Selecting the correct LED lighting can be a difficult challenge for those new to machine vision and even those with years of experience. We often receive calls for assistance in selecting the right lighting solution for a specific application. The challenges we face can be related to the environment, working distance, limited space, or even just an unusual object. With over 20 years of experience manufacturing for the machine vision industry, I am surprised that customers still request red LEDs for use with monochrome camera applications. I believe the only possible reason is that customers continue to buy what they are most familiar with. The mentality seems to be that since we have always bought red LED lights there is no reason to change.

While there are always cases in which red would be the best choice, including distinguishing items in the same color spectrum, making items less distinguishable by using the opposing spectrum (opposite on a color wheel), and having a monochrome camera with a higher spectral response in red. Typically these are not the reasons a customer selects red LEDs. In fact, I have found that a better LED color is often overlooked, blue. I hope to shed a little light on why blue is a superior LED choice over red now.

To understand the reason why red LED lights are so widely used you need to look back at the history of LEDs and early monochrome vision systems. The first practical use of LEDs was red which was introduced in 1962. At this point, the intensity was so low that they were only useful for indicators. Years went by and new developments helped to make them useful for lighting in vision applications. The price was relatively inexpensive when you factored in the life cycle and downtime due to intensity degradation and failure of other types of light sources. Red LEDs could be strobed at a higher current helping to reduce camera pixel blur of parts in motion and extend the life of the LEDs even further. Red LEDs were also the perfect fit for most of the monochrome vision systems because at that time they had a higher spectral response in the infrared to red range. Blue LEDs were higher priced and the spectral response of the CCD in most if not all of the monochrome vision systems was low in the 470nm (blue) range not to mention the lumen output of blue LEDs was very low. Why would you ever buy a blue LED for anything other than an indicator?

Now enters the invention of the white LED. The white LED came onto the scene in the mid ‘90s. This new technology was based on using a blue LED with a phosphor coating to create perceived white light. This new technology had its share of problems because of phosphor degradation that caused short LED life expectancies. White LEDs were also a premium price compared to red LEDs usually 5-10 times the cost. Just when things looked bleak for white LEDs a company called Philips came along with their new Lumiled Luxeon LEDs. They were advertising LED life retaining an average of 70% output after 50K hours for a 1-watt white LED with a 21-lumen output. This was the turning point for white LEDs in the vision industry especially when it came to color applications, but it also should have lit the way for blue high-output LED usage in monochrome applications.

What started to emerge should have moved a good percentage of monochrome applications over to blue high-output LEDs. The newer vision camera systems have imaging chips with the highest spectral response in the cyan (505nm) range (see Chart 1). I know that cyan isn’t blue but that will be explained later. The problem has been that LED manufacturers have been specifying the output of their LEDs in luminous flux (lumens). The following is taken from wikipedia.org and defines the terms currently used to measure LED output.

“In photometry, luminous flux or luminous power is the measure of the perceived power of light. It differs from radiant flux, the measure of the total power of electromagnetic radiation (including infrared, ultraviolet, and visible light), in that luminous flux is adjusted to reflect the varying sensitivity of the human eye to different wavelengths of light.”

Based on the previous definitions and an understanding of the LED manufacturer’s primary market it is not hard to understand their output unit is and continues to be luminous flux or lumens. In lighting for machine vision or vision systems in general, almost all the LEDs being sold are geared toward human viewing. Manufacturers are primarily building their products to replace car headlights and taillights, facility lighting, home lighting, museum lighting, flashlight bulbs, indicator lighting, and a variety of other types that involve human viewing. For these reasons, the LED manufacturers measure their products in lumens. Vision systems however do not rely on the human eye. The imaging chip in a vision camera has a spectral response curve that differs from the human eye (see Chart 1) so the better unit of measure would be radiant flux (watt).

Vision lighting vendors should adopt the radiant flux (watt) or a milliwatt unit of measure so that a uniform standard can be used for light output comparison. The milliwatt standard would be based solely on the power output of the LED and because LEDs have a narrow spectral output a customer could apply that information directly to a camera’s spectral response curve (see Chart 1). This brings me to the point that I made earlier, most of the newer vision systems produced have imaging chips with the highest spectral response in the cyan (505nm) range. You would think cyan would be the best choice but because blue LEDs have 2.4 times more mW output than cyan LEDs it is not (see Table 1). The point is that the LED manufacturers are not making many improvements in cyan LEDs but they are making gigantic strides in blue (470nm) LEDs due to the race for higher-output white LEDs. You will notice in Chart 2 that the spectral response curve of either the white or neutral white LED is highest in the 450nm blue range because of how they are manufactured.

If you follow Table 1 above you will notice that the typical output of a blue LED is only 35 lumens and the red LED is 74 lumens but these values are based on the human eye spectral response. Because the lumen output of LEDs is the norm for light output and because milliwatt output is harder to calculate most vision lighting companies don’t feel they need to change. When you look at the actual milliwatt output you notice in Table 1 that the blue LED has 1.8 times higher mW output than the red LED. Based on this information and the fact that most of the imaging chips in newer monochrome vision cameras have a higher spectral response in the blue range, you would agree that blue (470nm) would be the best choice.

Another reason to choose blue LEDs is their stability over temperature. In Chart 3 the blue LED output is stable over the operating temperature range, but the red LED has a 14% drop in output at 50°C and a 43% drop in output at 100°C. This creates issues where changes in intensity cause false failures due to dim images as the light warms up or even having to adjust inspection parameters from winter to summer months. Once again showing that blue LEDs are superior to red with today’s camera and LED advances.