Designing a Heatsink for Multiple Devices and Packages

Designing Heatsinks and Thermal Solutions for Xilinx Devices (XAPP1377)

Document ID
Release Date
1.0 English

Many systems have multiple devices that need thermal management. Often it is desirable to simplify the heatsink design by having a single heatsink touch all these devices. There are many unique considerations for these designs, and this section is not intended to cover many of these aspects. It is, however, intended to highlight some focus areas common to this design.

The primary trade-off of this approach is that often thermal performance is sacrificed due to accommodating the different height tolerances of all the devices the heatsink must contact by increasing TIM thickness, and thus thermal resistance to the heatsink. The general suggestion here is to optimize TIM thickness for the most thermally critical component to get the best thermal transfer to the heatsink for that device, often at the cost of slightly increased thickness of the less critical devices. This can be a tricky balance, particularly if multiple devices have tight thermal margin. However, if the devices are expected to be close to or exceed the temperature limits, this is a necessary step to get the best system performance.

When it comes to lidless devices, managing the contact resistance to the heatsink is more critical, which generally lead to one of two choices. The first and most thermally efficient choice is to construct the proper island in the heatsink and optimize the BLT to the lidless device. The second option is to use a floating lid or an attached intermediate heat spreader connected to the top of the lidless device as a means to ensure good thermal contact to the die of the lidless device. It is not suggested to use a thick TIM or gap filler on lidless devices, as it is more susceptible to voids and larger thermal resistance resulting in poor thermal performance and potential reliability issues.

Due to the larger size of the heatsink base for multiple device thermal attachments, another consideration is that managing heat spreading within the base becomes more important. This generally leads to a higher importance to consider two phase cooling in the heatsink base as a means to maximize cooling and minimize the impact of one device heating another.

Lastly, it is often an even greater consideration where the attachments for the heatsinks exist on multiple device thermal management solutions. Xilinx devices still require even application of 20 to 50 PSI for lidded and lidless packages of the thermal management solution, and thus, having attachments not located very close to the device can make it difficult to manage this. If it is uncertain or impossible that the minimum pressure can be applied to the device, you should consider installing an attached heat spreader that can ensure the proper pressure on your Xilinx device.