# Pulsing LEDs - possible higher efficiency ?

#### Klaus

##### Flashlight Enthusiast
One of these days in my local electronic shop - the one I visit regularly now since gotten bitten from the CPF virus - we had a little talk about LEDs - white LEDs in particular - suddenly the guy mentions the little circuit they came up with to get 3 times the brightness at just a "little" higher current.

They had been running the LED at 100ma pulsed with 1/16 of 1Khz (62,5 hz) and measured the brightness at 3 x times.

Nobody was really exactly remembering the actual current this setup was consuming but everybody was somehow shure it was just a little bit more and that the increase in current was way lower than the increase in brightness. Also someone remembered something like a little higher voltage they had to run it at when pulsing it ...

He handled me a datasheet of a white LED which looked pretty much like the Nichias except it mentiones another duty cycle for the pulsed current.

Datasheet reads : I(FP) 100ma duty 1/16 1Khz

Nichia Web site: I(FP) 100ma; Pulse width <= 10ms, duty ratio <= 1/10

As I sometimes feel as having passed this part of my education on roller-blades I have no real idea how this data relates ....

But believing these guys it seems that if you spent a little time on this that every LED has a point where the right mix of pulsing frequency and voltage should give a benefit in brightness/consumption compared against the standard 5600mcd/20ma.

I don´t know if this is old and common knowledge and that the circuitry for pulsing is just too complex / big and therefor uneconomical for flashlight use or if this really is interesting for the LEDaholics around here - just wanted to share

Regards

Klaus

#### MrAl

##### Flashlight Enthusiast
The 'overall' efficiency always drops when
you pulse the LED. For 50% duty cycles
it doesnt drop too much, but it does
drop.

For example, if you drive an LED
at 20ma constant dc current and
drive another LED at 20ma average
current at 25% duty cycle(which takes
4 times the peak, or 80ma peak)
the efficiency at 80ma maybe puts out
about 3 times the light energy, but since
its on only 1/4 of the time, to get the
light energy total you have to divide that
by 4, so you end up with 3/4 or 0.75 times
the light output that you would have gotten
if you just ran it with a smooth dc current.

Since this is an interesting subject and
see is believing, you might wish to try
doing a comparison between an LED driven
at 20ma smooth dc and an LED driven at
80ma peak dc for 25% duty cycle. All
you have to do is get two led's and
hook them up, one pulsing and one
not, and compare the light by shining
them both on a white wall or ceiling
at the same distance from the ceiling.
If you want more quantitative results,
build this simple light meter:
http://hometown.aol.com/xaxo/page1.html

and compare the two outputs. Just to make
sure both led's you use are the same,
run them both at 20ma first and check
for equal light outputs, then run them
under pulse/non-pulsed and compare
them again, then switch the two (pulse
the one that was not pulsing in the
second comparison and dont pulse the
one that was pulsing) and compare them
for a third time to verify that they
both run the same under pulsed conditions.

Let us know how you make out?

Good luck with it,
--Al

#### Klaus

##### Flashlight Enthusiast
Dear Al,

as I had mentioned already - my electronic knowledge is quite limited - therefore I already would run in problems to have to design the pulsing circuit - so far for that

When posting the thing my intention was to possibly give some (valuable?) input for some of the more experienced guys like you (or one of the others working with you on the driver circuits) - as I wrote this is what I got from that shop and the guy talking to me was one of the owners and they are quite good on what they are doing (as far as I can tell
)

Following your points would just be the opposite of what they seem to have tested - higher brightness at less relative consumption - higher efficiency - your points would lead to higher brightness at higher relative consumption - lower efficiency.

I guess this really would needed to be tested and measured for a definitive answer - or some answer from someone who looked into this already - I could imagine Peter G. from ARC having looked into this already - but he might not be willing to share his knowledge with some competitors also lurking here.

Klaus

#### Gransee

##### Flashlight Enthusiast
Klaus, welcome to the CPF!

MrAL's numbers are right on the money in my opinion. There is no gain in flashlight brightness from pulsing an LED compared to running it with DC. There is at least one other thread here covering this topic plus other websites on the web if you want to read about this further.

Peter Gransee

#### Klaus

##### Flashlight Enthusiast
Thanks MrAL, Peter G. and PeLu on clearing this up for me and possibly others as well - and sorry if this was discussed in another thread I overlooked.

Klaus

#### PeLu

##### Flashlight Enthusiast
<BLOCKQUOTE><font size="1" face="Verdana, Arial">quote:</font><HR>Originally posted by Klaus:
They had been running the LED at 100ma pulsed with 1/16 of 1Khz (62,5 hz) and measured the brightness at 3 x times.
<HR></BLOCKQUOTE>

Increasing brightness by pulsing LEDs is a popular myth.
On early red and IR LEDs it was true that they had a better efficiency at higher currents. For our white LEDs it is the opposite.
Other sources say that the perceived brightness is higher with pulsed light. But also this is not true for illumination applications.
See Don Klipstein's page for further info.

In your friend's setup the flaw may have been in the measuring.

Mral's estimation about getting about 75% efficincy in between 20 and 80 mA is even a little bit underestimated. Efficiency for a normal Nichia LED will be from 30 to 6lm/W between 1 and 100mA. This is for DC and half of the efficiency loss is due to the higher junction temp, so it will not be that bad at a pulsed current.
And when you run your LED at 80mA it will need a higher voltage (and more power therefore) than at 20mA due to resistive losses.

But some lights pulse their white LEDs for another reason:
They just change the duty cycle to dim the LED and still get the same light colour.

#### PeLu

##### Flashlight Enthusiast
I just remebered another case:
Using Nichia white LEDs, you could dim them in two ways:
1. decrease the current (voltage)
2. pulse them and leave the current as it is.

The second one was made for some high end LED lights to keep the light colour constant. It was explicitly stated that this was done although it decreases efficiency.

#### papasan

##### Enlightened
not being an engineer (least not of the electrical kind =)) i may be way off base, but...

doing research before on making a specialized headlamp (for caving) that was dimable, i found that using pulse width modulation (changing the pulses duty cycle to change light output) was *much* more efficient than using something like a variable potentiometer (sp?) or a series of resistors with switchs and whatnot...

#### arab

##### Enlightened
<BLOCKQUOTE><font size="1" face="Verdana, Arial">quote:</font><HR> I just remebered another case:
Using Nichia white LEDs, you could dim them in two ways:
1. decrease the current (voltage)
2. pulse them and leave the current as it is.
The second one was made for some high end LED lights to keep the light colour constant. It was explicitly stated that this was done although it decreases efficiency.
<HR></BLOCKQUOTE>
In the case of the Photon 3, the 2 levels of reduced light output are produced by pulsing the LED at 2 different frequencies. But if this method "decreases efficiency" then why don't they just reduce the voltage to achieve the same affect? If ARC can do this with their AAA (albeit increase rather than reduce) in such a small light, why not LRI?

Please bear in mind that I am not too well versed in electronics ...

#### php_44

##### Enlightened
<BLOCKQUOTE><font size="1" face="Verdana, Arial">quote:</font><HR>Originally posted by arab:

In the case of the Photon 3, the 2 levels of reduced light output are produced by pulsing the LED at 2 different frequencies. But if this method "decreases efficiency" then why don't they just reduce the voltage to achieve the same affect? If ARC can do this with their AAA (albeit increase rather than reduce) in such a small light, why not LRI?

Please bear in mind that I am not too well versed in electronics ...
<HR></BLOCKQUOTE>

There are several factors involved here:
How efficient the LED produces light at a given current.
Nichia whites are usually rated at 20mA.
<UL TYPE=SQUARE><LI>More than 20mA produces more light at less efficiency. (so one LED at 40mA produces less light than two LEDs at 20mA each)
<LI>Less than 20mA produces less light but at higher efficency. (so two LEDs at 10mA each makes more light than one LED at 20mA)
[/list]

How efficently battery power is delivered to the LED:

<UL TYPE=SQUARE><LI>For dimming, pulsing (pulse width modulation) is very efficient for delivering battery energy
to the LED. As long as the current to the LED is 20mA or less, this is a very efficient way to convert battery energy to light. If the LED is being pulsed at >20mA then you loose LED efficiency.

<LI>If you dim via a resistor or linear regulator then you are loosing a good deal of battery power in the resistor or regulator. This lost power is not being used to make light - it instead produces heat. This is generally not a good method to use, though it is simple and inexpensive.
[/list]

Now, the photon 3 has a complicated tradeoff at work. I'll bet that they simply pulse the current to the LED at whatever level the battery will deliver. This may well drive the LED over 20mA reducing the LED efficiency, but it IS more efficient than dimming the LED via a resistor. At least they are not using your expensive lithium cells to warm a resistor. So the photon 3 is minimizing waste.

Separately, there are lights that must match the battery voltage to the LED(s) being lit. If the light has one 1.5V cell (like the arc or infinity) then a voltage boosting circuit is used to generate the correct voltage to drive the LED. The most common way to do this is to use the inherent characteristics of an inductor with a circuit that alternately charges the inductor, then forces it to discharge into the LED. This is called a switch mode power supply. They can produce a pulsed output, but generally are constructed to produce a rather filtered and smooth output. Because this type of circuit can be designed to power the LEDs at their most efficient current level, this type of circuit can be very efficient. Most lights, though, power the LED at much more than 20mA to produce more light - regardless of the reduction in LED efficiency.

Hope this helps!

#### arab

##### Enlightened
Thanks for the very detailed reply!

One question - According to the Light Technologies web site , the PAL uses a "Current Controller" to vary light output http://www.lightechnology.com/history.htm. So why didn't they go for pulsing?

D

#### **DONOTDELETE**

##### Guest
It's very likely their current controller uses pulsing, since pulsing is the most efficient way to control average LED current.

Since fast pulsing isn't visible, there's no reason for them to say "we use pulsing".

In fact, saying so might be a turn off to certain buyers concerned that the PAL blinks or flickers.

#### Chris M.

##### Flashlight Enthusiast
It's very likely their current controller uses pulsing, since pulsing is the most efficient way to control average LED current.

Actually it uses a linear regulator- a little surface mount LM317. I doubt they much care for efficiency though, since those things drain your batteries down all the time anyway with that always-on thingy, so what`s an extra hour or so lost due to the regulator....

D

#### **DONOTDELETE**

##### Guest
Wow, a linear regulator with a 9-volt battery? Almost as bad as using resistors.

#### Chris M.

##### Flashlight Enthusiast
I think there`s resistors in there as well! One high value one bypassing the switching circuit (for that annoying always-on-can`t-turn-it-off mode) and, I believe, one to do main current limiting to the LED- with the 317 providing dimming only. I don`t know for sure, I havn`t seen a schematic and it`s all so small.....