There are only 3 things that can affect thermal performance and long term reliability. Thermal compound is a 100% mechanical process so first review Application and Contact, Pressure links below and recheck your set up.
This Thermal Choking is a well understood effect, In engineering a common device used to limit heat flow to sensitive components in an assembly is a “Heat Dam” shown hereas an example in Figure #3. Where the heat flow is thermally constricted by reducing the contact area to material limitations.
Efficient heat transfer relies In part on the amount of contact, the more contact you have the lower the thermal resistance. In Figure 1 the optimal configuration heat is passed fron the CPU to the heat spreader and flows in the "B" then in the "A' direction to the sink
Many manufactured sinks have reduced contact areas, contact can also be influenced also by the variable contours of the IHS. In Figure 2 with the reduced contact configuration the the heat spreading is is minimized in “B” direction, in effect choking the available “pipe” area to flow/transfer heat.
Thermal Choking when it occurs will compress differences between compounds as copper material limits are ecountered downstream of the choking you may see only a degree or perhaps no difference at all between thermal compounds. This Thermal Ckoking effect explains a good portion of reviewers and endusers differences in test results.
Troubleshooting Contact with a Straight Edge.
The easiest and fastest way to check the contact quality of a sink or IHS in a mount is to use a straight edge like a ruler. Lay the ruler accross the piece and try a number of posisitions ,crosswise edge to edge and corner to corner with the piece back-lit .
The test piece below has a contour of a low profile pyramid with a flat top so the straight edge would noticably rock back and forth in either cross directions. This was confirmed with a dial indicator measuring from center to the four sides and found on average a .006 difference from center flat to edge.
Top and Bottom Comparison
Side to Side Comparison
Contact Area Impression
From our straight edge comparison we put together the model In the Figure 4 as shown on the left .
The heat sink base is shaped somewhat like a pyamid with a flat top and then tapering from there to the 4 edges.
So with an on-center mount in position 5 you would end up with an increasing layer of compound thickness 2X to 3X from center to edge creating conditions for thermal choking.
This sink also has a 2 point mounting system so due to the shape of the base in a secure mount it would only be possible to mount full contact in one of three positions 1,5,or 2.
As this particular sink is mounted some care would have to to avoid a missfire on your mount by sequentually tightening the mounting screws to avoid hanging up on position 1 or2 resulting in an unbalanced mount.
The Contact Area photo is a fuji prescale impression made with a glass slide and pony clamp to illustrate the flat area on-center. The two mounting points are top and bottom for reference.
As a practical matter most dedicated hobbyiests would bypass the analysis and just note poor contact from the straight edge test and either live with it or take the time to lap the sink flat.
The detail was added here to highlight the potential of a improperly load balanced mount with irregular sink contact.
Good pressure is critical for optimum performance, pressure and contact are inextricably linked as pressure improves contact hence thermal performance. In the curing process in which the compound is compressed the particles line up and resolve down to a Bond Line Thickness (BLT) that is equal to the average particle size of the compound's particular mix of particles.
The BLT thickness in part is determined by the amount of liquid combined with the particles to give it a particular viscosity which can be noted as it's resistance to compression. Pressure plus viscosity determines the BLT which determines thermal performance.
Every thermal compound has a pressure/performance curve as the IC Diamond one below.
A liquid metal for example being of low viscosity and containing no particles will hit it's optimum BLT at extremly low pressure, perhaps as little as 10 psi. However performance would not improve much beyond that point and would be reflected as a straight line from 10 psi to 70psi.
A stock compound might hit it's 90% performance @ 40psi then with little performance increase beyond that point. At the lighter pressure loads performance comparisons tend to compress temperatures to within a degree or two to those with higher viscosity like IC Diamond where performace advantages are realized at higher pressure levels. So for example IC Diamond and stock are compared at 45lbs and you might see perhaps 2-3 C improvement with IC Diamond. With higher pressure both compounds will improve but IC Diamond will improve more as it approaches it's optimal BLT to nearly a 6C advantage in performance for IC Diamond.
Note the graphic below - Analyzing pressure and contact is fairly straight forward by dissassembling the mount you can observe the paste pattern of contact and pressure which closly mirrors the Fuji Pre scale impression.
The area of high pressure in the center and edges was high enough to resolve the compound down to where it is pretty close to it's optimum BLT. This is approximately 1/2 the thickness of a sheet of paper and can be noted as a thin glaze and would be a desirable outcome if the entire IHS this "glazed" appearance.
Overly thick compound are indicators of low pressure and low contact with high thermal resitance. Thin compound are areas of high pressure high contact with low thermal resistance. It's that simple
Pressure and Contact Comparison
Some problems are an easy fix like application of compound or an unbalanced mount. Other issues such as sink and mounting hardware are a little more complex but can be resolved with a little effort.
We highly recommend user group forums as we have noted many creative simple solutions generated there for solving specific hardware issues. Due to the shear number of different products it would be to cumbersome to deal with in this format and exceeds the scope of this effort.
Miscellaneous Notes and Tips
1. With screw systems when the compound is cool it has a higher viscosity and occasionally on installation it can limit how far the screws can be torqued down.
On system start up the compound's viscosity wll thin with the addition of heat and start to flow and with the screws set into a fixed position will leave a slight gap. After burn in or cure for a couple of hours re-torque the screws.
What to Expect in Practice
On one of our Survey's a cluster of results was observed so we sorted the data to highlight them. In the OCUK Data Sort Chart below The hilighted bars are water blocks and the ones in blue are air cooled sinks.
The 10 WB on the right range from +2.25 to -1.8 C are generally under performers as far as thermal compound testing goes. The 4 on the left were a couple of blocks of similar branding perhaps indicating better mounting schemes or flatter bases.
The group of Photos underneath the chart are contact tests were done by Gilgamesh of overclockerstech.com on 5 water blocks purchased by IC for the tests. This is fairly typical of you you can expect in practice for water blocks.
All were compass tested (N-S, E-W) in two directions on the same CPU, darker the color the higher the pressure. The snapshot pictures display a range of contact, the first being the least and the last being most contact. The range of temps spanned several degrees C comparing the best result of 2 tests on each with the last in the series having the best performance with coincidentaly the best contact and pressure print.
Water Block Contact and Pressure Impressions
How flat a sink or IHS is can affect thermal results as any high spots will add to the thickness of the adjacent layer. The high spot layers will be thinner and closer to the optimized BLT due to increased pressure loading on the reduced area. That being said the primary mission of applying thermal compound is to fill those surface irregularities and is noted here for those seeking the full optimization experience.
Playstation contact/pressure images taken from the main page reliabiliy example.
Playstation 2 contact and pressure images could only be acquired after shimming the mounting screws to increase pressure as it was under the 28 psi lower limit of the film and even at that it was still a marginal pressure. This does not reflect a defect in design as surface area is increased over what you would see on a typical CPU perhaps 2X on average psi would drop in half.
On a contact basis it is the best we have have ever seen on a manufactured product, after running a dial indicator across both dies and sinks all were found to be within a .001 of an inch. It's only a one off sample but looked pretty good from our perspective.
No sink thermal analysis was done here but as in any thermal stress situation a quality thermal compound can obviate many thermal problems.