ASRock X870E Taichi OCF Review and complete Teardown: Focus on the Essentials or more with less is almost impossible

1 - Introduction, unboxing and technical data 2 - Topology of voltage regulators and their cooling 3 - Teardown: USB 4 sub system, PC audio and WiFi 7 4 - Teardown: Chipset topology and other components 5 - Backplate, cooler, pads, and thermal conductivity 6 - UEFI, overclocking and own experience 7 - Performance and conclusion

Cooling concept, backplate and mechanical-thermal consideration

The backplate of the ASRock X870E Taichi OCF takes on a much more active role than is usual with many motherboards. It not only serves as an optical or mechanical element, but is also clearly functionally integrated into the overall thermal concept. On the rear of the PCB, thick thermal pads are used over a large area and in a targeted manner to cover precisely those areas where performance-relevant hotspots are located on the front. These include in particular the VRM zone, adjacent power stages, parts of the voltage filtering and other thermally stressed ICs. The pads couple these zones directly to the solid backplate and thus enable additional heat dissipation across the entire rear surface, which is particularly relevant for long-lasting loads.

In addition to its thermal function, the backplate also fulfils an important stabilizing function. The PCB is braced across its entire surface via several defined screw connection points, which reduces deflection under mechanical load. This structural reinforcement should not be underestimated, especially with heavy coolers, large graphics cards or frequent mounting on open benchtables. The even tensioning counteracts stresses in the PCB material and ensures reproducible contact forces between coolers, pads and components. Indirectly, this also contributes to the long-term stability of the solder joints, as micromovements under thermal expansion are minimized.

Another aspect is the electrical function of the backplate. Due to its large metallic surface and the defined connection via screw points, it can act as an additional ground plane. This improves the current return flow, particularly in high-frequency loaded areas, and can reduce parasitic effects. Although the backplate does not replace a dedicated ground plane in the multilayer PCB, it is a useful addition, particularly around the power supply and fast I/O signals.

The actual VRM cooler continues this concept consistently. ASRock uses a massive, L-shaped heat sink here, which extends over the entire area of the power supply and at the same time fully integrates the I/O shield. Heatpipes are deliberately omitted, which may seem unusual at first glance, but in this case is technically well justified. The sheer mass of material and the large surface area of the heat sink allow direct heat distribution within the block without introducing additional thermal transitions. This is often more efficient and robust than complex heatpipe designs, especially for moderate to high continuous loads. The cooler for the chipset ICs, which is concealed under a cover, has an identical solid design.

The VRM cooler is not mounted in isolation, but is mechanically screwed to the backplate. This creates a continuous thermal path from the front of the PCB via the voltage converters and coils, through the PCB, to the rear. This sandwich construction significantly increases the effective heat capacity and distributes load peaks both temporally and spatially. In addition to the VRMs and the associated inductors, the USB 4 controller is also actively cooled, which also generates significant power loss under high I/O loads. The contact surfaces are cleanly designed, the pads are sufficiently dimensioned and evenly compressed, which indicates that the pad thicknesses have been carefully matched.

The material analysis using LIBS confirms the constructive design. Both the backplate and the VRM cooler are made of aluminum, which is clearly confirmed by the detected elements. Aluminum offers a sensible compromise between thermal conductivity, weight and machinability. The microscopic images also show a typical, slightly roughened surface structure, which primarily results from manufacturing processes, but secondarily also slightly increases the effective surface area. No coatings or composite materials can be detected, nor are there any foreign elements that would indicate inferior alloys.

We therefore see a cooling concept that relies less on spectacular individual measures and more on mass, contact quality and mechanical integration. The combination of supporting backplate, large pad connection and solid VRM cooler without heat pipes appears technically consistent and designed for continuous load. Especially in the context of a board that is primarily aimed at stability and reproducible behavior instead of short-term record attempts, this approach is comprehensible and functionally cleanly implemented.

The thermal pads: composition and performance

The thermal pad used on the VRM, chipset and backplate is recognizable as one and the same type of pad, which has merely been manufactured in different thicknesses to reliably bridge the respective gaps between the component surfaces, heat sink and backside support. The microscopy images clearly show a silicone-based matrix with a high proportion of mineral filler, the gray base mass is homogeneous, while at the same time numerous spherical to roundish inclusions are noticeable, ranging from a few micrometers to well over 50 µm, depending on the field of view. In the examples measured, the marked particles and aggregates lie roughly in the spectrum from around 3.6 µm to 82.3 µm, with several clusters in the range around 10 to 15 µm and 35 to 45 µm, plus individual larger inclusions around 50 µm and above. This distribution is consistent with a pad that not only works with fine filler fractions, but also deliberately contains larger particles and agglomerates in order to support thermal conduction in the volume while maintaining a defined compressibility.

The LIBS analysis confirms the image of a mineral-filled, silicone-containing pad, with significant proportions of oxygen, zinc, silicon and aluminum, hydrogen is also present as a matrix indicator, carbon does not appear in the evaluation as a supporting component. In practice, oxygen plus zinc very often indicates zinc oxide-based fillers, silicon and aluminum also indicate silicate or aluminum-containing components, both of which are common in this material class if you want to combine mechanical stability, temperature resistance and robust processability. It makes sense that the composition remains identical for different pad thicknesses, as this means that the thermal behavior per material volume does not change; the adjustment to different construction heights is then made exclusively via the thickness and compressibility, not via a change of material.

With the ASTM measurement data, it is important that the evaluation here shows the bulk thermal conductivity of the pad material and the surface-related contact resistance separately, which is exactly what can be seen in the linear dependence of the thermal resistance on the thickness. In the series of measurements shown, which extends from approximately 1000 µm to 1600 µm, the regression line is very clean, the determination is practically 1, which indicates consistent material behavior and a stable measurement. The reported thermal conductivity is 6.929 W/mK with a scatter of ±0.095 W/mK, which is a clearly superior value for a soft, compressible pad of this class and explains why the manufacturer dares to cover not only VRMs and coils, but also the chipset and the backplate coupling.

The area-related interface resistance also stated is 26.5 mm²K/W with ±2.5 mm²K/W. The amount is not atypical, but it is thermally relevant as soon as you get into areas where the thickness is reduced or the contact pressure fluctuates in practice, i.e. exactly where pads have to work on rear sides, on component edges or on components that are slightly offset in height. For the overall effect in the system, this means that the effective heat dissipation in practice is not only determined by the 6.929 W/mK in the volume, but also by the actual contact quality, the surface roughness and the surface pressure actually applied.

The advantage of using the same material in different thicknesses is that the mechanical tolerance chain can be beaten with the thickness without having to buy in a second thermal variable. The disadvantage is that very thick pads inevitably have a higher overall resistance despite good thermal conductivity because the thickness component dominates, while very thin pads benefit in terms of volume but can react more sensitively to contact deficits because the interface component is then more important in percentage terms. However, the measured ASTM values in combination with the very clean linearity are a good indication that the pad material is not only suitable for marketing, but also consistent in terms of measurement technology, and that the cooling design with backplate coupling and VRM block also makes sensible use of this basis, provided that the mechanical screw connection keeps the surface pressure sufficiently even across all zones.

And what about the pads on the coils? These pads clearly show that ASRock does not use the same materials as the large pads for the VRM, chipset and backplate. Even visually, the surface appears more homogeneous, but at the same time “softer” and less particle-heavy, i.e. more like a classic silicone elastomer with a comparatively moderate filler density. In the microscope images, it is also noticeable that the structure appears less “grainy” and less densely packed, but with clearly visible inclusions. Such inclusions are not unusual in soft pads, but they reduce the effective heat conduction because they act locally like small islands of insulation.

The material classification from the LIBS evaluation, which classifies the whole thing more as a favorable silicone compound, i.e. a silicone compound in which the heat conduction essentially comes from the proportion and quality of the fillers, not from the base material, also fits in with this. And here there are no thermally conductive metal oxides, only a high carbon content, which should explain the dark color.

This is very clearly reflected in the ASTM data. At around 1.214 W/mK, the determined thermal conductivity is at a level that can be described as functional, but not as high quality, especially if you place it directly next to the significantly better pads from the VRM environment, which land at around 6.929 W/mK in their measurement.

Nevertheless, this is not automatically a drama for the coils, because the heat flux densities there are typically lower than at the hotspots of the MOSFETs, and because part of the power loss in the coil is dissipated to the environment via convection and radiation. The practical disadvantage is more of a systemic nature: a weaker pad forces a greater temperature difference between the coil and heat sink due to the higher thermal impedance, which means that the entire cooling system works less “evenly”. In combination with very thick pads, this effect is further intensified, as the thermal resistance of such materials scales strongly with the thickness, and it is precisely this thickness that is usually high in the design of coil pads because height differences and tolerances have to be bridged.

Incidentally, the negative interface resistance in this context is not a “physical cooling turbo”, but a measurement artifact that can occur more frequently with very soft, thick pads and unfavorable compression. The background to this is that in the ASTM evaluation, the total thermal resistance is fitted linearly as a function of thickness, and the intercept of this fit is interpreted as the interface component. However, if the thickness is not recorded cleanly and reproducibly, or if not only the thickness changes under load, but also the effective contact area, the internal structure and thus the apparent bulk conductivity, then the linear model assumption is violated. This is particularly critical with these 2 mm pads because they can often only be partially compressed in the relevant load range and not linearly, but at the same time “flow” at the contact surfaces, displace laterally and thus change the real contact geometry with each load point. In such a situation, the fit can mathematically provide a negative intercept, although the real interfaces naturally always represent an additional resistance. The fact that the value here also carries a large uncertainty is a further indication that although the regression under these boundary conditions produces a nice straight line, the parameters can no longer be interpreted separately in a stable manner. In practical terms, this means for the evaluation: The 1.214 W/mK is plausible as an order of magnitude and fits the observed material appearance, whereas the negative interface value is a warning signal that the 2 mm pads can only be mapped unfavorably in the measurement setup and that the separation into bulk and interface should rather be interpreted conservatively and with a view to the compression mechanics for this material.