PWM-dimming of RGB LEDs

Here I have tested a single individual lamp, of a single brand and type. From the ubiquity of identical or very similar lamps sold under different brand names by numerous sellers in eBay and Aliexpress, this seems to be a “typical” or popular type. The intention is for it to serve as an example of one specific technology for adjustment of the light output of LED lamps. This is not a comparative review.

If you haven’t yet read the post on dimming of LEDs, I recomend that you read it before continuing.

In recent times cheap RGBW LED bulbs with four colour channels and their own small remote control have become widely available, both from well known brand names, and cheaper local and Chinese sellers. I tested one branded LEDSAVERs RGB, from a Swedish supplier. Although labelled RGB, the lamp has four independent colour channels red, gren, blue and white (RGBW).

For the test shown below in a video, I used a PicoScope 2204A USB oscilloscope with the PicoScope 6 software running under Windows 10. These oscilloscope has two channels. I connected to each channel a TSL251 light-to-voltage optical sensor, one filtered with red acrylic (Plexiglas ) and the other with blue acrylic (Plexiglas ). I powered the sensors with a stabilised 5V power supply.

The test includes an initial section showing dimming of the white channel of the lamp. Later sections show how different colours are created by varying the length of pulses from pairs of channels. In the video only colours formed by mixing red and blue are shown, as well as how dimming in this case.

In the video the red oscilloscope trace, channel B of the oscilloscope, corresponds to the red-filtered sensor, while the blue trace, channel A, corresponds to the blue-filtered sensor.

At the lower edge of the window appear live measurements of some parameters: pulse frequency is approximately constant at 164 Hz, and duty cycle changes in steps of approximately 10%, with a minimum of 10%. (To see this number you will need to set the video to full screen and HD resolution in YouTube.)

From the trace in the video we can see that one cycle lasts for 6 ms or 1/166 s of a second. At 50% duty cycle, we have periods of 3 ms or 1/333 s light followed by 3 ms or 1/333 s darkness. At 10% duty cycle, we have 0.6 ms or 1/1666 s followed by 5.4 ms darkness. We do not see flicker at this rate, but a camera will see the pulses. With fast shutter speeds we should get a whole assortment of over- and under exposed images and differences in colour, plus broad bands. Even mechanical shutters do not expose the whole sensor exactly at the same time. For shutter speeds somehow slower than the length of the cycle, there will be some variation from frame to frame, unless the shutter speed used is a multiple of the cycle period.

Be aware that what is shown in the images below is completely invisible to the human eye. The wall looked as having a constant and even colour at each of the settings used.

With 50 Hz or 60 Hz AC line frequency or 100 Hz or 120 Hz light pulsing most digital cameras will adjust the sampling rate of the live-view screen and electronic viewfinder (EVF) to prevent flicker. This LED lamp, however, uses a different frequency for dimming and resulted in very strong flicker on both the screen and EVF. 

Compared to the usual oscillation in light output as a result of AC mains power, these square pulses and dark periods are more disturbing for photography. Here is a white wall illuminated with red + blue mix, dimmed down. Each photograph in the series was taken with a different shutter speed using the “silent” shutter of the olympus E-M1 camera.

This slideshow requires JavaScript.

No dimming, mixed blue + red at a different %.

This slideshow requires JavaScript.

In the case of this lamp, because of the way the colours are mixed, when photographing under a mixed colour, the colour will vary from frame to frame. Now using the mechanical shutter, a burst of 24 images all using a shutter speed of  1/1000 s. This demonstrates that even with the mechanical shutter not all the frame is exposed simultaneously.

This slideshow requires JavaScript.

Even when using only the white channel of the lamp, dimming to about 50%, resulted in artefacts, and variation in exposure and uneven exposure.

This slideshow requires JavaScript.

This lamp reaches a maximum duty cycle of 100% at full output, giving under this setting perfectly steady light output. This is better than in the case of many non dimmable LED bulbs whose output fluctuates to some extent at twice the mains AC frequency. Consequently, if one uses one pure colour, R, G, B, or W buttons, and full output, exposure will be even at any shutter speed, with either type of shutter.


A similar, but weaker problem can affect other types of lamps whose light output varies at twice the frequency of mains AC line frequency, see the post on shutter speed and light flickr.

A caveat is that modern dimming driver circuits for LEDs not always  use PWM. In some cases they use constant current dimming, which is an alternative technology that completely avoids light output fluctuations.

Some LED lights meant for use as light sources in photography, avoid the effect of output fluctuations by using PWD dimming at a much higher frequency than in the LEDSAVERs  lamp described and tested here. As an example, I measured dimming of an Amaran AL-H9 LED light source. The manufacturer advertises it as using a type of dimming that maintains constant illumination. This is not completely true, as dimming used in this light source is based on the same approach as in the LEDSAVERs lamp. However, by using a frequency of 40 kHz (240 times faster) most of the problems of PWD are avoided. This is so because it is unlikely that anybody will use a shutter speed close to 1/40000 s, with a light source that has rather weak light output.

Those readers interested in photobiology, or in growing plants, may find interesting that the effect of pulsed light and its frequency is a question of practical importance for which we do not yet have good answers. Furthermore, how plants respond to alternating pulses of light of different wavelengths, their frequency, relative duty cycles and phase shifts in mixed illumination remains as a big unknown. An unknown related to a type of illumination that has no relation to that available in the natural habitat, but which is becoming important for plant cultivation indoors.

As a conclusion, a photograph of of a white wall evenly illuminated with yellow light from the LEDSAVERs lamp, dimmed to about 50%. “Silent shutter” at 1/320 s.

Yellow = Red + Green, dimmed.

All illustrations, text and measurements are of my own authorship, and copyrighted.

(c) 2017 Pedro J. Aphalo



Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s