Protecting high brightness LEDs from thermal stress
Simple and low-cost electronics can protect costly high-brightness LEDs from adverse thermal conditions, write Alan Buxton and Ho Wong of Zetex Semiconductors.
Since even the best possible thermal design practice can be thwarted by an inappropriate lighting installation, the responsibility of maintaining safe LED operating conditions and minimising the impact of thermal effects on LED life must surely fall to the drive electronics.
A quick review of the product specifications provided by manufacturers of high brightness LEDs serves to identify some key design parameters, which need to be taken into account and which illustrate the negative effects of running such components at high temperatures.
The effective life of the LED is inversely related to the power dissipation and temperature of the LED junction. Manufacturers show mean time before failure (MTBF) figures of about 100 million hours at at junction temperature (Tj) of 80°C. In practical systems LED failure is not likely to be a problem. However, in systems where heat is not adequately removed and Tj rises to 120°C or beyond, then LED life will be significantly shortened. In extreme conditions the LED could suffer immediate failure.
Thermal design can certainly apply some overcompensation to take into account worst-case installation scenarios, but in certain instances this may not be possible. Consider a downlight installed in a well-insulated ceiling void. The void not only acts to prevent adequate heat dissipation but also doesn't offer sufficient housing space to fit a suitable heatsink.
Relative light intensity is inversely related to junction temperature. While data sheets vary, manufacturers quote a reduction of light output of up to 30% at maximum junction temperature. Lumen maintenance over time is inversely related to junction temperature. An LED can typically lose 30% of its light output over 50k hours when run at 70°C junction temperature -presumably the reduction is greater at higher temperatures, but figures are not published.
In reality, the reduction in light output over time - whatever the cause - is actually not a great issue. In fact it may not be all that noticeable and in any case LED performance will be comparable with that experienced when using alternative light sources.
In considering the factors discussed, the single most important goal for a prudent designer is to remove heat from the LED, in order to keep the junction temperature below the maximum rating, thereby avoiding premature failure.
The electronics used to generate the required LED current can easily incorporate methods of detecting over temperature conditions. The techniques serve to reduce the LED drive current in order to maintain a stable operating temperature.
Clearly this would reduce the light output, but the LED would survive to enjoy a long and bright future.
By way of an example, the circuit shown in Figure 1 incorporates temperature control in a buck converter configuration. The circuit is designed to drive an LED with a drive current up to 1 A. The supply voltage is from 4 to 6 volts.
When the switch Q1 is on, current flows through the LED and L1. This current builds up to a point where the voltage across Rsense reaches the threshold of U1. The ZXSC300 controller then removes the drive to Q1 which turns off. The energy stored in L1 is then discharged, flowing through D1 and the LED. The ZXSC300 has a fixed turn off period of 1.7μs, after which time Q1 turns on and the cycle is repeated. In this application the switch frequency is about 150kHz.
Temperature detection is achieved using a 150kΩ NTC thermistor, which is located in close thermal contact to the LED. The current flowing through the thermistor is multiplied and summed with the peak switch current in order to regulate the LED current.
As the temperature increases the thermistor resistance reduces and a larger current flows through the thermistor. This produces an increase in the Isense voltage, which causes the controller to turn off at a lower LED current. Suitable values of thermistor, Rgain and Rsense are chosen to maintain the LED operating temperature within the safe operating limits.
The circuit shown in Figure 1 uses components calculated using the following simple formula: Vcc-Vsense = Ithermal x RNTC
Vsense of ZXSC300 is 20mV and thus is negligible relative to Vcc, so Ithermal = Vcc/ RNTC
(Assuming Ipeak>>Ithermal and Rgain >> Rsense)
Vsense = (Ithermal x Rgain) + (Ipeak x Rsense)
Ipeak = (Vsense - Ithermal x Rgain)/Rsense = (Vsense - Vcc/ RNTC x Rgain) / Rsense
In this circuit example a Yuden 150kΩ thermistor is used as the temperature sensor. The target control temperature is 75°C and the output current is 833mA. Rgain is 10Ω, Rsense is 20mΩ and Vsense is 20mV.
The table in Figure 2 shows the thermistor temperature characteristics and the effect on peak current with a 6 V supply. These results are shown graphically in Figure 3 for several voltages in the 4-6 V range, which demonstrates that supply voltage variation has only a small effect on the temperature control.
This example shows the components required for driving an LED current of 833 mA. The circuit can be easily adapted to drive lower currents by changing the values of Rsense and a different temperature break point can be chosen by changing the value of Rgain.
It has been shown that costly high-brightness LEDs can be protected by the addition of some relatively simple and low-cost electronics. This technique can be applied to many different control systems in both buck and boost operating modes using any of the ZXSC series LED driver ICs.
Using this method of thermal protection allows the lighting designer to achieve solutions which are smaller and less expensive to produce. In some cases a form factor can be used which would not be achievable without thermal protection.