An LED is a welcome replacement for traditional light sources. On an aircraft, LEDs are gradually being used more and more for the purposes of indication and warning.
Anti collision lights (the flashing lights seen on an aircraft)are either red or white in colour. White anti collision lights use white LEDs, and red anti-collision lights use either red, or red orange LEDs.
The one problem with an LED is the relationship between light output and temperature. The luminous flux falls with temperature, as shown below:
The CREE XPE Red LED, for example, is about 30% efficient. The remaining 60% energy is given off as heat, and this heat results in the rise in temperature of the LED.
For anti collision lights, the effective intensity must be determined in accordance with the Blondel Rey equation, which, simply put, mathematically states that as the pulse duration is increased, the light appears brighter. This relationship between apparent brightness and the “on” duration of the pulse is non linear.
So we are faced with a conflicting situation: Increasing the pulse “on” duration will make the light appear brighter, allowing for fewer LEDs. But increasing the pulse duration increases the average heat dissipated per LED. There must be a pulse duration that strikes the best balance between the two conflicting requirements. That’s what we shall uncover here.
To begin with, let’s assume that we have a design wherein each LED contributes 50cd of light. Lets further assume that we are building a light that has an effective intensity of 400cd. If we were to flash the light for only 5ms, we would need an LED intensity of 400*41 = 16,400cd. (The 41 coming from the inverse of the Blondel rey factor for a 5ms pulse = 0.0244) This would be met by 16,400/50 = 328 LEDs. On the other hand, if we were to flash the light for 1s, we would need an LED intensity of 480cd. This would be met by 9.6 LEDs.
Now each red LED drops about 2.3V, at 700mA. The power dissipated per LED is 1.61W. For a flash rate of 42 flashes per minute, the gap between two successive flashes is 1.428s. Thus, the average power in the 5ms case is = 0.005s/1.428s * 328 LED* 1.61/LEDW = 1.85W. The average power in the 1s case is = 1s/1.428s * 9.6 LED * 1.61W/LED = 10.82W.
While we may have thought that 328 LEDs generate more heat than 10 (9.6) similar LEDs, the above calculations show that the pulse duration also determines the average heat dissipated.
The graph below shows three curves: Required number of LEDs, Power dissipated, and LED-Power product. The number of LEDs are determined using the Blondel Rey equation, with the 400 effective candela target, and the 50cd/ LED value. The power is determined by as the average power dissipated by the LEDs. The power increases with the pulse duration, despite the fact that the number of LEDs are dropping (as seen in the blue curve). However, the product of number of LEDs required and the power dissipated hits a minima at 200ms. That is the optimal flash duration.
The interesting thing about this is that, the minima is always reached at 200ms, irrespective of the flash rate. So if you were to flash a light designed inaccordace to the Blondel-Rey equation with a flash rate of 60 flashes per minute, the minima at 200ms will yield a minimum power-LED product of 82.4Watts-LED. If one were to flash this same system at 42 flashes per minute, the 82.4Watts-LED product is now seen at 685ms, but the minima is still at 200ms, with 57.7Watts-LED. The 200ms is reached irrespective of the optical design of the system , or the target intensity.
Infact, most LED anti collision lights (see any LED anti collision you can lay your eyes on), flash at 200-300ms. Simple, isn’t it?
Conclusion: Flash your LEDs at around 200ms pulse width. Reduce the flash rate to the minimum possible. That way, you will save on LEDs, and have the lowest heat for the given thermal system.