118.4F |
48C |
| Power Cooler PCH137* |
90.9 |
32.7 |
122.2F |
50.1C |
131.0F |
55C |
| *Delta Fan Used |
Conclusions:
At first the temperatures may seem a little high. First note
that the ambient temperature was rather warm, due to having so much computer
hardware running combined with the Texas heat and wimpy A/C. Second, the tests
were conducted with a true target of about 68 watts, which some other sites
don't quite get up to that much heat. It was interesting to see how the bottom
thermistor varied from the top thermistor in some cases. I can't explain the
cause for this, however the results were reproduced several times just to make
sure it wasn't a fluke / glitch. I trust that the top thermistor temperature is
much more accurate than the bottom, but it is nice to know that in most cases
the bottom thermistor is fairly accurate.
I'm very happy to see that the new Vantec CCK-6030D has come out
on top. I know Vantec has been working real hard lately to produce some top
quality heat sinks. However even the GlobalWin's, Thermoengine, and Power Cooler
were within a degree of the Vantec, so I would have to call them all a tie. The
new Thermaltake Dragon Orb 3 has performed admirably too, quite a leap from the
previous orb's. The Dragon Orb just looks really cool & colorful compare to
the other heat sinks, and I noticed the fan wasn't as loud as the Delta that was
used on all the others.
The new Copper Vantec heatsinks should be available soon at your
favorite stores!
Back To The Main Page
|