Page One
Like most of you probably reading this, I've been around the heatsink block more than a few times and remember the 'milestones' in recent years regarding thermal interface materials development. We've all used the "white stuff", and I remember taking several trips to the local RadioShack for more than a couple of the little tubes of the white stuff.
Within the last 5 years, we've seen a tremendous amount of development and refinement in heatsink designs to the point that any improvements are more and more difficult to acheive. The difference in temperatures acheived be today's heatsinks are also *much* closer together then they were even a year ago, with differences usually 7c or less between a $50 heatsink and a $15 one. Waterblock development has also matured and grown to almost mainstream acceptance with more than a handful of companies providing excellent products compared to what was available not that long ago.
It's no wonder then that thermal interface materials has also been developed and refined to the point that it has. Not only have CPU dies gotten much smaller, but heat output has also come pretty close to being tripled in some instances making a heatsink's job that much harder. Anything that helps the heatsink is a welcome addition.
Arctic Silver created quite a stir a while back with their alternative to the traditional 'white stuff', and honestly it was better. So much so, that it became the de-facto standard. Reviews started hitting the web showing some pretty unbelievable improvements in temps-I've seen reviews claim almost a 10c difference. Unscientific testing aside (if you were to take all reviews at face value), it's very, very difficult for me to believe that a change in the thermal interface material can exponentially increase a heatsinks effeciency - or conversely, a poor application can exponentially decrease it's effeciency.
Admittedly, I have always applied thermal paste wrong--and still do. After a few discussions with Bil Adams and watching how he applies paste, I felt pretty ignorant in hindsight at my previous application methods. As has been explained in countless tech articles over the years, the purpose of thermal paste is to fill in the gaps and voids to provide a little more surface contact between the silicon and the heatsink. All things being equal, even the generic white stuff has a lower thermal resistance than air does.
For a long time I applied paste to the bottom of the heatsink with some masking tape and a razor blade; it gave me a uniform thickness of paste that's repeatable for testing. But what alot of us forget is the scale of the gaps that we are trying to fill between the heatsink and the chip; Even on the worst finished machined extrusion you've ever seen, the gap is much smaller then the thickness that most people apply paste (myself included). I can remember distinctly early heatsink articles focusing on the "paste imprint" on the bottom of heatsinks showing a big schplodge of paste where the CPU contacted the heatsink.
|
