Increase LED driver efficiency without a sense resistor (MAGAZINE)
Seeking to improve driver efficiency for portable power applications, Chris Glaser and Jim Chen explain a voltage-regulation technique known as high-side sensing, which eliminates the lossy sense resistor component by relying on a MOSFET.
This article was published in the October 2013 issue of LEDs Magazine.
Visit the LEDs Magazine archive page and view the E–zine version in your browser. You can download a PDF of the magazine from within the browser E–zine.
Unlike most DC/DC converters that regulate the output voltage, an LED driver regulates current in the LED. The LED current is proportional to the light output, and the light output is the key concern for an LED driver. To regulate the LED current, traditional LED drivers have regulated the voltage drop across a discrete current sensing resistor. Since the entire LED current must be routed through this resistor to develop a measurable voltage, this sense resistor is an important loss term in these LED drivers. This article discusses a technique for regulating LED current without the need for this lossy sense resistor. This increases LED driver efficiency, which is critical for portable power applications in consumer and medical end equipment.
The light output of an LED is proportional to the current through the LED. So to achieve a controlled light output, the LED's current must be controlled. The user also typically wants to dynamically adjust light output during operation. Practically, this might mean adjusting the backlight's brightness on a smartphone or glucose meter in order to better read the display in bright light or at night. This dimming capability requires a tight control over the regulated current in the LED, in order to accept user inputs and make slight changes to the light output based on ambient lighting conditions.
Regulation using a sense resistor
Regulating the output current accurately presents a challenge for most DC/DC converters, which typically are configured to regulate the output voltage. Measuring the output current usually requires generating some voltage drop across a sensing element or measuring the magnetic field created by a flowing current with a Hall-effect sensor. Hall-effect sensors are bulky, expensive, and not suited for portable applications. Measuring the voltage across sense resistors creates an additional loss in the system; the resulting lower efficiency creates excess heat and a temperature rise, while reducing the battery's runtime.
Measuring the LED current through a sense resistor is frequently chosen because the approach is simple and straightforward. Since the current in the sense resistor is DC, accuracy can be high. However, there is a loss of efficiency with the sense resistor approach.
Fig. 1 shows a typical LED driver block diagram that regulates the current through a sense resistor. The voltage across the sense resistor is compared to a reference voltage, VREF, by an operational amplifier (op amp). The op amp drives a power stage that converts the input voltage, VIN, to an output current, IOUT.
MOSFET in the power stage
A more efficient approach is to measure the LED current in the power stage. Almost all power stages in LED drivers have at least one MOSFET between VIN and the LED to accomplish the necessary switching action for the given power conversion topology. Since a MOSFET is present, the same MOSFET (or a smaller, sensing MOSFET in parallel with it) may also be used to sense the current flow. Unlike the traditional approach with a discrete sense resistor carrying only a DC current, the current flow in the MOSFET typically is AC. So further processing and manipulation of the MOSFET's current signal occurs to generate an equivalent LED current. Fig. 1b shows this type of topology.
Known as high-side sensing, the MOSFET approach eliminates the sense resistor component and its losses from the system, while allowing the LED's cathode to be connected to ground. This is generally a preferred connection; the system ground is often a large plane and ideal for spreading heat created in the LED. This reduces the LED temperature, increasing its lifetime and increasing the efficiency of the system by eliminating one of the losses (the sense resistor). However, configuring the op amp and current sensor circuitry to handle the common-mode voltages that occur with high-side sensing is more difficult compared to the sense resistor method.
Comparing the options
To provide a fair comparison, the same LED and power stage should be used to compare the sense resistor and MOSFET in the power stage methods of current regulation in an LED driver. A device that supports both methods is the TPS61260 boost converter. Since the same device is used, the power stage is the same. It is configured to boost 1.2V, 1.8V, and 2.4V input voltages to drive a single LED dimmed from 10 mA to 100 mA. The LED's forward voltage is about 3.1V at 100 mA of current. The op amp and current sensor circuitry for the MOSFET in the power stage method of current regulation is integrated in the device, so this complexity does not appear to the product designer.
The efficiency is calculated by computing the power delivered to the LED itself, which is then divided by the average input power. By using the LED's forward voltage and current as the output power, the losses in the sense resistor are kept out of the output power equation, resulting in lower efficiency. This is an accurate calculation for the system efficiency; the power loss in the resistor is truly a loss because it does not produce light output. Using the LED driver's output voltage in the efficiency calculation is inaccurate; the output voltage includes both the LED forward voltage and voltage drop across the sense resistor.
Fig. 2 shows the efficiency difference of the two methods; Fig. 3 shows the resulting difference in input current drawn. The 0.5V that the TPS61260 regulates across the sense resistor is significant compared to 3.1V forward voltage of the LED, so the efficiency in the sense resistor configuration is significantly lower — by more than 10%. This results in a higher input current drawn for a given light output, which reduces the system's battery runtime.
Dimming accuracy for each current regulation method is compared in Fig. 4. In both graphs, pulse-width modulation dimming is used to reduce the LED current from its 100 mA full level. Both curves are highly linear; there is some offset in the slope to the ideal curve. Sensing the LED current through a MOSFET in the power stage provides the same dimming linearity as the sense resistor method. This allows the user to have the same control over adjusting the light output to changing ambient conditions.