Today more and more traditional magnetic ballasts for fluorescent lighting are replaced by electronic solutions to increase overall efficiency, and to achieve higher performance at lower weight and volume.

The main reason for replacing the 50/60 Hz transformer or magnetic ballast is energy efficiency. Depending on the application, the return on investment when replacing magnetic with electronic ballasts can be as short as one year, the electronic version being much more expensive though. At the same time, an electronic ballast provides higher performance and comfort at lower weight and volume, a great advantage in professional applications like fluorescent lighting for offices and buildings.

The gas discharge in a fluorescent lamp has negative differential resistance; this means current increases while operating voltage drops. Consequently, any fluorescent lamp needs a current limiting element in a series (the ballast). The traditional magnetic ballast together with the starter as shown in Figure 1 perfectly implements the requirements to operate a fluorescent lamp. Initially, the starter switch S1 is closed and a current flows through L1 and the filaments of the lamp. When the starter opens after a certain time – this is achieved thermally by means of a bimetal element – the filaments are at high temperature and the abrupt change in current induces a high voltage in the inductor and across the lamp. After ignition the impedance of the inductor limits the discharge current.

Some disadvantages of this simple ballast are obvious, others not. First the starter switch may open close to zero crossing of the line voltage. Current flow is small at this time and the same is true for the ignition voltage. The lamp may not strike and one can easily identify a magnetic ballast by recognizing several attempts to start the lamp. A less obvious disadvantage is the poor system efficiency due to two reasons. First for the sake of cost, a high loss in the inductor itself is accepted. Second the ions in the discharge recombine during zero crossing of the line voltage and have to be reionized in the next halfcycle. The latter effect results in con-siderable loss of energy.

One of the key advantages of the electronic ballast is that the lamp is driven with a much higher frequency of 30 to 60 kHz typically. Due to the higher operating frequency, the recom-bination of ions does not happen and the efficacy of the lamp itself increases about 10% compared to operation at 50/60Hz. Moreover the electronic ballast it-self is designed to achieve effi- ciency of more than 90 % and together with state of the art, high efficiency lamp technology (T5 lamps) the energy savings can easily achieve 30 % compared to magnetic ballast at line frequency.

Other advantages that electronic ballast may have are ‘perfect’ preheating of the filaments making lifetime of the lamp virtually independent from the number of switching cycles, flicker free start and operation, constant light out-put with variable input voltage and high power factor. Finally, import-ant for emergency lighting, the electronic ballast can be operated with DC input voltage i.e. from batteries.

The topology for fluorescent lamp ballast most popular in Europe is the voltage fed series resonant half-bridge shown in Figure 2. The half bridge is driven with variable frequency and a duty cycle close to 50 %. At startup, as long as the FL is not ignited, the ballast controller generates frequency well above the resonant frequency of L1/C1. Thus a high current flows through the lamp filaments heating them up to the desired temperature.

After a time that is normally determined by external components the controller starts to lower operating frequency towards the resonance. A high voltage across the lamp is generated as a result and the lamp will ignite. After ignition the impedance of the FL damps the resonant circuit fairly well and the voltage across the lamp drops close to the operating voltage. In most applications the lamp current is sensed directly or indirectly and the operating frequency is adjusted until the setpoint is met. As long as the operating frequency is above the resonant frequency of L1/C1 the MOSFETs are soft switched and switching losses are negligible while at the same time EMI is reduced.

MOSFETs with fast recovery body diode (FRFET®) are perfectly suited for this application. There are 500 V and 600 V Q-FET™ available with fast body diode as well as 600 V SuperFET™. Since the gate of the upper MOSFET needs high voltage drive, a high-side gate driver is needed. The FAN7380, FAN7383 and FAN7384 high voltage drivers as well as in some the FAN7382 meet all requirements with best in class noise immunity. Finally there are pure ballast controllers available, like the FAN7544 that implements the control and safety functions, as well as controllers with integrated high voltage gate drive like the FAN7532 ( Figure 3).

Current international standards demand for power factor correction in lighting equipment if consumed power is above 25W. This mainly has two reasons: One is that the classical incandescent bulb behaves like a resistor i.e., voltage and current are in phase. The second reason is that lighting con-sumes 10 to 12 % of the total power produced and is normally operated for several hours a day, quite long compared to other equipment. Consequently if lighting electronic was not power factor corrected this would cause significant additional losses in the power grid.

Since most fixtures have a total power below 150W, critical (or boundary or transition) mode PFC is the most economic solution. In this mode the peak current through the inductor is controlled such, that this peak value is proportional to the rectified input voltage. During off-time, the inductor current falls back to zero and zero crossing of the current i.e. de-magnetization of the inductor initiates the next switching cycle. It is easy to see that the average inductor current is proportional to the input voltage, the desired result. There are mainly two different approaches to control the peak inductor current. In the so called current mode as implemented in the FAN- 7527, the rectified line voltage is sensed to generate the current reference that sets the actual value of the peak current. The necessary divider network can cause considerable loss, something that ballast designers badly try to avoid. In voltage or constant on-time mode implemented with the FAN7529, the on time of the switching device is kept constant during one or more line half-cycles. Keeping ontime constant, peak switch current is again proportional to input voltage as can be easily derived from the basic differential equation dI/dt = V/L. Common to both modes is the sensing and regulation of the output voltage.

In a low-cost ballast one may find different PFC topologies. Either a high inductance value iron core choke is used to smooth input current or, more often, the power switch and the control IC are omitted and a so called charge-pump PFC is used. In this topology the half-bridge is used to drive the fluorescent lamp and the PFC at the same time as shown in Figure 4. Since the lamp power has to be regulated and there is no additional degree of freedom that could be used to control the PFC, it is very difficult to find proper L and C values that result in good power factor and stable lamp operation over a wider input voltage range. That is the main reason why this topology is not used more often, though it is an inexpensive solution.

In a gas discharge there is a region close to the cathode where the discharge voltage drops steeply and no light is emitted, the so called ‘Cathode Fall’. Due to the voltage drop and the current flow there is a certain amount of power dissipated in this region. With increasing operating time the filaments of the lamp become less and less emissive and the cathode fall increases. In turn, the power dissipation close to the cathode increases and this region of the lamp is heated up more and more. If the diameter of the lamp-tube is small, it could be heated up to the melting point. Therefore, the thinner the tube, the more important is a feature called EOL (End of Lamp Life) detection. Especially for T5 this feature is indis-pensable and contained within by European safety standards for fluorescent lighting.

Normally the FL is operated with AC and each filament is the cathode for 50 % of the time. Fortunately one of the two filaments will lose emissivity first and lamp behaviour becomes asymmetrical. Thus, it is possible to detect EOL by e.g., monitoring overall lamp voltage or symmetry of operating voltage or current.

CFL contain a electronic ballast integrated with the lamp. Since these replace incandescent bulbs traditionally they are thrown away after a defect. That’s why the electronic of a CFL does not need to have the extraordinary long lifetime of a fluorescent lamp ballast (up to 50,000 h). In addition the power is limited due to limited space and since PFC shall be omitted. Altogether, while having the same basic structure, the CFL uses a slightly different inverter circuit compared to the fluorescent lamp ballast. Instead of a control IC most CFL still use a self oscillating half bridge.

Newer integrated circuits like the FAN7711, a controller and high voltage gate driver, as well as the FAN7710 with additionally integrated Power MOSFETs help to simplify the design of CFL while being competitive in cost. This is especially true if additional performance and safety features implemented by the integrated controllers are considered.