The word is officially out—inefficient light sources are not cool. Pardon the pun, but we’re rapidly becoming aware of the impact our energy consumption has on the environment. Lighting applications consume a large portion of our overall energy consumption. According to the U.S. Department of Energy, lighting consumers 12% of residential energy and 25% of commercial energy. So, there’s certainly a significant amount of energy savings to be made by using lighting technology that’s more efficient.
The incandescent bulb has hardly changed since it was initially designed, and it has an efficacy of approximately 8 lumens per watt. Efficacy is a measurement of efficiency used by lighting gurus that specifies how many lumens of light output are produced for each watt of input power. Approximately 95% of the power put into an incandescent bulb creates heat, not light. Alternatives to the incandescent bulb, on the other hand, can easily provide efficacy values that are two to 10 times higher.
I’m not trying to pick on the incandescent bulb. If efficacy were the only important quality of a light source, incandescent bulbs would have long since vanished. Other parameters such as lifetime, durability, and quality of light are important, depending on the type of lighting application.
For optimal efficiency, all lighting technologies can benefit from switch-mode power-supply (SMPS) systems. Furthermore, intelligent control can be applied to any lighting technology to minimize energy loss through active conservation. Therefore, intelligent embedded-control systems that include SMPS-control features are a necessary component for developing energy-efficient lighting applications.
“Dim the Lights, Please”
Incandescent bulbs have remained desirable for their high color-rendering index (CRI). The CRI of a light source is a measure of its ability to faithfully reproduce the colors of an object that’s illuminated by the source. A monochromatic light source would have a CRI of 0, since only one color can be reproduced. Incandescent bulbs have a CRI very near 100, the maximum possible value.
It’s obvious that we have grown accustomed to the warm, pleasing light that incandescent bulbs provide in our homes. Incandescent technology is also popular in retail lighting applications, where it increases the appeal of products on display.
Adding an embedded processor to a lighting application doesn’t have to be complex (Fig. 1). Simply dimming an incandescent bulb can save lots of energy. And your mother was right—you could even turn it off when you don’t need it! The circuit in Figure 1 uses a six-pin PIC10F200 MCU to control a triac circuit. The triac controls the light intensity by controlling the amount of conduction time in each half-cycle of the ac-input voltage. In effect, the triac performs a pulse-width-modulation (PWM) function on the incoming ac voltage. Waiting longer before turning on the triac at the start of the ac cycle reduces light intensity (Fig. 2).
Two I/O pins are required to control the triac. The MCU monitors a sample of the ac-line voltage on an input pin to obtain zero-crossing information. It can then use the zero-crossing information to implement a variable triac firing delay.
Some engineers might say they don’t need an MCU to control a triac, since they can do that with a simple RC-delay circuit. However, the MCU offers some advantages. A triac requires a certain amount of gate-bias current to enable current to flow. Also, a triac has a minimum holding-current specification. When the amount of current flowing through the triac exceeds the holding current, the gate bias can be removed and the triac will continue to conduct.
What this means is that the triac can be energized with just a short pulse on the gate when using a MCU. Therefore, the bias circuit will have a very low current when averaged over each ac cycle. This makes it possible to use smaller and less expensive bias-circuit components. The MCU will need a 5-V power supply as well, but it turns out that a very inexpensive resistor and Zener-diode circuit can be employed, since the MCU draws less than 500-µA average current. Two 11k, 1/8-W resistors are used in this example application to generate the MCU bias supply.
Now that you have an MCU in the circuit, you can add additional functions, including remote control, motion sensing, and timing-related functions. In addition, the dimming control can be made linear. Because the ac voltage has a sinusoidal profile, you won’t get a linear relationship between the triac firing delay and light intensity. This can be easily fixed using a lookup table to translate the requested lamp intensity into an appropriate firing angle.
If you want to control the circuit with a photocell or IR sensor, it’s best to power these devices using an I/O pin on the MCU. This way, the sensing devices can be enabled only when required to conserve power drawn from the 5-V bias circuit.
Digital Lighting
Fluorescent lamps offer a much more efficient source of light over incandescent bulbs. Although the quality of light isn’t as pleasing (due to its lower CRI), the efficacy of a fluorescent bulb is typically 10 times higher than that of an incandescent bulb. Fluorescent bulbs are most widely used in commercial applications, where energy costs must be kept low. However, they’re also finding increased use in residential applications as consumers become more interested in energy savings.
Fluorescent ballast designs have traditionally been based on magnetic (inductor) circuits, but they’re rapidly moving to electronic designs to increase system efficacy. In fact, legislation such as California’s Title 24 has been put in place to ensure that inefficient magnetic designs are gradually phased out of production.
At a minimum, the fluorescent ballast must regulate the bulb current. The resistance of the bulb varies widely, depending on the operational state. The fluorescent bulb consists of a glass tube filled with a small amount of mercury vapor and an inert gas. A tungsten filament is located at each end of the bulb.
Before the bulb is lit, the gas will have a very high resistance. To start the bulb, current is passed through the filaments (not through the gas) so they’re heated and begin to emit electrons. Then, a high voltage potential is applied across the two filaments to strike an arc in the gas mixture. Once the arc is struck, the resistance of the gas mixture drops significantly, due to an avalanche effect. The ballast must lower the voltage across the filaments to maintain the proper current flow through the mixture.
A resonant circuit is commonly used to control the bulb current for switch-mode ballast applications. An inductor and capacitor are placed in series with the bulb (Fig. 3). A second capacitor is placed across the filaments.
A square-wave, variable-frequency oscillator (VFO) drives the resonant circuit through a pair of power transistors connected to a dc bus. A dead-time generator provides complementary signals for the power transistors and ensures that shoot-through currents are eliminated.
The frequency of the VFO regulates the current flow through the lamp. To start the lamp, a high frequency is applied to the circuit. This causes current to flow through the filaments and the filament capacitor CF. The high-frequency operation heats the filaments so the lamp can be started.
After heating the filaments, the VFO’s frequency is changed to a lower frequency. The voltage across the filaments rises rapidly and strikes the arc. When the bulb is lit, the VFO’s frequency can be adjusted to obtain different bulb currents and light-output levels.
A rectifier and a filter-capacitor circuit are usually the first things you’ll find in an electronic ballast circuit to convert the incoming ac voltage into a dc bus for the resonant converter. Unfortunately, this causes the ballast to consume current only at the peaks of the incoming ac voltage. Power-factor correction (PFC) is required in an electronic ballast to increase efficiency and eliminate input-current harmonics.
In many ballast designs, separate ICs are used for the PFC, ballast-control, and external-control functions. However, a digital signal controller (DSC) can be used to implement a complete digital-ballast solution (Fig. 4). This circuit employs the dsPIC33F DSC because its 16-bit CPU has the calculation performance required to simultaneously perform PFC, control the resonant mode inverter, and respond to external control signals if required.
There are many ways to describe how a PFC circuit works, but basically the PFC circuit tries to make the input-current waveform follow the same sinusoidal profile as the input voltage. One of the most popular ways to implement PFC is with a voltage-boost circuit. The inductor current, and therefore the input current, can be controlled by the duty cycle that’s applied to the inductor switch. Ultimately, the rectified ac voltage is boosted to a higher value, usually around 400 V dc. So, the PFC circuit is a boost-voltage regulator. The voltage-regulation function can easily be performed by a digital control loop.
There is a special requirement for this voltage regulator. The regulator uses an inner-current control loop that controls the current profile in the inductor of the PFC circuit. The output of the voltage control loop provides a command to the current control loop, which sets the amount of input current. Before the current command is provided to the current control loop, it’s mixed with a sample of the rectified input voltage. This mixing forces the input current to have the same shape as the input voltage (Fig. 5).
The ballast application uses two PWM channels to implement a fully digital ballast solution. One PWM channel drives a half-bridge circuit connected to the lamp, while the other controls the PFC-boost circuit. The analog-to-digital converter (ADC) monitors two voltages and two currents. The dc-bus voltage, ac-input voltage, and input current are monitored for the PFC function. The lamp current is monitored to control lamp brightness and detect bulb failures.
Proportional-integral-derivative (PID) controllers are used in the PFC algorithm to regulate the bus voltage and input current. The behavior of each PID controller (and the PFC algorithm) can be modified by changing software coefficients. The DSC device has enough CPU bandwidth to execute these PID controllers. In particular, the inner-current control loop will be executed at the same frequency as the PWM signal applied to the boost-circuit switch. A frequency of 100 kHz or more is often used in the boost circuit to keep the inductor size small.
The circuit currents are measured using simple shunt resistors in series with the power switches. This lowers the circuit cost, but requires a little extra work to get the data. The voltage on the shunt resistor only indicates the circuit current at certain times in the PWM cycle. Therefore, the PWM timebase automatically triggers the ADC measurements to ensure that the shunt resistors are sampled at the correct time.
Whatever Color You Want
The world’s longest-lasting light bulb, now 105 years old, is installed in a firehouse in Livermore, Calif. Unfortunately, most modern-day light bulbs don’t last as long as this one. When lifetime, efficacy, and durability are important, power LEDs have the most to offer.
Present power LED technology provides efficacy values that rival fluorescent technology. In addition, the power-LED industry expects to double present efficacy values in the next few years. The lifetime of LEDs can exceed 50,000 hours, which is a huge benefit in commercial applications where the cost of changing the light bulb isn’t free.
Although LEDs have tremendous efficiency and lifetime advantages, the CRI of white LEDs can be very low. Many white LEDs have a bluish color; white light is produced by covering a blue or infrared emitter with yellow phosphorus to shift the light to the desired wavelength. The LED’s final color spectrum is limited by the emitter’s initial wavelength and by how the phosphorous spreads the light energy across the visible spectrum.
One emerging application for power LEDs is LCD video-display backlighting. Many current LCD-panel designs use fluorescent technology to provide the backlighting. LED technology improves LCD image quality by using separate red, green, and blue (RGB) emitters. The use of separate emitters allows for much greater control of the produced color spectrum over white LED technology. Within a range defined by the three component colors, any color can be generated.
RGB LEDs allow the LCD panel to produce a broader range of colors than a typical fluorescent design. In addition, the LEDs can be modulated on and off using the video scan information. LEDs have instant on and off times when compared with other lighting technologies. The scan modulation allows the LCD panel to produce a sharper image.
Figure 6 shows an RGB LED-control circuit that could be used for an LCD panel application or even a general-illumination application. A source of constant current must drive the LEDs. When multiple LEDs are employed, they’re typically connected in series so each LED receives the same amount of current. The choice of current-drive level will be a tradeoff between the amount of light produced, efficacy, and possibly thermal limitations.
Three MCP1630 switching devices are used to implement the constant-current drivers. These devices are MCU peripherals that contain the analog components necessary for an SMPS control loop. The MCU provides a clock signal to set the switching frequency and limit the maximum duty cycle. A buck or boost topology could be implemented in this circuit, depending on the available input voltage and the forward voltage across the string of LEDs.
The PIC18F1330 MCU was selected because it has three 14-bit PWM channels. These PWM channels are used to modulate the outputs of the three constant-current drivers and set the light intensity of each color. The high PWM resolution is required so there’s accurate color control over a wide range of brightness levels.
The wavelength and light intensity of LEDs can change with variations in manufacturing process, age, and drive-current level. In most backlighting applications, active color control is needed to ensure a consistent color, and brightness is produced. An RGB sensor is used to detect each component of the light output. The MCU or DSC calibrates the output of the sensor, determines the amount of color error, and calculates three PID routines that set the R, G, and B component levels.
Efficient lighting applications require power-circuit control and intelligence. You can integrate both of these functions with a MCU or DSC, decreasing circuit complexity and increasing flexibility.