Electronic ballasts for fluorescent lamps have already overtaken magnetic ballasts in both volume and value. The same trend is now taking place in the high-intensity discharge (HID) lamp ballast market. HID lamps deliver a high-brightness output and typically serve indoor applications such as retail accent or ceiling lighting, and outdoor applications such as street lighting. New applications including automotive headlamps, front projection for meeting rooms and rear projection (DLP TVs) are also now using HID ballasts.
HID lamps have unique electrical characteristics and require a careful and specific control method. There are basic HID lamp requirements the designer must consider, as well as key protection requirements necessary for safety and to prevent destruction of the lamp or ballast. Let us look at the various methods of controlling each ballast subcircuit, with emphasis on the full-bridge output stage. A fundamental understanding of these concepts will help the designer gain further insight to the nature of HID lamps and the circuits that control them.
HID Lamp Requirements
HID lamps are available in the form of metal halide, mercury or sodium vapor. These lamps are popular because they are efficient and have a high-brightness output. HID metal-halide lamps are typically five times as efficient as incandescent lamps and last 20 times longer. In the case of sodium vapor, they are twice as efficient as normal fluorescent bulbs. HID lamps produce light using a technique similar to that in fluorescent lamps in which a low-pressure mercury vapor produces ultraviolet light that excites a phosphor coating on the tube. In the case of HID lamps, the gas is under high pressure, the distance between the electrodes is short and the lamp produces light directly without the need for a phosphor.
HID lamps require a high voltage for ignition, typically 3 kV to 4 kV, but more than 20 kV if the lamp is hot. The lamps also require current limitation during warmup and constant power control while running. It is important to tightly regulate lamp power with respect to lamp voltage to minimize lamp-to-lamp color and brightness variations. Also, HID lamps use an ac-voltage drive to avoid mercury migration. They operate at a low frequency, typically less than 200 Hz, to prevent lamp damage or explosion due to acoustic resonance. A typical metal-halide 70-W HID lamp has the following requirements: a nominal wattage of 70 W; a warmup time of 1 min to 2 min; and a cold-start ignition voltage of 4000 VPK.
Fig. 1 shows the typical startup profile for HID lamps. Before ignition, the lamp is an open circuit. After the lamp ignites, the lamp voltage drops quickly from the open-circuit voltage to a very low value — typically 20 V — due to the low resistance of the lamp. If otherwise unimpeded, this characteristic causes the lamp current to increase to a high value; therefore, the ballast must limit the lamp current to a safe maximum level. As the lamp warms up, the current decreases as the voltage and power increase. Eventually, the lamp voltage reaches its nominal value, typically 100 V, and the ballast regulates the power to the correct level.
To satisfy the lamp requirements and different operating modes, an electronic-ballast topology must efficiently convert the ac mains voltage to the appropriate ac lamp voltage, ignite the lamp and regulate power.
More on Buck Converters• Buck-Converter Design Demystified
• Optimizing Voltage Selection in Buck Converters
• Power Conversion Synthesis Part 1: Buck Converter Design
• Improving Efficiency in Synchronous Buck Converters
HID Ballast Topology
A typical HID ballast (Fig. 2) performs eight basic functions. An electromagnetic interference (EMI) filter blocks ballast-generated noise. A full-wave rectifier provides the high-voltage bus power. A power-factor-correction (PFC) block ensures sinusoidal input current. A buck converter controls the lamp current. A full-bridge output stage provides the ac lamp drive. An ignition circuit strikes the lamp. Control circuitry manages each stage. Finally, protection circuitry safely deactivates the ballast in the event of a lamp- or ballast-fault condition. Currently, this is one of the most popular approaches to powering HID lamps with a low-frequency ac voltage.
The PFC stage is a boost converter that operates in critical-conduction mode with a free-running frequency. This is a standard topology that many power-supply and ballast applications use for power levels below 100 W. The PFC stage maintains a sinusoidal current that is in phase with the ac line input (to attain a high power factor and low total harmonic distortion) and regulates the dc bus output to a constant level, typically 400 Vdc. When the PFC switch (M1) turns on the current, the boost inductor (LBOOST) ramps up linearly to a peak value. Switch M1 then turns off and the inductor current discharges back down to zero. When the current reaches zero, M1 turns on again and the cycle repeats itself. The amount of current necessary to keep the dc bus regulated at a constant level for a given load power determines the on-time. Since the input voltage to the PFC stage is sinusoidal, the resulting current will be triangular within each switching cycle, with the peaks following a sinusoidal envelope (Fig. 3).
The on-time will be approximately constant and the off-time will vary depending on how high the peak is for each switching cycle, resulting in a free-running frequency system. When the EMI filter at the input smoothes these triangular-shaped currents, the result is a sinusoidal current that is in phase with the ac input voltage (the dashed line in Fig. 3).
The buck stage controls the amount of current that the ballast delivers to the lamp load while warming up and running. Immediately after the lamp ignites, the lamp resistance drops and the lamp passes a large current. The buck controller should supply adequate current to keep the lamp from extinguishing, but the current limiter must prevent the buck inductor from saturating while the lamp is warming up.
While the lamp is running, the controller manages the buck's on-time to keep the lamp power constant. Current flows from the dc bus through the buck inductor to the load when the buck switch (M2) turns on. During the on-time, the current in the buck inductor (LBUCK) increases linearly as it supplies load current.
When the on-time ends, the buck switch turns off and load current continues to flow in the buck diode (DBUCK) and the buck inductor. The current through the buck inductor decreases linearly for the duration of the cycle. The controller adjusts the on-time depending on how much current the load needs to regulate the power. The time it takes for the buck inductor current to discharge to zero determines the off-time. A standard PWM circuit can control the buck stage and a high-voltage level-shift IC (such as the IR2117) boosts the gate-drive signal up to the buck switch's gate-to-source potential.
The output stage includes a full-bridge circuit for driving the lamp with a low-frequency square-wave voltage and an ignition circuit for striking the lamp. The top of the full-bridge circuit connects to the buck output voltage and the two half-bridge midpoints oscillate 180 degrees out of phase from each other to produce the necessary ac voltage.
During the ignition phase, the lamp is an open circuit and the buck output voltage is limited to a maximum value. The ignition circuit comprises a diac (DIGN), transformer (TIGN), capacitor (CIGN), resistor (RIGN) and switch (MIGN). When the ignition controller turns on switch MIGN, capacitor CIGN discharges through resistor RIGN.
When the voltage across the diac reaches the diac threshold voltage (Fig. 4), the diac turns on and a current pulse flows from the buck output, through the primary winding of the ignition transformer (TIGN) and into capacitor CIGN. This arrangement generates a high-voltage pulse on the secondary to ignite the lamp. The capacitor CIGN charges up until the diac turns off, and CIGN then discharges down through resistor RIGN until the diac voltage again reaches the device's threshold and another ignition pulse occurs. When the lamp ignites, the buck output voltage decreases quickly to the lamp voltage as the converter provides the lamp current. The ignition controller disables the pulses after the lamp has ignited by turning switch MIGN off.
An IRS2453D full-bridge-control IC manages the lamp-drive bridge. This high-voltage IC contains all of the necessary circuitry for the full-bridge oscillator and high- and low-side gate drivers. The IC also contains a nonlatched and latched shutdown pin as well as integrated bootstrap diodes for the high-side driver supplies. The timing diagram shows the CT oscillator timing pin, the gate-driver outputs, and the resulting midpoint and lamp voltages (Fig. 5). The IC also includes an internal 1.5-µs dead time between the low-side (LO) and high-side (HO) gate-drive outputs. This dead time prevents external MOSFET shoot-through and allows for each half-bridge voltage to self-commutate for zero-voltage switching.
The HID ballast should include specific protection circuits to detect various lamp- and ballast-fault conditions and safely shutdown or reset the ballast. These fault conditions include ac-mains interrupt or brownout, lamp ignition failure, lamp warmup failure, lamp open circuit, lamp short circuit and lamp end-of-life. A summary of these conditions appear in the table along with the proper ballast response to each fault and the possible outcome if the ballast does not protect against the fault.
Voltage and current signals within the various stages can serve as detection points to realize the protection circuitry. The ac line or dc bus voltages can reset the ballast if a brownout condition occurs. Timers are typically necessary to deactivate the ballast after a predetermined time period should the lamp fail to ignite or warmup. A lamp voltage or power monitor can detect if the lamp is unstable or is reaching end of life. Regardless of the specific methods a ballast design implements to detect each fault, the protection circuits should be robust and reliable to ensure proper safety in the ballast application and to prevent catastrophic field failures should fault conditions occur.
New applications and lamp types are continuously emerging in the marketplace and each includes its own unique design challenges. International Rectifier and other manufacturers in the industry will continue to improve and simplify control methods and ICs in the field of HID lighting. Designers will need to stay on top of the rapid changes that are sure to take place in the coming years.
|Fault condition||Ballast action||Unprotected outcome|
|AC mains interrupt or brownout||Reset ballast and restrike lamp||Lamp can extinguish and remain off|
|Lamp does not ignite||Deactivate ballast if lamp does not ignite after a maximum time period||High-voltage safety hazard at output terminals|
|Lamp does not warm up||Deactivate ballast if lamp does not reach nominal power after a maximum time period||High current stress on buck and full-bridge stages can cause component failures.|
|Lamp is not connected or has a broken connection||Deactivate ballast after a maximum time period||Ballast will try to ignite lamp|
|High-voltage safety hazard at output terminals|
|Short circuit at ballast output terminals||Deactivate ballast after a maximum time period||High current stress on buck and full-bridge stages can cause component failures|
|Unstable lamp or end of life||Deactivate ballast after maximum time period||Lamp can extinguish, flicker, conduct in one direction, or encounter color or brightness shifting |
Unstable lamp can cause disruption or failure of ballast circuits