The choice between analog or digital drivers depends on the application requirements. Whichever route you choose, you've just passed the first of many decision points.
No longer relegated to simple indicator lights on consoles or Christmas decorations, LEDs are completely transforming the lighting industry. Their high efficiency, low implementation cost, and long operating life continue to supplant existing lighting technologies. In addition, the silicon and design tools available for LED development make it possible to take advantage of these devices in applications across every industry.
The challenge for engineers new to this technology is that LEDs are driven by analog systems that can be quite complex. Systems can be based on current or voltage, and advanced compensation is required to maintain consistency and efficiency across loads, color changes, and string lengths. Further complicating design is the fact that LEDs from different manufacturing runs can vary significantly enough to require calibration for some applications.
There are also many approaches for implementing LED technology, varying from very simple analog driver circuitry for light bulb replacement, to advanced digital MCU-based systems for stage or street lighting, requiring adaptive compensation and remote control through a wired or wireless communications link. Which implementation is optimal, however, depends upon many factors. This article will look at some of the more important elements of an LED-based design and their impact on cost, power efficiency, and design simplicity.
Analog versus digital
A basic LED system takes an AC or DC input and converts it to an appropriate DC voltage or current to drive the LED (see Figure 1). One of the most common applications utilizing a pure analog topology is light bulb replacement, where the LED system needs to be as simple as the product it is replacing. For such an application, advanced features such as a communications link or internal intelligence are unnecessary, making an analog approach ideal.
Within a digital system (see Figure 2), each LED string is driven by an MCU using pulse width modulation (PWM). If the MCU can source enough power, it can drive LEDs directly. For systems driving high brightness LEDS, or strings of LEDs, the PWM drives a MOSFET that performs the actual power switching.
Figure 1: An analog implementation of an LED system. (Source: Texas Instruments. Used with permission.)
One issue developers may face is a lack of familiarity with LED technology. For example, fixture manufacturers moving from incandescent to solid state lighting do not necessarily have much experience with electronics in general. The prospect of struggling to work out digital power conversion algorithms and power supply design, especially without in-house power expertise, makes a simple analog approach appealing.
Figure 2: A digital implementation of an LED system. (Source: Texas Instruments. Used with permission.)
Another advantage of analog systems is the ability to integrate the higher voltage components. For example, it is possible to reduce the overall number of MOSFETs required in a system compared to a digital implementation. Simple feedback is also less complex with an analog-based approach. A digital system requires an ADC to measure current or voltage to be used in a software-based compensation algorithm. For an analog system, the feedback loop consists of an error amplifier that adjusts the operating voltage across the LED to match an internal voltage reference.
However, while the simplest approach is often the more robust from an engineering perspective, it is not always the most attractive to the marketplace, or the most cost-effective. A digital approach to driving LEDs provides a more flexible topology compared to analog systems. For example, a PWM configured through software can reliably drive a wide range of voltages and currents. This is especially useful for applications that may need to support a variety of LED types to match the brightness requirement of each customer. Reconfiguring the PWM and compensation loop, for example, allows developers to replace a 350 mA string with a 700 mA string without having to change the base system design. If the system has a communications link and supports remote upgrading, string replacement can take place in systems already deployed in the field.
The primary advantage of taking a digital approach is that an MCU can perform much more than just the power management function of driving LEDs. Advanced features such as power factor correction (PFC), temperature sensing, and communications tend to be significantly more cost-effective and easier to implement compared to an analog approach. The more of these features a system requires, the more likely a digital approach will be the best option. In addition, once the shift to an MCU is made, all of the other advanced digital features become available, allowing better differentiation from competitive products.
- Color accuracy: Certain lighting applications, such as stage lighting, require accurate color mixing. Note that color accuracy also applies to white LEDs when the ability to maintain a specific or consistent shade/intensity is required. Compensation is required for fixed offsets (e.g., LEDs from different manufacturing bins may perform differently and will need to be calibrated) and for dynamic changes (e.g., color shifts that occur with varying temperature).
- Color mixing: Some applications need the ability to blend LEDs to create varying shades of color, such as a light that washes a wall with a shifting rainbow of colors. Dynamic coordination and color accuracy between LEDs can potentially become important factors.
- Dimming: Triac dimming can be difficult to implement with analog components. In contrast, an MCU can detect the leading and trailing edge of dimmers to greatly simplify design.
- Multiple strings: As the length and number of strings increases, the overall power requirements of the system increase. With an analog approach, each pair of strings requires a transformer, as well as individual rectifiers (see Figure 3). As the number of strings increases, so does the component count. At some point, it becomes advantageous to shift to a digital topology, given the fewer number of components required.
- Sensors: An LED application may need to sense its environment for a number of reasons. Measuring the temperature around an LED, for example, allows for advanced color correction. A motion sensor enables intelligent control. An ambient light sensor allows the system to turn itself off when there is no need for it to be on. In each of these cases, an MCU is required to monitor the sensor, process the data, and decide how to act.
Figure 3: With an analog approach, each pair of strings requires a transformer, as well as individual rectifiers. (Source: Texas Instruments. Used with permission.)
Many systems will require an MCU to handle system management tasks, sensing, and perhaps a communications interface. The key question to ask is whether the MCU can also efficiently support drive functionality. Implementing digital LED control may be as simple as specifying a higher performance device within the same MCU family. It is also worth confirming that the MCU roadmap has devices with enough performance to introduce advanced LED functionality in next-generation designs.
If, however, the MCU is a low-end device being used solely for simple system housekeeping, it likely won't have the real-time performance to drive LEDs. In such a case, a completely different MCU family may be required to handle driving the LEDs. To reduce cost in this system, the functionality handled by the low-end MCU will then need to be migrated to the higher performance MCU, effectively requiring a complete redesign of the system. Developers may find it to be more cost-effective — and faster to market — to implement the LED subsystem separately from the system MCU, using an analog-based topology.
Analog LED systems are typically tuned to provide optimal efficiency at a specific load. When engineers begin to design an LED system using an analog approach, they have the freedom to choose what this load will be. For the highest efficiency, the load selected should be the load at which the system will operate the majority of the time. For lights, this is often the highest brightness. One disadvantage of this approach is that when the light operates at a different load, such as when the light is dimmed, efficiency diminishes as well. Another disadvantage is that if an application needs a configuration of strings with a different number of LEDs, the optimal load value changes as well, requiring a redesign of the analog circuitry. With a digital approach, a more complex driving algorithm can be used to dynamically adjust the drive power to the current load. This allows the system to operate optimally — not just at the optimal load, but across the entire operating load range.
Some efficiency techniques are possible with an MCU that are generally not available in analog component form. An example of such a technology is buck PFC. For a variety of reasons, power for a traditional industrial lighting application goes through a two-stage conversion process. First, the 120 VAC input is boosted to 400 VDC or higher. Since LED strings operate at a lower voltage, such as 40 V, the voltage then needs to be bucked down.
Unfortunately, raising and lowering the voltage results in unnecessary losses and therefore reduced efficiency. Today's MCUs can bypass these inefficiencies through buck PFC techniques that convert the input AC voltage directly to the required DC voltage. Not only does this eliminate the DC/DC conversion stage, it yields higher efficiency and uses smaller capacitors.
Power factor correction
PFC is an advanced technology that is being required in more and more applications. Power, by nature, is resistive. The circuitry to drive LEDs utilizes elements such as inductors and capacitors that change the profile of power supplied to the circuit. Effectively, the transmission line has to be sized for a larger current than is actually being used within the circuit. PFC makes LEDs and their circuitry appear like a resistor, similar to how incandescent lights appear to the transmission line, thus reducing the impact on transmission line size.
Useable power is something that the utility companies care very much about. For example, if equipment with a PFC of 1.0 is connected, 100 percent of the plant's power is available for use. With a PFC of 0.5, however, only half of the plant's power is available as useable power. In other terms, without PFC technology implemented in electronic devices, utility companies will have to build bigger plants to provide sufficient useable power.
Whether a system needs PFC or not can be a hard question to answer because PFC doesn't necessarily bring direct benefits to users of LED-based equipment. This can make it difficult to justify the expense of adding PFC. To aid in its adoption, many countries are mandating the use of PFC. In the US, low wattage residential lighting does not have PFC requirements. For commercial lighting, PFC is generally required.
Depending upon the application, developers may find it useful to integrate a communications link into an LED-based lighting system. The link can be wired or wireless, and serves as the foundation for intelligent and remote control of the system. Note that to support a communications link, the system will need an MCU to transmit/receive data and act upon it. This MCU may or may not be used to drive the LEDs as well.
For equipment such as industrial lights or home automation lighting that are plugged into an outlet, powerline communications (PLC) technology provides a robust link leveraging the existing power infrastructure of the home or factory building. For a connection that is suitable for portable devices that run off of a battery or cannot support PLC, a wireless technology like ZigBee can provide a cost- and power-effective implementation. If the LED system is part of a larger system that already has a network connection which can be leveraged, communication can take over an interprocessor interface such as I2C or SPI. A system could also have a manual communications link like USB. In such a case, the system could store performance data for periodic manual downloading by a technician performing scheduled maintenance.
Certain applications may also need to support a protocol layer. For example, DALI and DMX512 enable developers to build lighting equipment that is compatible with existing control systems. Other systems may collect performance statistics that are useful for a variety of reasons. For example, a system could store statistics on each string of lights it controls. When a light burns out, the system then triggers a maintenance alert. A more advanced system could track the operating temperature of each light and how long it has been lit to predict when the light will fail, so that preemptive maintenance can be performed.
Self-monitoring capabilities are becoming increasingly important for a wide range of commercial and industrial applications. For example, cities pay a set fee to operate street lights, regardless of how much power the lights actually consume. If each street light has its own meter, the city can pay based on measured usage, resulting in substantial operating savings. Over time, these savings would more than offset the added cost of the communications link and management infrastructure.
Whether you decide to go analog or digital, you'll have a myriad of component choices. In the past, each LED string had its own controller. Today, analog LED drivers and digital controllers provide a more cost-effective approach through their ability to drive multiple strings with a single controller.
The variety of analog LED drivers is quite extensive, offering a range of capabilities depending upon the application. For example, the TLC5960 LED driver from TI can drive up to eight channels and uses intelligent headroom voltage control to optimize the forward voltage of each string, thereby allowing developers to accommodate the subtle differences between LED strings due to binning. The TLC5960 also has inputs to measure the current/voltage through each string and provide feedback to dynamically adjust the output of the DC/DC converter to maintain a constant current through strings.
For digital designs, TI offers the TMS320F2806x for advanced LED applications. The MCU is part of the company's Piccolo family and offers integrated floating point processing to support features such as PFC and PLC.
Cypress Semiconductor provides a hybrid approach to LED design with its PSoC family of microcontrollers. Reprogrammable analog and digital blocks within the PSoC architecture allow developers to implement drive functionality in hardware to maximize performance and resolution.
Developers also have access to evaluation kits to accelerate product design and development. Development kits provide engineers new to LED technology with a fast way to evaluate topologies and components. The Piccolo Multi-DC-DC Color LED Kit from TI, for example, introduces developers to designing with color LEDs (see Figure 4). Alternatively, TI's UCC28810-EVM003 drives multiple LED strings and shows developers how to efficiently scale the magnetics of a design. Rather than having multiple DC/DC converters that need to be regulated individually to achieve consistent lighting, the kit uses transformers with secondary windings to reduce component count. In addition, developers new to LED design can work with powerful design tools like TI's controlSUITE. Developers can quickly configure multiple strings of LEDs, adjust control loop frequency, and tune performance without having to write a single line of code.
Figure 4: The Piccolo Multi-DC-DC Color LED Kit from TI introduces developers to designing with color LEDs. (Source: Texas Instruments. Used with permission.)
It may be worth taking the time to explore several of the available options before making a choice. Microchip, for example, offers a variety of demonstration boards based on different topologies — including driving LEDs with a charge pump, boost regulator, SEPIC regulator, or PWM controller — which highlight some of the different ways a system can be built (see Figure 5). Developers can also rely upon reference designs offered by vendors. While these designs are targeted for specific applications, they can be useful learning guides, as each reference design will have some characteristic for which it has been optimized, such as efficiency or intelligent control.
Advances in LED technology will continue to challenge our preconceived notions of how we can use them and in what applications, as well as how we design with them. By adopting new design topologies, manufacturers can take advantage of leading LED technology to design lighting systems that are both more efficient and cost-effective.
Figure 5: The MCP1630 Boost Mode LED Driver from Microchip demonstrates how to design an LED system using a PWM controller. (Source: Microchip. Used with permission.)