Now we come to the actual core topic of my article. It is well known that good pastes are also defined by the fact that the differences between the first and the umpteenth cycle are as small as possible. The frequent temperature differences of less than one Kelvin in many GPU tests are certainly also defined by the high initial performance of some pastes, which, however, cannot be maintained for long over several cycles. For this very reason, I’m going to take the Thermal Hero Quantum and simply run it through 6 time-consuming cycles. At around one hour per cycle, that’s just over 6 hours for this paste for the following test.
At 30°C, the Thermal Hero Quantum loses a whopping 38 percent in thermal conductivity between the first and sixth run and at 100°C it still loses a good 9 percent! Of course, the changing layer thicknesses (BLT) also play a role here, as the cheap paste literally melts when cold, but we’ve already discussed this with the minimum BLT. Good pastes lose a maximum of 1 to 4 percent at high temperatures, but this is more than double that!
It is not the actual length of a measuring cycle that is decisive, but the frequency of the respective temperature changes, which unfortunately also take a lot of time. This is because the warm-up and cool-down process does not take place abruptly, but rather slowly increases and decreases. Otherwise it’s useless. Of course, the BLT also plays an important role, so I’m going to take a closer look at that too. And for a direct comparison, I will now add a second, usable paste.
Now let’s take a look at the sobering measurement result and be amazed! At around 60 °C, both curves meet, then the Thermal Hero Quantum literally drifts apart! Depending on the test setup, however, such a curve also contributes to the fact that not only the values of different reviews differ so much, but that especially the reviewers who found the Thermal Hero Quantum so good usually had lower temperatures in the paste. This is because the CPU temperature read out is by no means the paste temperature and so in the chain of all thermal resistances between die and water, despite similar CPU temperatures, different paste temperatures and thus also BLT can occur from build to build! This makes such tests hardly comparable with each other as long as no uniform conditions can be guaranteed.
The Thermal Hero Quantum, which is only filled to a medium level, can be squeezed down to an astonishing 15 µm even at 30 °C, while the Thermalright TF8 only manages around 25 µm when cold. However, if you slowly heat the pastes up to 100 °C, you can see that the Thermal Hero Quantum increases the layer by almost 127 percent (!) and thus more than doubles it! This is good evidence of bubble formation, possibly also water residue and a low-quality silicone oil that is already outgassing. I won’t go into the exact processes here, as I have already discussed this several times.
The Thermalright TF8 starts with 25 µm, but ends at 100 °C with only 31 µm BLT, which means an increase in layer thickness of only 24 percent. So instead of 2.3 times more, we only have just under a quarter more. This already says much more about the quality of a paste than a single test run with CPU and stress test program over 30 minutes. The blue curve indicates that the Thermal Hero Quantum is core scrap, even if some testers praise it to the skies. That was nothing, is nothing and will be nothing!
But now I have another 6-hour special of the Quantum, which shows us that the delta between the maximum and minimum BLT remains almost the same, but shifts downwards in relation. In the end, however, this also means that the paste starts to dissolve after just a few hours at 5 µm. But I’ve already been through that.
That would certainly refute this myth and all the accompanying phenomena.
55 Antworten
Kommentar
Lade neue Kommentare
Urgestein
Veteran
Urgestein
Veteran
Urgestein
1
Mitglied
Veteran
Urgestein
Urgestein
Urgestein
Urgestein
1
Mitglied
Urgestein
1
Urgestein
Urgestein
Veteran
Alle Kommentare lesen unter igor´sLAB Community →