Now that forum members have a pretty good idea of what PWM is, and lots of neat ways it is used in practical applications, let's mention a couple of points that might have been overlooked. The first point is on the human eye. It is not clear to me and maybe to many researches as well, that the brain itself pieces individual light pulses together to manufacture what looks like a steady beam. It is also possible that the light receptors in the eye actually have a kind of persistance similar to a resonating chamber that continues to output a signal after the stimulus has been removed. There is no morality or goodness factor associated with this effect in either case, but it is interesting to speculate nonetheless.
Carrying the idea of persistence a little further, I have yet to see convincing research that specifies how an LED behaves once the driving current has been removed. Phosphors of the type used in florescent tubes definitely have measurable persistance which allows the tubes to actually emit a relatively steady light output if their drive signal frequency is high enough. The same principal is the foundation of picture tubes of all kinds. Modern electronic florescent ballasts pulse the tube phosphors at 30KHz - 50KHz, faster than the phosphors' light output dies out. What we do know from simple testing is that LED's do not appear to have any persistence, but I suspect they do even though it is not easy to measure it by simple means. The result of the LED's relative lack of persistence is the potential for flicker, if the LED is driven by a anything other than a battery. How that flicker affects humans that work underneath such lighting is hinted at by earlier studies of florescent tubes that operated at 50/60 Hz and had noticeable flicker which caused headaches as well as more serious side effects. But, the subject at hand is really flashlights which usually have relatively short-use tasks. Concerns about health are really confined to eye damage which can be inflicted by exposure to high intensity light output, rather than flicker caused by a PWM driver.
It is interesting to speculate about whether PWM is philosophically desireable, more or less efficient, wasteful of energy and so on. These questions do not benefit from speculation. Instead, what is required is information relating to particular combinations of battery voltage, battery current capacity, specific LED emitters with known forward voltages and known heating co-efficients, and lastly the manufacturer's spec sheet on the electronic PWM chip itself. Given all this information on a particular set of components, various facts about average run time vs light output, emitter heating, residual heating in the electronics, etc can all be known. However, calculations for one set of parameters will not transfer to another group of flashlight components. Generalization just isn't possible. Manufacturers can provide these figures for their products, but few consumers will understand them. The closest thing to useful is still run time vs light output. Beyond that, get your test gear out and share the results with everyone else!
Finally I would like to point out that there is a fork in the PWM road that has been ignored so far. That avenue has to do with the physics of each particular LED emitter. Each part will have a maxium voltage that the emitter can tolerate, and a maximum amount of heat (generated by the emitter) that the emitter can stand without degrading its life and lumen output (under various conditions of ambient temperature and amount of heat sinking attached.) We really have 3 things that work to reduce light output over time. These are firstly normal aging of the emitter, which is a fixed quantity. Heat generated by the emitter is an important factor in overall life span. Thirdly, the application of higher voltages than the emitter can tolerate will shorten its life. Specifying "PWM" without qualifications ignores what the peak applied voltage is going to be, and it is the peak voltage that does the damage. The average voltage vs time can surely control the light output. It is the same average which can produce damaging heat in the emitter if it is poorly controlled. But, lots of heat-sinking won't do any good if the emitter physical structure is damaged by an over-voltage situation.
Thus, PWM applied to the available drive voltage (from the batteries) directly might not be a great idea unless the specs on the emitter are known. The LED may not blow up instantly (which is easy to do in any case), but its life span will definitely be shortened by over-voltage pulses. Some combinations of battery voltage and emitter forward voltage will work OK, others not so well. In any combination however LED's must be current limited. The limiting can be done with a simple resistor, which will also dissipate some heat (and thus waste some energy) or it can be done electronically by chopping the battery voltage. As a general rule, the chopping action of a PWM circuit is a good deal more efficient than a simple resistor by nature of how the voltage is supplied to the LED emitter. Voltage is never supplied at some intermediate level as it is when using a resistor. Switching transistors internal to the chips turn on and off very quickly, and spend little time acting as resistors. If the chip doesn't present a resistance to the current supply, it won't create a voltage drop and thus won't dissipate power (in the ideal case).
But returning to the over voltage theme. Sophisticated electronics includes more than simple PWM control. The use of tiny switching supplies is what allows a combination of specific emitters to be useable with flashlights having different numbers of cells. If the available voltage is too high for the emitter, the supply must buck (or reduce) that voltage to a useful level, not simply chop it up into little pieces. If the available voltage is too little the supply must boost (or increase) the available voltage to run the emitter. Part of a switching supply resembles a simple PWM circuit, but the overall result of this circuit - smps for short - is to regulate current by regulating the applied voltage. The emitter may still see short pulses of current, but a smps on a tiny chip can do a lot more, with more efficiency than either a resistor or a strict PWM chip. And the principal benefit of this type of supply (which is internal to the flashlight emitter assembly) is to limit damaging over-voltage pulses from getting to the LED. As always things aren't as simple as they first seem. Fortunately, lots of useful things can be done on the head of a pin.
King7
Carrying the idea of persistence a little further, I have yet to see convincing research that specifies how an LED behaves once the driving current has been removed. Phosphors of the type used in florescent tubes definitely have measurable persistance which allows the tubes to actually emit a relatively steady light output if their drive signal frequency is high enough. The same principal is the foundation of picture tubes of all kinds. Modern electronic florescent ballasts pulse the tube phosphors at 30KHz - 50KHz, faster than the phosphors' light output dies out. What we do know from simple testing is that LED's do not appear to have any persistence, but I suspect they do even though it is not easy to measure it by simple means. The result of the LED's relative lack of persistence is the potential for flicker, if the LED is driven by a anything other than a battery. How that flicker affects humans that work underneath such lighting is hinted at by earlier studies of florescent tubes that operated at 50/60 Hz and had noticeable flicker which caused headaches as well as more serious side effects. But, the subject at hand is really flashlights which usually have relatively short-use tasks. Concerns about health are really confined to eye damage which can be inflicted by exposure to high intensity light output, rather than flicker caused by a PWM driver.
It is interesting to speculate about whether PWM is philosophically desireable, more or less efficient, wasteful of energy and so on. These questions do not benefit from speculation. Instead, what is required is information relating to particular combinations of battery voltage, battery current capacity, specific LED emitters with known forward voltages and known heating co-efficients, and lastly the manufacturer's spec sheet on the electronic PWM chip itself. Given all this information on a particular set of components, various facts about average run time vs light output, emitter heating, residual heating in the electronics, etc can all be known. However, calculations for one set of parameters will not transfer to another group of flashlight components. Generalization just isn't possible. Manufacturers can provide these figures for their products, but few consumers will understand them. The closest thing to useful is still run time vs light output. Beyond that, get your test gear out and share the results with everyone else!
Finally I would like to point out that there is a fork in the PWM road that has been ignored so far. That avenue has to do with the physics of each particular LED emitter. Each part will have a maxium voltage that the emitter can tolerate, and a maximum amount of heat (generated by the emitter) that the emitter can stand without degrading its life and lumen output (under various conditions of ambient temperature and amount of heat sinking attached.) We really have 3 things that work to reduce light output over time. These are firstly normal aging of the emitter, which is a fixed quantity. Heat generated by the emitter is an important factor in overall life span. Thirdly, the application of higher voltages than the emitter can tolerate will shorten its life. Specifying "PWM" without qualifications ignores what the peak applied voltage is going to be, and it is the peak voltage that does the damage. The average voltage vs time can surely control the light output. It is the same average which can produce damaging heat in the emitter if it is poorly controlled. But, lots of heat-sinking won't do any good if the emitter physical structure is damaged by an over-voltage situation.
Thus, PWM applied to the available drive voltage (from the batteries) directly might not be a great idea unless the specs on the emitter are known. The LED may not blow up instantly (which is easy to do in any case), but its life span will definitely be shortened by over-voltage pulses. Some combinations of battery voltage and emitter forward voltage will work OK, others not so well. In any combination however LED's must be current limited. The limiting can be done with a simple resistor, which will also dissipate some heat (and thus waste some energy) or it can be done electronically by chopping the battery voltage. As a general rule, the chopping action of a PWM circuit is a good deal more efficient than a simple resistor by nature of how the voltage is supplied to the LED emitter. Voltage is never supplied at some intermediate level as it is when using a resistor. Switching transistors internal to the chips turn on and off very quickly, and spend little time acting as resistors. If the chip doesn't present a resistance to the current supply, it won't create a voltage drop and thus won't dissipate power (in the ideal case).
But returning to the over voltage theme. Sophisticated electronics includes more than simple PWM control. The use of tiny switching supplies is what allows a combination of specific emitters to be useable with flashlights having different numbers of cells. If the available voltage is too high for the emitter, the supply must buck (or reduce) that voltage to a useful level, not simply chop it up into little pieces. If the available voltage is too little the supply must boost (or increase) the available voltage to run the emitter. Part of a switching supply resembles a simple PWM circuit, but the overall result of this circuit - smps for short - is to regulate current by regulating the applied voltage. The emitter may still see short pulses of current, but a smps on a tiny chip can do a lot more, with more efficiency than either a resistor or a strict PWM chip. And the principal benefit of this type of supply (which is internal to the flashlight emitter assembly) is to limit damaging over-voltage pulses from getting to the LED. As always things aren't as simple as they first seem. Fortunately, lots of useful things can be done on the head of a pin.
King7
Last edited:
