2013 IEEE 17th International Symposium on Power Line Communications and Its Applications The Impact of thhe CFL Lamps on the Power-Line P Com mmunications Channel A. Emleh, A.S. de Beer, H.C. Ferreira A.J. Haan Vinck Department of Electrical and Electronic Engineering E Science University of Johannesburg, P.O. Boox 524, Auckland Park, 2006, South Afrrica {aemleh, asdebeer, hcferreira}@ujj.ac.za Institute for Experimental Mathematics Duisburg-Esssen University Ellernstr. 29, D-453326, Essen, Germany vinck@iem m.uni-due.de Abstract – A number of measurements show w the results of the harmonics and conducted emissions from “Energy-Savings” Compact Fluorescent Lamps (CFL) when connected to the power-line communications channel. Differrent CFL’s from different manufacturers were investigated. Th he paper covers the CENELEC band, as well as the broadban nd communications channel: 150kHz – 30MHz. The obtained resu ults show the levels of harmonics and interference that these typess of lamps produce. It shows that CFL’s produce interference in the t 3kHz - 150kHz band, but this pose no risk for PLC. Some CFL’s do however produce interference in the 150kHz – 30MHz band that can interfere with power-line communications. II. MEASURE EMENT SET-UP Fig.1. shows the measurem ment set-up for the low-frequency CENELEC band, where the lamp l is supplied with 220VAC through an isolation transfformer and Line Impedance Stabilization Network (LISN). Index Terms – CFL, Compact Fluorescent Lamp, Harmonics, Powerline Communications, PLC, EMC, Interference, noise. I. INTRODUCTION Distortion to the voltage and current waaveforms in the inbuilding powerline communications channell has increased due to the harmonic impact caused by the CFL L’S [1] where the current total harmonic distortion can exceed 100% [2]. To analyze the harmonic impact of largee-scale in-building loads, we investigate the effects of the CFL L’S when seen as interference sources on the powerline commuunications channel since they are one of the main power-quuality concerns in residential areas [3]. This paper investigates the effects when compact fluorescent lamps are seen as interference souurces on the wiring system of the powerline communicationns channel. Two different measurement set-ups (depending on o the interference band) are given. The procedure of the measurements wass made to comply with CENELEC narrowband rules in one casse, and to serve the broadband signals in the other case [4]. Since compact fluorescent lamps range beetween high to low quality lamps, harmonics are more likely to be b produced by the low quality ones. The main cause to the harmonics (in the C use in their CENELEC band) is the rectifiers that the CFL’s normal operation. [5]. The cause of noise in the 150kHz – 30MHz band is a power electronics converteer employed by all CFL’s. 978-1-4673-6016-6/13/$31.00©2013 IEEE Fig.1. Measurement set-up for measureements in the 3kHz – 150kHz range. The isolation transformer is used as the LISN causes an earth-leakage current to flow,, which makes the supply trip. Floating the LISN rectifies thhis fault condition. The LISN is used in this set-up to meet the following f needs: 1. It supplies a standardized noise load to the conducted interference created byy compact fluorescent lamps. At higher frequencies (typpically > 1MHz) the noise load impedance presented by b the LISN (and seen by the CFL’S) is 50Ω. 2. It filters noise from thee AC supply. The measurement side (current probe annd CFL in Fig. 1) is therefore clean from any noisee on the power supply and an accurate assessment of o the noise produced by the CFL’S can therefore be made. A clean 50Hz 220VAC is supplied to the CFL. Measurements and conclussions in this paper are made for two regions of the emission speectrum: • 3kHz – 150kHz: This is the frequency range of the so called CENELEC bandds as defined by EN 50065-1 [6]. Measurements for thesse bands were made in the time 225 • domain and a Discrete Fourier Transform (DFT) performed to obtain harmonics in i the frequency domain. A Tektronix DPO7254 oscilloscope and w used. It was Tektronix TCP0030 current probe were assumed that the Common Mode (CM) ( currents are negligible in this band and that all interference is in mption also used in Differential Mode (DM) – an assum EN 50065-1. Results were downloaded to a PC for processing. This measurement set-upp is shown in Fig. 1. 150kHz – 30MHz: This is often thhe frequency range for Broadband PLC. It spans the range r traditionally used to measure conducted emissionns as per CISPR-16 [7]. In this case, measurements in the frequency domain were directly made using a Rhode & Schwarz S-Lindgren 94111FSH323 Spectrum Analyzer and ETS 1L 1GHz bandwidth current probe. Since interference on a PLC system occurs in DM, a sppecial arrangement of cables with respect to the current probe p was used for measuring in this range. This speciaal arrangement can be seen in Fig. 2. It cancels the CM current and measures the DM only. The set-up shows an isolation transformer connected to 220V from one end annd to a LISN from the other end. A CFL lamp is connnected to the LISN through a current probe in differrential mode. The current probe is linked to a specctrum analyzer to provide the computer with the obtainned results. w power converter and filter as consists of an active rectifier with shown in Fig. 3. Fig.3. Rectifier and converter type CFL L driver. VAC input is rectified and filtered In the CFL the 50Hz, 220V (by C). This high voltage DC C is then converted and current limited by an active high frequuency switching power electronic converter. Although there are different d topologies for this kind of converter (which falls beyoond the scope of this paper) it is sufficient to say that this circuuit will produce high frequency switching harmonics in the currrent drawn from the supply, over and above the lower networkk harmonics due to the rectifier action. The filter between the reectifier and AC supply filters the high frequency (150kHz – 30 MHz) noise being produced by the converter. As it will be shown, manufacturer’s designs vary in the amount of this noise beinng filtered. Figure 4 shows measured line current waveforms for four f trace shows the line current compact fluorescent lamps. A fifth for all the lamps switched on o at once. These lamps were manufactured by four differennt companies. The lamps have similar waveforms which varry slightly in amplitude due to differences in the brands. IV. HARMONICS - CENELEC BANDS In order to determine whhat effect the harmonics of the compact fluorescent lamp has on the power-line communications channel, thee currents in Fig. 4 must be represented in the frequenccy domain. This is done by performing a Discrete Fourier Transform T (DFT) on exactly one period of data from Fig. 4. Perrforming a DFT on one cycle of data yields the harmonics (or frequency domain components) w that make up the time domain waveform. Fig.2. Measurement set-up for measurements in the 1500kHz – 30MHz range. URE III. CFL’S DRIVER STRUCTU In this section we show that, for noise geeneration in CFL’s, there is a common structure of the CFL’S drivers, which 226 Fig. 4. Time domain line current waveforms for fourr different compact fluorescent lamps. The voltage and line current with all the lamps switched on together are also shown. The current harmonics (magnitude) for the waveforms in Fig. 5. Frequency domain line current harmonics (D DFT) for the four different CFL’S shown in Fig.4. A fifth spectrum shows s the harmonics when all the lamps are switched on simultaneously. 227 The DFT’s of Fig. 4 are shown in Fig. 5. The DFT’s were performed for each of the four waveforms in Fig. 4 with the addition of a fifth analysis which is the harmonics of the current of all the lamps from the differentt brands combined (turned on simultaneously). This measurem ment of the fifth analysis was performed to show the interferrence that the CFL lamps cause to the powerline channel. Thee fundamental for each lamp is at 50Hz. The “all lamps combined” fundamental is the largest as can be expected, since all thhe lamps combined draws the largest amount of current. The harmonics h (starting close to a hundred mA at the fundamental) rooll off to the tens of µA’s around 20kHz which is close to the noisse floor. In order to compare the current harmonics to typical powerline channel signal voltages, the current maagnitudes in Fig.5 must be multiplied with typical power-line chhannel impedances for each of the harmonic frequencies. This power-line p channel impedance can be approximated by using the values of the LISN that is used for measurement (Fig. 1 and 2). The LISN characteristics are specified in EN 50065-1. Figure 6 gives the results when the “all lamps combined” harmonics (in voltage) are plotted against thhe CENELEC EN 50065-1 standards for maximum power-linne communications signal and Electromagnetic Compatibility (EMC) levels. The “all lamps combined” is the worst case noise for the lamps. At 135dBµV (around 5V) the allowablee wideband signal strength for a PLC signal is very high. It is around 40 - 60dB s of all of the higher than the noise harmonics from the signal CFL’S tested together. The noise harmonnics are also well below the allowable EMC limit as stated in EN E 50065-1. It can therefore clearly be seen that in the CENE ELEC bands from 3kHz – 150kHz, CFL’s are unlikely to inteerfere with powerline channel communications. CFL’S are shown. All those foour incorporate power electronics (of the structure shown in Figg. 3.). In terms of noise in the Broadband PLC spectrum (1150kHz – 30MHz) the lamps generate noise in this part of o the spectrum. – due to the switching action of the power electronics e converter. Fig. 7 shows the interfeerence voltage from the lamps versus the EMC average and peak/quasi-peak disturbance level limits in the band 150kHz – 30MHz. In the CENELEC bands from 3kHz – 150kHz there are dedicated maximum signal n exist in the 150kHz – 30MHz transmission levels. These do not band and maximum signal trannsmission is assumed to be at the EMC limit levels. It can be seeen in Fig. 7 that for some of the lamps the interference voltage is close to the EMC limit in the beginning of the band. This can c be expected as some of the manufacturers only filter noisee to the EMC limit in order to save on manufacturing costss. In stark contrast from the CENELEC bands, PLC signals in the 150kHz – 30MHz band have to directly compete with noise from devices with power C and may have a zero SNR electronic converters, such as CFL’s, at certain frequencies. NCLUSION VI. CON In this paper, four brands of “Energy-Savings“ compact fluorescent lamps were testeed for noise generation in the power-line communications chhannel. This group of CFL’s uses active power electronic conveerters to drive the lamps. They produce interference in the 3kH Hz - 150kHz band, but this pose low risk for PLC. Some of o them do however produce interference in the 150kHz – 30MHz 3 band. As the noise level and PLC signal level are goverrned by the same EMC standard, the CFL’s can cause noise at the t levels of PLC transmissions, effectively drowning the com mmunication signals Unless this standard is revised and PLC signals allowed to exceed the munication signals will have to EMC limit, power-line comm compete with zero signal to noiise ratios. M V. BROADBAND SPECTRUM In Fig. 4, the line current drawn by four different brands of mps combined” with the EMC standards and maximum allowable PLC P signal strength in the CENELEC Fig. 6. Comparison of the harmonics of “all the lam band 3kHz – 150kHz. 228 Fig. 7. High Frequency, including the 150kHz – 30MH Hz band, spectrum results. [3] Chen Jiang, D. Salles, W. Xu, W. Freitas, “Assessing the Collective Harmonic Impact of Modern Residential Loads—Part II: Applications” IEEE Tranns. On Power Delivery, Publication Year: 2012, Page(s): 1947 – 1955. 1 [4] H. C. Ferreira, L. Lampe, J. Newbury and T. G. Swart, Power Line Communications: Theorry and Applications for Narrowband and Broadband Communicattions over Power Lines, Chichester, England: John Wiley & Sons, 2010. [5] M. Ghafouri, A. Razi Kazeemi, P. Dehghanian, M. Vakilian, “Investigation of the Effects of Compact Fluorescent Lamps in Power Distribution Systems”” 10th International Conference on Environment and Electrical Engineering (EEEIC), Publication Year: 2011 , Page(s): 1 - 4. [6] EN 50065-1: Signaling on loow-voltage electrical installations in the frequency range 3 kHz to 148,5 kHz - Part 1: General requirements, frequency bands and electromagnetic disturbances. European Staandard, CENELEC, Ref. No. EN 50065-1:2011 E, Brussels, Appril 2011. [7] CISPR 16: Specification forr radio disturbance and immunity measuring apparatus methods, International and Electrotechnical Commissionn (IEC). By comparing the results obtained from the CFL’s measurements to those obtained from classsical lamps, it was observed that using the CFL lamps resullts in having less harmonics and interference to the powerlinne communications channel. REFERENCES [1] R.A. Jabbar, M. Al-Dabbagh, A. Muhammadd, R.H. Khawaja, M. Akmal, M.R. Arif, “Impact of compact fluorescent f lamp on power quality,” Power Engineering Confereence, 2008. AUPEC '08. Australasian Universities, Publication Year: Y 2008 , Page(s): 1 – 5. [2] Jing Yong, Liang Chen, A.B. Nassif, Wilsunn Xu,”A FrequencyDomain Harmonic Model for Compact Fluorescent F Lamps” IEEE Trans. On Power Delivery, Publication Year: 2010, Page(s): 1182 – 1189. 229