Flicker in the Capture, LED light source

Question

Incident:

I'm trying to capture small particles at about 20-30 cm distance to the camera at ~60-70 fps. Was using SMD32 LED with rechargeable battery as light source. By using that light source, there was no flicker in the video (almost no light intensity variance between frames), but the scene was a bit dark.

SMD32 LED with rechargeable battery

To increase intensity, I tried cheap light bulbs of all kind, such as Philips Classic Tone 60W, and Energy efficient Philips Economy Stick 23W. As expected, intensity levels were quite reasonable, but the flickering was quite obvious and problematic in the captured video. Then I tried 2 LED light bulbs: Wiselite Warmwhite LED A60 8W, and GoldMaster 9W LED. Interestingly, LED light bulbs DID flicker too.

Energy efficient Philips Economy Stick 23W, Philips Classic Tone 60W, Wiselite Warmwhite LED A60 8W

Please note that voltage levels in the region is quite fluctuating, sometimes electrical home appliances stop operating and lights dimming although the electricity is on.

Question:

What made SMD32 LED non-flickering compared to LED bulbs? Is it the battery, or multiple light sources, or the type of the LEDs? What should I do for a non-flickering but enough intensity capture environment?

---

Edit :

After the responses, I have further investigated the SMD32 LED, by only capturing the light projected from SMD32 LED on to the screen. Fps was variable between 59-61, exposure time was 1.5 ms, CV_8UC1 color scheme (grayscale) was used. I have saved 0th, 40th, 80th, 120th... frames and combined them to build animated gif images. Each gif image consist of 35 jpg images. As side note, each animated gif image below is at 4 fps.

1st animated gif one on the left is captured when SMD32 LED is not plugged into mains power (while it is only using battery power). It is almost steady.

2nd one on the right is captured when SMD32 LED is plugged into mains power. Overall intensity compared to first one increased but the flicker is quite obvious.

Consequently, there is something with the mains power. And I still don't know how to obtain the desired lightning conditions in a cheap way.

[ [

Answer 1

Here is a blog posting from one of the leading LED fixture manufacturers, SORAA. (Their chief scientist won the Nobel Prize for the invention of the blue LED.) One of the design problems is that a rippler-suppressor reduces LED output efficiency by 2%-3%. If you are a consumer looking for absolute Lumens from a given form factor, or highest Lumens per Watt, you'll end up buying a product that compromises on flicker in favor of output. Moreover, while the industry used to agree on a definition of a flicker index, they are walking away from that and going to very application (and person)-specific definitions. In the referenced article, mention was made that in the state of California, the only recognized flicker test is performed by the person sensitive to the phenomenon. Finally, if you are using these lights for anything other than scientific motion studies, i.e., if color is actually important to you, Color Rendering Index (CRI) may also be important.

Obviously, cameras have very different sensitivity than humans. And specialty instruments (made for high-speed video) exist, with more expensive power supplies and lower lumen/watt output. I have been impressed with the SORAA lights I've purchased (CRI 95, and no visible flicker at framerates I've used). If they don't test out for you, the best alternative is to get a specialty light that does specifically advertise flicker-free (usually by using a true DC power supply).

Answer 2

I can't speak to the specific models / part numbers, but it's often the case that an LED 'array' is not continuously lit, but rather scanned or modulated.

That is, each element (or row, or other group) is briefly supplied full current, then switched off while another element is addressed, and so on. This reduces the total amount of current required, and if done above a 'flicker rate' determined by human persistence of vision, is undetectable in normal use. This rate is anything above about 30 Hz for a full cycle of elements, and is usually higher.

However, this scan rate is not synchronous with camera scan rates or shutters times, so you may get 'beats' or strobing in captured video. A similar effect can occur when shooting video displays and computer monitors.

A trick that can sometimes show this is to blink rapidly while looking at the light off-axis (at the edge of your vision), or quickly moving the light up/down or right/left in the same area. Persistence of vision is lower at the periphery.