Logistics of driving a high-powered LED for low-power applications

Thread Starter

xox

Joined Sep 8, 2017
608
I'm working on a project right now that has some very low power consumption requirements. I had originally planned to use ~20 ma LED's, but now it looks like my only option is the ~350 ma variety. I was thinking I could perhaps pulse them at a very high frequency using a 5% duty cycle, but I'm just not sure that they would still be visible. Could that possibly work, or maybe there's some other way that would draw an equivalent amount of power?
 

dendad

Joined Feb 20, 2016
3,939
Why are you constrained to using a 350mA LED now?
Some more info would help. What is your power supply limits. And is the LED an indicator or for illumination?
The LEDs can be pulsed at low duty cycle quite ok and in fact that is a good way to get a brighter light from them on restricted power. But we still need to have details.
 

Thread Starter

xox

Joined Sep 8, 2017
608
Why are you constrained to using a 350mA LED now?
Some more info would help. What is your power supply limits. And is the LED an indicator or for illumination?
Those were the only one's I could find capable of producing the range of acceptable wavelengths (roughly 660 nm up to just short of near infrared). Power supply current limit is basically 20-30 ma. LED will essentially provide three functions: transmission, detection, and illumination. The first two will require very little current to operate, the illumination will consume the vast majority of the circuit's power.

The LEDs can be pulsed at low duty cycle quite ok and in fact that is a good way to get a brighter light from them on restricted power.
Okay, thanks, that's a relief to hear. I can just experiment with the frequency setting, but do you off chance happen to have any idea what sort of frequencies we're talking about here? I realize of course that it may be different for various wavelengths, but a ball park guess might be helpful in any case. :)
 

dendad

Joined Feb 20, 2016
3,939
Put a large capacitor as a reservoir so you can pulse the LEDs and not pull the power supply down.
I'm doing similar as I'm having a go at a laser tag system for my son.
The pulse frequency is on no concern with respect to the light wave length as they are so far apart, but will be determined by your "transmission" as it sounds like you are sending data.
The laser tag will be packets of 40kHz and received by TSOP1738 IR receive modules. I'll use there LEDs..
http://www.ebay.com.au/itm/3W-3535-...-10pcs-/162019790228?var=&hash=item25b921cd94
An easy way to try it out is to use an Arduino PWM out and increase the PWM duty cycle while looking at the current draw. Stop when you get to your 20mA.
 

Wendy

Joined Mar 24, 2008
22,646
Something I like to point out is you never have to drive LEDs at full power. Most cases they will performe well with a fraction of their rated current. As a bonus they last much longer too.
 

crutschow

Joined Mar 14, 2008
28,156
Also note that for maximum efficiency you want to use a switch-mode current drive for the LEDs with no series resistors to dissipate power.
You can use PWM as long as you use an inductor as the current limit element, not a resistor.
I can just experiment with the frequency setting, but do you off chance happen to have any idea what sort of frequencies we're talking about here?
If all you are concerned about is visible flicker than anything above a 100Hz or so should be fine.
Higher frequencies will however, allow the use of a smaller inductor for a given average LED current if using a switched-mode supply.
 

DickCappels

Joined Aug 21, 2008
7,953
Provided the frequency is not so low that flicker can be perceived but not so high that the turn-on time of the LED is not a factor (MHz) the perceived luminance will be proportional to duty cycle. This means for a given perceived brightness you need the same average current as you would need with DC drive.

Related to Wendy's comment on current, LEDs age as a nonlinear function of current so with a duty cycle lower than 100% you get fewer lumen-hours over the life of the LED.
 

Thread Starter

xox

Joined Sep 8, 2017
608
Also note that for maximum efficiency you want to use a switch-mode current drive for the LEDs with no series resistors to dissipate power.
That makes sense, a resistor is only going to eat up precious power after all.

You can use PWM as long as you use an inductor as the current limit element, not a resistor.
Sorry, so you mean the constant current source circuit supplying the LED's itself should be implemented with inductors instead of resistors?

If all you are concerned about is visible flicker than anything above a 100Hz or so should be fine.
Yes, that would be great if flicker could be reduced or eliminated, but my major concern here is simply visibility (a 5 or 6 meter radius would be acceptable).

Higher frequencies will however, allow the use of a smaller inductor for a given average LED current if using a switched-mode supply.
I wonder if an op amp based gyrator in the inductive configuration would suffice? Or wait, that would only model the current/voltage phase relationship of an inductor - not the magneto-resistive characteristics, correct?
 

GopherT

Joined Nov 23, 2012
8,012
Also note that for maximum efficiency you want to use a switch-mode current drive for the LEDs with no series resistors to dissipate power.

You can use PWM as long as you use an inductor as the current limit element, not a resistor.

If all you are concerned about is visible flicker than anything above a 100Hz or so should be fine.
Higher frequencies will however, allow the use of a smaller inductor for a given average LED current if using a switched-mode supply.
Could you please clarify Sentence in bold. That is only if the user is implementing a switch mode constant current source AND PWM, correct?
 

Thread Starter

xox

Joined Sep 8, 2017
608
Provided the frequency is not so low that flicker can be perceived but not so high that the turn-on time of the LED is not a factor (MHz) the perceived luminance will be proportional to duty cycle.
Or would it be the other way around? I mean, pulsing an LED at 350 ma for 1/17th of a second per second would average out to ~20 ma over time, but obviously that wouldn't be so much visible (if at all). On the other hand, pulsing at a higher frequency for a smaller fraction of time should saturate the receptors in the eye in such a way as to achieve a persistence of vision effect, wouldn't it?

Related to Wendy's comment on current, LEDs age as a nonlinear function of current so with a duty cycle lower than 100% you get fewer lumen-hours over the life of the LED.
Interesting, well that's rather counter-intuitive! Is there a fairly accurate approximation formula for this effect on life-expectancy of the bulb, or is it just something that's heuristically determined on a case by case basis?
 

crutschow

Joined Mar 14, 2008
28,156
On the other hand, pulsing at a higher frequency for a smaller fraction of time should saturate the receptors in the eye in such a way as to achieve a persistence of vision effect, wouldn't it?
Once you get above the flicker frequency of the eye, the frequency has little effect on the perceived brightness (for a given duty-cycle) since the eye receptors act pretty much like integrators of the light photons.
 

Thread Starter

xox

Joined Sep 8, 2017
608
Could you please clarify Sentence in bold. That is only if the user is implementing a switch mode constant current source AND PWM, correct?
Just to be clear on my part, I'm talking about a battery-powered supply that if anything will be stepping up it's voltage, and straight DC.
 

Thread Starter

xox

Joined Sep 8, 2017
608
Once you get above the flicker frequency of the eye, the frequency has little effect on the perceived brightness (for a given duty-cycle) since the eye receptors act pretty much like integrators of the light photons.
It was determined by looking at data sheets.
But then wouldn't a 5% duty cycle be too faint to see? I just don't understand now how that could possibly work. :confused:
 

DickCappels

Joined Aug 21, 2008
7,953
But then wouldn't a 5% duty cycle be too faint to see? I just don't understand now how that could possibly work. :confused:
That is only a reduction of 95%. Whether or not you can see it depends on how bright it would have been at 100% duty cycle, the ambient illumination, the extent of adaptation of the illumination, and the contrast with whatever surrounded the LED.

The best and only reliable way to know how hard to drive a given LED is try a design and see how it looks under all viewing conditions.

Incidentally, everything else being equal, green LEDs give more luminance per watt than other colors. The eye is most sensitive to wavelengths around 555 nm in the daytime and 505 nm under very low light levels.
 

Thread Starter

xox

Joined Sep 8, 2017
608
That is only a reduction of 95%. Whether or not you can see it depends on how bright it would have been at 100% duty cycle, the ambient illumination, the extent of adaptation of the illumination, and the contrast with whatever surrounded the LED.

The best and only reliable way to know how hard to drive a given LED is try a design and see how it looks under all viewing conditions.
You're right, I think it's high time to do a little experimentation. And cruschow's most excellent PWM circuit looks to be a good tool for that job. Now if you'll excuse me...:)
 

Tonyr1084

Joined Sep 24, 2015
6,451
On "Flicker": Cheap cameras may record at 28 FPS (Frames Per Second). Any slower and the viewer of said video will see flicker. 30 FPS seems to be more pleasing to watch. Any faster and you're just using up film (or bits or whatever medium you're recording onto). I don't think you're going to see flicker at 100 times per second. Even if the duration of the flash is extremely low.
 

DickCappels

Joined Aug 21, 2008
7,953
Another artifact is the speed of motion of a flickering image across your retina. I have seem the tail lights of many cars with pulse width modulated tail lights so they can switch between normal running and braking, bouncing down the road at night, leaving a trail of dots on my retina.

Similarly, our eyes move in little jerky saccades which can easily exceed 10,000 degrees per second, each little saccadic jump can accentuate temporal changes in illumination or color. This is why field sequential color displays have always seemed weird to those who have to use them.

To avoid these artifacts refresh rates need to be very high, depending upon conditions. The more light falling on the retina the more sensitive to flicker.
 
Top