I'm trying to make a little carry lantern which uses a super bright 3.7V LED at 1Watt (those hex shaped ones) powered by a Li-ion 3.7V battery. Now I've seen the idea of pulsing an LED to reduce the overall power consumption on the battery but I'm wondering if there is any way of doing this without the use of any IC.

I've read up on something such as an avalanche transistor which seems to be pretty close on what I'm aiming for. I've noticed people talk about the fact that during pulses you can go over the LEDs rating depending on the duty cycle (IE 50% cycle can go to 200% of the LEDs current rating) as long as you have a way of mitigating the heat properly.

I'm not too bothered if I over supply the LED and reduce it's lifetime. I would really like to know how to easily pulse an LED without an IC (or easy to build). Could I also control the current to the LED?

  • \$\begingroup\$ Do you realize that by pulsing you will reduce the intensity of the light? If you do, than why not just use a less powerful LED and save the power by that? \$\endgroup\$ – Eugene Sh. Nov 21 '16 at 17:36
  • \$\begingroup\$ Well I know that in the case of some head lamps they use the idea to create very bright lights that use much less power. They pulse the LED at such a high frequency that the loss in current consumption compared to the loss of light makes it worth it. Sort of like how the idea of a Joule thief makes an LED flicker but to us it looks like its on constantly. Sorry if I'm been simple but it's the best way of explaining myself ^.^ \$\endgroup\$ – Kuera Nov 21 '16 at 17:41
  • \$\begingroup\$ Kuera, read my post here: electronics.stackexchange.com/questions/267674/… Those laws regarding human perception of intensity are pretty well established. The Joule Thief is terribly inefficient. It's claim to fame is the simplicity and the fact that it is fairly bulletproof, so poor choices in toroid or bad wiring, etc., and it still mostly works. It is true that at LOW frequencies (not high), that we perceive somewhat brighter than expected results. So you might get some "efficiency" there -- but you will notice the blinking. Is that okay? \$\endgroup\$ – jonk Nov 21 '16 at 17:58
  • \$\begingroup\$ I know using the the term "Joule thief" was bad when I typed it. Bad example I know. very interesting read that. I noticed at the end you said that the higher the pwm the better for vision sake. I don't mind experimenting a bit if there are some factors I could change to alter the outcome of said frequency or current. so yes noticing the blinking wouldn't bother me as I intend to probe and mess around with the idea for the sake of learning if nothing else. Thanks for the reply and the read. \$\endgroup\$ – Kuera Nov 21 '16 at 18:13
  • \$\begingroup\$ Our eyes integrate so, generally, the apparent brightness of a light source stays constant if the average power into the light source stays constant, which means that using a readily available pulsed constant current driver to flash the LED will dim it Why do you not want to use \$\endgroup\$ – EM Fields Nov 21 '16 at 18:23

I'm going to recommend that you simply not worry about efficiency, right now. You've agreed that it could work to use a flicker frequency that is noticeable. And this means you need to do some experiments first before you can settle on a workable solution. Frankly, I'm not even sure there IS a workable solution for you. But there might be. Before I go there, just a short comment about why operating an LED with PWM, broadly speaking, won't help you at all:

As I wrote in this link, PWM frequency for LED dimming, there's quite a number of studies about human perception of intensity when varying frequency and duty cycle. The Talbot-Plateau says that if a light source blinks fast enough so that it passes beyond the critical flicker fusion frequency (CFF), then the brightness will appear to be the same as if the light source were steadily operated at the time-averaged luminance. The upshot of this is that you won't gain any operating efficiency from operating an LED at a frequency past the CFF. For example, if you use a duty cycle of 50% and a frequency of 200 Hz, and operate the LED at some luminance (current \$I_x\$, for example), then it will appear to have the same luminance as that same LED when operated with DC at half the current, or \$\tfrac{I_x}{2}\$.

In short, no efficiency gains here. In fact, you'd be worsening the efficiency because a semiconductor switch drops some power, itself, as does whatever is operating it.

But there is a possibility trapped between the Broca-Sulzer effect and the Bruecke-Bartley effects, which you might consider exploring for your own purposes. I hadn't discussed these effects much in the above-mentioned link of an answer I gave earlier. So now might be a time to add a few more details to help explain.

The Broca-Sulzer effect found that \$50\:\textrm{ms}\$ flashes appeared to be some 30–40% brighter than \$50\:\textrm{ms}\$ flashes of equal luminance. (This effect was broaded in scope by Ganzfeld's results, which aren't important here.) So, here it seems that the effective brightness begins to rise when the frequency rises from 1 Hz towards 10 Hz.

The Bruecke-Bartley effect covers the other direction, this time lowering the frequency from the CFF. It found that the effective brightness begins to rise when the frequency declines from the CFF and towards 8 Hz to 10 Hz.

The technical details of these two effects would cause a practitioner in these fields to want me to be a lot more precise about what the results say and don't say. But that's a rough summary for the purposes here.

In short, it would seem to be the case that somewhere around the area of 8-10 Hz, you could expect some human perception effects to allow you to use less average power for a similar visual sensation.

This is why I'd suggested you consider this as an option, if you are searching for improved efficiency through PWM. And it would seem, perhaps, that you might be able to expect something like a 30% improvement. That is, if the flicker effect isn't too annoying to use. So you do need to experiment here.

Efficiency isn't ONLY found through PWM. LEDs themselves do exhibit different efficiencies in converting electrical energy into light. And humans are vastly more sensitive to wavelengths near \$550\:\textrm{nm}\$ than to blue or red, for example. But I'm assuming here that things being equal on these points, you want to explore how PWM in the right range might also help out.

But I want to point out that you should have already done the work needed to identify your best LED option before starting down that path. So if you feel you have already got the right LED, then I think you can consider the idea of exploring the boundaries between the Broca-Sulzer effect and the Bruecke-Bartley effect.

For such experiments, I think it would be helpful to have a way to easily vary the frequency from about 1 Hz to close to the CFF. The CFF itself varies a bit when the light source or your eyes cause the light to hit different parts of your retina, over time. So, personally, I'd want the ability to go from 1 Hz to 100 Hz to cover most situations. Plus, I'd want a push button that would immediately over-ride the flicker frequency and go straight to 100 Hz, so that I could quickly make comparisons that way. I'd also want the ability to vary the duty cycle (mostly for fun.) But in your case I think a fixed 50% duty cycle would be good enough for the experimental purposes and you don't actually need a control for that.

Regarding current? If you accept my suggestion for only 50% duty cycle, then you probably just need two control currents -- one at the standard DC current recommended for your LED and one at twice that much, when operating at 50%. I don't think there would be serious harm in allowing you to set it for twice-normal even when operating at 1 Hz (or DC.) The worst that happens is that you damage the LED and those are cheaply replaced.

But you need to figure out what experiments you want to perform. You can read what I wrote here and consider that. Or come up with your own ideas. But either way, this can lead you to a specification that I think you could post here for a useful reply.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.