Linus Tech Tips PTM7950 Review – Original, OEM Product or Fake?

1 - Introduction and overview 2 - Microscopy, pad thickness and homogeneity 3 - Microscopy, particle analysis and chemical composition 4 - Burn-in and performance measurements 5 - Summary and conclusion

Application, burn-in and tear-off image after around fifty cycles

After completing around fifty burn-in cycles, the tear-off image clearly shows how homogeneously the material has worked its way into the microstructure of the two contact surfaces after repeated heating and cooling. The wetting is complete, the filler particles are densely packed in an evenly distributed polymer matrix and there are no significant cavities or unevenly compacted zones. The typical, finely rastered tear-off patterns correspond exactly to what you would expect from a correctly baked phase change material.

The LTT pad behaves practically identically to the usual PTM7950 and differs only in minor details of the surface pattern, which are more likely to be due to different initial thicknesses and slight deviations in the rolling process. Internal tests show that Laird TPCM 7000 sometimes has even greater long-term stability when subjected to extremely high thermal cycles. However, the difference remains in the range far removed from practice. For the end user, whether in private or semi-professional use, all the pads tested are within such a narrow performance window that the difference is hardly relevant from a technical point of view. The actual performance of the Honeywell PTM7950 and Laird TPCM 7000 is close together, and according to the measurements, the LTT pad also fits seamlessly into this spectrum.

Technical examination and evaluation of the material quality

The analyses show a largely consistent picture. The microscopy shows a homogeneous filler distribution of aluminum oxide and zinc oxide, the LIBS measurement confirms the known element pattern of a silicone-free PTM material, and the particle sizes are exactly in the range that is typical for PTM7950. The burn-in characteristic with an early melting start at around 35 degrees Celsius, identical end point at 55 degrees and a final bond line thickness of 14 micrometers also corresponds to the behavior of OEM PTM.

However, the uneven initial thickness remains striking. While the packaging indicates a nominal thickness of 0.25 millimetres, the measured values are actually in the region of 0.20 millimetres and also show local fluctuations. The partially poorer rolling and the edge zones with decreasing thicknesses indicate that the material originates from web areas that are not usually intended for precise, industrially specified blanks in the OEM process.

This can have various causes. It is possible that there is less selectivity in the assembly for the end customer market, in which sections are used that no longer meet the full tolerance in the industrial series process. It is also conceivable that offcuts from larger production runs are used, which are functionally perfectly fine but do not meet the tight OEM specifications in terms of thickness tolerance. In terms of technical performance, this hardly plays a role in the private sector because the pad largely levels out the thickness itself during the burn-in process anyway. In industrial use, on the other hand, greater attention is paid to reproducible output parameters, which is why such sections would probably be rejected there.

Putting this into perspective for the end customer market, it should be noted that the uneven thickness has only a minimal influence in practical use, as all PTM pads are pressed together under comparable forces to almost identical final thicknesses. What ultimately counts for the user is the thermal resistance in the baked state, and this is in the upper performance range of the PTM family for the material tested.

Conclusion and classification of the results

In the overall evaluation, the LTT-Pad delivers a performance that has nothing to hide on CPU-DIEs and GPU processors. The thermal resistance after burn-in is low and in the measurements even slightly lower than the reference PTM7950 and the Laird TPCM7000. The GPU temperatures follow this trend and are in the range of hundredths to tenths of a degree below the reference values after a few cycles. Microstructure, particle morphology, LIBS signature and the characteristic early onset of melting behavior clearly correspond to a material design typical of PTM.

The decisive question remains as to whether the LTT pad is actually an original PTM7950 or a closely related product with the same chemical basis but a different finish or tolerance class. The available data predominantly indicate that it is very close to the original. The fillers, the polymer matrix, the burn-in behavior and the final performance are almost completely identical to what is known from previous OEM samples. The only clear deviation is the uneven initial thickness and the lack of a pronounced kink at 45 degrees Celsius.

Both can be plausibly explained by the fact that the material originates from web areas that no longer meet all the tight OEM tolerances, but are functionally identical. The smoothed burn-in curve could also indicate a slightly different matrix-particle interaction or a narrower melt distribution without changing the basic characteristics. The totality of all characteristics therefore leads to the conclusion that the LTT pad very probably comes from the same material family as Honeywell PTM7950 and is practically equivalent to it in terms of thermal performance. For the normal user, this means a reliable, very efficient solution which, after correct burn-in, represents one of the best available thermal interfaces for CPUs and GPUs.

Finally, I would like to make a comparison that spontaneously elicits an almost nostalgic smile from me. Because Honeywell may have done something here that I always looked forward to as a child: Back then, the baker’s shop sold for little money the edges of cakes that were actually unsaleable and were left over from the rectangular baking tray when the cakes were cut to size. They didn’t have the perfect shape, were sometimes a little crooked or uneven, but tasted exactly like the original and were often even the better choice for me because they had the same contents, just without the flawless appearance.

This is exactly the impression left by the tested pad. The microstructure is excellent, the thermal behavior is identical to the high-quality product and the performance is in no way inferior to the original. The only real deviation lies in the external shape, the slight variations in thickness and the less than perfect rolling, which would be put to one side in an industrial environment but is completely uncritical in the end customer market. So, if you like, this pad is the edge of the cake among the PTM7950 products. Not intended for industrial showcase sales, but fully-fledged, powerful and, in practice, at least as good as what is otherwise lying perfectly cut in the display. And sometimes, as I knew even as a child, these are the pieces with the best bite. Incidentally, I have contacted Linus and they will check the batch again. But it doesn’t change the fact that the pads work. In the end, the appearance is rather marginal