Thermal Design Parameters

LED Cooler
In many of the write-ups that are available from array manufacturers and thermal management companies, there is a tendency to suggest that the purpose of thermal management for LEDs is to avoid a catastrophic failure. The approach taken is to estimate the highest expected ambient temperature and then calculate the heat sink resistance necessary to keep junction temperature below its rated maximum of (typically) 150°C.

At Cooliance, we take a different view: Our approach is to create a thermal environment that guarantees rated performance under normal operating conditions for extended periods of time.  That means running at junction temperatures well below the rated maximum. In order to maintain color quality and maximize flux over the life of the array, we recommend junction temperatures in the 85° to 100° range. This in turn typically means running case temperatures at 65° to 75°.

That higher junction temperatures lead to reduced lighting performance over time is an accepted fact in the industry. Measured in terms of the number of hours before “L70” (the point at which light output has dropped by 30% or more), a forty degree higher ongoing junction temperature can lead to system life reductions of thirty-five to forty-five percent. In other words, if your array is expected to deliver 70,000 hours at a junction temperature of 85°C, running it at 125°C may reduce those hours to as little as 38,000 hours.

Light quality is also affected by heat. In some applications, such as museums and galleries, maintaining stable light quality over time is of primary importance. Ultraviolet lighting applications are also especially sensitive to heat. Building reserve cooling power into these applications is important.

Generally speaking, cooler is better, and the incremental cost of getting that extra margin of design reserve is small relative to the boost in long-term lighting performance.

To help understand the interplay of all the parameters that come into thermal management design, we have made a simple model available. Inputting data about your particular array, you can experiment with different assumptions about ambient temperature, target junction temperatures and the use of different interface materials.