The thermal resistors Rth
Let’s start with the most important aspect, the thermal resistance Rth. The most important property of Rth is that it correlates nicely linearly with the layer thickness, while the thermal conductivity describes a completely different curve and remains anything but linear. But the experienced reader already knows all this, of course. We are interested in layer thicknesses of 200 µm and less on the CPU; on the GPU it is usually even 100 µm and significantly less, depending on the bending.
The curve of the TF8 is almost congruent with the curves of the DOWSIL TC-5888 and TX-5500, which is not only astonishing, but could also be the reason why some pastes are said to contain the DOWSIL TC-5888, although it is actually more likely to contain what Thermalright buys for the TF8. It’s not for nothing that this paste has long been the front-runner in my measurements. However, the TF9 and TFX should be evaluated with caution below 50 µm, as the pressures required for this are simply pointless and impractical.
I have now compared the relevant layer thicknesses from 50 to 400 µm as a bar chart for Rth. And why don’t I compare down to 25 µm? The explanation is simple, because the TF9 and the TFX have a problem with the minimum layer thickness! Both pastes can hardly be compressed any further below 50 µm with normal contact pressures and 35 µm is actually the limit. I will write more about this later, including the causes.
Minimum possible layer thickness
This is exactly why I wanted to know how far you can go with a little pressure and how much a paste can still be compressed. I use the usual 60 Psi (41 N) on the measuring surface of 1 cm², which is completely sufficient and is more than what a GPU cooler, for example, achieves. Except for the TF9 and the TFX. Here I made an exception and even doubled (TF9) or tripled (TFX) the power. Not even an Intel cooler on the current sockets can achieve that! Actually, the limit would be 35 or 50 µm. That’s why it’s not down to the user’s skills, but to the heat-conducting concrete. The layer is clearly too thick, even under heat. In this diagram, however, I am not listing the values generated by force, but the real values at a maximum of 60 psi (41 N per cm²):
Interface Resistance
What also seems interesting is the contact resistance, in our case the interface resistance. Here you can see how well the surface of the material “clings” to the contact surfaces (IHS, heatsink). These values are also easy to compare and meaningful, as they are always the same calibrated reference blocks. Coarser degrees of grinding or a less favorable microstructure can also be a negative factor, which then influences the effective thermal resistance and thus also the conductivity, as can too low temperatures and too high a viscosity.
82 Antworten
Kommentar
Lade neue Kommentare
Veteran
Urgestein
Urgestein
Mitglied
1
Mitglied
Urgestein
1
Urgestein
1
Veteran
1
Mitglied
Mitglied
Mitglied
Mitglied
Urgestein
Veteran
Veteran
Alle Kommentare lesen unter igor´sLAB Community →