Turn-On and Turn-Off Characteristics of Incandescent and Light-Emitting Diode Signal Modules REPLACING INCANDESCENT INTRODUCTION APPROACH MAY CREATE Traffic signal light sources either are incandescent or light-emitting diode (LED). LED-based traffic signal modules have a higher capital cost but offer tremendous energy savings because they consume less power and do not require annual preventive replacement. Consequently, LED modules are rapidly replacing incandescent modules. However, using LED modules from different manufacturers on a given intersection approach may create confusion for drivers because the same color indicators on different signal heads may not start (or end) simultaneously. It was reported that, at an intersection approach with LED modules from different manufacturers, motorists noticed that one of the green modules came on before the other. The authors conducted this study to investigate the turn-on and turn-off characteristics of incandescent and LED traffic signal modules. CONFUSION FOR DATA COLLECTION AND REDUCTION TRAFFIC SIGNAL LIGHTS AT INTERSECTIONS WITH LIGHT-EMITTING DIODE (LED) MODULES IS BECOMING MORE COMMON IN THE UNITED STATES. HOWEVER, USING LED MODULES FROM DIFFERENT MANUFACTURERS ON THE SAME INTERSECTION Incandescent lights (DURA-TEST TRAVELERS AND lamps) and LED modules manufactured by Gelcore and Dialight were used in this ADVERSELY AFFECT investigation. LED modules from these two manufacturers were evaluated TRAFFIC SAFETY. because they are the main suppliers of LED traffic signal modules for the Illinois Department of Transportation. At the Traffic BY MADHAV V. CHITTURI, RAHIM F. BENEKOHAL, PH.D. Operations Laboratory at the University AND MONTTY GIRIANNA, PH.D. of Illinois, a frame was set up with three signal heads next to each other (one incandescent and two LEDs). They were programmed to turn on and turn off simultaneously using a National Electrical Manufacturers Association (NEMA) controller. ITE JOURNAL ON THE WEB / OCTOBER 2005 Data were collected for 114 cycles of 92 seconds in duration. The operation was videotaped to achieve one-thirtiethsecond time accuracy. More detailed information on data collection and experimental setup can be found in Benekohal, Girianna and Chitturi.1 The videotapes were time-coded and played at a slow speed so that each frame (one-thirtieth-second) could be viewed individually. Traffic signal modules do not completely illuminate or darken instantaneously. For each module, the observer recorded the following times: • Time the module started to turn on; • Time the module was fully on; • Time the module started to turn off; and • Time the module was fully off. The terms used in this study are defined in the following section. A 95-percent confidence level was used for all statistical tests. DEFINITIONS Fade-in time is the time required for a module to reach full brightness. In Figure 1, fade-in time is represented by the hashed area between the points labeled “start on” and “fully on.” Fadeout time is the time required for a module to reach complete darkness. Fade-out time is represented by the hashed area between the points labeled “start off ” and “fully off.” Turn-on (or turn-off) delay quantifies whether the LED modules start turning on (or turning off) earlier or later than the incandescent module. Turn-on delay, dn, is the difference between “start on” times; turn-off delay, df, is the difference between “start off” times for the LED modules and the incandescent module. 69 FINDINGS Fade-In and Fade-Out Time The ITE LED Vehicle Traffic Control Signal Modules Draft Specification defines turn-on time as the time delay between application of nominal operating voltage and the module reaching 90 percent of full light output.2 It defines turn-off time as the amount of time required after removal of the nominal operating voltage for the LED signal module to show no visible illumination. The ITE Draft Specification requires that turn-on and turn-off times for LED signal modules not exceed 75 milliseconds. Fade-in (and fade-out) times, reported in milliseconds, for the three types of traffic signal modules are summarized in Table 1. T-tests with 95-percent confidence were performed to determine if fade-in (fade-out) times were statistically different from zero. The test results indicated that fade-in (fadeout) times for all three modules were greater than zero. Although the ITE turn-on (turn-off) times were similar to the fade-in (fadeout) times defined in this study, there were certain minor differences. The ITE turn-on time requires the signal module to reach 90 percent of full illumination. In this study, fade-in time requires the module to reach full brightness as perceived by an observer. The ITE Draft Specification procedure uses a two-channel oscilloscope for measuring time delays. In this study, the times were measured by an observer viewing the video at a slow speed. Therefore, fade-in time would be expected to be higher than turn-on time for any module. The evaluated LED modules met the fade-in time fade-out time Fully on Start on Fully off Start off Incandescent Start on Start off Fully on Fully off LED I dn Start on df Start off Fully on Fully off LED II df dn dn = Turn-on delay; df = Turn-off delay fade-in time fade-out time fully dark fully illuminated Figure 1. Definitions and terms used to determine turn-on and turn-off characteristics. ITE Draft Specification requirements for turn-on time. Statistical tests indicated that the average fade-in times for all LED modules were less than 75 milliseconds. In fact, the maximum fade-in time for all LED modules, except green Gelcore, was less than 75 milliseconds. The average fade-in time for the red incandescent module was statistically less than 75 milliseconds; however, the average fade-in times for the yellow and green incandescent modules were greater than 75 milliseconds. The average fade-out times for all the modules were statistically greater than 75 milliseconds. These high values could be due to the fact that human observers made the measurements and that the fade-out time defined in this study was measured from the time the modules start dimming to the time they were completely dark. Comparing fade-in (fade-out) times of incandescent lamps to the LED modules indicated that, for any given color, the fade-in time for incandescent was greater than that for the Gelcore LED, and the fade-in time for the Gelcore LED was greater than that for the Dialight LED. The differences were statistically significant (see Table 2). Similar comparisons for fade-out times indicated that the red incandescent had the longest fade-out time, followed by the Gelcore LED and Dialight LED. For green indications, the Dialight LED had the longest fade-out time, followed by the incandescent and the Gelcore LED. These differences were statistically significant. For yellow, the fade-out time for incandescent was the longest and statistically different from the LED modules; the fade-out times for the two yellow LED modules were statistically the same. Table 1. Fade-in and fade-out times for incandescent and LED modules. Incandescent Color Fade-in time (in milliseconds) Fade-out time (in milliseconds) Gelcore LED Minimum Average Maximum t-value* Red Yellow Green Red Yellow Green 33 133 100 500 433 333 71 169 143 602 501 377 100 200 167 767 600 467 36.88 83.55 89.72 120.68 139.98 164.94 Dialight LED Minimum Average Maximum t-value* Minimum Average Maximum t-value* 33 33 33 367 167 167 40 39 63 403 222 227 67 67 100 467 300 267 32.14 33.28 38.72 137.34 92.21 96.45 33 33 0 133 200 433 34 34 36 208 220 475 67 67 67 267 267 533 82.39 115 35.91 53.88 104.73 247.56 * Note: Critical ‘t’-value = 1.6584. 70 ITE JOURNAL ON THE WEB / OCTOBER 2005 Turn-On and Turn-Off Delay Turn-on delay quantifies whether LED modules start to turn on earlier or later than the incandescent module. Similarly, turn-off delay quantifies whether LED modules start to turn off earlier or later than the incandescent module. The times at which the incandescent module starts to glow (start on) and starts to dim (start off) are the benchmarks for measuring turn-on and turn-off delays, respectively. Positive values for turn-on (turn-off) delays mean that the LED modules start to turn on (turn off ) later than the incandescent module. Negative values mean that they turn on (turn off) earlier. In Table 3, the turn-on (turn-off ) delays for the LED modules are documented. All turn-on delays were statistically different from zero. With the exception of the red Dialight, the turn-on delay for all LED modules was zero or positive. In other words, the LED modules started to turn on some time after the incandescent module. For the red Dialight LED, the delay was negative. On average, the turn-off delay for all LED modules other than the green Gelcore was negative. In other words, LED modules other than the green Gelcore start turning off earlier than the incandescent module. For any color, the Dialight LED module had a smaller turn-on (turn-off) delay than the Gelcore LED module. When the video was played at a regular pace, it was observed that for the green color, the Dialight module came on earlier than the Gelcore module. This supported the problem drivers had reported about the intersection approach with different LED modules. However, there was no such perceptible difference for the yellow and red LED Table 2. Summary of statistical tests for fade-in and fade-out times. Fade-in time Fade-out time Color Statistical results Red Yellow Green Red Yellow Green Incandescent > Gelcore LED > Dialight LED Incandescent > Gelcore LED > Dialight LED Incandescent > Gelcore LED > Dialight LED Incandescent > Gelcore LED > Dialight LED Incandescent > Gelcore LED = Dialight LED Dialight LED > Incandescent > Gelcore LED modules. The practical implication of this finding is that one should not mix green LEDs from Gelcore and Dialight on the same intersection approach, but mixing yellow or red LED modules from the two manufacturers would not create a noticeable delay. Overlaps and Gaps If modules have perceptibly different turn-on delays when the signal changes from one phase to another, motorists might find an intersection with no signal glowing or signals of two different colors glowing at the same time. Phase transitions from green to yellow and yellow to red were studied. These two signal changes are critical from a safety point of view. Two situations that could create a potential traffic safety hazard are: • No signal indications are on at the intersection for a very brief time (the intersection is momentarily dark). This situation would happen if the modules for the first phase are fully off and the modules for second phase have not yet started to turn on. This instance is represented by a gap in case 1 (fully off and start on are the reference points for phases 1 and 2, respectively), as shown in Figure 2. • Two different color signal indications are on simultaneously for the same approach at the intersection (dual indication). This situation arises when the modules for the first phase have not turned off completely and the modules for the second phase are fully on. This situation is represented by an overlap in case 2 (start off and fully on are the reference points for phases 1 and 2, respectively), as shown in Figure 2. The time difference between the reference points for phases 1 and 2 is computed as . The overlaps/gaps, , between the signal phase transitions green to yellow and yellow to red are presented in Table 4. The first two columns show the nine possible combinations of signal modules. Gaps and overlaps result in negative and positive values for , respectively. All the overlaps of case 1 and all the gaps of case 2 are statistically different from zero. Potentially unsafe traffic conditions could be created when gaps occur in case 1 or when overlaps occur in case 2. In both the phase transitions studied, no gaps were observed in case 1 and no overlaps were observed in case 2. Therefore, for any combination of the modules investigated, Table 3. Turn-on and turn-off delays (in milliseconds) for LED modules. Gelcore LED Color Turn-on delay Turn-off delay Dialight LED Minimum Average Maximum t-value* Red Yellow Green Red Yellow Green 0 33 100 -133 -100 -33 46 69 149 -16 -3 17 100 133 200 33 0 67 27.46 58.46 86.66 -5.46 -2.41 9.23 Difference Minimum Average Maximum t-value* -33 0 33 -133 -100 -67 -21 36 36 -32 -7 -6 0 67 67 33 33 33 -13.66 34.08 44.54 -9.1 -3.86 -3.59 in average# t-value* 67 33 113 16 4 23 29.56 20.74 59.83 6.81 2.75 12 * Note: Critical ‘t’-value = 1.6584; # critical ‘t’-value = 2.414. ITE JOURNAL ON THE WEB / OCTOBER 2005 71 Figure 2. Overlaps and gaps for cases 1 and 2. two signals were not glowing simultaneously and there was no dark intersection. Some readers may be concerned about the overlaps observed in case 1 and the gaps observed in case 2. To address these concerns, the investigators viewed the videotapes at regular speed and observed that in none of these cases could the viewers see multiple signals glowing simultaneously or none of the signals illuminated. CONCLUSIONS This feature examined the fade-in and fade-out times, turn-on and turn-off delays and overlaps/gaps of colors between consecutive phase intervals for incandescent lamps and LED modules. For all colors, the LED modules had shorter fade-in times when compared with the incandescent module. The fadein time for the Dialight LED was shorter than for the Gelcore LED, for all the three colors. The fade-in times of all the LED modules evaluated in this study were less than the ITE Specification requirement for turn-on time. The Dialight LED module had the longest fade-out time for green. The incandescent module had the longest and the Dialight LED modules had the shortest fade-out times for yellow and red. The fade-out times for all the evaluated mod- ules were greater than the ITE Specification requirement for turn-off time. This difference could be due to the difference in the measurement approach. Except in the case of the red Dialight, all LED modules started to turn on a fraction of a second later than the incandescent modules. All of the LED modules other than the green Gelcore started to turn off a fraction of a second earlier than the incandescent modules. In the lab, an observer could note that the green Dialight module came on before the green Gelcore module. However, for yellow and red, the difference in LED modules was not noticeable. Two situations could create unsafe traffic conditions: when approaching motorists see two different color signal indications turned on simultaneously, and when no signal indication is on at the intersection. Statistical tests and visual observations confirmed that these situations did not happen. That is, none of the combinations of the traffic signal modules studied could result in multiple signals indications being displayed or none of the signals being illuminated. In Table 4. Summary of gaps/overlaps (in milliseconds) between consecutive phases. Signal module combination Phase change Green Yellow Green to yellow Incandescent Incandescent Incandescent Gelcore LED Gelcore LED Gelcore LED Dialight LED Dialight LED Dialight LED Incandescent Gelcore LED Dialight LED Incandescent Gelcore LED Dialight LED Incandescent Gelcore LED Dialight LED Yellow Red Incandescent Incandescent Incandescent Gelcore LED Gelcore LED Gelcore LED Dialight LED Dialight LED Dialight LED Incandescent Gelcore LED Dialight LED Incandescent Gelcore LED Dialight LED Incandescent Gelcore LED Dialight LED Yellow to red Case 1: Overlaps/gaps from fully off to start on Case 2: Overlaps/gaps from start off to fully on Minimum Average Maximum t-value* Minimum Average Maximum t-value* 300 233 267 167 100 133 400 333 367 358 289 322 226 156 189 450 381 414 433 367 400 300 233 267 500 433 467 117.56 92.4 104.68 102.65 67.51 85.73 211.47 173.55 185.59 -233 -167 -133 -233 -167 -133 -233 -167 -133 -188 -127 -89 -171 -110 -72 -194 -133 -94 -100 -67 -33 -100 -67 0 -100 -100 -33 -72.37 -61.79 -45.98 -57.49 -51.98 -38.35 -72.89 -88.71 -68.82 400 333 400 167 100 167 167 100 167 483 437 504 201 154 222 194 148 215 633 567 633 300 233 300 233 200 267 106.95 95.96 113.73 73.36 54.65 81.39 95.73 72.66 112.69 -133 -167 -100 -133 -167 -100 -133 -167 -67 -89 -105 -32 -92 -108 -35 -97 -112 -39 0 0 67 -33 -67 0 -33 -67 -33 -30.99 -42.08 -16.27 -37.06 -51.78 -25.13 -43.29 -58.66 -33.28 * Note: Critical ‘t’-value = 1.6584. 72 ITE JOURNAL ON THE WEB / OCTOBER 2005 other words, the unsynchronized turn-on and turn-off characteristics did not result in these two situations that could compromise traffic safety. ■ References 1. Benekohal, R., M. Girianna and M. Chitturi. “A Comparison Between the Turn-On and Turn-Off Characteristics of Incandescent and LED Traffic Signal Modules.” University of Illinois Civil Engineering Studies, Transportation Engineering Series, No. 121, December 2003. 2. Purchase Specification of the Institute of Transportation Engineers. “Vehicle Traffic Control Signal Heads—Part 2: Light Emitting Diode Vehicle Traffic Signal Modules.” December 2003. MADHAV V. CHITTURI is a doctoral candidate in the Department of Civil and Environmental Engineering at the University of Illinois at Urbana–Champaign. He graduated from the Indian Institute of Technology in Kharagpur, India, with a bachelor’s degree and the University of Illinois at Urbana–Champaign with a master’s degree in civil engineering. His research interests include traffic simulation, traffic modeling and traffic safety. ITE JOURNAL ON THE WEB / OCTOBER 2005 RAHIM F. BENEKOHAL, Ph.D., is a professor of civil engineering at the University of Illinois at Urbana–Champaign. Previously, he worked for the consulting firm of RKA Inc. He received a B.S., M.S. and Ph.D. from the Ohio State University. He teaches and conducts research in traffic modeling, ITS evaluation, transportation safety and transportation systems analysis. He is the winner of the Arthur Wellington Award and ITE Illinois Section Past President Award. He is director of the Traffic Operations Lab at the university as well as director of the Illinois Traffic Engineering and Safety Conference. He is a member of ITE. MONTTY GIRIANNA, Ph.D., graduated from Bandung Institute of Technology in 1985 with a degree in civil engineering. For several years, he worked for the National Development Planning Agency (BAPPENAS) of the Republic of Indonesia before pursuing graduate studies in urban planning at Massachusetts Institute of Technology and a Ph.D. in transportation systems at the University of Illinois at Urbana–Champaign. He is now working for BAPPENAS as chief of infrastructure financing. 73