When yellow and green LEDs became available, the number of applications increased. Now the color could change, to give additional information, or could indicate more urgent alarms. For example, green = OK, yellow = requires attention, and red = faulty. Most important was the ability to have LED lamps in traffic lights.

One characteristic of the light from an LED is that it occupies a narrow spectrum about 20-nm wide; the color is fairly pure. By contrast, a semiconductor laser used for telecommunications occupies a spectrum about 2-nm wide. The very narrow spectrum of a laser is important because, when used with optical fiber systems, the narrow spectral width allows a wide-system bandwidth. In general-purpose LED applications, the spectral width has very little effect.

Another important characteristic of LED light is that current is converted into light (photons). This means that doubling the current doubles the light amplitude. So dimming lights by lowering the current is possible. It should be noted that the specified wavelength emitted by an LED is at a certain current; the wavelength will change a little if the current is higher or lower than the specified current. Dimming by pulse width modulation is a viable alternative used by many people. pulse width modulation dimming uses a signal, typically a frequency range of 100–1000 Hz, to turn the LED on and off. The pulse width is reduced to dim the light, or increased to brighten the light.

The “holy grail” was blue LEDs, which are made from indium gallium nitride (InGaN). When adding colored light, red, green, and blue make light that appears white to the human eye. The reason for only “appearing” white is that the eye has receptors (cones) that detect red, green, and blue. There are big gaps in the color spectrum, but the eye does not notice. White LEDs are usually made using blue LEDs with a yellow phosphor dot over the emitting surface. The yellow phosphor creates a wide spectrum and, when combined with the blue, appears white. Different phosphors produce a range of white from cold white (bluish white) to warm white (yellowish white).

An interesting application for blue LEDs is in dentistry. Illuminating modern resins used in filling materials with blue light can cure them (make them change form, from liquid into solid). The 465-nm wavelength has been found to be close to optimum for this application, although the intensity of the light must be high enough to penetrate through the resin.

Some interesting applications rely on the purity of the LED color. The illumination of fresh food is better with LEDs because they emit no ultraviolet light. Photographic dark rooms can use colors where the film is insensitive; traditionally red-colored incandescent lamps. Even traffic lights must emit a limited range of colors, which are specified in national standards. Incandescent (filament) lamps in these applications were inefficient because they produce white light that is filtered, so most of the light energy is blocked by the filter and thus wasted as heat.

It should be noted that the color of an LED would change as the LED’s temperature changes. Some colors change more than others: a customer once told me that the amber-colored LED for a traffic light is very susceptible to color change and it is difficult to keep within specification. The LED temperature can change due to ambient conditions, such as being housed adjacent to hot machinery, or due to internal heating of the LED because of the amount of current flowing through it. The only way to control ambient temperature is to add a cooling fan, or by placing the LED away from the source of heat. Mounting the LED on a good heat sink can help to control internal heating.

In the early days of LED production, all LEDs had a current rating of 20 mA and the forward voltage drop was about 2 V for red, higher for other colors; later low-current LEDs were created that operated from a 2-mA source (for battery-powered applications). One LED manufacturer called Lumileds was created by HP and Philips in 1999 and produced the first 350-mA LED (Luxeon Star). Now there are a number of power LED manufacturers, rated at 350 mA, 700 mA, and 1 A or higher. Power LEDs have been used in increasing numbers; in channel lighting (signage), traffic lights, street lights, automotive, agricultural lighting, mood lighting (color changing “wall wash”), and also in theatres for lighting steps and emergency exits.

However, low-current LEDs in the 20–100 mA range are more efficient: the light output per watt (effectiveness or efficacy) is better for lower-current LEDs. Low-current LEDs are also very much cheaper than high-current LEDs. Since 2010, the trend is to use an array of many low-current LEDs instead of a single high-current LED, to produce more lumens per watt. The power rating of these LED arrays can reach 150 W (e.g., Cree CXA3590 LED). In some lamps this has the benefit of distributing the light over a larger area and thus avoiding the need for lenses or light diffusors. Another advantage is that the distribution of heat is easier with multiple low-current LEDs rather than a few high-power LEDs.

Channel lighting is so called because the LEDs are mounted in a channel (Fig. 2.2). Typically this channel is used to form letters, for illuminated company name signs. In the past, channel lighting used cold-cathode or fluorescent tubes, but these had reliability problems. Health and safety legislation, such as the RoHS Directive, banned some materials, such as mercury that is used in the construction of cold-cathode tubes. So, to cope with the shapes and environmental conditions, the most viable technology is LED lighting.

Traffic lights have used low-power LEDs for some years, but now some manufacturers are using a few high-power LEDs instead. Traffic lights are usually powered from the AC mains supply and good power factor correction is required. Having a high power factor (PF) means that the supply current is in phase with the supply voltage and the harmonic distortion is low. Keeping the PF high helps to reduce the cost of the AC supply because power supply companies charge a premium if the PF is low. Also, some traffic light control systems need a load that has a good PF, for fault detection purposes.

One problem with traffic lights is controlling the wavelength of the yellow (amber) light. Yellow LEDs suffer from a greater wavelength shift than other colors, and this can cause them to operate outside their permitted spectral range. Another problem is making them fail-safe; authorities permit some degree of failure, but usually if more than 20% of the LEDs fail, the entire lamp must be shut down and a fault reported to maintenance teams.

High-ambient temperatures inside the lamp housing can lead to LED driver failures. This is particularly true if the LED driver circuit contains electrolytic capacitors, which vent when hot and eventually lose their capacitance. Some novel LED drivers have been developed that do not need electrolytic capacitors and can operate for several years at high temperatures. Failing LED drivers can give LED lights a bad name; why have LEDs that can work for over 100,000 h if the LED driver fails after a 10,000-h operation?

Streetlights have been built using medium- and high-power LEDs. Although this would seem to be a simple application, high-ambient temperatures and relatively high-power LEDs can give rise to driver problems.

White LEDs are made using a blue LED and a yellow phosphor. The blue photons have high energy and when they combine with the yellow phosphor, the photon energy is reduced. The resultant light has a broad spectrum, with peaks in the blue and yellow regions. One problem is that the high-blue content produces a “cold-white” light. In some cases, white and yellow LEDs are used together to create a “warm-white” light. However, phosphor technology has developed to allow “warm-white” LEDs. In general, warm-white LEDs have a lower efficacy (fewer lumens per watt) because the phosphors used in these LEDs absorb more photons than they emit.

Automotive lighting has many applications; internal lights, head-lights, stop lights, daylight running lights (DRL), rear fog lights, reversing lights, etc. The greatest problem with automotive applications is that the EMI specifications demand extremely low levels of emissions, which are difficult to meet with a switching circuit. The driver circuit is often shielded with a metal box to achieve low-emission levels. Linear drivers are sometimes used to reduce costs, provided that the efficiency is not a critical requirement. Connecting a linear driver to the metal body of the vehicle can be used to dissipate the heat generated.

Automotive stoplights using LEDs have a significant safety advantage over those using filament lamps. The time from current flow to light output in an LED is measured in nanoseconds. In a filament lamp the response time is about 300 ms. At 60 mph (100 km/h), a vehicle travels 1 mile (1.6 km) per minute, or 88 ft./s. In 300 ms, a car will travel over 26 ft. (8 m). Stopping 300 ms sooner, having seen the previous car’s brake lights earlier, could save death or injury. Also, LED brake lights are less likely to fail than filament lamps, due to being able to withstand greater levels of vibration.

Agriculture has benefited from using LEDs because the plant growth is affected by color. Green light has little effect on plant growth, but both red and blue light do. So immediately, the efficiency is improved because green light does not need to be produced; we only have to produce wavelengths that affect the plants. One effective wavelength is ∼660 nm (red) because this wavelength corresponds to an absorption peak in chlorophyll. So red light helps leaf growth, whereas adding blue light (∼450 nm) helps with stem and seed growth (wheat would grow with just red light, but the addition of blue increases the size and quantity of the seeds).

Mood lighting is an effect caused changing the color of a surface and uses human psychology to control people’s feelings. It is used in medical facilities to calm patients, and on aircraft to relax (or wake up!) passengers. Blue light is sometimes used by long-distance truck drivers, to help them stay alert. Note that the LCD backlight in some handheld computers produces a significant amount of blue light, which can lead to insomnia if the computer is used at bedtime.

Generally mood lighting systems use red, green, and blue (RGB) LEDs in a “wall wash” projector to create any color in the spectrum. Other applications for these RGB systems include disco lights! One mood lighting application is an alarm clock that has an LED lamp to simulate dawn, with the light level starting very low and gradually increasing. The same manufacturer makes seasonally affected disorder lights to help people affected by depression during the long-winter months (in the Northern hemisphere at least).

Backlighting displays, such as flat screen televisions, also use RGB LED arrays to create a “white” light. In this case the color changes little, ideally not at all. However, a control system is required to carefully control the amount of red, green, and blue to create the exact mix for accurate television reproduction. To create a wide-dynamic range, the intensity of the backlight is dynamically controlled, so for dark scenes the backlight is dimmed and for bright scenes the backlight is brightened. Global dimming means the whole backlight brightness has a single control. Some backlights are divided into sections, for example, six-horizontal bands, so that each section’s brightness can be controlled independently. This latter scheme works because most outdoor scenes have a dark ground and a bright sky.

LEDs are also used to backlight computer screens, cashpoint displays, ATM machines, etc., but here the exact color or brightness is less important. In some cases the backlight is used to help create a “corporate” color scheme.

The term candela is also used. This is the light produced by a lamp, radiating in all directions equally, to produce 1 lm/sr. As an equation, 1 cd = 1 lm/sr. A steradian has a projected area of 1 m², at a distance of 1 m from the light source. The light from a 1-cd source, at a meter distance, is 1 lux or 1 lm/m² (Fig. 2.3).

Light emission efficiency (luminous efficacy) from LEDs is described in terms of lumens per watt. There is some competition between LED manufacturers get the highest luminous efficacy, but when comparing results it is important to make a note of the electrical power levels used. It is easier to make an efficient 20-mA LED, than an efficient 700-mA LED. As already mentioned, this is why using LED arrays of multiple low-power LEDs are now preferred over using a few high-power LEDs.

If there are temperature differences between two LEDs, this will cause differences in their forward voltage drop. As the junction temperature rises, it is easier for electrons to cross the energy barrier. In fact, the voltage drop reduces by approximately 2 mV/°C as the temperature rises. This is a problem if, for example, two LEDs are connected in parallel: the hotter LED will have a lower forward voltage drop, which will mean that is takes a greater share of the total current, which means that it gets even hotter and its forward voltage drop becomes lower, and the current rises further, etc. In other words, thermal runaway. In LED arrays, which have series/parallel connections, good thermal coupling between LED die is essential, and is achieved by mounting the LED die on a ceramic substrate.

LEDs are not really a constant voltage load. As the semiconductor material is not a perfect conductor, some resistance is in series with this constant voltage load (Fig. 2.4). This means that the voltage drop will increase with current. The equivalent series resistance (ESR) of a low-power, 20-mA LED is about 20 Ω, but a 1-W, 350-mA LED has an ESR of about 1 Ω. Manufacturers are continually working to reduce the ESR. The ESR is roughly inversely proportional to the current rating of the LED. The ESR will have production variations too.

The ESR can be calculated by measuring the increase in forward voltage drop divided by the increase in current. For example, if the forward voltage drop increases by from 3.5 to 3.55 V (a 50-mV increase) when the forward current goes from 10 to 20 mA (a 10-mA increase), the ESR will be 50 mV/10 mA = 5 Ω.

In Fig. 2.4, the Zener diode is shown as a perfect device. In reality, Zener diodes also have ESR, which can be higher than the ESR of an LED. For initial testing of an LED driver, a 3.6- or 3.9-V Zener diode (rated for 5-W power) can be used to replace a white LED. If the driver is not working as planned the Zener diode may be destroyed, but this is far less costly than destroying a power LED. As the Zener diode does not emit light, the test engineer will not be dazzled. However, be aware that Zener diodes can get very hot. I built an LED driver test jig using wire-ended Zener diodes mounted in block terminal strips (connected in series for easy connection). The first block terminal strips that I used were made from molded nylon, which melted due to the heat generated by the Zener diodes.