The next big leap in lighting technology will be adding it to the “Internet of Things” – devices that can operate autonomously and, in the case of lighting, be controlled remotely by humans using smart phones, tablet computers, and PCs. To become part of the Internet of Things, every lamp must have its own IP address, a RF/MCU chip, and a lightweight protocol stack – in addition to the traditional power supply, lamp driver and passive components that populate lighting control circuits today.
Lighting – including LED-based lighting that will be the subject of this article – has some technology accommodations to make that are largely due to control system diversity in a competitive market. Besides smart phones and tablets, lamps must communicate with sensors, wall switches, and universal remotes that control TVs, network boxes and set-top boxes. These devices use a variety of network protocols including RF4CE, 6LoWPAN, and three different ZigBee® application profiles – Home Automation, Light Link, and Green Power. There are also a number of proprietary protocols for home automation that could be included in a universal wireless lighting control system.
Viewed from protocol perspective, ZigBee is a well-established standard with carefully developed application-layer profiles that can reduce development time. ZigBee’s profiles help ensure that radios from different vendors use standard commands and provide standard commissioning and security capabilities. Smart grid/smart meter applications are likely to use IEEE Std. 802.15.4 for energy management, actuated by public utilities acting in accordance with agreements with individual consumers. ZigBee’s willingness to accommodate the preferences of utilities in its smart energy application profile gives it an advantage in this use scenario.
There are, however, intellectual property costs including ZigBee Alliance membership and certification fees, along with the potential for third-party licensing royalties.
6LoWPAN is IPv6-based, which greatly simplifies software development and maintenance for designers already familiar with IP interfaces. 6LoWPAN is basically a header compression scheme for IPv6 packets defined by the Internet Engineering Task Force (IETF) and promoted by the Internet Protocol for Smart Objects (IPSO) Alliance. When designing network functionality into a light bulb, a significant engineering decision point for designers is that the smaller packets used by 6LoWPAN require small memory and memory footprint. Memory size influences cost and energy consumption, but in this case the footprint takes on additional significance because the control module is embedded in the base of a bulb.
The RF4CE specification has been developed in partnership with ZigBee. It is a lighter protocol, which means it requires smaller memory configurations for lower cost devices, such as remote control of consumer electronics.
Unlike the numerous options available for protocol stacks, IEEE Std. 802.15.4 appears to be the favored physical (PHY) and media access (MAC) layer technology for all of the protocols previously mentioned.
The most likely lighting-specific solution for melding the segmented state of the networking world is a gateway device for IP signaling (smart phones and tablets) and a direct link to the LED lamp for the 802.15.4/ZigBee/RF4CE/6LoWPAN communications (see Figure 1).
Figure 1: A TCP/IP gateway expands the functionality of the system to smartphones and tablet PCs (Courtesy of NXP Semiconductors).
Design considerations overview
Designing a robust lamp control system with communications capability that can be successful in the competitive lighting market depends on several high-level considerations. The system must be small enough to fit inside the lamp housing, produce very little heat (due to the heat sensitivity of LEDs), be energy efficient (particularly when the lamp is not switched on), and be inexpensive enough to make integration cost feasible at a fairly low price point.
To this end, the RF transceiver and MCU should be integrated on the same chip and offer a number of features such as software upgrade ability and security. Specific features include:
- Low operating power (ideally approximately 20 mA Tx and Rx)
- Low sleep current (ideally approximately 0.1 μA deep sleep with I/O-wake-up)
- 128-bit AES encryption security embedded in the processor
- Extensive set of peripherals for design flexibility, including I2C, SPI, UART, ADC, DAC, PWM, timers, and GPIO
- On-chip temperature sensor
Several radio/MCU chips could be used for this application including NXP Semiconductor’s JN5148, which is designed for ZigBee software stacks. NXP also offers the JN5142 for its 6LoWPAN software stack (JenNet-IP). Texas Instruments’ CC2531 can be used with both ZigBee and RF4CE. Microchip Technology’s MRF24J40 and Freescale Semiconductors’ MC13213 are also possibilities.
Adding wireless control also introduces other issues that lighting system designers may be unfamiliar with, including antenna design and EMI considerations. To avoid interference, EMI must be kept to a minimum, which affects the selection both of the systems’ other main components – the voltage regulator and the lamp driver, which are both EMI sources.
In addition to the radio/MCU chip, the other two major components in the design are the lamp driver and the AC/DC converter. A block diagram of a typical remote-controlled smart lighting system is shown in Figure 2.
Figure 2: A smart lighting solution for LED lamps must fit in a restricted area inside the lamp (Courtesy of NXP Semiconductors).
Under normal operating conditions, the converter (smart supply) powers the logic circuits of the other two chips (nominally three volts for the MCU/RF and 15 volts for the lamp driver). An additional function is to provide standby power for the lamp driver when the lamp is off, and in its standby power role, the converter has much to contribute to the overall energy conservation goals. A five watt low-power SMPS converter with ultra-low standby power and a number of integrated functions is the preferred choice for a typical LED lamp application of less than 11 watts.
Primary characteristics to look for include: Low EMI (because the RF chip is nearby), no-load power consumption of less than 15 mW, universal mains compatibility, an integrated 700 volt MOSFET, high efficiency over its operating range, primary sensing to eliminate an opto-coupler, and several common voltage and temperature protection circuits. Because of the constrained design footprint, look for a converter/standby power supply with start-up directly from the rectified mains voltage (i.e., without any external bleeder circuits). LEDs are very sensitive to current instability so the converter – which operates as a regulated voltage source – should be able to deliver maximum current over a broad output voltage range.
A few parts fit the bill including NXP Semiconductors’ TEA1721 and ON Semiconductor’s NCL30000.
Although there is plenty of competition between the individual chips that can be integrated into the remote control system, few semiconductor companies excel in all three critical areas – ultra-low-power wireless chips, AC/DC converters and lamp driver technologies. Moreover, integrating remote control functionality on PC boards, small enough to fit in an LED lamp, is not trivial. A few reference designs are becoming available for lighting system designers and, among the earliest, targets a remote-controlled, non-isolated 11 watt LED driver using NXP Semiconductor parts. The design consists of two small PC boards: the MCU/RF IC and the controller are integrated onto a SSB, which is approximately 28 mm x 16 mm and the lamp driver is integrated onto the motherboard, which is approximately 49 mm x 20 mm.
The AN11126 reference design is available here.
NXP Semiconductors’ TEA1721AT and JN5148, used in the reference design, fit the general specifications described in this article. So does their SSL21082 lamp driver. A block diagram is shown in Figure 3.
Figure 3: The SSL21082 lamp driver maintains LED current to an accuracy of within five percent (Courtesy of NXP Semiconductors).
The SSL21082 offers the small footprint, low cost and high efficiency (up to 95 percent) required for a remote-controlled lamp application. It starts directly from the mains supply through an internal high-voltage current source. Thereafter, a dV/dT supply is used with capacitive coupling from the drain or an auxiliary supply. This dual supply scheme provides flexibility in the application design. The SSL21082 consumes 1.3 mA of supply current with an internal clamp limiting the supply voltage.
The lamp driver’s output current maintains LED current accuracy within five percent, and its numerous protection features include LED temperature feedback that is easy to implement and a unique “valley detect” function that increases efficiency.
Powering the SSL21082 has three operating modes:
- Under normal operation, the voltage swing on the dV/dT pin is rectified by the IC, which provides current for the VCC pin.
- At start-up, an internal current source is connected to the HV pin. The current source provides internal power until either the dV/dT supply, or an external current on the VCC pin provides the supply.
- An external voltage source can power the VCC pin.
The IC starts up when the voltage at the VCC
pin is higher than a predetermined startup value. It locks out when the voltage at the VCC
pin is lower than another predetermined value (stop value). The hysteresis between the start and stop levels allows the IC to be supplied by a buffer capacitor until the dV/dT supply is settled. The SSL21082 has an internal VCC
clamp, which is an internal active Zener that limits the voltage on the supply VCC
pin to the maximum value. If the maximum current of the dV/dT supply minus the current consumption of the IC (determined by the load on the gate drivers), is lower than the maximum value of the supply current, no external Zener diode is needed in the dV/dT supply circuit.
The dV/dT pin is connected to an internal single-sided rectification stage. When an alternating voltage with sufficient amplitude is supplied to the pin, the IC can be powered without any other external power connection. This solution provides an effective method to prevent the additional high-power losses, which would result if a regulator were used for continuously powering the IC. Unlike an auxiliary supply, additional inductor windings are not needed.
During supply dips, the input voltage can drop too low to supply the required IC current through the dV/dT pin. Under these conditions, if the VCC
voltage drops lower than the regulator switch-on supply voltage level, another regulator with a current capability up to the value of high supply current on pin HV is started. The job of the regulator is to fill in the required supply current, which the dV/dT supply does not deliver, thus preventing the IC going into an under voltage lockout state. When the VCC
voltage is higher than regulator switch-on supply voltage level, the regulator is turned off.
The converter in the SSL21082 is a Boundary Conduction Mode (BCM), peak-current controlled system that operates at the boundary between continuous and discontinuous mode. Energy is stored in an inductor during each period that the switch is on.
The basics of BCM operation are shown in Figure 4.
Figure 4: The SSL21082’s converter operates at the boundary between continuous and discontinuous mode. Energy is stored in an inductor during each period that the switch is on (Courtesy of NXP Semiconductors).
When the internal MOSFET is switching, the stored energy in the inductor modulates the current through the LEDs in the lamp. During the primary stage, the current through the inductor is sensed by a resistor and, when the limiting voltage is reached, the internal MOSFET is switched off.
When the internal MOSFET switch is on, the amplitude of the current build-up in the inductor is proportional to VIN
. When the MOSFET switch is switched off, the current continues to flow through a freewheel diode and an output capacitor. The current then falls at a rate proportional to the value of VOUT
. The LED current is almost equal to half the peak switch current. A new cycle is started as soon as the inductor current is zero.
At the beginning of each new cycle, the switch has a substantial voltage over it due to capacitance on the drain arising from the state of other components. As a result, the switch heats up and the efficiency decreases. To overcome this, the valley detect feature has been integrated that is unique for NXP Semiconductors’ converters. The valley detect circuitry senses when the voltage on the drain of the switch has reached its lowest value and this is used to trigger the next cycle resulting in a significant reduction in switching losses.
In this design, the conversion frequency must remain below 200 KHz so the inductance must be chosen to maintain that frequency under the operating conditions dictated by the supply voltage, LED voltage and component spread.
Many lighting designers are unfamiliar with antenna selection and implementation, but a few guidelines can make the job easier. The antenna is typically connected directly to the mains supply. In most remote controlled LED designs, the antenna must be located beyond the outer edge of the socket of the LED lamp because the socket is fully enclosed by metal and acts as a shield preventing RF radiation from being effectively transmitted. It is important be make sure that the antenna cannot be tampered with or touched from the outside of the housing because it is powered by the mains. Enclosing it in a plastic cover is a common and effective solution.
Ideally, the antenna signal is guided through a 50 ohm coaxial cable from the small signal board (SSB) solder pads to an external antenna. The solder connections to the coax cable must be kept as short as possible because each millimeter of wire adds one nanohenry of inductance which is already significant in the 2.4 GHz band in which 802.15.4 operates.
The first – and most cost effective – option is a small PCB that contains either an SMD chip antenna or strip line antenna. An antenna with a good tolerance to surrounding metal should be chosen and the system design must conform to the recommended clearance. While it is always a good idea to consult the antenna data sheet, the clearance is typically 8 mm to 10 mm. Care should be taken to keep the coaxial cable tails as short as possible to maintain the 50 ohm impedance of the system.
PCB antennas work well in many cases, but when they do not provide the performance the system requires, a second option is the wire loop antenna. This solution exhibits better RF performance and a well-defined radiation pattern, but may cause shadow effects if the LEDs cover the complete metal front plate of the lamp. Because the loop antenna is not in free space and is affected by the dielectric properties of the plastics that cover the bulb, it is important to tune the antenna’s diameter in order to obtain optimal antenna performance at 2.4 GHz.
Wireless remote control of lighting is more likely to achieve market success if each lamp can be controlled individually by sensors and by IP-based devices such as smartphones, tablet computers and other devices that give users control. Components are available to create these systems with the performance criteria driven by factors such as ultra-low power operation, small footprint, low EMI, and low cost. Reference designs are also becoming available that make what could be a steep learning curve for a lighting engineer into a much easier task.