# Thread: Constant Current vs. PWM dimming Revealed

1. ## Re: Constant Current vs. PWM dimming Revealed

THAT is great data!

I remember a patent that was granted for using PWM for tint management in dimming LED's. Can't seem to find the patent now

2. ## Re: Constant Current vs. PWM dimming Revealed

evan - thanks for the clarification. All PS designs I've ever seen have output filters (not for flashlights tho). I'm surprised filters aren't used for flashlight drivers...the cost of a cap is minimal.

3. ## Re: Constant Current vs. PWM dimming Revealed

This is a great thread. Thanks for bumping it

But i think a basic point is missed in all this analysis. Of course, PWM is less efficient than CC dimming talking in radiometric units.

But it not necessarily mean it is less efficient in the whole visual effect.

Think you are in front of the flashlight, and at 100% power, you receive and perceive 50 cd. With CC dimming at 50%, you receive and perceive, say for example, 30cd (due to increase efficiency).

But, if you dim at 50% by PWM and the frecuency is enough high, you receive in average, 25cd (using the corresponding 100% power current in the light pulse). But you still notice 50cd of light, with flickering (depending of the frecuency used). This is why PWM was so widely used in the past, specialy with red leds, wich increase light output in a near linear way with increasing current, so the efficiency loss is only due to higher voltage applied during the pulse.

So a well designed PWM may be more efficient, in terms of human perception, than CC dimming.

Hope you understand what i want to mean with my reduced english.

4. ## Re: Constant Current vs. PWM dimming Revealed

But, if you dim at 50% by PWM and the frecuency is enough high, you receive in average, 25cd (using the corresponding 100% power current in the light pulse). But you still notice 50cd of light, with flickering (depending of the frecuency used).
No, this is not true. You still percieve 25cd. Above a certian frequency your eyes integrate the total output into a non flickering beam equal in brightness to the average of the pulses and off times. Everyone keeps saying that your eye see the maximum brightness of the pulses, but it is simply not true.

Think about this - if at lower duty cycles, PWM still appeared as bright as a single pulse, just how would PWM achieve dimming at all? Ponder that question and you will realize that your eyes average everything out.

5. ## Re: Constant Current vs. PWM dimming Revealed

Good reasoning, evan9162, you convinced me.

As you said, everybody says it, and its wrong. I read it so many times i believed it without questioning.

Its possible its one of the false things with, due to repetition, sounds true.

Or maybe it depends of the frecuency used?

(im going to research in this topic a bit more, as im not sure human eye averages out all, in my understanding, eye response isnt always linear)

6. ## Re: Constant Current vs. PWM dimming Revealed

The effect of CC dimming is even more dramatic if you go on down towards 3% dimming, where you get 240% more light, if the LED I used was driven at 1140mA (data from that example on page 1).

So, if we assume you had 50 candelas at 1140mA
-PWM would give you 1.5 candelas
-CC would give you 3.6 candelas

For the same power consumed.

Or another way to do it is to dim both down to 1.5 candelas, and consume 58% less power.

Another piece of the puzzle is the effect on battery efficiencies when you take a full blown current pulse out of the battery (PWM), vs. sipping ever so lightly on the cell. The battery will deliver quite a bit more energy if you just pull a light current off of it, instead of yanking a full 1140mA pulse out of it.

Now, a typical switcher starts getting less efficient down at those levels. Most chips these days have burst modes/pulse skipping, which will greatly increase the efficiencies, and if you have a proper sized capacitor, you can keep the ripple down at 0.8%. Another technique is to use two phase switchers, and just switch off one phase. If you think a little deeper, you can put heavy MOSFETs (higher gate charge, lower on resistance) on one phase, for the higher currents, and put light MOSFETs (low gate charge, higher on resistance), and keep the efficiency of the switcher up. You can also lower the frequency, to reduce losses due to gate charge and gate drivers at light loads. Often you will save additional energy on the very light load side, if you replace the one of the MOSFETs (depends on buck or boost), with a schottky diode.

As you start going up in PWM frequencies, you start taking additional losses due to the 700-800pf of capacitance in the LED. Adding just a capacitor on the output of the PWM, without an inductor, will greatly increase the losses in the transistor.

CM-

The paper on the PWM thing was done by Henry of HDS, where he used PFM to get around the PWM LED flashlight dimming patent.

Another technique is to dim with CC, on the upper end, where efficiency is important (high power draws), and then switch to PWM/PFM of the CC mode, where you convert down to lower the current, but then pulse that.

One note about dimming of LEDs, is that if you use something like an X1/WF bin, the green tint will become pretty obvious and re-inforced at the lower currents. Of course, this also depends on where in the X1/WF bin it is actually at- within the X1/WF bin.

Both PWM and CC have tint shift issues when dimming, and there is a chart on page one demonstrating the effect. PWM has a bit less shift, but notice how when CC dimming, it started shifting back towards the cyan direction at the very low levels.

7. ## Re: Constant Current vs. PWM dimming Revealed

I did some research about the possible enhancement of photometric effect using PWM, but i was unable to find any solid info at the time this thread was active.

"A research group at Ehime University developed a pulse drive control method to make LEDs look twice as bright by leveraging the properties of how people perceive brightness. The group was led by Masafumi Jinno, an associate professor of Dept of Electrical and Electronic Engineering at Graduate School of Science and Engineering of Ehime University.
When a short-cycle pulse voltage with a frequency of approximately 60Hz is applied to an LED at a duty ratio of about 5%, the LED looks about twice brighter to human eyes than that driven by a direct voltage, the research group said.
Based on an evaluation test using subjects, the group reported that a blue LED looks 1.5-1.9 times brighter while green and red LEDs look 2.0-2.2 and 1.0-1.3 times brighter, respectively.
"With this method, the brightness of LED with a luminance efficiency of 100lm/W can be simulated by using a 50lm/W LED," Jinno said.
The test result was unveiled at the "New Technology Presentation Meetings by Four Universities in Shikoku Region" sponsored by Japan Science and Technology Agency (JST).
There are two principles, the Broca-Sulzer effect and the Talbot-Plateau effect, involved in how human eyes perceive brightness. The Broca-Sulzer effect refers to a phenomenon in which light looks several times brighter to the eye than it actually is when exposed to a spark of light, such as a camera flash.
In addition, the Talbot-Plateau effect is a principle where human eyes repeatedly see flashes and sense the average brightness of the repeated lights. Thus far, "it has been believed that, due to the Talbot-Plateau effect, the brightness perceived by human eyes would not change even if an LED is pulse driven," Jinno said.
"The Talbot-Plateau effect is a principle found in the days when fluorescent mercury lamps and other light sources driven by a power supply with a longer voltage cycle of about several hundred milliseconds were used," Jinno said.
Thus, the group decided to drive the LEDs using a power supply with a shorter voltage cycle of about several hundred microseconds. As a result, the group discovered that, when a pulse voltage with a frequency of approximately 60Hz is applied at a duty ratio of about 5%, the impact by the Broca-Sulzer effect becomes greater than that of the Talbot-Plateau effect so that the light emitted from the LED looks brighter to human eyes.
The LEDs in three colors used in the evaluation test were all manufactured by Nichia Corp. The model number of the 464nm blue LED is NSPB500S, the number of the 520nm green LED is NSPG510S, and that of the 633nm red LED is NSPR510CS."

So finally, the enhancement of photometric effect has been proved. Human eye is an imperfect integrator of light. From the article, i take the conclusion the effect is more noticiable when using short pulses, below 1ms.

The problem then deciding what is more efficient photometrically, if PWM or CC, depend on if the enhancement is higher than the radiometric losses, wich have been well caracterized along the thread (we miss you, Newbie ). As the article report different enhancements depending of the wavelenghts tested, i think its going very difficult to do accurate maths about it, almost impossible.

It would be required to do a function wich takes into account frecuency, pulses duration, lux level of the pulse (perfectly integrated) and wavelenght. Too difficult, IMO.

Ive found too a graph wich explain brightness enhancement perception for different lux levels and pulses duration:

It seems 60 ms is the limit of pulses duration to get noticiable effects and that the effect is way higher as higher is the flux level. And from the previous article, that human eye integrates better as longer the wavelenght (so strong bluish spectrum benefit more than reddish).

8. ## Re: Constant Current vs. PWM dimming Revealed

Hi there,

First off, this is interesting. It was known that the eye integrates light but
it was always suspected that there may be some combination of pulse and
frequency that could upset this characteristic. Apparently, if what they say
is true, we have something new now.

There are some problems however.
The first one i notice is that 5 percent is a pretty low duty cycle.
The second one is that the effect is different for different wavelengths.

The difference with wavelenghts is going to cause color shift, which will
shift white into the sky-blue color. This isnt good, unless the white can
be first shifted into pink, but that would require different LEDs than
the white that we have now. I guess it's doable though.

The duty cycle problem is a little harder to get around. With a 5% duty
cycle and a pulse 20 times as high as normal, there will be great efficiency
problems that will reduce the light anyway, so even if there is a twofold
increase at low currents that doesnt mean the same at high currents.
Also, a pulse 20 times as high as normal means the manufacturers specs
will be exceeded for max pulse amplitude.
Ok, so say we drive it with a pulse 5 times as high as normal to stay within
(some LED) specs. At 5 percent duty cycle that means we're down to
1/20th of the normal light output, then increased it my 5 times times the
eff factor, increased by the magic factor (of the article). This leads to
a rough equation like this:
E2=E1/20*5*e5*m
where
E1 is the light output of a white Nichia at 20ma,
E2 is the new light output
e5 is the efficiency decrease factor for driving at 5x nominal current
m is the magic factor, a vector really.

To get an idea in percent, we can set E1=1.
e5 would be around 0.75
m is 2 for blue and green, and 1 for red (approximately)

This gives us:
E2=1/4*e5*(2+2+1)/3
so
E2=1/4*0.75*5/3
so the result is:
E2=0.3125
which means the max light output is 31 percent of what an LED is when
run at nominal 20ma, and the color is sky blue.

Now if we tint the LED plastic we also reduce light output, so that's a
problem too. Thus, we would end up with white light, but even less
than 31 percent of the full brightness.

The average current, however, would only be 5ma. This is 1/4 of the
nominal current. The new efficiency over the old eff could be expressed
as:
Eff2=Eff1*4*0.31
so after normalizing the old efficiency to 100 percent (1) we get:
Eff2=1*4*0.31
so
Eff2=1.24, which means 124 percent.
Thus, we have gotten a little more light out of it, at that light level.

The final analysis:

The max light level is 31 percent of normal brightness (about 1/3 of normal).
The color is tinted sky blue.
The efficiency is better, meaning about 20 percent less energy is expended
in lighting the LED, at 1/3 brightness.
The small Nichia LEDs could be driven like this, but the larger high powered ones
are not rated for over 2x nominal current so technically speaking, they cant be
powered in this way.
The LED can not be lit at full brightness without going way over manufacturers specs,
and this may not even be possible due to serious efficiency issues at *much* higher
peak currents.

9. ## Re: Constant Current vs. PWM dimming Revealed

Wouldn't a light with a pulse frequency of 60Hz actually be potentially flickery, especially for low-level lighting where some dark/peripheral vision could be expected to be operating?

In the study, I wonder if the people were looking at the LEDs, or using the LEDs to actually light up a scene.
Personally, looking at coloured 5mm LEDs, I've found it very hard to make meaningful brightness comparisons. Having identical LEDs side-by side running at significantly different currents, apparent differences (if they exist at all) are much smaller than the differences in drive currents. The LED is the brightest thing around of its colour, and the end result is that it seems to practically saturate a spot on the retina +/or brain.

I'd wonder if pulse vs. constant brightness results from looking at (relatively very bright) LEDs would differ from looking at (relatively much darker) scenes lit by LEDs?

10. ## Re: Constant Current vs. PWM dimming Revealed

Hi again,

In addition to my rough analysis above i'd like to add the following...

60Hz is pretty fast. It's not super fast, but it's usually fast enough to fool
the human eye, or at least mine (chuckle) because i have built white LED
lights that run from half wave rectified waves and they seem ok to me
(although i prefer 120Hz for other reasons). If you turn your head fast though
you can see the pulsing, so i guess it depend on how you are using the light,
if it will work for you or not. Even displays are sometimes run at 50Hz,
and some monitors operate the vertical sweep at 60Hz, although i dont like
them because you do notice some slight blink effect of some kind. But then
i dont know the phosphor decay properties anyway.
I'd also be very surprised if they were looking right at the LED, as that would
not do much for anybody, although i dont know for sure.

11. ## Re: Constant Current vs. PWM dimming Revealed

Very good point, MrAl. I think you nailed up the main problem: as the higher enhancement happen as higher is the light pulse (higher the surge current), the highest, and probably, the most noticiable effect, when using short duty cycles and huge surge currents.

But high power leds have a low marging respect to max pulsed current, usually up to 2A. For a 350mA rated led, its about 6 times higher than the nominal continous current. Meaning we can use a 20% duty cycle as shorter. And when using such high pulse currents, the radiometric losses are way higher, too.

Its not a problem for low current leds, though. And maybe further improvements on high power leds may do viable very high pulse currents. Im thinking on increasing die sizes specifically. The way the Osram choosed with the Diamon Dragon, wich use a 2 sq mm chips (double than actual high power dies), wich can by runned up to 4,5A (manufacturer rating, we know very often we can surpase it). This first device using a 2 sq mm chip havent a top end efficiency at all, but having doubled area, doubling current only increase current density at the die level by half, wich mean half the radiometric losses, given a good cooling (the DD has only 2,5 K/W of thermal resistance due the large die).

So probably, in the practice currently is going to be difficult to take advantage of the Broca-Sulzer effect, but its an effect we must keep in mind when deciding between PWM and CC for an application.

Of course, any application where flickering are a problem is a candidate for CC drive. But there are many applications where flickering isnt a relevant issue.

Personally, i hate flickering, gives me headache. I need 100hz frecuency to avoid it.

Originally Posted by uk_caver
In the study, I wonder if the people were looking at the LEDs, or using the LEDs to actually light up a scene.
Personally, looking at coloured 5mm LEDs, I've found it very hard to make meaningful brightness comparisons. Having identical LEDs side-by side running at significantly different currents, apparent differences (if they exist at all) are much smaller than the differences in drive currents. The LED is the brightest thing around of its colour, and the end result is that it seems to practically saturate a spot on the retina +/or brain.
I wondered the same when read it. Hopefully, somebody will get the original study and could clarify it.

Think than 660nm leds have a LER (lm produced by 1 optical watt) below 60 lm/w, and mostly, actual efficiencies around 20 lm/w (input watts), so they cant produce high lux levels, so the Broca-Sulzer effect is almost unnoticiable.

But orange-red leds have LERs in excess of 200lm/w, wich arnt so far from white leds, mostly below 300lm/w actually. The effect should be still noticiable when using them.

All i wanted when i posted the article is add it to this excelent reference thread, so people wanting to decide between PWM and CC drive keep in mind all the factors wich may affect its relative perfomance.

And be aware that luxometers dont measure the Broca-Sulzer effect at all!

12. ## Re: Constant Current vs. PWM dimming Revealed

The LEDs in the study weren't exactly high-power - green and red were somewhere in the 4000-8000 mcd range at 20mA, with a peak drive current of 200mA for the red and 100mA for the green, so at a 5% duty cycle, the red would be at roughly half the 20mA brightness, and the green at a quarter.

At those light outputs, to actually light up any kind of target, you'd think it'd have to be a close target in a room with subdued lighting.

13. ## Re: Constant Current vs. PWM dimming Revealed

Hello again,

For a comparative light meter that i think will work

http://hometown.aol.com/xaxo/page1.html

Kinnza:
You bring up some good points...
First, many people dont want any flicker at all and i think
the amount of flicker noticed varies by person...for example,
i like at least 75Hz myself.
Second, i dont know of any light meters that take the B-S
effect into account either, and we dont know how this
varies from person to person either, or if it varies
with age even. What if people over 60 years old dont
notice anything? I wonder about if they took that into
account with the study.

14. ## Re: Constant Current vs. PWM dimming Revealed

Originally Posted by uk_caver
The LEDs in the study weren't exactly high-power - green and red were somewhere in the 4000-8000 mcd range at 20mA, with a peak drive current of 200mA for the red and 100mA for the green, so at a 5% duty cycle, the red would be at roughly half the 20mA brightness, and the green at a quarter.

At those light outputs, to actually light up any kind of target, you'd think it'd have to be a close target in a room with subdued lighting.
What Caver said twice now, the study doesnt mean much to OUR purposes, cause 60Htz is horrible light,
and Green leds have no Phosphors, so there is no phosphor persistance used in the equasion, or phosphor averaging occuring. so ya phosphor based LEDs like white and quantum dots would be different again.

besides if they start using that on signal lights, and set off some epeleptic into a fit, that aint good. heck 60htz leds already give me a fit :-)

15. ## Re: Constant Current vs. PWM dimming Revealed

Originally Posted by VidPro
What Caver said twice now, the study doesnt mean much to OUR purposes, cause 60Htz is horrible light,
and Green leds have no Phosphors, so there is no phosphor persistance used in the equasion, or phosphor averaging occuring. so ya phosphor based LEDs like white and quantum dots would be different again
At 60hz pulses, phosphor persistence is utterly irrelevant. I was able to look at my LF2x (7.8kHz, as low as 0.2% duty cycle) output using a photodiode and oscilliscope , and did not see any trailing edge due to phosphor persistence. And on the lowest setting the pulses are only 250 nanoseconds.

The biggest problem with this effect though is that it only works at 60Hz, 5% duty cycle, and probably really low illuminance. I couldn't tolerate my Fenix L0D at 100Hz/15% duty cycle. A shorter duty cycle and lower frequency would be horrible.

I noticed they were using pretty weak LEDs. They probably also had all other lights off -- whcih would imply extremely low illuminance. I highly doubt it would work the same way as illuminances high enough for general purpose room lighting, or even outputs on the order of existing high power LEDs, where driving at duty cycles like that would be less efficient than simply CC at low current due to power supply reasons, anyway.

16. ## Re: Constant Current vs. PWM dimming Revealed

Originally Posted by 2xTrinity
At 60hz pulses, phosphor persistence is utterly irrelevant. I was able to look at my LF2x (7.8kHz, as low as 0.2% duty cycle) output using a photodiode and oscilliscope , and did not see any trailing edge due to phosphor persistence. And on the lowest setting the pulses are only 250 nanoseconds.

.
right, even phosphor based leds have very fast responce times, but testing without phosphor based leds, and forming a conclusion FOR phosphor based leds , is how we got other totally eronious information and specs and data.

17. ## Re: Constant Current vs. PWM dimming Revealed

Hi again,

Some very good points being brought up in this thread.
It's starting to look like the applications, at least for now, will be limited
to either very special purpose (cockpit illumination) or simply indicator lamps.
It may also help in emergency situations where there is no other lighting available
and the available power source (battery) is a very valued resource, such as in
emergency caving.

#### Posting Permissions

• You may not post new threads
• You may not post replies
• You may not post attachments
• You may not edit your posts
•