The proper approach and knowledge of application issues is vital when upgrading lighting systems. Energy savings and improved employee productivity are the results of a good project.

EC&M's April '99 issue features an article on lighting efficiency and how to improve it. However, without understanding the possible consequences when upgrading lighting systems, projects can end up badly. If the issues addressed below are not considered, the results can be disappointing and costly. Knowing the potential pitfalls and how to avoid them will lead to a successful upgrade.

Ballast factor. The biggest mistake made in lamp and ballast upgrades is failing to take into account the ballast factor (BF). This aspect of lighting determines light level; the higher the BF, the higher the light level. You should select electronic ballasts based on the light level required, not on maximum energy savings. It's common to choose a ballast with a low input watts rating. When this is done, the ballast factor is usually too low, producing a low light level. Although the initial light levels may seem satisfactory when the lamps are new and surfaces are clean, over time, lamp lumen depreciation and dirt depreciation factors will reduce the light level, sometimes to a level that's too low.

In 1-for-1 upgrades (with the same number of lamps and fixtures) from T12 lamps to T8 lamps, a ballast factor of 0.88 ensures equivalent light output. If the old lighting system provides too much light, consider reducing the light level by specifying a lower BF. To ensure lamp life isn't affected, the lower limit for instant-start ballasts is 0.7. For systems that produce too little light or where the number of lamps will be reduced, consider a higher ballast factor to make up some of the light loss. However, a higher ballast factor comes at the cost of higher power input. The rapid-start ballasts have a higher input power because of the additional power required to heat the cathodes. You can use ballasts with BF up to 1.2 without fear of overdriving the lamps, and they can provide up to 30% more light. Fig. 1 shows the relationship between power input and ballast factor for rapid- and instant-start ballasts.

The color of light. Correlated color temperature (CCT) is a numerical measurement of a lamps color appearance. It's based on the principle that any object will emit light if it's heated to a high enough temperature and the color of the emitted light will shift in a foreseeable way as temperature increases. CCT is based on the color changes of a theoretical "blackbody radiator" as it's heated from a cold black to white-hot state. With increased temperature, the radiator shifts gradually from red to orange to yellow to white and, finally, to blue white. Therefore, a lamp's CCT is the temperature (in degrees Kelvin) at which the color of the blackbody correlates with the color of the lamp.

Lamps with low color temperature are called "warm," while lamps with higher color temperature are called "cool." These descriptions have nothing to do with the temperature of the blackbody radiator, but rather refer to the lamp appearance. Colors from the red/orange/yellow end of the spectrum are described as warm and colors in the blue end of the spectrum are referred to as cool. Looking at standard colors of fluorescent lamps, we see the appearance is warm (3000K), cool (4100K), or neutral (3500K).

Lamp CCT should match the type of activity occurring in the lighted area. For offices, schools, and other work environments, designers usually choose a cool lamp color (4100K). For stores, showrooms, and supermarkets, most prefer a neutral color (3500K). Restaurants and lobbies usually use a warm lamp color. Neutral (3500K) works well for most interior applications and doesn't emphasize either end of the color spectrum.

Avoid mixing lamp colors when upgrading lighting systems. For task specific lighting, ensure the light source has the required portion of the color spectrum you need. For example, if the visual task is sorting blue and black objects, avoid incandescent sources, which are blue deficient.

The color rendering index (CRI) is a system derived from visual experiments and determines how colored objects appear when lighted. It uses eight standard pastel color samples illuminated by the lamp and then by a blackbody matched to the same color temperature. If no samples shift in appearance, the lamp is rated at 100. If there is a shift, the result is a lower CRI rating of the lamp. The CRI uses a relative scale; the higher its value, the closer the lamp is to the standard colors.

CRI values of 80 or above are high. A CRI of 75 works well for offices, whereas, a CRI of 85 is recommended for high attention areas. Changes from lamps with lower CRI values to lamps with high CRI may be perceived as "too bright." This apparent increase in brightness commonly occurs with one-for-one retrofits of 4-ft T12 lamps with 4-ft T8 lamps. The T8 lamp, being smaller in diameter with the same lumen output as the T12, has a higher surface brightness. In addition, because the CRI of the T8 lamp is usually much higher than the T12 it replaces, the T8 is perceived to be brighter.

There is generally a high incidence of occupant complaints on one-for-one upgrade projects that the lighting is "too bright." In these cases, maintenance personnel usually solve the problem by removing one lamp. Relighting projects, where new 3-lamp fixtures relight areas that formerly used 4-lamp fixtures with T12 lamps, have fewer occupant complaints.

Lighting quality. Productivity can be diminished in workspaces where lighting quality is low. There is no one measure of lighting quality, but there are several issues affecting this lighting component that you should recognize such as light level, glare, visual comfort probability (VCP), veiling reflections, flicker, and spacing criteria (SC).

Light level. A host of issues must be given careful consideration before any hasty conclusion is reached regarding the appropriateness of measured light levels. These include, but are not limited to: the type of space and how the space is used, flexibility of use, the desired lighting effects, the visual tasks, owner's liability, and several weighting factors such as occupants age, the importance of speed and/or accuracy in performing a given task, and surface reflectance.

A light meter is typically used to determine if light levels are appropriate. These values can be compared with current recommendations made by the Illuminating Engineering Society of North America (IESNA). These are consensus recommendations made by lighting professionals. However, there are no single value recommendations for any particular space or specific task.

In spaces that are overlighted, lighting requirements must be changed to provide a lower level of light. For spaces that are found to be underlighted (a common occurrence, in stairwells, where owner liability is high), you should ensure that sufficient light is recommended for the safety and security of occupants and visitors. Highlighting safety concerns will strengthen your credibility. These unsafe conditions can become a serious legal liability for you, your employer, or your client if not addressed.

Glare. Controlling glare produced by unshielded bulbs and tubes, and from poorly shielded fixtures, is important to avoid loss of visual performance. Smaller diameter lamps have higher bulb wall brightness that must be taken into account to avoid lamp glare. The T5 lamps are best used in coves and other indirect applications. Use louvers, such as those found on deep-cell parabolic fixtures, to block lamps from view, or use low-glare acrylic lenses to reduce surface brightness at high viewing angles. Indirect fixtures reduce high luminance ratios in interior spaces by uniformly lighting the ceiling cavity.

Visual comfort probability. Visual comfort probability (VCP) is a rating of interior fluorescent lighting systems that is expressed as the percent of people, who, when viewing the system, find it acceptable in terms of discomfort glare. VCP can be used to predict glare potential. You should select VCP values based on the visual tasks. The IESNA recommended practice for office lighting (RP1) recommends a minimum VCP of 70% for offices and 80% where there are a lot of computers and screen glare needs to be reduced. VCP tables are provided in fixture manufacturer's catalogs.

Veiling reflections. Veiling reflections are reflections of incident light superimposed on diffuse reflections. One common example is when you place a white paper on the dashboard of a car. The image of the paper reflects on the windshield and appears as a veil between the viewer and objects outside the car. A similar condition occurs in interior lighting when lens fixtures are in front of and above workstations. Lighting with well-shielded luminaries, located perpendicular to and at sides of workstations, avoids veiling reflections. Test sheets with special ink and paper are available to detect this condition.

Flicker. Flicker is produced by fluorescent or HID lamps when connected to magnetic ballasts that turn lamps on and off 1202 per second. It can cause distraction, eyestrain, fatigue, and nausea. Some people notice effects of flicker more than others do, and it can be especially troublesome in high-level lighting where industrial inspection is being performed. Use of electronic ballasts that operate lamps at high frequency will eliminate any perceptible flicker.

Spacing criteria. Specular reflectors can affect lighting quality. Occupants often perceive lack of quality lighting when retrofits are based on energy savings alone. The problem is not that reflectors are used, but that they are not used correctly. The incentive of a low first cost often drives the selection process to commodity grade reflectors. These "one-size-fits-all" inserts can be placed in nearly all four-lamp fluorescent 224 troffers, and two lamps removed. However, in most cases, the remaining two lamps are not relocated and the spacing criteria (SC) of the unit is severely reduced.

When the SC of luminaries already mounted on wide centers (commonly found in speculative buildings) is reduced, the effects are noticeable on walls, and less noticeable between luminaries. The point at which the light intercepts the wall is lowered when the S.C. of the units closest to the wall is reduced. The result is that walls appear darker. The solution is to either use custom reflectors or provide an asymmetrical designed reflector in those luminaries located in rows closest to the walls. Custom reflector designs do not reduce the SC. Using commodity reflectors results in dark walls, and a dim environment, and most occupants do not believe that anyone has improved their workspace.

Lamp life. The average-rated life for lamps is defined as the point at which 50% of a large group of lamps is still burning. A number of factors relate to lamp life. Also to be considered is the time interval when lamps should be changed. The lamp mortality curve (see Fig. 2, on the original article's page 38) looks like the right hand side of a bell curve.

The rating of fluorescent lamps is based on a test cycle of 3 hr/start (3 hr on, 20 min off). Most fluorescent lamps have an average-rated life of 20,000 hr (at 3 hr/start). For longer burn cycles (greater than 3 hr/start), average-rated life goes up. Burn cycle is the average length of time a lamp operates each time it's turned on. Average-rated life is a function of the number of times that a lamp is started.

The calendar life of lamps is the time between lamp replacement and includes the time lamps are turned off. For example, a standard F40 RS lamp operated continuously results in an average-rated lamp life of 34,000 hr (calendar life is the same; 34,000 hr4 3.9 yr). Turning off this lamp for half of the time (12 hr/day), decreases rated lamp life to 30,000 hours, but extends calendar life to 6.8 years.

Out-of-tolerance line voltage (either too low or too high) can affect lamp life. Short cycling lamps can also affect lamp life. Recent field observations uncovered some causes for short life lamps, including occupancy sensors set to short delay times (or left in the test mode), use of instant-start ballasts in switching applications, or the combination of short-delay time sensors and instant-start ballasts. Other causes are sometimes mismatched lamps and ballasts.

Instant-start vs. rapid-start operation. With instant-start operation, an external current doesn't heat the lamp cathodes. Electrons are supplied by "stripping" them off the emissive coating of the cathode through the arc discharge process. A "brute force" high voltage is applied across the lamp to start it. The high voltage is necessary to initiate the discharge between the unheated cathodes. Since the lamp cathodes are not heated, instant-start operation results in lower electrical losses and the input power to the ballast is reduced by approximately 2.5W per 4 ft lamp. The disadvantage of instant-start operation is a reduction in average-rated lamp life due to the rapid erosion of the cathode emissive coating, especially at short burn cycles. When T12 lamps operate in an instant-start mode they experience approximately a 25% decrease in lamp life than lamps in rapid-start mode. Parallel lamp operation is the prevalent method of wiring on instant-start ballasts. You can operate up to four lamps on electronic, high frequency ballasts.

In rapid-start operation, cathode heating is supplied by a separate set of windings on the ballast, which adds approximately 2.5W/lamp. This additional power reduces the efficiency of the system, but normal lamp life is obtained, even with switching applications such as occupancy sensors. Series wiring is the prevalent method of wiring lamps on rapid-start ballasts. Rapid-start ballasts can normally operate up to three lamps per ballast. For the two-lamp, rapid-start ballast, when one lamp fails the second lamp will become dim, signaling that it is time to change the lamps. A few rapid-start, electronic ballast designs operate the lamps in parallel to eliminate the problem of lamps dimming when the second lamp fails.

The choice of rapid-start or instant-start operation usually relates to the application. If the application has long burn hours, and little switching, then an instant-start ballast is the best choice. However, when there is a lot of switching, the rapid-start ballast is the better choice. When converting from rapid-start to instant-start ballasts, ballast manufacturers advise shorting the socket (or using an internally shorted-J socket) to avoid dramatically reduced lamp life. Instant-start operation only requires one pin and bi-pin sockets should be shorted so they appear to the ballast as one pin.

Harmonics. Some harmonic distortion is produced by lighting equipment. The line current can flow at the fundamental frequency (60 Hz in the U.S.) or it may be combined with harmonic currents (multiples of the fundamental) such as 180 Hz (3rd harmonic). Harmonics are generated by core and coil ballasts due to saturation of the magnetic steel core. This power deformation is usually in the range of 12% to 22% total harmonic distortion (THD). Specification of equipment having less than 20% THD is advised. However, when ballasts with a THD that is lower than 10% are specified, a new problem can be created, inrush current.

The low impedance of electronic ballasts rated with THD less than 10% creates inrush current. The problem is compounded when several ballasts are on the same lighting circuit. Although inrush current is of very short duration, the inrush peak current of an active front-end ballast can be 1002 or more its operating current. For passive front-end designs that incorporate a series input choke coil, the inrush peak current is about 302 the operating current. The actual magnitude of the inrush current depends on several factors, including the electrical characteristics of the lighting circuit and where the switching occurs.

The effects of inrush are usually long-term caused by the arcing of metallic contacts that promotes contact erosion that eventually causes them to weld together. Occupancy sensor contacts and building management system lighting relays are the most commonly affected switching equipment. Most occupancy sensors have been redesigned to incorporate zero-crossing relays to prevent the problem. Zero-crossing relays sense the 60 Hz waveform of the power line and delay closing the contacts until the next zero crossing. No current flows at zero, preventing inrush.

Transient protection. Electronic ballasts are sensitive to the surges and spikes normally found in building power distribution systems. Many of the failures of early electronic ballast designs have been attributed to a lack of transient suppression. Most electronic ballasts now have some degree of protection built in. ANSI/IEEE Standard C62.41 addresses surge-withstand provisions. This standard defines a set of test waveforms. However, internal protection may not be sufficient for some situations. For example, for those facilities in areas experiencing high lightning activity, a separate transient voltage surge suppressor (TVSS) is recommended. A practical location for this protective device is at the lighting distribution center (breaker box). Installations in high-rise buildings encountered failures from the large transients created by elevator motors starting and stopping. You should also protect these with surge suppressor devices located at the lighting distribution center.

Avoiding common mistakes. There are several easy to use computer programs for calculating desired light levels. But remember, design for maintained light level instead of initial light level. Lamps depreciate in their light output over time and light reflecting surfaces get dirty. These light-loss factors must be taken into account. Consider a relighting project when fixtures are very old or in bad condition or when they are no longer meeting the lighting needs of the current visual tasks.