Thermal pads — forced reality

Content

• Introduction
• Searching for the best
• Testing methodology
• Testing results
• Adding thermal adhesive
• Fixing errors
• Conclusion

Introduction

The use of thermal pads as a thermal interface for memory chips and power elements instead of a regular thermal compound is necessity caused by design features. But little of you thought of how well they perform and what temperatures they bring.
This idea to test the thermal pads (T-pad) wasn′t spontaneous. I was building up my first WC system and faced some problems. The GPU power element temperatures were outrageous and it caused card to malfunction — artifacts during gaming. For my ATi Radeon 4870×2 video card I used a fullcover type waterblock Watercool Heatkiller GPU-X2 4870-X² LT. The thermal pads are placed between the radiator and the power elements, voltage regulators and memory chips as it is common for majority of such devices. However, stock air cooling systems are not an exclusion. A T-pad kit of different thickness cut for faster and easy installation comes together with the waterblock. Strictly following the instructions I placed each T-pad on an appropriate element. Small square-shaped pieces and strips of 1 mm thickness were used for some power elements and memory chips whereas 0.5mm thick strips were used for a PCI-E bridge and the rest of the power elements (VRM), particularly, 0.5mm thick strips were used for Vitec voltage regulators. Having an idea about temperature level of a GPU while using the stock cooling system, I began testing the water cooling system (WCS) which was a novelty for me. And there came the first disappointment. The VRM element temperatures were high. Even in idle the temps reached 55°С whereas the temperatures of both GPUs won′t go over 43-44°С. During gaming VRM element temperatures easily reached 80-90°С — I had never seen such results even with the stock air cooling system. When I launched the Furmark stress test the temperatures of the power elements hit 120°С and over within several minutes, the same happened in other synthetic stress tests. That is where I stopped all the tests. I dismantled the WCS partially and started the investigation on the reason that was causing such high temperatures. First of all, I checked the pressing imprints on the waterblock and video card′s elements — everything was OK. The T-pads were made flat and had an imprint of the element marking whereas the thinner ones were stuck for good to the surface of the radiator. It is obvious that there is something wrong with the T-pads because the element temperatures can′t be higher if cooled with water using the fullcover waterblock than if cooled by stock system. I compared the outlook and structure of both thermal pads: the ones that came together with the stock cooling system and the ones that came together with the waterblock – they were totally different. I didn′t have any other T-pad kits for that moment so I came up with the idea to add some thermal compound on both sides of the pads. I had lots of Thermalright Chill-Factor thermal compound syringes, so I used it. After WCS reassembly and start, I went straight to temperature testing. In Idle the temperatures of the card′s power elements decreased a to 50°С, but in Load there was a significant temperature drop to 100-105°С after a 30 min run of the Furmark test. Unfortunately, the desired result wasn′t reached. From that moment I decided to stop testing and leave everything as is. There were regular texture artifacts, notably in Valve games — CS:S and Left 4 Dead, which pointed to the memory cooling problems. It is worth of noting, than I didn′t add the thermal compound on memory chips, did it only for power elements. And as it turned out in vain, but it′d be discussed later.