Sensitivity of modern lighting technologies at varying flicker severity levels

Sensitivity of modern lighting technologies at varying flicker severity levels. J.J. Gutierreza, P. Saiza, I. Azcarate∗,a, L.A. Leturiondoa, K. Redondoa, S. Ruiz de Gaunaa, Digna M. Gonzalez-Oteroa Affiliations and addresses: aCommunications Engineering Department. University of the Basque Country UPV/EHU. Alameda Urquijo S/N 48013 Bilbao, Spain Corresponding author: ∗ Izaskun Azcarate email: [email protected] Tel. : +34946018209 Fax. : +34946014259 Abstract1 Efficient lighting technologies are not necessarily less sensitive to voltage fluctuations2 than the incandescent lamp, and therefore a procedure for controlling the immunity of lamps3 to voltage fluctuations was defined in the IEC 61547 standard. This procedure checks that a4 lamp is not more sensitive than the incandescent lamp to voltage fluctuations corresponding5 to the Pst = 1 curve. For a lamp that behaves linearly, these tests are sufficient to guarantee6 1 This is the accepted manuscript of the article that appeared in final form in International Journal of Electrical Power & Energy Systems 92 : 34-41 (2017), which has been published in final form at https://doi.org/10.1016/j.ijepes.2017.04.010. © 2017 Elsevier under CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/) that a lamp is less sensitive than the incandescent lamp at any voltage fluctuation level. This7 paper analyzes the linearity in the response of a set of lamps with both simulated and real8 voltage signals. For a given input voltage signal containing fluctuations, a new signal was9 generated with a voltage fluctuation whose amplitude was proportional to the original one.10 Both signals were passed through an illuminance flickermeter and the obtained flicker severity11 values were compared. The results showed that not all the lamps behaved linearly. Some12 lamps were less sensitive than the incandescent lamp at the reference level, and with other13 voltage fluctuation amplitudes produced flicker severity values higher than the incandescent14 lamp. Moreover, the nonlinearity shown with real voltage signals was not reflected with the15 same nonlinear behavior with simulated fluctuations in all cases. These results lead to the16 conclusion that the current immunity protocol is insufficient for guaranteeing that a lamp is17 less sensitive to voltage fluctuations than the incandescent lamp at every voltage fluctuation18 level.19 Keywords20 Energy-efficient lighting, Power quality, Flicker, Voltage fluctuations, Linearity.21 2 1. Introduction22 Energy efficiency and sustainability have become priorities in a world moving toward23 reducing carbon dioxide emissions and energy costs. To this end, the lighting industry has24 addressed important changes with the replacement of incandescent lamps by other efficient25 lighting technologies [1–3]. However, this change poses different challenges in terms of power26 quality and more specifically in terms of flicker. The standardized flicker measurement pro-27 cedure [4] and hence the compatibility levels to voltage fluctuations are based on the response28 of the incandescent lamp. The progressive banning of inefficient lamps could compromise29 the utility of the existing flicker measurements [5].30 Several studies have focused on the sensitivity of new lighting technologies to voltage31 fluctuations [6–11]. The two most representative modern lighting technologies, compact32 fluorescent (CFL) and light emitting diode (LED) lamps, usually present low sensitivity33 to voltage fluctuations [7]. However, depending on different factors such as the brand and34 model of the lamp or the complexity of the voltage fluctuations, these technologies present35 even more sensitive behaviors than the incandescent lamp [8–11]. The diversity of sensitivity36 between the different technologies available on the market means that the increase in lighting37 efficiency does not imply an enhancement in the immunity to voltage fluctuations [12].38 In view of these results, it has become necessary to keep the existing compatibility levels39 to voltage fluctuations as the reference for the flicker measurement [13]. Consequently, it is40 essential to guarantee that the new lighting products are less sensitive to voltage fluctuations41 than the incandescent lamp. In this scenario, a procedure for testing the immunity of new42 lighting equipment to voltage fluctuations has been developed within the working group43 IEC TC34. This procedure has been published as a technical report [14] associated with44 the immunity standard for lighting equipment IEC 61547 [15]. The procedure for testing45 the immunity of a lamp consists of the comparison of the flicker produced by the lamp46 under test and the incandescent lamp, when both lamps are affected by the same voltage47 fluctuations corresponding to the compatibility threshold, i.e., the Pst = 1 curve [16]. The48 lamp under test that produces flicker severity values below 1 is considered immune to voltage49 3 fluctuations.50 Nevertheless, being less sensitive than the incandescent lamp for voltage fluctuations51 at the level of Pst = 1 does not directly imply the same behavior at higher flicker levels.52 For lighting technologies that are less sensitive than the incandescent lamp at the level of53 Pst = 1, the maintenance of this behavior regardless of the voltage fluctuation amplitude54 will depend on the linearity in the response of the lamp. The incandescent lamp provides55 a linear relationship between the amplitude of the voltage fluctuation and the correspond-56 ing illuminance fluctuation [17], which is also linearly related to the flicker perception by57 the human visual system. As a consequence, an increase in the amplitude of the voltage58 fluctuation will produce a proportional increase in the flicker severity value produced by an59 incandescent lamp. Preliminary experiments using sinusoidal voltage fluctuations pointed60 to a nonlinear behavior of some lamps [10]. The current work goes further by analyzing61 the linearity when the lamps are supplied by more complex voltage signals, i.e., simulated62 rectangular voltage fluctuations and real voltage signals.63 In Section II, the linear relationship between the relative amplitude of the voltage fluc-64 tuation, ∆V V, and the flicker severity value is demonstrated through a numerical example65 following the block diagram of the IEC 61000-4-15 standard [4]. Section III presents the66 set of lamps, test setup, and methodology used to perform the linearity study. The results67 are presented in Section IV and these are discussed in Section V. Concluding remarks are68 provided in Section VI.69 2. Linearity of the existing flicker measurement70 The IEC 61000-4-15 standard defines the functional and design specifications of a flick-71 ermeter that provides an objective measurement of the level of annoyance produced by72 the flickering of the illuminance of a lamp [4]. This measurement procedure is based on73 a simplified physiological model of the behavior of the lamp–eye–brain system, taking the74 incandescent lamp as the reference. Fig. 1 depicts the scheme of the IEC flickermeter where75 the input signal is the supply voltage, u(t), and the output parameters are the short and76 long term flicker severity values, Pst and Plt respectively.77 4 u(t) uk(t) B1 VOLTAGE ADAPTOR B2 SQUARING MULTIPLIER B3 0.05 35 8.8 u2(t) u2k(t) DEMODUL. AND WEIGHTING FILTERS u3(t) u3k(t) B4 SQUARING MULTIPLIER + SLIDING MEAN FILTER u4(t) u4k(t) B5 STATISTICAL EVALUATION Pst,Plt Pstk,Pltk Figure 1: Scheme of the IEC flickermeter according to the IEC 61000-4-15 standard [4]. The characteristics of the incandescent lamp and the eye–brain system provide a linear78 relationship between the amplitude of the voltage fluctuation and the flicker severity value.79 For the incandescent lamp, there is an empirical relationship between the luminous in-80 tensity and the supply voltage:81 L Ln =V Vnγ 3.4 < γ < 3.8 ,(1) where Vnrepresents the rated rms value of the voltage and Lnis the corresponding luminous82 intensity [17]. For small voltage changes, the relative response of the lamp can be expressed83 as follows:84 ∆L Ln =γ·∆V Vn .(2) A linear increase in the relative amplitude of the voltage fluctuation means a proportional85 increase of the relative amplitude of the illuminance fluctuation. This behavior is modeled86 jointly by blocks 2 and 3 of the IEC flickermeter (Fig. 1). For a sinusoidal voltage fluctuation,87 u(t) has the form:88 u(t) = A√21 + 1 2·∆V Vsin(wmt)sin(w0t),(3) where A= 230 V, w0= 2π·50 rad/s, and wm= 2π·fmrad/s, with fmbeing the frequency89 of the voltage fluctuation. The fluctuation at the output of the demodulation filters of block90 3 is characterized as follows:91 u2(t)≃C2·1 2·∆V V·sin(wmt+φ2),(4) where C2and φ2are the magnitude and phase of the frequency response of the demodulation92 filters at fm. This signal is then weighted by the band-pass filter of block 3, obtaining the93 5 illuminance fluctuation, u3(t), that presents a fundamental component linearly related to94 the voltage fluctuation as follows:95 u3(t)≃C3·C2·1 2·∆V V·sin(wmt+φ2+φ3),(5) where C3and φ3are the magnitude and phase of the frequency response of the weighting96 filter at fm. The term C3·C2·1 2·∆V Vrepresents the relative amplitude of the illuminance97 fluctuation linearly weighted by the eye response.98 The remainder of the chain models the brain response to illuminance fluctuations and also99 presents a linear behavior. The square of the output signal of block 3 is linearly processed by100 the sliding mean filter of block 4, obtaining the instantaneous flicker perception as a linear101 combination of two main frequencies, the direct current component (DC) and 2fm:102 u4(t)≃C2 3·C2 2 2·1 2·∆V V2·C4−C5·cos(2wmt+ 2φ2+ 2φ3+φ5),(6) where C4and C5are the magnitudes of the frequency response of the filter at DC and 2fm 103 and φ5is the phase at 2fm. Finally, block 5 implements the statistical evaluation of u4(t)104 providing the Pst value.105 To demonstrate the linear behavior of the brain system, the input voltage fluctuation106 was changed so that its relative amplitude was ktimes higher than the previous value:107 ∆V V0 =k·∆V V.(7) Consequently, the new output signal of block 3, u3k(t), also presents a single component108 at fmwith an amplitude ktimes higher than u3(t):109 u3k(t)≃k·C2·C3·1 2·∆V V·sin(wmt+φ2+φ3) = k·u3(t).(8) 6 In the same way, the output signal of block 4 can be expressed as:110 u4k(t) = u3k(t)2=k2·u2 3(t) = k2·u4(t),(9) presenting an amplitude k2times higher than u4(t). As the Pst value is obtained as the111 square root of the statistical data of u4(t), the corresponding Pst is linearly related to the112 previous value as follows:113 Pst k=k·Pst .(10) Since the Plt value is obtained by the cubic average of the Pst values obtained during 2 hours,114 the Plt value is also linearly related to the previous value as follows:115 Plt k=k·Plt ,(11) showing that the relative amplitude of the voltage fluctuation and the flicker severity values116 are linearly related.117 This behavior, which has been demonstrated for the incandescent lamp model included118 in the IEC flickermeter, can only be assumed for lamp–eye–brain systems involving lamps119 that present linear responses. As the eye–brain system responds linearly to illuminance120 fluctuations, it has to be proved whether the modern lighting technologies present a linear121 behavior, as the incandescent lamp model does.122 3. Experimental setup123 This section describes the set of lamps used for the experiments, the system implemented124 to generate the supply voltage and to record the illuminance, as well as the procedure for125 sensitivity analysis.126 3.1. Set of lamps under test127 Table 1 shows the main characteristics of the set of commercial lamps used in this128 work. The lamps are from different manufacturers and lighting technologies, including linear129 7 Table 1: Set of lamps under test. ID. Lamp PoweraLumen Efficiency Brand Technology (W) Classb F1cLFL 18 1050 B Sylvania C1dCFL 11 570 A Philips C2dCFL 23 1380 A Lexman C3cCFL 18 1050 B General Electric C4d,e CFL 12 600 A Philips C5d,e CFL 11 570 A Megaman L1 LED 12 650 A Osram L2 LED 8 470 A Philips a230 V / 50 Hz bMost efficient (A); Least efficient (G) cElectromagnetic ballast dElectronic ballast eDimmable lamp fluorescent lamps (LFL), CFL using electromagnetic or electronic ballast, and LED lamps.130 These lamps present different values of power, lumen, and energy efficiency ratings. Two of131 the lamps included in the study were dimmable.132 3.2. Test setup and equipment133 Fig. 2 shows the system implemented to generate the supply voltage and to record134 the illuminance of the lamps. The system can generate simulated voltage fluctuations,135 characterized by the relative amplitude ∆V V, and the frequency fm, or real voltage signals,136 previously recorded on the LV network. Once the simulated voltage fluctuation is configured137 or the real voltage signal is selected, the NI USB-6211 card is used to convert the digitized138 signal into an analog signal, at a rate of 6400 Hz. The analog signal is then amplified by139 a 7500 Krohn-Hite amplifier (75 W, from DC to 1 MHz) and a 120/230 V transformer to140 achieve a level of 230 V suitable for supplying the light source. The lamp under test and141 the light sensor of the luxmeter (E4-X Hagner Digital Luxmeter) are placed together within142 a white box. The output of the luxmeter provides the illuminance signal, l(t), which is then143 digitized by the NI USB-6211 card at a rate of 6400 Hz and 16-bit resolution. Finally, the144 illuminance signal is stored for later analysis.145 3.3. Illuminance flickermeter146 The analysis of the response of the lamps was performed in terms of the flicker severity147 values generated by each lamp. To that end, it is necessary to use an instrument based148 8 Voltage signal configuration (simulated or real) D/A conversion Amplification u(t) Voltage generation Lamp Light sensor Luxmeter l(t) A/D conversion Data storage Illuminance acquisition (a) (b) Figure 2: Block diagram (a) and photograph (b) of the system implemented to generate the supply voltage and acquire the illuminance of a lamp. 9 C4 lamps, the nonlinear response was detectable at some specific time intervals, whereas L2243 lamp presented a relatively constant nonlinear response during all the analyzed time period.244 It is also remarkable that the C4 and L2 lamps presented a higher nonlinearity as the flicker245 severity level increased. This behavior is also represented in Table 3 with the increase in the246 deviation values of these lamps for increasing values of k.247 It can be concluded that the study of the linearity using simulated voltage signals is not248 always representative of the responses of the lamps subjected to real voltage signals. Taking249 the response of L2 lamp as an example, with simulated voltage signals this lamp presented a250 quite linear response with low deviation values, whereas under real voltage signals the lamp251 showed a clearly nonlinear response. These differences can be related to the different spectral252 complexity of simulated and real voltage signals. The effect of the spectral complexity of253 the input voltage signal on the response of the lamps should be analyzed depending on the254 electronic implementations of the basic operation principles of each lamp [21].255 5. Discussion256 The nonlinear response of lamps is very relevant when it comes to analyzing the immunity257 of the lamps to voltage fluctuations. The procedure for testing the immunity of a lamp was258 defined taking the Pst = 1 curve as the reference [14]. This threshold was obtained based259 on the response of the incandescent lamp. The situation becomes critical when the lamp is260 identified as immune but, because of its nonlinear response, at higher flicker severity levels261 the sensitivity of the lamp exceeds the sensitivity of the incandescent lamp, producing a262 greater annoyance. Fig. 7 represents the Pltkvalues of L1, C4, C5, and C2 lamps normalized263 with respect to the values obtained for the incandescent lamp model:264 Pnorm. ltk=Plamp ltk PIEC ltk .(17) That is, if Pnorm. ltk>1, the lamp being tested presents a higher sensitivity to voltage fluctua-265 tions than the incandescent lamp model.266 L1 lamp presented a linear response under real voltage signals (Fig. 6g). Comparison267 16 k=4k=3k=2k=1 Time (h) Pnorm. ltk 24 2220 1816 14 1210 86 42 0 0.2 0.4 0.6 0.8 1.0 1.2 (a) L1 k=4k=3k=2k=1 Time (h) Pnorm. ltk 24 2220 1816 14 1210 86 42 0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 (b) C4 k=4k=3k=2k=1 Time (h) Pnorm. ltk 24 2220 1816 14 1210 86 42 0 0.2 0.4 0.6 0.8 1.0 1.2 (c) C5 k=4k=3k=2k=1 Time (h) Pnorm. ltk 24 2220 1816 14 1210 86 42 0 0.2 0.4 0.6 0.8 1.0 1.2 (d) C2 Figure 7: Pltkvalues for L1, C4, C5, and C2 lamps normalized with respect to the Pltkvalues of the incandescent lamp. of this response with the incandescent lamp model shows that this lamp presented a lower268 sensitivity than the incandescent lamp model for k= 1 (Fig. 7a). This behavior was269 maintained for all the kvalues, because of the linear response of the lamp. C4 lamp presented270 a nonlinear response under real voltage signals with a negative deviation from the expected271 values (Fig. 6e). This lamp was more sensitive than the incandescent lamp for k= 1 as can272 be seen in Fig. 7b. For increasing values of k, the ratio between its sensitivity and that of the273 incandescent lamp model decreased but it remained more sensitive than the incandescent274 lamp. C5 and C2 lamps presented nonlinear responses under real voltage signals and in both275 cases the measured flicker severity values exceeded those expected (Fig. 6f and 6c). However,276 comparison of their responses with the incandescent lamp model shows noticeable differences277 (Fig. 7c and 7d). Despite the nonlinear response of C5, the ratio between the sensitivity of278 C5 and that of the incandescent lamp model remained below 1 at all flicker severity levels.279 In contrast, the ratio between the sensitivity of C2 and that of the incandescent lamp model280 increased as the proportionality factor kincreased, even exceeding 1 at some time intervals.281 17 The nonlinear response of C2 lamp compromised its immunity to voltage fluctuations at282 flicker severity levels higher than that used as the reference Pst = 1 in the current immunity283 test procedure. C2 lamp would be classified as immune even though at high flicker severity284 levels its sensitivity could exceed the sensitivity of an incandescent lamp. To avoid this285 situation, the immunity test procedure should include a test to analyze the sensitivity of286 the lamps at different flicker severity levels. However, this is not an easy task, as simulated287 voltage signals do not reflect the high spectral complexity of the real voltage signals to which288 the lamps are exposed in real scenarios.289 6. Conclusion290 This paper presented an analysis of the linearity in response, in terms of flicker, of a set291 of lamps representative of different lighting technologies. The study was performed using292 both simulated voltage and real voltage signals. The results showed that not all the lamps293 had a linear response to voltage fluctuations; hence, their sensitivity with respect to the294 incandescent lamp depended on the flicker level. Therefore, it is possible that a lamp less295 sensitive than the incandescent lamp at the compatibility threshold, i.e., Pst = 1 level, would296 be more sensitive at other voltage fluctuation levels.297 This behavior cannot be detected by the existing immunity protocol as the tests are298 performed for a single flicker level corresponding to the compatibility threshold. Therefore,299 the immunity testing should be performed at different flicker severity levels. However, the300 definition of an immunity protocol that considers the possible lack of linearity of the lamps301 under test is not easy. The difficulty stems from the fact that some lamps that present a302 nonlinear behavior with real voltage signals do not reproduce this behavior when subjected303 to simulated voltage fluctuations. As a future work, it would be convenient to explore the304 origin of these differences in terms of the different electronic implementations of the basic305 operation principles of the lamps.306 18 Acknowledgment307 This work received financial support from the Spanish MINECO through project DPI2014-308 53317-R (cofinanced with the European Regional Development Fund); from the Government309 of the Basque Country (BFI-2012-315 and IT1087-16) and from the University of the Basque310 Country UPV/EHU (UFI11/16 and PIF2011/169).311 References312 [1] European Comission (EC), Communication from the Comission to the European Parliament, the Coun-313 cil, the European economic and social committee and the committee of the regions, Energy Efficiency314 Plan, Brussels, COM(2011) 109 final.315 [2] European Comission (EC), GREEN PAPER, Lighting the Future: Accelerating the deployment of316 innovative lighting technologies, Brussels, COM(2011) 889 final.317 [3] A. De Almeida, B. Santos, B. Paolo, M. Quicheron, Solid state lighting review - Potential and challenges318 in Europe, Renewable Sustainable Energy Reviews 34 (2014) 30–48. doi:10.1016/j.rser.2014.02.029.319 [4] IEC 61000-4-15ed2.0, Electromagnetic Compatibility (EMC) Part 4: Testing and measurement tech-320 niques - Section 15: Flickermeter functional and design specifications (2010).321 [5] CIGRE WG C4.108, Review of flicker objectives for HV, MV and LV systems, Final WG report, Tech.322 rep. (2009).323 [6] K. Taekhyun, M. Rylander, E. Powers, W. Grady, A. Arapostathis, LED lamp flicker caused by interhar-324 monics, in: Proceedings of IEEE Instrumentation and Measurement Technology Conference (IMTC),325 2008, pp. 1920–1925.326 [7] R. Cai, J. F. Cobben, J. M. Myrzik, J. H. Blom, W. L. Kling, Flicker responses of different lamp types,327 IET Generation, Transmission and Distribution 3 (9) (2009) 816–824. doi:10.1049/iet-gtd.2008.0429.328 [8] K. Chmielowiec, Flicker effect of different types of light sources, in: Proceedings of 11th International329 Conference on Electrical Power Quality and Utilisation (EPQU), 2011, pp. 1 –6.330 [9] H. Sharma, F. Sharp, M. McGranaghan, Flicker/voltage fluctuation response of modern lamps including331 those with dimmable capability and other low voltage sensitive equipment, in: Proceedings of 22nd332 International Conference and Exhibition on Electricity Distribution (CIRED), 2013, pp. 1–4.333 [10] I. Azcarate, J. Gutierrez, P. Saiz, A. Lazkano, L. Leturiondo, K. Redondo, Flicker characteristics of334 efficient lighting assessed by the IEC flickermeter, Electric Power Systems Research 107 (2014) 21–27.335 doi:10.1016/j.epsr.2013.09.005.336 [11] I. Azcarate, J. Gutierrez, A. Lazkano, P. Saiz, K. Redondo, L. Leturiondo, Experimental study of337 19 the response of efficient lighting technologies to complex voltage fluctuations, International Journal of338 Electrical Power & Energy Systems 63 (2014) 499–506. doi:10.1016/j.ijepes.2014.06.039.339 [12] I. Azcarate, J. Gutierrez, A. Lazkano, P. Saiz, K. Redondo, L. Leturiondo, Towards limiting the sen-340 sitivity of energy-efficient lighting to voltage fluctuations, Renewable and Sustainable Energy Reviews341 59 (2016) 1384–1395. doi:10.1016/j.rser.2016.01.022.342 [13] CIGRE WG C4.111, Review of LV and MV compatibility levels for voltage fluctuation, Final WG343 report, Tech. rep. (2016).344 [14] IEC TR-61547-1:2015, Equipment for general lighting purposes - EMC immunity - Part 1: An objective345 voltage fluctuation immunity test method (2015).346 [15] IEC 61547-ed2.0, Equipment for general lighting purposes - EMC immunity (2009).347 [16] IEC 555-3, Disturbances in supply systems caused by household appliances and similar electric equip-348 ment - Part 3: Voltage fluctuations (1982).349 [17] UIE, Flicker measurement and evaluation, Tech. rep., International Union for Electroheat, Disturbances350 WG (1992).351 [18] D. Gallo, C. Landi, N. Pasquino, Design and calibration of an objective flickermeter, IEEE Transactions352 on Instrumentation and Measurement 55 (6) (2006) 2118 –2125. doi:10.1109/TIM.2006.887122.353 [19] I. Azcarate, J. Gutierrez, A. Lazkano, L. Leturiondo, P. Saiz, K. Redondo, J. Barros, Type testing of354 a highly accurate illuminance flickermeter, in: Proceedings of IEEE 15th International Conference on355 Harmonics and Quality of Power (ICHQP), 2012, pp. 897 –903.356 [20] J. J. Gutierrez, J. Ruiz, U. Irusta, S. R. d. Gauna, A New Alternative for the Input-Voltage Adaptor of357 the IEC Flickermeter, IEEE Transactions on Instrumentation and Measurement 57 (5) (2008) 923–930.358 doi:10.1109/TIM.2007.913820.359 [21] S. Uddin, H. Shareef, A. Mohamed, Power quality performance of energy-efficient low-wattage LED360 lamps, Measurement 46 (10) (2013) 3783–3795. doi:10.1016/j.measurement.2013.07.022.361 20