I bought an LED stripe from näve (very similar to this model), but found that it created flicker in video recordings. The flicker got less visible with lower frame rates, but even at 24 fps it was there. I found that the frequency of the flicker was certainly not 50/60 Hz, but I believe the controller varied it to set the brightness (as I tuned it brighter or darker, the number of flicker "bands" caught on video changed) - and as I moved towards the maximum brightness, the flicker got less perceptible in the video recording.

I was able to fix this by connecting the stripe to another controller. As far as I can tell, this controller does not pulse the light, but rather adjusts the current. It causes no flicker at any frame rate (24/25/30/50/60Hz), and no bands at any shutter speed (up to 1/4000 s). If I go over a certain current, the LEDs just stop getting brighter, and the controller emits a faint humming noise - but no part of it seems to overheat.

I am wondering - why would LED bands not use this approach? Is there any detriment to me using it?

  • \$\begingroup\$ FYI, you should be able to smooth out flicker by adding a capacitor. I think something around 470-1000μF should eliminate the flicker. \$\endgroup\$ – Bort Dec 28 '16 at 15:20
  • \$\begingroup\$ @Bort a capacitor by itself might be not so healthy for the supply, if it tries to pump a PWM signal into the line. A decoupling inductance and a capacitor would do. \$\endgroup\$ – leftaroundabout Dec 28 '16 at 15:41
  • \$\begingroup\$ @leftaroundabout - I am not sure in which way you mean 'unhealthy'. Do you have a link to information about that? I am interested, because I see PWMs with simple cap filters often. \$\endgroup\$ – Bort Dec 28 '16 at 15:43
  • \$\begingroup\$ @Bort a cap only becomes a filter if you combine it with an Ohmic (or inductive) output impedance, which many sources have. But a LED driver is a regulated source, and if you use this to drive a load very different from the one envisioned by the designers, you may end up with strange oscillations. A pulsed constant-current source should cope fine with a capacitive load, but without further information this is a bit speculative. \$\endgroup\$ – leftaroundabout Dec 28 '16 at 16:31

This answer contradicts the assumption that constant current degrades performance for colour rendering. As far as the LED is concerned I assert there is no compromise using Constant current whereas PWM may cause "blue shift" from phosphor photon "quenching " OR insufficient dwell time ( two distinctly different phosphors and time sensitivities (Too long or too short. )

Phosphors for LEDs start with those typically used in traditional TV CRT's and Fluorescent tubes where used was UV to excite the phosphor for a fairly long time so that flicker at 60Hz would be less noticeable, thus the time constant of decay was in the 1~10 ms range. Although LEDs use Blue as a substrate so phosphors in the blue range are not needed.

Why is an LED white?

Think of White LEDs with massive thick transparent BLUE emitting substrate and a hundred atoms thick coated of phosphor that is critically thick to vary the colour from cool (thinner) to warm ( thicker). Like some FL tubes they may be tri-phosphor for broader spectral and higher CRI quality levels which also affect other factors.

The decay time < shutter time causes the dark bands in the image.

The same phosphors were used in the Blue LED's with a very thin layer of phosphor which converts 10 to 20% of the Blue light into a smooth spectrum of yellow,green,red using yttrium aluminum garnet (YAG) phosphor (doped with cerium). The decay times are slower but the Red content was weaker, so mostly used in "Cool" white LED's. This decay time is affected by PWM choices from 1k to 10kHz, where <=1Khz may be noticeable on camera but > 1kHz might not and depends on both the PWM and the LED phosphor and CCT,CRI values.

Now with higher efficacy (>120LPW) White LED's they have added more dopants to improve CRI as well with more red emissions. Phosphor selection will continue to evolve with quality indicators like CCT/CQS that also affect cost. They include oxynitride phosphors is the MSi2O2N2:Eu2+ (where M = Ca2+, Sr2+, Ba2+) compositions whose emission ranges from 575 to 675 nm.

The phosphors are selected for cost, efficacy, thermal stability and longevity more so than response time, so PWM rates may have to increase for camera work or use filtered PWM.

Filtered PWM has some losses associated with the real conductance or resistance losses in Chokes and Capacitors (ESR), but this is the most effective way to eliminate PWM flicker and may be added to any controller. This is the same as the LC low pass filter (LPF) used in Class-E PWM audio power amplifiers.

For Linear voltage controlled current sources or sinks, there is of course the linear loss across the regulator. This only becomes a disadvantage if there is a large voltage drop in the regulator.

  • Typically LED brightness range is limited by the quality or ESR of the power LED and thus the voltage range required from say 10% to 100% brightness. In very high Efficacy LEDs this can be from 2.8V to 2.95V and in lesser efficacy LEDs, this can vary from 2.8 to 3.2V and worse. Thus this range can be 5% to 12% (150mV/3V=5%, 400mV/3.2V~=12%). This implies the input supply voltage for the string must be buck,boost or series regulated to create this voltage drop to dim the LEDs and its pass current * voltage drop of 0 to 5% or 0 to 12%V * unknown I [Amp] implies the power loss Vdrop*I=P of this regulator, which can be quite good <5% if well designed to match the string voltage +/-5 to 10%. This requires good DC-DC design principles with a desired response speed for modulated light with current sense and fast current regulated intensity.

Thus the PWM ripple current amplitude and time interval and the LED epiwafer phosphor decay times all affect the dark bands found in time aperture captured images.

To prevent flicker the current ripple must be reduced, as the sliding shutter capture method may alias or interfere wit the PWM dead time from phosphor decay, when PWM is used. A line filter may be used to match the current peaks used with sufficient attenuation of the PWM frequency being 20~30dB or more depending on the camera image quality desired on grey scale background..

Linear or filtered PWM is the ideal way to control LED's to prevent the issues associated with diode saturation, efficacy loss, colour shift, but adds a little cost with the LC filter depending on quality and depth (-dB@1kHz) of the filter., (similar to an AC line filter cost)

There are also phosphors designed for LED photoflash that require a minimum of 50ms duration for optimal CRI, that may also reduce flicker effect with PWM. http://www1.eere.energy.gov/buildings/publications/pdfs/ssl/chowdhury_phosphors-panel_sandiego2014.pdf

enter image description here enter image description here


Analog dimming tends to result in a larger color temperature shift than PWM dimming. Additionally, the linearity of the brightness change tends to be better with PWM dimming. Finally, nearly any buck current regulator (especially hysteretic) can be shunt FET dimmed even if it doesn't have onboard dimming hooks.

  • \$\begingroup\$ @M_D Can you please backup the color shift claim? From my reference info saturation effects of PWM shift the specified colour coordinates more than continuous current. This may be related to thermal shift and has to do with reduction of blue LED and phosphor efficacy above continuous current. I understand old reports indicated PWM was more stable. I do not believe this to be true. \$\endgroup\$ – Tony Stewart Sunnyskyguy EE75 Dec 28 '16 at 18:46
  • \$\begingroup\$ @Tony Stewart I admittedly haven't looked at this in quite a while, and am happy to be corrected if there are newer findings. There were several articles written in Power Electronics about it a few years ago. As I recall understanding it, color shift can happen due to die temp or current through the LED. With PWM dimming, you avoid the shift due to current because you are operating the LED at a single set point when on. \$\endgroup\$ – M D Dec 28 '16 at 19:17
  • 1
    \$\begingroup\$ I have known this info for 10 years being in the custom white LED business with awareness to critcal colour specs. google.ca/… This cites 3 Mfg's and there may be more with different results now. \$\endgroup\$ – Tony Stewart Sunnyskyguy EE75 Dec 28 '16 at 19:37

Very simply pwm is cheaper than current regulation. The number of people that need to worry about video synch is much lower than the number of people that just want their LEDs to be less bright.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.