Tech in progress...
Updated: 01/15/2004

I have a lot of useful information about my products in particular and neon versus LEDs in general, but I just don't have the time to write it all out yet. I have started with some basic background information, but the format of this page will slowly become more organized as the content level increases.

Much of the information is quite technical in nature, but I have added in some of my own commentary (in yellow), where appropriate, to highlight some of the more useful points.

Some basic LED history...

Commercial research into LED technology started in the early 1962s, notably at Bell Labs, Hewlett-Packard, IBM, Monsanto, and RCA. Work on gallium arsenide phosphide (GaAsP) led to the introduction of the first commercial 655nm red LEDs in 1968, by H-P and Monsanto. In 1971 H-P released the 5300A 500MHz portable frequency counter using a GaAsP LED display. LED displays flourished in the early 1970s as numeric displays in pocket calculators by H-P, Texas Instruments, Sinclair and others. For a short time, LEDs appeared in digital watches, but were soon replaced by LCDs. Meanwhile, LEDs replaced incandescent and neon lamps as status indicators and became the standard numeric and alphanumeric display choice for instrumentation.

The LED's hottest competition in the 1970s and 1980s for consumer goods came from vacuum fluorescent displays (VFDs), whose bright blue-green display offered high intensity and high contrast when viewed through a green or blue filter. VFDs were first developed by ISE Electronic Corporation in 1967. ISE, often known by the division name of Noritake, together with Futaba and NEC, offered display tubes from the late 1960s and early 1970s, starting with simple single digit displays used in rapidly growing desktop calculator market. Multi-digit display tubes appeared soon, reducing manufacturing cost, and these are possibly best remembered for their appearance in the popular Casio pocket calculators. Later, Samsung started making tubes for their own consumption for use in consumer goods. In 1993, NEC sold their complete manufacturing line to ZEC in China, and between them Futaba, ISE, Samsung, and ZEC produce around 95% of the world's VFD tubes production.

In the 1980s and onward, monochrome LCDs competed strongly with LEDs and VFDs for consumer devices, instrumentation, and automotive panels. LCDs have the advantage of lowest power and easily customization, and became the obvious choice for battery operated applications. Although LCDs don't emit light, there are many applications where ambient light can be guaranteed. Alternatively, the light from a couple of green, orange, or yellow LEDs can be diffused and spread behind a small (10 square centimeter) LCD with an opaque plastic molding, to provide a cheap and pleasant backlight.

What are LEDs?

A light emitting diode (LED) is a PN junction semiconductor diode that emits photons when forward biased. The light emitting effect is called injection electroluminescence, and it occurs when minority carriers recombine with carriers of the opposite type in a diode's band gap. The wavelength of the emitted light varies primarily due to the choice of semiconductor materials used, because the band gap energy varies with the semiconductor. Not all injected minority carriers recombine in a radiative manner in even a perfect crystal; non-radiative recombination occurring at defects and dislocations in the semiconductor can give rise to wide variations in useful emissions in seemingly identical diodes. This means in practice that manufactured batches of LEDs are sorted and graded for intensity matching.

Well, that's definitely a technical description, but unless you've had some training in semiconductor physics, it probably doesn't make a whole lot of sense. ;) The main point one needs to understand is an LED is nothing more than a basic diode with the added advantage that it emits light when power is running through it. As with any diode, an LED will only allow electricity to pass in one direction. In fact, you could replace any standard diode with an LED without effect, as long as the basic specs are the same (which will be described later on). Since the specs of an LED can vary slightly, even when the LEDs are of the same type, manufacturer, production lot, etc., there can (and will) be variations in brightness from one LED to the next when placed into a circuit. This should be considered when more than one LED will be used in the same locale.

Anatomy of an LED A typical LED

LEDs are processed in wafer form similar to silicon integrated circuits, and broken out into dice. Chip size for visible signal LEDs generally fall in the range 0.18mm square to 0.36mm square. InfraRed (IR) LEDs can be larger to handle peak powers, and LEDs for lighting are larger again.

The simplest packaged LED product is the lamp, or indicator. The basic structure of an LED indicator consists of the die, a lead frame where the die is actually placed, and the encapsulation epoxy which surrounds and protects the die, and also disperses the light (Figure 1). The die is bonded with conductive epoxy into a recess in one half of the lead frame, called the anvil due to it's shape. The recess in the anvil is shaped to throw the light radiation forward. The die's top contact is wire bonded to the other lead frame terminal, the post.

LEDs come in various package sizes and types with the T1-3/4 (5mm) probably being the most recognized. As you'll see shortly, the light's dispersion pattern is due in part to the shape of the anvil, but a significant portion of the pattern is due to the surrounding epoxy packaging. A domed lens will concentrate the light into a narrow beam, whereas an inverted cone will concentrate the light towards the sides of the lens. For simple projects, the top of the domed lenses can be cut off to increase the dispersion angle.

Radiation patterns LED radiation pattern

The mechanical construction of the LED lamp determines the dispersion or radiation pattern of the light. A narrow radiation pattern (Figure 2) will appear very bright when viewed on-axis, but the viewing angle will not be very wide. The same LED die could be mounted to give a wider viewing angle, but the on-axis intensity will be reduced. This tradeoff is inherent in all LED indicators, and can be easily overlooked. High brightness LEDs with a 15 to 30 viewing angle are a good choice for an information panel directly in front of an operator, but a wide direction indicator or automotive dashboard might require as wide an angle as 120.

The radiation pattern is an oft overlooked specification. In many cases, particularly with enthusiasts without engineering backgrounds, the brightness rating (mcd) is the sole value upon which a purchase decision is made. When choosing an LED for a specific application, an LED's viewing angle (half angle) should also be considered. The next section describes this in more detail.

Viewing angle

Viewing angle is one of the most confusing terms regarding LEDs. Most people assume that the viewing angle is the angle at which you can view an LED's light. In reality, viewing angle is related to brightness. The LED industry defines viewing angle as the full angle at which brightness is half of the brightness from dead center. More scientifically, if (angle theta) is the angle from off center (0) where the LED's brightness is half, then 2 is defined as the full viewing angle, as represented by Figure 3 below. The entire yellow portion represents a 70 viewing angle, while the orange portion represents a 90 viewing angle.

LED viewing angle 1
Figure 3: LED Viewing Angle.
Brightness in relation to LED viewing angle
Figure 4: Example brightness based on a 70° viewing angle.

Figure 4 illustrates how brightness is influenced by a viewing angle of 70. An LED with a brightness of 8,000 mcd (millicandela) viewed at 50 from off center will be just as bright as an LED with a brightness of 2,000 mcd viewed head on. While an LED is not as bright when viewed at an angle beyond the stated viewing angle, there is still sufficient brightness to see the LED.

The above paragraph illustrates a VERY important point, one which I hinted to in the last section. The unknowledgable purchaser might assume that a 5,000 mcd LED is better than one rated at 4,000 mcd, but to be sure the viewing angle must be taken into account. If the first LED has a viewing angle of 10, and the second has a viewing angle of 100, the first LED will only appear brighter when viewed nearly directly on. At just 5 off center, the first LED only appears to be a 2,500 mcd emitter while the second LED still appears to be nearly a 4,000 mcd emitter. At 45 off center, the first LED will hardly have a visible glow while the second LED will appear to be a 2,000+ mcd emitter. If the application is for a light flood (such as when replacing neon tubes), the second LED is far superior and will appear significantly brighter overall compared to the first LED.

Optical Characteristics of Our Eyes and LEDs Human eye daylight color response

There are two definitions appropriate to this discussion:
1) Radiometry - The measurement of radiant energy at all wavelengths (visible and invisible).
2) Photometry - The measurement of apparent brightness to the human eye.
The human eye "sees" the range of light wavelengths from 380nm to 740nm as the familiar color spectrum. The Commission Internationale de l' Eclairage (CIE) formalized standards for the measurement of light, and the response of the human eye or "standard observer", back in the 1930s. These standards characterized the variation in eye response over the entire visible range under a variety of lighting conditions, such as daylight and night. The CIE also defined the primary colors (red - 700 nm, green - 546.1nm, blue - 435.8nm). I'm not sure why the CIE was so specific in their determination of color wavelengths, but my guess would be the values are merely averages of the wavelength chosen by a large sample of people as being "red", "green", or "blue".

These standards and definitions have been controversial, and other standards exist. The points of interest for LEDs are that the human eye response peaks roughly at green 555nm, is sensitive to yellow, falls off sharply towards blue at 400nm, and also towards red at 700nm. This can be seen in the 1931 photopic (day vision) chromaticity diagram, which is shown in a simplified form in Figure 5. The curve for scotopic (night vision) is quite different, peaking at about 512nm (blue).

Some interesting little tidbits...

If you would like to be seen as easily as possible in a large crowd during the day, wear green. Since our eyes are most sensitive to that wavelength, our eyes tend to pick out that color from a jumble of colors. If you really want to get technical, our eyes are most sensitive to green for historical/biological reasons...since we've evolved over the millenia from simple, vegetation-eating animals (in one form or another), our eyes adapted to pick up on the bright green reflection of leaves.

During the nighttime, our eyes are most sensitive to blue (512 nm). So it should come as no surprise that the blue LEDs many car alarms are using these days can be seen from a greater distance compared to some of the red LEDs. While it's true that the LEDs themselves are significantly brighter than those of years past, our eyes' sensitivity to those wavelengths gives an extra boost.

How is light measured?

Radiant light intensity (all wavelengths) is measured in lumens. The lumen is defined such that 683 lumens of light is provided by 1 watt of monochromatic radiation at a wavelength of 555nm. Luminous intensity, in candelas (cd), results from the application of the CIE color response to the radiant flux and provides the measurement for the visible portion of a light source. LED intensity, therefore, is described in cd or mcd to indicate the light output useful to the observer.

A candela is a unit of measurement of a light's intensity. An ordinary wax candle generates one candela. More precisely, candela is the luminous intensity of a monochromatic source which has a frequency of 540 TeraHertz (555 nm, or green).

If you stop and think about this, it is somewhat of a conundrum. If the intensity of a light is measured based upon the amount of light it emits at 555 nm (green), how is anything measured for LEDs that are NOT green? The answer to this one is reasonably simple...LEDs emit light at a range of wavelengths, but the dominant wavelength is the color seen when viewing the LED. Since the measurement is then weighted based upon our eyes' ability to perceive different colors at different levels of brightness, the candela rating is a good measure of perceived brightness, even between LEDs of different wavelengths (i.e., colors).

But what about lasers since they are a single wavelength of light? Let's not get ahead of ourselves...

LED electrical characteristics LED forward voltage varies with color and current

The electrical behavior of LEDs is similar to other semiconductor diodes. The forward voltage is higher and is dependant upon the different materials used for different colors, as shown in Figure 6. The forward voltage rises with current and falls with temperature by about 2mV/C. And, like all semiconductors, the LED must be derated at higher operating temperatures.

The optical behavior of the LED varies significantly with temperature. First, the amount of light emitted by the LED falls as junction temperature rises. This is because of an increase in the recombination of holes and electrons that make no contribution to light emission. Also, the emitted wavelength also changes with temperature, mainly due to the semiconductor energy gap changing with temperature.

Again, more technical jargon, but the gist of the matter is the LEDs characteristics are affected by color, temperature, and current. Since each LED color is made using different materials, the voltage drop and light output will vary from color to color. For example, red LEDs typically have a negative output change with increases in temperature (i.e., as the temperature goes up, light output goes down). Blue and green LED outputs are less affected by temperature, but their forward voltage is more sensitive to temperature changes. Since resistors have a positive coefficient (i.e., as the temperature rises, so does their resistance), they can be used to offset the negative coefficient of red LEDs, but this will probably not work with other color LEDs.

LED Life Expectancy

LEDs have a MTBF (mean time between failures) usually in the range of 100,000 to over 1,000,000 hours. This is a long time for continuous operation, considering a year is 8760 or 8784 hours (Can't forget leap years). In practice, the useful measure of LED lifetime is its half-life, that is an LED is deemed to have reached the end of its life when the light output falls off to half the original.

When current flows through an LED junction the current flow is not uniform, resulting in small temperature differentials within the chip. These temperature differentials exert stress on the lattice, causing minute cracks to occur. These lattice defects accumulate with use, and reduce the photon conversion efficiency of the chip, so reducing light output. The attrition rate varies according to the LED material, temperature, humidity, and the forward current.

Skipping over all of the techno-babble, LEDs last a long, long time. While the MTBF may be in the 100,000 hour range, the half-life of a typical quality LED is more along the lines of 10,000-20,000 hours. This is easily 1-2.5 years before the LED is noticably more dim, but this also assumes continuous operation at 24 hours a day. Most lighting systems will only be run 1-2 hours a day for vehicular systems and 8-12 hours a day for architectural systems.

White LEDs

There are essentially two technologies for generating white light from LEDs. One way is to mount a red die, a green die, and a blue die very close together within a package, and mix the light outputs in the correct proportions to achieve white light. The problem with this approach, ignoring the technical issues of setting the correct LED drive levels, is the cost of 3 dice. Nonetheless, a major advantage to this approach is the user can set the color to any hue desired.

I tend to prefer the tri-color LEDs myself since they allow me maximum control over their displayed color.

The cheaper approach, pioneered notably by Nichia, involves including a phosphor with the blue LED that absorbs some of the blue light and fluoresces in a second color to achieve a near-white. Some early white LEDs using this technique showed a noticable blue tinge, but the most recent developments are excellent.

If you plan on using this style of white LED by itself, you'll be OK. However, do not try to place a colored filter in front of the LED to change it's color. While the light appears white in color, it is still a mixture of mostly blue and yellow light. So, a red filter will actually show up as a pink color, a yellow filter will show up as a lemon-lime color, a green filter will show up as an aqua color, and an orange lens will show up as yellow. Another strong reason why I use the tri-color LEDs.

Information on this page has been gleaned from multiple sources in various formats (the internet, trade magazines, research papers, etc.). Below is a list of any site from which I have pulled information from. I have attempted to keep this list as up to date as possible, but things will slip through the cracks. Should you happen to come across any information or have a useful piece yourself, feel free to send it to me and I'll credit the source, as appropriate.

Maxim App. Note #1883 -- LEDs Are Still Popular (and Improving) After All These Years
Grandwell Industries -- LED Display Viewing Angle and Brightness