Why are LED pulsed?

But each LED will then be (no more than) 1/10th as bright as if it was driven at 100% duty cycle at the same current. To get them back up to the same brightness as if each were driven at 100% duty cycle, you have to hit them with 10 times as much current during their 10%.

Talking about electrical power without talking about light output is pointless. If you ignore light output you could say that lowering the current to zero would save all the power, but what is the point of that observation.

At 50 percent you have to doublee the current to get the same brightness,so where is the advantage?? Or do you accept halve the lightlevel?? Because thats wat you get .

because

the

I can also throw another one at you that you will not believe. The driver chip can only dissapate so much current so the leds are pulsed for that reason.

AS to the first part, the eye is tricked into believing the led is on all the time at the same or near the same brightness if the led is pulsed. That trickery lets you see a motion picture from film.

Assume you have 10 LEDs, all appearing to be on at once. But if you drive them one at a time for < 10% of the time you have reduced the drain.

A while back (a couple of years ago) I bought up every copy of the HP Optoelectronics Fiber-Optics Applications Manual, 2nd ed, that I could find -- including other countries. Then shipped them to interested parties. But you might look for another copy out there.

It does discuss the idea that LEDs will have a point of maximal efficiency in pulsing, when considered from human perception (and not from examining radiant power with instrumentation.) If you get a copy in hand, look to page 5-20. They provide a curve of "relative luminous efficiency" and it illustrates an example where you might see about some improvement in brightness perception by multiplexing and using higher currents. Not a lot, though.

And they only provide the one curve and it is probably for some old red LED they had at the time. So this isn't anything but a data point for a particular class of LEDs, I suspect. If that.

For example, assume your LED is spec'd for 5mcd @ 10mA. You choose to pulse it at 40mA at 25% duty cycle, so the average current is the same as the continuous would have been. They would show calculations like this:

10mA 1.58 5mcd * ------ * ------ = 6.4mcd 10mA 1.24

So they say. The last fraction represents picking two different numbers off of that chart, for pulsed versus continuous. By the way, that curve pretty much flattens off at about 1.62 or so, no matter what you pulse at as you go higher than about 50mA. So you get nothing more after that, according to the chart. And remember they are saying this is about perception, not instrumental radiance.

But this was all two decades ago. This is around the same time that Edwin Land was reporting out a whole new perspective on color perception in humans. There has been a lot of progress since this time in the science side of things and I'd bet there is much better information to be had, today. And more authoritative as well as being less ad hoc and more theoretical.

Jon

I should have also mentioned (because I had written a note in the margin about this) that the drive voltage also increases slightly with increased current (v = (kT/q)*ln(1+I/Is), or something like that.) Usually not much more than 5% or 10% more, though, for factors of 4X or 5X.

What HP may have noticed could have been a due in some part to a shift in emission wavelength distribution towards the shorter end where our eyes are more sensitive (maximal somewhere in the 555nm area, I think.) If not all of the slight increase in voltage is simply ohmic overhead due to the larger current, and some of it might actually lead to an even slighter increase in activation voltage/energy, then it's a possible consideration. Can't say, though. HP doesn't spend much time documenting their methods in understanding this phenomenon.

Jon

I should also note that I'm not suggesting much change. For certain places in the spectrum there is quite a shift in sensitivities in the red area with just a 1nm change. For example, there is about an 8% improvement in sensitivity in just shifting from 690nm to 689nm.

That's the order of change I was wondering about.

Jon

Journal

don't

I also have a vague recollection that the early, deep red LED displays had an energy loss mechanism that competed with photon production, but the proportion of energy that went to photon production increased at higher drive current. So multiplexing the LEDs improved their current to photon ratio, slightly. But today's LEDs are quite linear, down to unusable light levels so this effect is no longer operational.

Many LEDs do have a dominant wavelength change with respect to current, and, as you say, at some wavelengths, this can have a significant effect on their visibility. Today, with so many wavelengths to choose from, you might as well just select a device that produces the desired wavelength. But this factor is something to be aware of.

It would have helped some if I'd written this as:

(40mA*.25) 1.58 5mcd * ------------ * ------ = 6.4mcd 10mA 1.24

That way you can see better how they say to derive that top value.

Jon

The main reason I mention this is that it's important to be aware of possible ways by which an apparent brightness change might occur. Otherwise, in ignorance, all manner of random guesses about it might be supported by apocryphal observations. I'm in full agreement with your general points here on the subject, but knew of a few details that might explain an authoritative source for argument.

In any case, if those disagreeing with you intend to say that the human eye is some kind of peak detector, they need to show this through both theory and experimental result. On theory, I don't know of a mechanism by which 'peak detection' would be selected in human perception -- and given that what would in fact _be_ selected for is the ability to see over a very wide range of lighting situations, from no more than star light to broad daylight and specular reflections, it's easy to understand why there is a rough log behavior in response

-- a linear system would be overwhelmed and useless or underwhelmed and useless, most of the time. I see no reason for peak detection in photopic vision. And given that some animals have acquired a tapetum in their retina in order to enhance scopotic vision, I don't find much argument for a mechanism there, either. Finally, I _have_ spent time playing with various pulsing schemes and have found very little upon which to base such 'peak detection' arguments, despite the perennial rumors about it. It just doesn't seem to help one way or another.

Which made me wonder a little about wave-shifting effects, because it doesn't take a lot of that in some cases to have an apparent change in intensity perception. But I'm sure all this has been researched and is available somewhere in the literature. I just haven't cared enough about it in practice to go look it up. If anyone has citations to some peer-reviewed science papers on the subject, I'd be interested.

And with the desired efficiency.

Mostly in just being aware that there may be more than one source for finding an apparent shift in brightness without changing the averaged drive current. I don't think pushing wavelengths slightly factors into much of anyone's design.

Jon

Yes ,and the light level is also 10 percent. As luck would have it,your eye does not interpreet in a linear way, so even a reduction to 10 percent might not look like much if you judge by memory. If you test 2 however with the same resistance and pulse one ,you will see the difference.

I dug out the article in question; it was published in the HP Journal in June 1972. Here's a quote:

"It was apparent early in the HP-35 planning that new display techniques would be required. Existing light-emitting-diode products used too much power and cost too much. HP Associates developed a magnified five-digit cluster which saves both power and cost and is packaged in a convenient

14-pin package (Fig. 5). Each digit has a spherical lens molded in the plastic over it. A slight reduction in viewing angle results, but for the handheld calculator this is not a problem.

LED's are more efficient if they are pulsed at a low duty cycle rather than driven by a DC source. In the HP-35, energy is stored in inductors and dumped into the light-emitting diodes. This technique allows a high degree of multiplexing..."

Of course, this is for LED's of the 1970 era, and there was no further explanation than the statement you see in the quote. Nothing about just how much more efficient pulsed drive might be.

This is largely a myth.

Many LEDs and most (not all) LED digital displays appear brighter for equal average current (of no more than a few milliamps per LED chip) when pulsed than when fed continuous current. However, the relevant nonlinearity is in the LEDs and not in human vision. Human vision has a significant nonlinarity, but that is after an "averager" that works impressively well when the pulse rate is fast enough to avoid visible flicker.

I have a web page on this -

- Don Klipstein ( snipped-for-privacy@misty.com)

I did experiments in this area. The second explanation is nearly enough the whole story.

I know a few LEDs that when pulsed look brighter to then the human eye than they do than when fed steady current for same reading by a solar cell

- but only when higher instantaneous current shifts their spectrum towards wavelengths that the human eye is more sensitive to. And most LEDs that do this noticeably are less efficient at higher currents.

- Don Klipstein ( snipped-for-privacy@misty.com)

Journal

don't

With LEDs that appear to get brighter by increasing instantaneous current without increasing average current, I have found radiometric increase also. The LEDs get more efficient.

The efficiency gain was at least mostly in the LED and not in human vision.

- Don Klipstein ( snipped-for-privacy@misty.com)

Journal

don't

I suspect that manufacturers have developed standard drive levels that they spec things at, on the basis of this kind of knowledge.

But the HP book I'm citing from specifically points out that the curve they present is __NOT__ congruent with radiometric changes. So that wasn't their point. (And I'm not defending what they said then about parts available now [or then], either. Just noting it.)

They weren't precise about what they meant, Don. It's possible that there was a slight wavelength shift attending the change and that this coupled with human perception -- and that might fit within what they were saying. Here's the quote: "Equation 5.2.4-1 applies to luminous intensity as perceived by the eye and is not applicable to radiant power output."

Frankly, I think this is much-ado-about-nothing-useful, though.

Do you have the book I mentioned?

Jon

That's been my experience.

The quote I provided is the only time I ever saw mention of such an effect.

I do remember that both of these eefects were conventional wisdom in the 1970s, and that they applied to LEDs made from GaAs and GaP. I don't remember whether I read it in manufacturer data sheets or in articles in trade magazines like Electronics.