INTRODUCTION
Light is a form of radiant energy that travels in waves made up of vibrating electromagnetic fields. These waves have both a frequency and a length, the values of which distinguish them from other forms of energy on the electromagnetic spectrum such as heat and radio.

Visible light represents a narrow band between ultraviolet (UV) light and infrared energy (heat). These light waves are capable of exciting the eye's retina, which results in a visual sensation called sight.

The most powerful light source in the universe, relative to us, is the sun. Light is also found in nature from lightning to fire to bioluminescent bugs and fish. Today, most of the light that people use to work and play is produced by electric lighting systems. The invention of electric lighting gives us unprecedented power and freedom that we take for granted. With Americans spending 80 percent of their lives indoors, most of the world we experience in life, as we see it, is defined by electric lighting systems.

An electrical lighting system converts electrical energy input into useful light output and heat. Nearly all lighting systems produce light by passing an electrical current through an element that heats until it glows, or through gases until they become excited and produce light energy. Incandescent lamps are an example of the first method. Fluorescent and high-intensity discharge (HID) lamps are examples of the gaseous discharge method. Gaseous discharge sources require a ballast, which is an electrical device that provides the initial voltage to start the lamp, then regulates the current flowing through it.

The first patented lamp (1880) was a carbon-filament vacuum incandescent lamp that operated using DC power, and for more than 100 years this light source has been the subject of development. However, its method of operation (incandescence) is relatively inefficient, converting only a small fraction of its electrical input into visible light. The rest is dissipated as heat. For this reason, fluorescent and HID sources were developed and the industry continues to improve these sources in a never-ending quest to increase efficiency and performance.

Light source development is one of the most dynamic aspects of the lighting industry. The introduction of a new light source can solve design problems, improve energy efficiency, trigger new fixture and ballast designs and result in entirely new industries and ways to light traditional spaces. For example, the introduction of the T5/HO lamp sparked a small revolution in indirect lighting, as fixture manufacturers developed smaller fixture designs to accommodate this slimmer lamp and distribute its high lumen package properly across the ceiling. The MR16 offered a new choice in display, museum and other applications. And LEDs are offering new colorful and highly efficient choices for accent, cove and other forms of lighting.

Today, thousands of lamp types are available to the specifier for almost every conceivable lighting need. Light source selection is determined by application requirements, economics and aesthetics. To meet these needs, the designer must consider the source's:

•  Efficacy
•  Color quality
•  Suitability for the operating environment
•  Light output
•  Service life (rated life)
•  Size
•  Electrical operating characteristics
•  Maintainability

In this article, we will review the major lamp types, including fluorescent, high-intensity discharge (HID) and incandescent/halogen. For the sake of brevity, mercury and low-pressure sodium lamps have been omitted from in-depth discussion. In addition, emerging light sources such as LEDs, OLEDs, sulfur lamps, electroluminescent lamps, cold cathode and electrodeless lamps have been omitted.

Before we review the major lamp types, we need to review the common characteristics of light sources that provide tools of comparison.

TOOLS OF COMPARISON
To properly compare light sources to determine the best lamp for the application, one must consider a range of factors, including type, efficiency, size, color, lumen depreciation, rated life and others. The important factors that warrant explanation are efficiency, color, rated life and lamp lumen depreciation.

Table 1.  Comparisons of major lamp types

Efficiency
Efficiency is a cost-benefit metric that describes the effectiveness of a light source or lighting system at producing light based on how much energy it consumes. Lighting efficiency, or efficacy, is expressed as the amount of light produced per unit of electrical input (lumens per watt or LPW or lm/W), can be viewed as a "miles per gallon" figure. The higher the LPW rating for a given lamp or system, the more efficacious it is.

The lamp manufacturers publish input watt ratings for their lamps in their catalogs. Rated input wattage is the amount of power the lamp requires to operate at any given instant in time. [Power should not be confused with energy. As the system operates over time, we take power (W) x Time (hours) to determine how much energy it consumes (kWh) and costs the owner (kWh x $/kWh charged by the power company).]

Any lamp that is not incandescent or halogen, however, requires a ballast, so we have to regard input wattage not for the light source, but both the light source and the ballast operating together as a system.

The lamp input wattage ratings in catalogs are ratings based on operation with a "perfect" ballast under laboratory conditions. No ballast is perfect, however, and the field is no laboratory. Therefore, the specifier cannot compare light sources using only the data in lamp catalogs. The specifier should use the ballast catalog to compare the systems that will actually be operating in the installation.

Ballast manufacturers publish up to three input wattage ratings for their ballasts operating various fluorescent lamps. The ANSI number is the result of a manufacturer test conducted using an ANSI-specified procedure (typically called the "bench test" because the lamps and ballasts are operated bare on a bench). The next one or two ratings are the manufacturer's ratings based on tests conducted in actual open and/or enclosed fixtures. The manufacturer's tests are considered more realistic simulations because they're based on fixtures operating in typical field conditions. When comparing the relative efficiencies of lighting systems, however, the ANSI number is desirable because it reflects results from a common test procedure and therefore is more helpful to make a pure comparison.

Figure 1.  Lamp-ballast system efficacies

Color Characteristics
Color is a critical characteristic of light sources and specifiers regularly rely on common metrics to compare light sources and determine the right source for the job at hand.

As the rainbow demonstrates, visible white light is made up of colors. Electric light sources are typically concentrated in certain colors, resulting a color appearance of the light and what colors these sources will bring out in objects. For example, under a low-pressure sodium lamp, which is monochromatic (completely saturated in yellow), all objects will appear either yellow or a shade of gray. Electric light sources also vary in their ability to render colors as they naturally appear. To understand how the light produced by a given light source will affect the color of objects in the space, three tools of comparison are used, including color temperature, color rendering and a tool called the light source's spectral power distribution. For a comprehensive discussion on the topic of light and color, click here.

Color Temperature
Color temperature, expressed on the Kelvin scale (K), is the color appearance of the lamp itself and the light it produces. Lamps are considered warm (3500K or less; red-yellow or orangish-white in appearance), neutral (3500K to 4000K; neutral or white appearance), or cool (4000K or higher; bluish-white appearance). Contrary to what we'd expect using intuition, the lower the color temperature is, the warmer the light source, and vice versa.

Figure 2.  Typical color temperature ratings for common lamp types

Color Rendering
Color rendering, expressed as a rating from 0 to 100 on the Color Rendering Index (CRI), describes how a light source makes the color of an object appear to human eyes and how well subtle variations in color shades are revealed. The higher the CRI rating is, the better its color rendering ability.

Figure 3.  Typical CRI values for common lamp types

Standard incandescent lamps have a CRI rating of 100 (and the lowest efficiency of all of the lamp types). Fluorescent lamps are in the range of 52 to 95, depending on the lamp. Advances in phosphor technology have enabled fluorescent and HID lamps to advance greatly in color rendering.

A common misperception is that color temperature and color rendering are the same; they are not. Again color temperature describes the color appearance of the light source and the light emitted from it. Color rendering describes how well the light renders colors in objects. However, the two metrics are interconnected: To compare the CRI ratings for any two given lamps, they must have the same color temperature for the comparison to have any meaning.

Spectral Power Distribution
Spectral power distribution shows the visible light spectrum and the wavelength composition for the light from the lamp (see Figure 4). The spikes indicate that the light is stronger in revealing certain colors.

Figure 4.  Spectral Power Distribution Curve for a 4100K fluorescent lamp with a triphosphor (red, blue, green) coating to improve color rendering (and lumen maintenance). Courtesy: Osram Sylvania, Inc.

Rated Life
Rated life is not the life expectancy of a lamp; it is a median or average value, defined as the point of time at which one-half of a large group of lamps can be expected to fail. This is because individual lamps do not fail in a predictable manner, while large batches of lamps fail somewhat predictably, expressed as the lamp's "mortality curve" (see Figure 5). The figure shows us that fluorescent lamps begin to fail just before 60 percent or rated life and after this point the rate of failure accelerates.

Figure 5.  Typical mortality curve for fluorescent lamps. Large groups of lamps tend to follow this curve

The exception to this rule is mercury lamps, an HID light source largely obsolete but still appropriate for some applications. After 24,000 hours of operation, 17-40 percent of a large group of lamps are expected to fail, and the rest generally will produce a depreciated level of light output.

Published fluorescent lamp life is based on three hours per start and HID lamp life is based on 10 hours per start. Lamp life is affected by how often the system is switched on and off, as most deterioration of the lamp electrodes occurs during startup. Fluorescent lamps generally experience a 30% greater average life at 10 hours per start; 50% greater average life at 12 hours per start; and 60% greater average life when operating continuously.

Designers must know the rated life of the lamps that will be deployed in their designs to select the optimal light source for the project and also to compensate for light loss that will occur as lamps burn out during operation. One should always design with maintenance in mind. For example, an incandescent lamp may provide the right color quality for a certain application, but if the lamp burns out frequently and is hard to reach, the fixture will become a major headache for the owner.

Lamp Lumen Depreciation
As all lamps age and near end of life, they produce less and less light. This phenomenon, called lamp lumen depreciation, is caused by several conditions, including deposits forming over time inside the glass bulb wall, deterioration of the electrodes or filament, and deterioration of the lamp's phosphor coating. The rate of deterioration for various lamp types occurs on a predictable curve, which is the lamp lumen depreciation (LLD) curve (see Figure 6).

Figure 6.  Lamp lumen depreciation curves for major lamp types

Various light types experience different degrees of lumen depreciation, as can be seen in Figure 6. Halogen lamps, for example, experience little lumen depreciation compared to mercury vapor. With fluorescent lamps, lower operating current and improved phosphor coatings typically improve relative lumen maintenance.

LLD is a light loss factor, expressed as a decimal (0.91, 0.88, etc.) that must be addressed during lighting design. This factor is considered "recoverable" because the value is not fixed; the actual depreciation value depends on the maintenance program that will be employed. In simple terms, if we have a lamp that produces 10,000 lumens of light output and operates with an LLD factor of 0.9, then over time the lamp will produce only 9,000 lumens for distribution to the workplane. This, of course, impacts light level and causes designers to build more light output (called a "depreciation cushion") into the system so that over time, with all light loss factors such as LLD accounted for, it will produce desired light levels.

How do we view the LLD curve and come up with an LLD factor? There are two simple methods. If we employ a planned maintenance strategy and group relamp (replace all lamps in a lighting system en masse to improve economy and the system's light output), then we first determine when we will group relamp (typically at 70 percent of rated life), then find our LLD factor on the lamp's LLD curve. So if we have a lighting system comprised of T8/RE70 lamps, and we group relamp at 70 percent of rated life, then, using Figure 6, we see that our LLD factor will be 0.94.

Many facilities, however, do not employ planned lighting maintenance, replacing lamps individually as they burn out. In addition, for a new facility the lighting designer may not know if the facility will have a planned maintenance program. Therefore, they must determine the LLD factor as a median. On a curve, the median is 40 percent of rated life.

FLUORESCENT LAMPS
Today, some three out of four lamps in commercial buildings are fluorescent lamps and they are available in a broad spectrum of shapes, sizes, wattages and color characteristics.

Figure 7.  Fluorescent lamp components

Fluorescent lamps are low-pressure gaseous-discharge lamps; being gaseous discharge, they require a ballast to operate. Named for the ballast used to stabilize old wooden ships in choppy seas, the lighting ballast provides the initial voltage to start the lamp, then regulates its operation. When switched on, the ballast transforms the supply voltage to the starting voltage required to initiate an arc between the lamp's electrodes (cathodes). This arc excites the gases contained in the bulb and produces UV light energy, which in turn excites the phosphor coating on the bulb wall and produces "fluorescence" -- visible light. Compared to incandescent and HID light sources, which are point sources, fluorescent lamps generally offer lower surface brightness and less shadowing.

Table 2.  Advantages and disadvantages of fluorescent lighting. Note that while low-pressure sodium (LPS) lighting is technically fluorescent, it is omitted from this table and this section for brevity, given its limited application

Figure 8.  Shapes of common fluorescent lamps

Figure 9.  Fluorescent lamp colors

Lighting Circuits
Fluorescent lamps may be designed to operate in preheat, instant start or rapid start systems. Rapid start systems are most popular.

In preheat systems, the ballast preheats the lamp's electrodes until sufficiently heated, at which time the ballast uses an additional component, a starter, to supply the starting voltage. Two or three seconds of warm-up time are required before the lamp is ignited. Today, most preheat lamps are small tubular fluorescents consuming 4-20 watts, and are capped with a two-pin (bi-pin) base. In preheat systems, up to two lamps can be operated by a single ballast; the ballast operates them in parallel, meaning if one lamp goes out, the other will continue to light.

In instant start systems, the ballast provides a higher starting voltage, heating the electrodes quickly to start the lamps immediately. While instant start lamps do not require a starter to ignite, a larger ballast is often required. Instant start lamps are even today called "slimline" because the first instant start lamps, introduced in 1944, were thinner in diameter than the preheat lamps used at the time. Instant start lamps are easily recognizable by their single-pin base capping each end of the tube. In instant start systems, up to two lamps can be operated by a single ballast. Lead-lag ballasts operate the lamps on a parallel circuit, meaning that if the first lighted lamp on the circuit fails, the second, or remaining, lamp will continue to light. Note that some electronic ballasts provide instant start operation for rapid start bi-pin lamps.

In rapid start systems, the ballast constantly heats the electrodes, which negates the need for a starter or a larger ballast. The lamp lights almost immediately. Rapid start lamps are capped with bi-pin bases. Up to four lamps can be powered by a single ballast on either a series or parallel circuit, depending on whether the ballast is electronic or magnetic. Series sequence ballasts operate the lamps in series, meaning that if the first lighted lamp on the circuit fails, its companion will fail to light. If the second lamp on the circuit fails, the first lamp will be stuck in starting mode and will light only dimly. Using a special ballast and control, the lamps are dimmable. In the rapid start family, high output (HO) and very high output (VHO) lamps (also called 1500-milliamp lamps or Power Groove lamps) operate at higher electrical currents (typically 800mA or 1500mA respectively) to produce higher light output for applications that require it. They feature a recessed double-contact base.

Table 3.  Comparison of fluorescent lamp types by starting method

* Note that compact fluorescent lamps (CFLs) feature up to a dozen base types in single-ended bi-pin, 4-pin and screwbase configurations.

** Rapid start HO and VHO operate at 800mA and 1500mA of current respectively. T12 lamps and T8 lamps share medium bi-pin bases but T12 lamps operate at 430mA and T8 lamps operate at 265mA; while T8 lamps were designed as an alternative to T12 lamps, they require a designated T8 ballast.

HIGH-INTENSITY DISCHARGE (H.I.D.) LAMPS
HID lamps are similarly constructed in that they feature an arc tube of stress- and heat-resistant material that contains gases, metals and the electrodes. HID lamps are identified via the predominant distinctive metals contained in the arc tube: high-pressure sodium (sodium), mercury (mercury) and metal halide (metallic halides). The arc tube is housed in a protective glass envelope. When starting voltage is applied to the electrodes from the ballast or ignitor, an arc is formed between them. Electrons in the arc stream collide with atoms of vaporized metals. The result of this action is the emission of light energy. Due to the high pressures of HID lamp operation, these wavelengths are concentrated in the visible light spectrum and therefore do not require a phosphor coating as a filter.

Table 4.  Advantages and disadvantages of HID lighting

Of the three types of HID lighting, high-pressure sodium and metal halide are the most efficacious and offer the best color, limiting mercury's use. Metal halide offers superior color quality with a bright white light, while most high-pressure sodium offer the greatest efficiency at the expense of color with an orangish light. For brevity, we will describe only high-pressure sodium and metal halide light sources.

High-Pressure Sodium (HPS) Lamps
The core operating components of an HPS lamp include two main electrodes within a slim arc tube that is constructed of a tough ceramic material that contains a small amount of sodium, mercury and zenon gas. Because the arc tube is very slim, the starting electrode's function is performed by a remote electronic starter. The arc tube is enclosed in an evacuated bulb made of borosilicate glass. The complete assembly is capped by a base that is most often of mogul design, although 35W lamps feature a medium base and wattages up to 150 watts are available with a medium base. Double-ended lamps are also available.

Figure 10.  HPS lamp

When an HPS lamp is activated, the ballast supplies a starting pulse through the ignitor (2,500-3,000 volts). Electrons in the arc stream collide with atoms of zenon gas and vaporized mercury and sodium, producing radiant energy. Due to operation at high pressures, light wavelengths that are produced are concentrated in the visible light spectrum. These wavelengths are saturated with colors that produce a warm yellowish or orangish light. Standard HPS lamps offer color temperatures from 2100K to 2700K. Typical HPS lamps are clear but lamps are available with a coating to diffuse light.

Startup is almost immediate, with full light output achieved in 3-5 minutes. A common characteristic of HID and low pressure sodium lamps, should the lamp be extinguished due to even a momentary power interruption, is that it must cool before restarting. This cooling time before the arc tube can be initiated again is called the restrike time. The hot restrike time for HPS lamps is less than one minute and full light output is reachieved in 3-4 minutes.

HPS lamps, like all HID sources, may be prone to a stroboscopic effect called "strobing." A gaseous discharge lamp operates by a magnetic ballast on a 60Hz power supply will cycle 60 times per second. Each cycle in theory appears as a sine wave; the voltage rises, then declines as the arc is operating and the lamp is on. The voltage reaches the zero point, at which time the arc ceases and the lamp does not produce light until the voltage curve reaches "on" again. The human eye generally cannot detect strobing, but when viewing a rotating machine such as an electric motor turning at speeds of multiples of 60 (such as 1,800 rotations per minute), the machine may appear to be motionless. This can be advantageous as it provides the opportunity to study the machinery. However, if unplanned, it can be seriously hazardous. To address strobing, typical options include staggering the lamps across the phases of a 3-phase power system, specifying supplementary incandescent lighting, or specifying an alternative light source such as metal halide, which does not experience strobing to the same extent.

Figure 11.   Typical HID lamp shapes

Metal Halide
The core operating components of a typical metal halide lamp include three electrodes within an arc tube that contains mercury and other metals in iodide form. The arc tube is constructed to withstand the internal high pressures and stresses of HID lamp operation. The tube is enclosed in a borosilicate glass bulb that is also highly heat resistant. The entire assembly is capped by a base that is most often of screw-shell design, although medium bases are available for lamps operating on 100 or fewer watts, and mogul bases are featured on most high-wattage lamps.

Figure 12.  Metal halide lamp

When a metal halide lamp is activated, the ballast applies starting voltage to the three electrodes, although electrical resistance is too high at this time to initiate an arc between the two main operating electrodes, the starting electrode is located near enough to one of them to initiate an arc. As the amount of vaporized metal in the arc stream and pressure builds, mercury and metallic halide atoms collide with free electrons in the stream, producing radiant energy. Due to the operation at high pressures characteristic of HID lamps, wavelengths are concentrated in the white light spectrum with only residual ultraviolet light and heat. Due to the advanced combination of metallic halides, these wavelengths are more evenly dispersed, improving color appearance and color rendering to a cool, bright white light at 4200-4500K. Some metal halide lamps feature a phosphor coating on the inside of the outer bulb that converts UV wavelengths to visible light to effect a warmer color at 3200K.

A metal halide lamp reaches full light output after 2-10 minutes. Certain low-wattage metal halide lamps require the help of an electronic starter, or ignitor, to perform the function of the starting electrode. Should the lamp be extinguished, the hot restrike time is 12-20 minutes, at which time the startup process will begin again. Light is not produced in the interim, which may require the specification of suitable standby lighting or selection of a different lamp type.

To reduce cold start and hot restrike times for metal halide lamps, specifiers can consider pulse start lamps. These lamps work with a high-voltage pulse ignitor that "jump starts" the lamp. According to Venture Lighting, 175/200/400/1000W pulse start lamps start 50 percent faster, start at temperatures as low as -40 degrees Celsius, and achieves hot restrike 50 percent faster. In addition, due to less wear on the electrodes with a faster start, an increase in lamp life and lumen maintenance may result. (Hot restrike is the restriking of the arc in an HID light source after a momentary power loss; the lamp restarts when the arc tube has cooled sufficiently.)

A variety of metal halide lamp designs are available, including conventional design (restricted to enclosed fixtures, can operate in any position), protected design (for use in open fixtures), safety design (extinguishes itself if the outer bulb is broken or punctured to prevent leakage of potentially harmful UV radiation), high light output design (phosphor-coated, 3200K color temperature), double-ended design (offering greater efficiency, restricted to a horizontal operating position in enclosed fixtures), and restricted burning position design (must operate in a given operating position, vertical base up, horizontal, etc.).

Should a metal halide lamp be operated continuously, the lamp should be turned off and allowed to cool for 15 minutes, then restarted, once per week to prevent "non-passive failure" at the end of life. Natural lamp failure is due to electrode deterioration that occurs during start-up and operation. The lamp will show signs of failure that include inability to light or reach full light output. Some metal halide lamps may fail in a non-passive manner, spraying hot glass from the shattered arc tube; these should be protected with a special design lamp and/or specified in an enclosed protective fixture. In addition, as most metal halide lamps approach end of life, their color appearance will shift to a bluish or pinkish light.

INCANDESCENT/HALOGEN LAMPS
Incandescent lamps achieve light through "incandescence." When voltage is applied across a metal element, this element will heat until it glows.

In most incandescent lamps, this element is a filament -- a coiled or double-coiled tungsten wire-housed in a glass bulb that contains either a vacuum or inert gas.

Incandescent lamps are sensitive to variations in the supply voltage that may occur, impacting service life, wattage and light output. Operating a 120V incandescent lamp on 126 volts, five percent overvoltage, will cut the lamp's life in half (while producing greater light output). The converse is also true; undervoltage increases life at the expense of light output.

Figure 13.  Common incandescent lamp shapes   Incandescent lamps are also sensitive to vibrations, which can break the filament. Rough-service lamps, featuring strengthened filaments, are available for applications where the lamp is exposed to vibrations and shocks, such as near a heavy metal door. Silicone-coated lamps are also available, which resist shattering when the lamp is damaged. In halogen (tungsten halogen or quartz halogen) lamps, a halogen gas is used inside a small quartz capsule that encloses the filament. During incandescent lamp operation, the current slowly evaporates the filament, resulting in bulb wall blackening as particles collect there; in halogen lamps, the halogen gas combines with these evaporated particles, which in turn are attracted back to the filament, prevent bulb wall blackening and enabling the lamp to operate at higher temperatures provide greater light output, better lumen maintenance and longer service life.

Table 5.  Advantages and disadvantages of incandescent lamps

Rough service incandescent lamps include additional strength in the filament for use in applications where the lamp is exposed to vibrations and shocks. Silicone-coated lamps offer resistance to breaking.

Incandescent and halogen lamps are available in a variety of reflector designs, including R (reflector), PAR (parabolic aluminized reflector), ER (elliptical reflector) and MR16 lamps. The built-in reflector shell aims the light and produces a range of beam patterns from very narrow (spot) to very wide (flood). ER lamps focus the light several inches in front of the bulb, making them ideal for use in downlights by enabling more light to escape the fixture. MR16 lamps are popular because of their small size; these halogen lamps require a transformer that reduces the voltage from 120V or higher to the right voltage required to operate the lamp.