Efficient Dimming?

[Light Thinking]

Hello . . .

Is there a hard rule for choice of dimming method; reducing the drive current, or PWM, that produces the greatest efficency?

Having read many design notes etc., I'm still unsure which method of dimming a LED is more efficient.

I have seen claims that PWM is more efficient due to a reduced duty cycle.

I've found others claiming since PWM is running full current WHEN on, the efficiency is reduced because LEDs run at higher efficiency when the current is not maxed out.

What is the real world answer?

Thanks!

 

[LukeA]

The LED is more efficient at lower currents, so constant current is more efficient.

The duty cycle of PWM doesn't make it more efficient. For example, you'll get 40% the light at 40% the input power of running at full current. With constant current, you get 40% the light of full power, but you get the light at slightly less than 40% of full input power. The exact difference varies for each LED, but the trend holds true.

 

[spencer]

I believe that PWM drivers are more efficient than constant current drivers but the LED is less efficient with PWM than constant current drivers. I think overall lumen for watt, a PWM setup would give you more lumens than a constant current setup. Depends on the specifics on drivers you use though.

 

[Marduke]

It also depends how dim you are going. If you are dimming a high power LED at <20mA, constant current efficiency tanks, and PWM is the only real option.

 

[Light Thinking]

This is why I asked.

LukeA . . .

The LED is more efficient at lower currents, so constant current is more efficient.

This is one way of thinking that I see often . . . .

spencer seems to agree . . .

the LED is less efficient with PWM than constant current drivers

But add in a driver, (what we are talking about) . . . .

I think overall lumen for watt, a PWM setup would give you more lumens than a constant current setup.

And Marduke adds the amount of dimming required . . .

So . . .

With three knowledgeable answers, It still seems less then clear in what cases constant current is preferred to PWM.

As with most electronic circuits, application and design matter. The reason I asked, I have found hard rules for both arguments on credible manufactures sites. (Both LED and driver makers!)

What factors should be used to decide, assuming the ability to do a suitable job exists?

Is there no clear answer that fits most applications, or is adjusting the current "normally" a better option?

Thanks for the answers so far!

 

[Wiggle]

Wouldn't an LED that was sent PWM perform similarly to one with a lower constant current? To me it doesn't make sense that the LED would think "ok I'm off, oh now I'm on but the current is high so my efficiency is a little lower". That might not make any sense but what I'm trying to say, isn't the LED less efficient due to the cumulative effect of the higher current integrated over the time causing higher losses? Do the same principles that cause the PWM to trick our eyes also "trick" the LED into staying out of the lower efficiency range?

 

[Sub_Umbra]

As a broad, generalized term, effeciency is kind of a squishy term when it is applied across the board to all LEDs. The OP does not limit the question to any particular LEDs, whitish or otherwise.

Aside from the above posts my own interests generally rest with monochromatic LEDs and in those cases PWM would seem to win hands down since producing the desired colors of most monochromatic LEDs is highly voltage dependant.

For example, I know of no other way to dim an LED rated at 495nm that will result in the desired wavelength at any lower output levels. The same would seem to be true for any who have very specific color rendering needs for whatever tint of "white" LEDs they require for mission specific tasks.

If, on the other hand, you only need some kind of white, and tint is unimportant in the application, then efficiency may be all that needs to be considered.

 

[Light Thinking]

Wiggle . . .

Wouldn't an LED that was sent PWM perform similarly to one with a lower constant current?

LEDs do normally have a point below their max. drive current, where they do produce more light for energy used, then at peak current. Manuf. have graphs that chart this. If I understand right, when using PWM at full drive current, it's on 100% then off 100%, without benifit of the better efficency of a lower drive current.

One reason I started this question, my confusion on conflicting claims . . . which produces the greater efficiency, (light produced for a given amount of energy used).

So . . . all things equal . . . which will produce more light per energy used?"

 

[bshanahan14rulz]

obviously current limiting (to the current that produces highest efficiencies with the particular LED) will produce more light per power consumed. a pwm circuit needs power too.

PWM has the advantage of changing brightness without changing the current or voltage so that tint stays consistent. Why not combine the two? limit the current to the LED's most efficient current and use a PWM to dim them as needed

 

[LukeA]

It's my belief that it's easier to construct from scratch an efficient variable PWM driver than an efficient variable constant-current one.

 

[Marduke]

PWM has the advantage of changing brightness without changing the current or voltage so that tint stays consistent. Why not combine the two? limit the current to the LED's most efficient current and use a PWM to dim them as needed

That is what is always done. A PWM circuit in a flashlight rarely runs the LED at "full" power. A level is chosen that corresponds to the intended high level, and you go down from there.

However, running at the optimum current is not usually ideal, as the peak current efficiency for a high power LED such as a Cree XR-E is ~20mA, which is quite dim.

Consequently, that is why the "low" on the Fenix lights is not very low, and keeps going up with each LED upgrade. The low current is right in that efficiency butter zone, and cannot go lower without modifying the constant current circuit to ALSO do PWM.

That is why ANY high power light capable of <~10lm is always using PWM and not constant current.

 

[Steve K]

Hello . . .

Is there a hard rule for choice of dimming method; reducing the drive current, or PWM, that produces the greatest efficency?
<....snip...>

I think there is some confusion because you haven't specified how you are measuring efficiency. Is it measured in photons per electron? Photons per watt delivered to the led? Photons per watt delivered by the battery?

In my mind, the power electronics is a big factor and needs to be considered. Is a regulator used? Is it a linear regulator or a switching regulator? What is the input voltage to the regulator? If it is a switching regulator, what data is available from the manufacturer regarding the efficiency versus input voltage and load?

None of this is trivial. If you are going to ask about efficiency, the question needs to be very specific if the answer is going to mean anything at all. Otherwise, it's like somebody asking "what flashlight is best?" without specifying all of the constraints on the solution, i.e. do you mind carrying a 100 pound light, or using 240vac power to energize it?

Depending on what you are trying to understand, perhaps the wording could be changed to "what are the effects of varying led brightness by pwm versus changing the DC current?"

regards,
Steve K.

 

[selfbuilt]

What is the real world answer?

As others have pointed out, "best" efficiency will depend on how you define the term.

For the real world, I suspect most care about how much runtime to expect for a given output on a given battery. In my testing, current-controlled lights invariably provide longer regulated runtime than PWM-controlled lights for the same overall output on the same battery source. That result is clear across all makes and models.

Of course, current-controlled is not feasible for really low drive currents, as Marduke has pointed out. So if you wish to go <10-20 lumens, you typically need to use PWM.

 

[Alan B]

This thread has a lot of good information. In summary, "it depends". As in much of engineering, the devil is in the details. There are tradeoffs and generalizations don't apply in all cases. Defining efficiency differently matters. For example it might be "photons per battery charge".

The energy the driver requires changes as the current changes. The LED light efficiency changes as current changes. The battery efficiency changes as peak current changes. Circuit complexity changes depending on what it must do, and how efficiently. PWM frequency makes a difference in efficiency and circuit complexity. Constant current won't produce the full range of dimming. A full range high efficiency system must include both constant current and PWM, and the complexity rises. PWM only systems are very simple and can be electronically efficient. The additional complexity and electronics tends to lower efficiency in the electronics while raising it in the LED, so which effect is larger? These things depend on the particular specs of the LED and the specific driver. More efficiency can be attained with larger components and more heatsinking, but is this compatible with the physical design and system cost. So there is not a general answer that will always be right. That's why you see credible sources that disagree.

The answer is .. it depends!