INTEGRATED SYSTEM FOR INTELLIGENT STREET LIGHTING

Maizonave, G. B.; Dos Reis, F. S., IEEE Member;Lima, J. C. M.;Bombardieri, A. J.;

Chiapetta F., E.; Ceccon, G.B.; Souza, R.R.N.; Tonkoski Jr., R., IEEE Student Member; Dos Reis, R. W.

Pontifícia Universidade Católica do Rio Grande do Sul

Avenida Ipiranga, 6681 – Porto Alegre, RS – Brasil CEP: 90619-900

Abstract –This reports the study and hardware implementation of a dimmable electronic ballast for high pressure sodium lamps, and a microprocessor-based system for control and energy measurement for this ballast, which uses a power line communications system to send and receive status and commands from another ballasts plugged in the same mains subcircuit. These ballasts are connected in the topology of a logic network, one of them being defined as the master of the subcircuit, and the others as slaves. The master unit distinguishes from the slaves by the additional communications system, which works through a cell phone, and enables the wireless connection to a PC-based central supervisory system. This way, any locale or town becomes able to control its entire main lighting system, in addition to the obtainment of more accurate data about energy billing, which together with the ability to control luminosity and the better power factor, will result in financial and energetic economy.

Index Terms—Street Lighting, High Pressure Sodium Lamp, HPSL, Power Line Communications, PLC, Cellular Telephony, GPRS, SMS.

IIntroduction

Before incandescent lamps came out, street lighting was obtained by the burning of gas or combustible oil. Street lighting was registered for the first time in London, in 1417, when Sir Henry Barton mayor, ordered torches and animal oil lamps to be hanged between Hallowtide e Candlemasse[1]. Benjamin Franklin achieved a drastic reduction on the number of murder deaths in Philadelphia, by means of a campaign for illumination of the streets, using whale oil.

The first electrical means of street lighting for large areas used a model of arc lamp developed by the RussianPavel Yablochkov in 1875. Such lamps became known by the names “electric candle” or “Yablochkov‘s candle”. With the creation of the incandescent filament lamps, during the 19th century, cheaper, trustful and highly bright, the arc lamps where quickly replaced in the street lighting, although its industrial application has lastedfor a long period yet.

The incandescent lamps, employed in street lighting up to the advent of the high intensity discharge lamps, operated normally as series circuits connected to an alternating high voltage source.

  1. State of Art

Nowadays street lighting uses mostly discharge lamps, usually high pressure sodium lamps. There are streets that use sodium lamps in one side and high pressure mercury in the other. There are still places that, in opposition to every technical recommendation, still use incandescent-mercury lamps on street lighting.

Almost all ballasts used in street lighting are of the electromagnetic type, since the market for the electronic ballasts is still incipient.

There exist several kinds of high pressure discharge lamps, whichcharacterize by the excellent luminous efficiency and long lifetime, overcoming the other types of lamp in both features. The HID lamps are divided in three distinct families: High Pressure Mercury (including the incandescent-mercury lamps), Metal Halide and High Pressure Sodium Lamps.

  1. The Negative Resistance Behavior

Arc discharge lamps must operate with a ballast which limits it’s current, because they present a feature usually named “Negative Resistance”.

After gas ionization inside the internal lamp bulb, the free electrons, emitted by an electrode (cathode) migrate to the other electrode (anode), and this migration causes collisions between the electrons and the atoms on the gas.

The collisions cause the transfer of electrons from internal orbits in the atom to more external ones, leading to an excitation state. As the electrons jump back to their original orbits, photon emissions occur, radiating in this way energy at several wavelengths, so much in visible bands as in the invisible ones. Another consequence of this process, almost always undesirable, is the chafing occurring in these collisions.

In a specific amount of these collisionsthe liberation of electrons from their atoms. These electrons, once free, collide with other atoms in the gas, originating again the described process. The repetition of this phenomenon causes an electronic flood, which initiates more electric current in the gas, making thus the conductance of the ionized path to be directly proportional to the intensity of its current. To this behavior, opposite to the one described in Ohm’s law, the name “negative resistance” was given. Consequently it becomes necessary to limit the current circulating through the lamp, by some impedance in series with the source circuit, without which the lamp would destroy itself.Fig.1 depicts the typical negative resistance behavior of a discharge lamp.

Fig.1 Negative resistance behavior.

The breakage voltage of the lamp is the required voltage for its ignition, and it depends on, among some other factors, the arc bulb geometry (basically the diameter and length), on the type of gases used in the lamp’s structure, on the pressure and temperature of these gases and on the type of the used electrode.Typical values of breakage voltage for low pressure lamps are between 50 V and 1.2 kV.

  1. The Adopted Inverter Topology

The power stage of the circuitry is accomplished by one full bridge inverter and a filter of the type series LC, parallel C. This kind of resonant filter allows limiting the lamp current by the series capacitor Cs and the L inductor, at the same time performing the lamp ignition, with the high voltage developed at the Cp capacitor terminals, which is connected in parallel with the lamp. This voltage appears at the Cp capacitor only if: a) the inverter operation frequency is equivalent to its resonance frequency or a multiple of this frequency; b)the lamp is off, since it can be considered an open circuit in this situation. Fig. 2 presents the complete circuitry of this ballast,which is composed by: a) a full bridge rectifier associated to a very small value capacitor; b) a full bridge inverter; c) an LC filter in the converter input, in order to minimize the conducted EMI generated by the structure. It is important to mention that the Rlamp resistor represents the circuit’s equivalent resistance when the lamp is on.

Fig.2 Full bridge resonant inverter with a series LC, parallel C filter.

The main advantages of the proposed structure are: a) high power factor (PF), in spite ofthe absence of a conventional PFP; b) an extreme topological simplicity if compared with conventional electronic ballasts; c) acoustic resonance minimization since the lamp current becomes zero at every mains semicycle (Fig.5) and remains in this state until a new cycle begins, in a way very similar to what occurs inelectromagnetic ballasts. The main disadvantage of this structure is the crest factor (CF) of two (CF 2) which is slightly higher than 1.8 (maximum value recommended to fluorescent lamps although there are no regulations with a maximum CF value for HID lamps).

  1. Experimental Results

From the implemented prototype, as Fig.3 illustrates, it was performed a set of waveforms acquisitions which are illustrated as follows.

Fig.3Implemented ballast prototype.

Fig.4shows the voltage and current waveforms at the electronic ballast’s input, with nominal power.One may conclude that the input current is in phase with the mains voltage waveform, what can result in high power factor.It can also be observed that during the reigniting periods that occur in every semicycle, the input current behaves exactly the same as an electromagnetic ballast, but it becomes perfectly resistive, as soon as the arc is established inside the lamp.

Fig.4Input voltage and current waveforms, with nominal power.

It is important to take note that the implementation of an EMI filter attached to the circuit significantly improves the input current waveform, which would behave as the one from a discontinuous mode converter, if it was not present between the mains and the ballast.

The obtained power factor at the device’s input was 0.98, with a current THD of 19.8%. The mains in the lab presented a THD of 2.5% during the performed measurements.

Fig.5 belowpresents the voltage and current waveforms at the lamp, during one entire mains cycle, at 60Hz frequency.Observance of the current waveform at the lamp leads to the conclusion that it turns off and on at each semi cycle.

Fig.5Waveforms of the voltage (above) and current (below) at the lamp, in the mains’ frequency.

Fig.6shows the voltage and current waveforms at the lamp in high frequency cycles.In this figure it is possible to remark the resistive behavior of the HPS lamp when operating in high frequency, because both waves are in phase and present almost the same form.Minor differencesmay be observed at the MOSFET commutation instant, due to the interference in the oscilloscope cables.

Fig.6Voltage (above) and current (below) waveforms at the lamp, in 68kHz cycles.

Fig.7displays the high frequency power waveform, acquired from the product of the voltage and the current at the lamp (Vlamp x Ilamp) in reduced intensity, during a period of mains peak voltage.The measured instantaneous average value was about 160 W.

Fig.7Voltage and power at the lamp with reduced intensity in high frequency cycles during a mains voltage peak.

Fig 8 Percent values of the ballast’s input current harmonics.

Fig. 9 Percent values of the mains voltage harmonics.

Fig. 10 Ballast’s electrical performance values at nominal power.

  1. Additional Features

Being an intelligent ballast, the developed device aggregates several functions, as control, supervision, power metering and communications with other ballasts of the same model, as well as with the supervisory central.

The metering of the power consumption for taxing purposes was implemented with a power meter, which was specially designed for this purpose. It uses a voltage transformer and a Hall Effect sensorin order to sample the input voltage and current. Such devices are linked to an analog multiplier which acquaints the power information.The proposed system is described in details by schematic diagrams and theoretical fundamentals.

The communication between ballasts, detailed in the fourth section, is accomplished through the mains power wires.The communication process with the supervisory central is made with cellular telephony. The main supervisory software can operate in any low cost personal computer with serial or USB communications, linked to a cell phone.

The control and supervision system bases itself in a microprocessor which permits high level programming (so enabling portability).The processor has a high frequency clock and high processing speed, what is necessary because of the great volume of tasks under its control.

IIThe Power Meter

  1. Introduction

The power meter circuit is used to measure the power absorbed by the lighting device and to protect the circuit, allowing it to turn off in case of overload.

The calculation of the instant power in a circuit is made by the product of instant voltage and current, thus the active power delivered to the load may be obtained by integration of the instant power, as shown in(1).

/ (1)

The following block diagram represents the power meter circuit and the steps that it implements to obtain the measured active power.The circuit protection is performed by the microprocessor, which receives the power measurement and decides whether to turn it off or not.It also continuously performs a second integration inthe power measurements, in order to compute the total energy drained by the lighting system.

Fig. 11 Block diagram of the performed operations.

  1. Sampling of the mains voltage

In order to sample the mains voltage, a voltage transformer with 220 VAC input and 9+9 VAC output and a resistive voltage divider were used, as shown inFig.12.

Several transformer models were tested, until a commercial model that causes the least possible distortion on the sampled voltage waveform was found.

A resistive voltage divider was added in order to make the sample compatible with the multiplier circuit. The transforming relation is given by(2).

Vout = 0,307Vin / (2)

Fig.12Voltage sensor.

  1. Input current sampling

A Hall Effect (MSA Control‘s model SC-50) was used to obtain the input current sample. Its use requires a current source, accomplished with a transistor. Fig.13 represents the proposed circuit, where one can remark an operational amplifier in differential configuration. The result of this circuit is the obtainment of a voltage signal, proportional to the measured current. Its amplitude is adjusted before the application in the multiplier circuit.

Fig.13 Current sensor.

The R21 trimmer can be used to correct offset voltages as much in the Hall Effect sensor, as in the operational amplifier itself. This adjustment is accomplished by placing a short-circuited winding through the sensor’s window, and adjusting the trimmer in order to obtain zero volts at the operational amplifier’s output (pin 14).

  1. The Protection Circuit

Since the power measurement is obtained from the voltage and current sensors, from multiplier and the integrator, the microprocessor achieves the tasks of totalization of the energy consumption and system protection.

The 370 W limit was used for this purpose, so it must not be passed beyond. As a security measure, the microcontroller has access to a control pin which can perform a shutdown operation, in case this limit is reached. The circuit’s protection is achieved by a direct action in the suitable pin in the LM3524 oscillator, in order to turn the inverter MOSFETs off.

The routines concerning the overload protection system in the microcontroller were made in order to have an inhibition period, thus not acting during the lamp ignition, since power consumption can be higer when the lamp is cold.

Since this whole task is only software-based, it may be easily changed in order to adapt to specific features of the conjugated ballast and lamp, thus granting much versatility to the system.

As the ignition process finishes, the microprocessor initiates the protection procedures, checkint whether the measured power does not reach the 370 W limit.

IIIThe Power Line Communications System

Due to the additional features of this electronic ballast, towards other commercial models, the necessity of communication of the various ballasts with the sub central became evident. In order to accomplish this, the mains power line was used as physical media for data transmission, hence eliminating the need of an additional physical network.

The development of a PLC (Power Line Carrier) was achieved, using only basic components, as passive elements and logical ports, based on a Microchip technical report.

Communication is established by modulation, transmission and demodulation of the concerning data. The first step (modulation) is carried out by the transmitter, having as transmission media the conventional AC power line. The second step (demodulation) is achieved by the receiver circuit. The basis of this modulation process is the sampling, which consists of instantaneous measurements of an analog signal at specific time instants, generating a sequenceof samples uniformly distributed in time.

The reception of this signal is made by the same filters, which eliminate the low frequency components that come from the mains power line. The tests carried out with the implemented modem demonstrated fairly good reliability, as much in the transmission as in the reception process.

Fig.14 The developed data transmission circuit.

This performance analysiscomprised a data transmission essay using the University mains network, and a power cable, a hundred meters long.

Fig.15The developed data reception circuit.

The transmission circuit injects a high frequency carrier in the power line, which is demodulated by the receiving side and again converted to a digital signal that can be normally read by the microprocessor. This modem uses a transformer-based coupling circuit, tuned in the same frequency of the other side.The carrier signal is generated by the microcontroller. When a “1” bit is transmitted, the carrier signal is injected on the power line. The demodulator circuit, perceivingthe presence of the carrier signal on the mains, demodulates it and places the logic level “1” in its output.

During the tests with both modems, it was noticed that the noisepresent in the mains affected their performance, corrupting the signal just when the transmitter did not inject the carrier signal on the network (which should correspond to logic “0”). In order to solve or get around this problem, a signal coding technique was applied, in such a way that each “0” or “1” symbol is represented by the injection of the carrier signal, but with different times.

The results were greatly satisfying, as Fig.16 illustrates the waveforms collected by a digital oscilloscope. The upper part of the picture displays the modulated and transmitted signal, while the one in the bottom depicts the already demodulated and received signal.