Silverstone Nitrogon NT06 Review

by sswilson     |     October 2, 2007

Effects of Case Orientation on Cooling Performance

To say that initial recorded results on this unit were puzzling would be an understatement. The results weren't “bad”, they just weren't living up to expectations. Numerous attempts were made at re-seating, and various application methods of applying the thermal compound were utilized in an attempt to explain the shortfall, yet the results continued to remain fairly consistent.

An examination of our testing methodology, and test bed suggested that the only obvious difference between the NT06 and the control unit was the orientation of the heat pipes due to its low profile design. E-mail correspondence with our Silverstone tech representative confirmed that their results showed an improvement with the heat pipes in a vertical position. Armed with this new information, a separate test was devised in an attempt to verify this theory.

We cannot change the standard testing methodology to improve results of a single review unit, to do so would defeat the whole purpose of a stable testbed and invalidate all other testing done on that testbed. On the other hand, in fairness to both the manufacturer, and those reading our reviews, it's important for us to provide all the relevant data we can get our hands on.
Testing Methodology
All hardware remains the same as used in the tests above (overclocked), except for installation into a fully open tower case vice the Highspeed PC Tech Station.
1 hour moderate load (Windows SMP folding) was applied with the tower case in the upright (vertical) position to allow optimal cooling performance, after which a speedfan log file was initiated to record CPU temperatures in 3 second intervals. 10 minutes after the log file was initiated, the tower case was placed on it's side (horizontal) to replicate the heat pipe orientation while installed in the test bed. 50 minutes after being placed on its side, the speedfan log file was stopped and data saved giving us a 1 hour snapshot of temperatures.
It must be stressed that the figures in the following chart are intended only for comparison within the chart itself. There is no comparison to the testbed control unit, and the use of a moderate load vrs our usual full Orthos load makes any direct comparison to our standard tests for the NT06 impossible.
This chart is a bit busy, but bear with us and hopefully all will become clear.

Those of you familiar with Speedfan's chart function will instantly recognize the red spikes of this chart. The only difference here is the 1 hour time span vice Speedfan's normal 13 minute display limit.

For those of you not familiar with Speedfan's chart function, the tip of each red spike represents the CPU temperature value in 3 second intervals. (Don't try to count them, there are 1200).

The blue line is a statistical “trend” based on the data in the red spikes. In a nutshell it smooths out the peaks and valleys giving us an approximation of average temperatures which is much easier to read than the original 3 second interval data.

Also indicated on the chart is the orientation of the case. The vertical orientation represents the first 10 minutes, and the horizontal, the last 50 minutes of the test.

If you're still with us on what the chart is displaying, you should now be able to see that there was an immediate raise in temperature when the case was moved from the optimal orientation, and that over a 50 minute time span it has mostly stabilized to a difference of between 2 and 2 ½ degrees Celsius.

A 2 deg. C difference might not seem significant, however we can surmise that since this orientation effect on temperatures can be measured, there is a good chance that it would be reflected in our main testing results if we were using a tower case as a test bed rather than the current horizontal mount open case.

If these results could be taken into account for our main benchmark tests, the medium RPM overclocked load chart would look a lot different when comparing the NT06 to the control unit.

We of course cannot make those kinds of changes for the benefit of a single review unit, so you'll have to draw your own conclusions based on the available data.