ABSTRACT
Contact: Wei Yan ee114868AciVu.edu.hk Introduction: It is known that the instability caused by acoustic resonance inside a HID lamp will be eliminated at extra high operating frequency. There are two possible circuit topologies for extra high frequency electronic ballast: one is called as LCC circuit and one is called as LCL circuit as shown in Fig. 1 (a) and (b). It is found that the starting process in a MH lamp at extra high operating frequency may affect the reliability of the electronic ballast and damage inverter switches [1]. This paper presents an experimental study on the starting process of a small wattage MH lamp and discusses the influence of the starting process upon the performance of different electronic ballast circuit. Experimental results on the starting process: The starting process of HID lamps has been extensively studied [2]. According to Ref. [2], the starting process inside a HID lamp should follow the procedures as: (1) breakdown, (2) mercury vapor arc, (3) glow, (4) glow-to-arc transition, (5) thermionic low pressure arc and (6) fully developed arc, where the mercury vapor arc is defined as the arc with high current value limited by ballast impedance and low voltage about 20 V. The mercury vapor arc is developed from the mercury condensed on the electrodes during the previous cool down. Our experiments for Philips CDM-T 35 W MH lamp driven by both 50 Hz magnetic ballast and 530 kHz electronic ballast give some different results. Fig. 2 (a) shows typical starting lamp voltage and lamp current waveforms at 50 Hz. Fig. 2 (b) is the expanded portion of the starting process. At 50 Hz, the starting process has two stages after the lamp arc has been broken down. The first one is mercury vapor arc that has big current and low voltage. As shown in Fig. 2 (b), the lamp voltage is about 20 V in most of period except the high re-ignition spike when the current changes its polarity. The lamp current is asymmetrical. Then, the mercury vapor arc stage is followed directly by the thermionic low pressure arc stage. The time from the striking of a discharge to the commence of the thermionic low pressure arc is over 200 ms. The thermionic arc has a symmetrical high current and voltage higher than 40 V with much lower re-ignition spike. The electrodes have been heated up by the mercury vapor arc to a temperature that can sustain thermionic emission. Figs. 3 (a) and (b) show typical lamp starting voltage and current waveforms driven by a 530 kHz LCC ballast. The time from the striking of the lamp to thermionic low-pressure arc is less than 80 ms. The period time of 530 kHz is about 1.89 us, which is only 1/10000 of the period time of 50 Hz. The glow discharge occurs after the lamp arc has been broken down, which has low current value and high voltage value (over 200 V). The high voltage across the lamp and parallel capacitor enhances the field emission in electrodes. The sudden reduction in the lamp resistance will cause the parallel capacitor to discharge through the lamp and will result in a current spike in the waveform. It seems like there is no mercury vapor arc phase under extra high frequency starting operation. Depending on the heating process in each electrode, one electrode will enter the thennionic emission phase first. The lamp current and lamp voltage are asymmetrical. When both electrodes enter the thermionic emission phase, the glow to arc transition ends. It should be noted that the lamp current and voltage might be highly asymmetrical during the glow to arc transition phase at extra high frequency. The effect of this asymmetry in the lamp current may affect the performance of the electronic ballast. Fig. 4 (a) shows the starting waveform for LCC ballast. Although the lamp current and the lamp voltage are asymmetrical, the inductor current is symmetrical. Therefore, the performance of the LCC ballast will not be affected by the transient starting process. Fig. 4 (b) shows the starting waveform for LCL ballast. Here, the lamp current is the same as the inductor current and the inverter switch current as well. An asymmetrical lamp current may cause the inductor to saturate. Therefore, special considerations should be taken in the inductor design in order to avoid core saturation. Moreover, the very small glow current cannot satisfy the requirement for zero voltage soft switching during the starting process. Hard-switching noise may cause the inverter switches damaged permanently. An auxiliary parallel resonant tank should be introduced to the LCL topology to improve the transient performance of the ballast [1, at the expense of an increase in the cost and components of the ballast.
