Underdriving LEDs (lower current)

[made2hack]

Hello all,

I want to use a ZLED7020 chip to drive an LED. However, rather than providing a PWM signal to choose between 50% and 100% duty cycle (approximately), for 50% brightness and FULL brightness, I want to under-drive the LED (by lowering the current). I want the light to have 2 power modes, 1/2 brightness and full brightness.

Using the ZLED7020 chip, there is a Set resistor which chooses the constant current the chip provides. Thus, with a 3 position knob / switch, I can have it choose between 2 resistor values and "off" if you will.

What are the potential problems of under-driving the LED? I want to state that I am aware that the proper way to drive the brightness of LEDs is with PWM, but I wonder if in my case I'm within the LEDs parameters?

I want Brightness 1 to be around 600mA and Brightness 2 to be around 1Amp. I have attached the LED FW Current / FW Voltage chart.

[capt bullshot]

There's no risk and no danger doing so.
Two things to consider: luminous intensity vs. current often is non-linear, and for white LEDs the colour (white) point may shift if the LED is operated at lower current than specified.

[danadak]

LEDs are roughly linear, intensity vs current, except at extreme ends
of operating range.







Regards, Dana.

[ogden]

LEDs are roughly linear, intensity vs current, except at extreme ends of operating range.

Yet light output graph is slightly curved. Following research states that in the middle sweet point current control gives up-to 10% more light compared to PWM (for same energy consumed):

http://pe.org.pl/articles/2013/11/68.pdf

There was some document stating similar facts from big led mfg (cree?), but I can't find it quickly enough. Only downside of current control is already mentioned color shift.

[T3sl4co1l]

The only hazard is evenness of light output (for arrays with parallel strings), and poor CRI (due to efficiency differences for the yellow phosphor).

Tim

[Zero999]

LEDs are roughly linear, intensity vs current, except at extreme ends of operating range.

Yet light output graph is slightly curved. Following research states that in the middle sweet point current control gives up-to 10% more light compared to PWM (for same energy consumed):

http://pe.org.pl/articles/2013/11/68.pdf

There was some document stating similar facts from big led mfg (cree?), but I can't find it quickly enough. Only downside of current control is already mentioned color shift.

You are right. The forward voltage decreases (logarithmically?) with current and the brightness decreases linearly, meaning the efficiency (V_F*I_IN)/P_OUT increases, up to a point of course.

However, the above only gives net gains, in terms of total system efficiency, if a switched mode power supply is used to drive the LED. If a resistor or linear current regulator is used to drive the LED, then reducing the current doesn't make any difference to the total system efficiency, because the driver's efficiency decreases (greater V_IN-V_OUT), as the LED's efficiency increases and the two effects cancel one another out.

The good thing is, the ZLED7020 is a switched mode constant current driver, so the efficiency of the whole system should increase, as the forward current decreases.

[danadak]

Yes the curve of brightness vs current is not exactly linear.

But 10% difference, given that human eye is not a linear sensor, probably not significant ?
For indicator type applications.


http://www.telescope-optics.net/eye_intensity_response.htm




Regards, Dana.

[made2hack]

Thanks everyone and thanks ogden for the article.

[james_s]

In higher power LEDs the intensity to current graph is even less linear. I have some LEDs that are made to run at 3 Amps and at that current they are only marginally brighter than at 1.5A although the eye is the largest non-linearity by far.

[Audioguru]

Since the sensitivity of our vision to brightness is logarithmic, then half the current looks almost as bright as full current. About 1/10th the current looks half as bright.

[mikeselectricstuff]

You may see more part-to-part variation as LEDs are binned at specced current, but unlikely to be an issue until you're below maybe 10% of nominal current

[Zero999]

Since the sensitivity of our vision to brightness is logarithmic, then half the current looks almost as bright as full current. About 1/10th the current looks half as bright.

Are you sure about the 1/10 of the current, for half the brightness? The eyes may have a logarithmic response, but it's not necessarily log₁₀. I believe it's also dependant on the wavelength, the lighting conditions and whether the eye is dark adapted or not. The peak sensitivity of the eye shifts from green, to the blue end of the spectrum, as it gets darker.

As far as the original question about LED running on lower currents. Some sources suggest, LEDs keep getting more efficient,  as the current is reduces, others say there's a sweet spot between 5% and 15% of the rated current, giving peak efficiency.
http://www.ti.com/lit/an/slyt617/slyt617.pdf
https://www.digikey.co.uk/en/articles/techzone/2011/oct/identifying-the-causes-of-led-efficiency-droop

[mikeselectricstuff]

Approx 50% perceived brightness at 25% current.

[made2hack]

Hello all,

So some test results. At 30 cm, the lux meter app on a phone showed approximately 1400 lux when the led was run at 440mA. The same led run at 1000mA showed approx 2800 - 2900 lux, so not quite 1:1 linear but close.

However, the brightness perception was not that the light was "doubled", perhaps maybe a 1/3 more. Granted this is highly subjective.

I might try as mike suggested around 25% current and see what subjective effect this will have.

[T3sl4co1l]

I used a logarithmic potentiometer in my dimmable LED flashlight, which feels very natural.

I feel like gamma correction of ~2.0 is most natural (i.e., 1/4 power --> 1/2 apparent intensity) -- a quadratic not logarithmic correction -- but log is better than lin, that's for sure!

Tim

[made2hack]

I used a logarithmic potentiometer in my dimmable LED flashlight, which feels very natural.
Tim

So, one can use a "Volume" potentiometer used in audio?

[cdev]

At lower current some LEDs last much much longer than they do at rated current.

In the past I used a pot, but now the current adjustment on my better (Sorenson) bench supply (which has resolution to 1 ma) to determine an optimal current target individually. LEDs vary enough when underpowered to make that the best way to do it.

[mercrazy]

FYI
the ZLED7020 chip is being discontinued at most supply houses.
replacement is PAM2861.