|80-way Thermal Interface Material Performance Test|
|Reviews - Featured Reviews: Cooling|
|Written by Olin Coles|
|Sunday, 14 June 2009|
Page 9 of 14
TIM Testing: Best Practice
Since this article has taken almost two years to complete (primarily because of testing and retesting products), there have been a lot of simple lessons learned (the hard way). For example, most enthusiasts don't realize that the thermal paste they use relies heavily on the coolers' surface finish. What's meant by that is a Heat-Pipe Direct Touch (HDT) cooler has a very rough finish with plenty of microscopic (and some clearly obvious) pits and gaps, while other coolers might receive a polished mirror finish. So picturing the differences in your head, do you think that one thermal paste performs exactly the same regardless of surface finish? How about contact pressure? This is where enthusiasts have really missed some critical aspects of thermal interface materials: no one thermal paste will work the same on every surface. You've got to match the TIM that works best with your surface finish!
After months of repetitive testing using seventy-five different Thermal Interface Material products, I've personally discovered some reproducible 'rules' in my research:
Those readers who have absorbed the previous pages of information will completely understand why these 'rules' seem common-sense, but allow me to briefly explain for those who haven't made the connection. Presume that both surfaces are lapped flat and polished smooth, and that contact pressure is optimal. In such a circumstance, using a TIM with low viscosity fills only the microscopic pits, and the pressure forces excess material off of the surface. Presuming that same surface and mounting pressure exist with high-viscosity TIM, the thicker material is going to resist bleeding and create a layer between surfaces. This layer is thermally conductive, but at a far less efficient level than if the surfaces were closer or even making contact.
The opposite is true for rough surfaces, and those with weaker mounting pressure (push-pin clips systems). Presume that processor has an unimproved Integrated Heat-Spreader (IHS) just like it comes from the factory, and the cooler is has a rough finish and/or uses HDT technology. In this situration, the thin low-viscosity thermal material will possibly bleed off and is not going to fill deep pits and crevices. A thicker high-viscosity thermal material will better fill large gaps and deep pits, but will also resist bleeding away from crevices.
How Results Are Reported
From the very beginning of this 80-way Thermal Interface Material review article, the list of TIM products to be tested was so overwhelming (from a test-person's standpoint) that a system had to be used to avoid wasting time. Since each test consumes an average 90-minutes from setup to tear-down, our test days were very long but also very productive. Throughout each test period, critical environmental temperatures were carefully measured and adjusted using constant air circulation inside the room paired with large styrofoam insulating partitions.
Testing a collection of thermal paste products of this magnitude is a daunting task, which is why the first series of test runs was designed to weed-out the non-hackers early and eliminate them from future testing. For this article, Benchmark Reviews has removed the end-result temperatures from our recorded scores, since so many enthusiasts make assumptions based on numbers. Instead, the overall performance determines the "Enthusiast Grade" each thermal paste receives, and the test collection is divided into catagories based on this grade. Why have we done this? The simple answer is because the results of our testing cannot be duplicated on another system elsewhere, making temperature scores irrelevant when grade ranking accomplishes the same goal. So here's how the products are sorted and assigned an Enthusiast Grade:
It's important to remember that no thermal paste produces exactly the same thermal conductivity on one test platform as it does on another. CPU and cooler metals and surface finish are determining factors, followed by compression strength. While there will always be a small degree of performance variance betwen TIM products used on the same hardware for a different system, the difference grows much more pronounced when the material is applied to differing surface finishes and contact pressures.
Benchmark Reviews would rate the contact pressure between the Intel Q9550 quad-core processor's integrated heat spreader (IHS) and the Intel push-pin thermal management system to have been moderately firm. The surface condition of the processor's nickel-plated copper IHS was smooth with light texture visible, while the CPU cooler featured a polished copper surface. The use of a standard-performance cooler with mounting clip system and unpolished IHS finish was intentional, as improvements would narrow the range of thermal performance. By testing with a mild yet fundamentally sound configuration, recorded temperatures were not as closely staged between products.