Thermal putty or thermal pad, which is better? Myths about thermally conductive materials, thermal conductivity and temperatures

1 - Overview and Introduction 2 - The real Chain of thermal Resistances 3 - The Benefits of a cooling Backplate 4 - All what you need is a better Cooling!

Backplate as a panacea? Yes and no!

I’ll take another look at how the backplate works, this time focusing on the realistic case of a cooler temperature of around 40 °C and a PCB hotspot of just under 70 °C. The key point is that although a backplate is an active component of the overall thermal system in this scenario, its effect does not unfold where many people assume it does. It only has an extremely small effect on the VRM junction temperature and instead primarily affects the PCB itself and all neighboring components located in the extended hotspot area.

The backplate is an additional, large-area heat path in the core. It absorbs energy that previously entered the PCB material through the bottom path of the DrMOS and distributes it over a much larger area, where it is released into the air via convection and radiation. It thus fulfills a protective function for the PCB structure and the large number of small components in this thermally stressed area. However, the actual limit of this mechanism is created by the serial resistors located in front of the backplate. The heat must first pass through the internal path ΨJB of the DrMOS, then through several copper layers and FR-4 layers, then through the putty or pad and only then into the backplate. This sequence significantly limits the performance of the backplate.

In the resistance chain, everything starts with ΨJB, the thermal resistance of the barrier layer to the underside of the DrMOS. It is typically around 2 K/W and determines how much heat reaches the PCB in the first place. This is followed by the actual board path, which consists of copper areas, vias and FR-4. In the thermal load situation considered here, this results in a dominant resistance in the range of around ten Kelvin per watt, which understandably leads to significant heating of the rear side. This is followed by the putty, which contributes further Kelvin per watt depending on its thickness and thermal conductivity. Only then does the heat reach the backplate, which distributes and emits heat very well, but can only absorb a limited amount of the total power due to the constrictions in front of it.

I have now shown the whole thing in two sets of curves, one for the rear of the PCB and one for the junction, both with and without the backplate. I set the cooler temperatures on the top side to 30, 35 and 40 °C, while the backplate remains constant at around 40 °C. The putty between PCB and backplate varies from 2 to 12 W/mK. In each case, the temperature is plotted against the putty conductivity so that you can see very clearly how the three cases relate to each other.

Without a backplate, the situation is trivial: the heat is only transferred to the air via the front side and the “naked” rear side, there is no putty to the backplate, so the resistances do not change. Accordingly, the lines above the putty conductivity are strictly horizontal. For a 40 °C cooler temperature, the PCB hotspot in the model is around 69 °C, for a 35 °C cooler around 67.7 °C and for a 30 °C cooler around 66 °C. Without the backplate, the junction is around 99 °C, 96 °C and 93 °C respectively, assuming the same temperature. The three lines are therefore “rigid” and only move in parallel if the cooler becomes colder or warmer due to a higher or lower fan speed.

With backplate and putty, a second set of curves is created, which now actually depends on the putty conductivity. For a cooler temperature of 40 °C, a relatively weak putty with around 2 W/mK on the back results in a PCB temperature level of just under 49.5 °C, with 6 W/mK the back drops to around 43.4 °C and with 12 W/mK again to around 41.7 °C. The same trend can be seen for 35 and 30 °C coolers, only shifted downwards in parallel because the entire heat path operates at a colder level. This means that the backplate pulls the hotspot down significantly overall and the effect scales with the putty conductivity, but flattens out towards higher values because the FR-4 component then dominates in the chain.

The curves for the junction temperature look much less spectacular. At 40 °C cooler, Tjunction with backplate and 2 W/mK putty is at around 88 °C, with 6 W/mK at around 84.6 °C and with 12 W/mK at around 83.7 °C. Although these are measurable gains compared to the case without a backplate at just under 99 °C, the curves only shift by a few Kelvin due to the variation in putty conductivity itself. If, on the other hand, the cooler temperature is lowered from 40 to 30 °C, the entire junction curve simply shifts downwards by around 10 Kelvin, regardless of the backplate, i.e. as one would expect from a predominantly air-cooled system.

Overall, this corresponds exactly to the theoretical expectation. The lines without backplate remain strictly linear above the putty conductivity and only show the effect of active cooling via the heatsink. The curves with backplate decrease with increasing putty conductivity because the additional path towards the 40-degree backplate becomes more usable, and they visibly flatten out towards high W/mK values because the FR-4 component then determines the total resistance. The effect on the back of the PCB is clear because the new path intervenes almost directly there, while the junction temperature only increases moderately, as it is still primarily determined by the internal package resistors and the path to the cooler on the top side.

We can see that the backplate only has a very small influence on the actual junction temperature of the VRM. Most of the temperature differences occur within the DrMOS package itself, in the path ΨJT upwards and ΨJB downwards. The board path also dominates the lower path, which is why the backplate can only reach a fraction of the total waste heat. Its effect therefore unfolds primarily where it offers the greatest advantage: It relieves the PCB thermally, distributes local hotspots over a larger area and thus prevents excessive heating of small components, traces and via clusters. The junction temperature of the DrMOS can hardly be changed, but the temperature of the entire hotspot area can.

This clearly outlines the role of the backplate. It is not a means of making the VRM itself significantly cooler, but an effective method of stabilizing the thermal structure of the PCB, distributing the hotspot more widely and reducing the load on all neighbouring components. It is indispensable in this function, even if it only marginally influences the junction. A large-area connection of the hotspot zone to a passive backplate via thermal putty would primarily reduce the temperature level of the PCB and the surrounding components on this board, but would only relieve the DrMOS junction itself to a manageable extent. The background to this is that the dominant heat path of the MP87993 from Monolith continues to go via the base pad into the front of the PCB and is distributed from there into the copper layers. The rear side is thermally the “end” of this path even without the backplate; the backplate basically only replaces or supplements the uncooled air as a heat sink.