Thermal Management of Light Emitting Diode Displays

By Russell McMahon and James Newton based on data from

High power LEDs are suitable for applications that need extremely high brightness or high output, such as automotive headlights, interior lighting, grow lights, etc. In many cases, LEDs are mounted in narrow spaces for applications that use mini power LEDs. In this case, heat management is one of the most important factors in the application. In contrast, for high power LED applications, the designer must consider how to manage heat, in order to enhance the performance of the LEDs. If heat management is not considered, the lifetime of the LED will be significantly decreased, or the LED will fail.

The maximum ambient temperature at which LEDs can be used is determined by the PN junction temperature (Tj). When the junction temperature reaches its upper limits it is called the maximum junction temperature (Tj_MAX ). So, in the heat management design, designers must know what Tj they are operating at, and whether the product will reach the Tj_MAX or not.

If designers can keep the Tj below Tj_MAX, they will achieve longer life for the LEDs.

Tj is practically impossible to directly measure, but it can be inferred from other measurements, and once the thermal properties of the application is understood, it can be calculated with some accuracy for any given ambient temperature, drive current or other factor.

Tj is related to the forward voltage of the diode (Vf). As the junction heats, the voltage drop will increase. It remains to be seen how consistent this relationship is between individual LEDs in a batch or between different LEDs. If these values remain well related over a range of parts, Vf can be used to predict Tj.

Tj is also related to Ta, the ambient temperature. As the junction heats up, that heat is radiated through the die, into the lead frame and out to the ambient surroundings.The rate at which the heat is released can be thought of as the thermal resistance of the materials in the circuit from junction to surroundings (Rth_a-j)

Tj can be estimated to be: Tj = Ta + W*Rth_a-j where W is the power consumed by the LED or W = Vf * I where I is the current flowing into the LED, Ta is the ambient temperature, and Rth_a-j is the thermal resistance of the materials between the diode junction and the surroundings. If Rth_a-j is not specified by the datasheet, then measuring Rth_a-j will allow us to calculate how hard we can drive the LED for a given Ta and rated Tj_MAX.

Tj = Ta after reaching thermal stability. So we can find Tj for a given drive power by waiting for some time in an enclosed space and then measuring Ta. If we heat an LED to say 70'C and leave it there for enough time the while LED will thermally stabilize to 70'C. We know that Tj is now 70'C

Tj does not change instantly. If we now drive the LED with small pulses at a low duty cycle at the LEDs rated current, and measure Vf when the current is on, then Tj will not have time to move very much due to the very short pulse and thus low thermal input, so we are able to measure ACTUAL Vf at this Tj. We can heat or cool the entire system, allowing the LED time to stabalize thermally so that Tj is still equal to Ta, and take another measurement of Vf with a short drive burst. The test pulses do not have to be small - just low energy (current x time) eg 350 mA x 1 uS at 1 Hz repeat rate = 0.35 microamp average. Using current pulses at a range of values across range will help plotting the curve.

Once we establish the relationship between Vf and Tj, we can hold Ta at some known temperature, drive the LED to produce a Tj higher than Ta, measure the Vf and lookup what the Tj must be, then apply the following formula to calculate:

Rth_aj = (Tj - Ta) / W

Dont be confused here: The Ta is the current ambient temperature while the LED is NOT at thermap stability, but instead while it is being driven and Tj is estimated by measuring Vf against our prior collection of data.

Notes:

In use at constant current die will be at thermal equilibrium almost immediately - small fraction of a second. Usually measured in mS.

It's really Tc (Temperature of the LEDs heat sink) rather than Ta that is used - ie you heat sink the lead and measure that. Tc is generally best mesaured at a point on the Cathode lead right at the entry to the LED body. The die is usually mounted on the Cathode lead structure. In the real world eg inside a light you will need to know overall thermal path resistance. But the path Tc to Ta can be dealt with by standard magic (PCB thermal issues etc)

While Tj_MAX should not be exceeded the usual aim is NOT to not-exceed it but to not exceed some lower value. So the object of this overall exercise is to know what Tj is in a given situation for a given LED. LED lifetime increases greatly as Tj drops below Tj_MAX. Lifetimes may double or more for every 10C drop. There are numerous factors to complicate this. Rate may be even higher than that.

It happens that Rja or any Rxx actually varies with drive and ambient temperature, but the variation is probably not usually large % wise.