This article was published in the July/August 2013 issue of LEDs Magazine.
View the Table of Contents and download the PDF file of the complete July/August 2013 issue, or view the E-zine version in your browser.
In many solid-state lighting (SSL) applications, such as architectural, area, and downlighting, color accuracy is very important. Moreover, an increasing number of products support dynamic color tuning — to set the white point or CCT and/or to enable dynamic full-color products. LEDs are ideal light sources for achieving a precise color. The color of LED lighting can be changed by mixing different colors of LEDs, such as red, blue, green, yellow, and white. When mixing LED colors, one or more strings of LEDs needs to be dimmed to realize the desired color mix. There are different methods for achieving LED dimming, so let's consider various dimming techniques used in LED color mixing.
A single LED die can only emit monochromatic light. To generate more colors, three primary color LEDs (red, green, and blue — RGB) can be used together for color mixing. Seven basic colors (red, green, blue, yellow, violet, aqua, and white) can be produced just by switching red, green, and blue LED channels. To produce more than seven colors, each LED channel must be able to change in brightness. Dimming an LED channel is achieved by adjusting the current through each LED string. Many colors can be produced by mixing three dimmable strings of RGB LEDs. To change the color temperature of white LED light, a commonly used method is to include a red LED string and vary the brightness relative to the white string to achieve the desired CCT.
Basically, there are two ways to achieve LED dimming: analog or linear current control and pulse-width modulation (PWM). Both dimming methods change the LEDs' brightness by controlling the average current through the LED string. Both can be implemented in either a switch-mode or linear LED driver. Fig. 1 shows a two-string LED driver containing one buck switcher and one linear regulator, based on the TPS92660. Both LED strings can be dimmed with either analog or PWM techniques. There are advantages and disadvantages to each. In most applications, the dimming method is chosen based on the color mixing performance requirement.
Analog dimming options
You implement analog dimming by adjusting the constant current level through the LED string. Analog dimming can be achieved by adjusting the LED current reference voltage inside the IC, or by adjusting the LED current sense voltage outside the IC. First, let's discuss dimming by changing the LED current reference voltage.
For most LED drivers, including switching regulators and linear regulators, the LED current is determined by the following equation:
Where VREF is the internal IC LED current reference voltage and RSNS is the current sense resistance.
The LED current can be adjusted by changing VREF in some cases. Note that not all LED driver ICs allow users to change the LED current reference voltages. For those ICs that allow the change of the LED current reference, there are generally two ways to do so. The first is to apply an analog voltage on the reference voltage adjust pin provided by the IC. One example is Texas Instruments' LM3409. The user can adjust the LED current by adjusting the voltage on the IADJ pin. The second way is to adjust the reference voltage through a digital communication interface like I2C. One example is Texas Instruments' TPS92660, which has an I2C interface that allows users to adjust LED current reference voltages through I2C commands.
Another popular way of doing analog dimming is to change the LED current sense voltage. In most applications, the current sense resistor is less than 1 Ohm. It is not practical to change the current sense resistor value by using a potentiometer. Instead, we change the voltage at the IC's CS (current sense) pin by injecting an external DC voltage. Fig. 2 shows a typical analog dimming circuit that works by changing the current sense voltage. The voltage at the CS pin is determined by the following equation:
At steady state, the CS pin voltage is equal to the reference voltage. The LED current can be changed by adjusting either the external DC voltage VADJ or the value of the variable resistor R2.
There is a disadvantage to using analog dimming in color mixing applications. The LED color temperature can change with the LED current. The LED's brightness and color may change during analog dimming, especially when the current change is significant. The system may not generate the desired color under these conditions.
Pulse-width modulation dimming actually turns the LEDs on and off at a fixed duty cycle and frequency. Assuming the switching or multiplexing is sufficiently fast — typically 200 Hz or greater — the human eye perceives the LEDs to be on continuously. The PWM-dimmed LED current is determined by the following equation:
Where IDIM is the dimmed LED current, D is the on duty cycle of the PWM dimming signal, and ILED is the constant current supplied to the switched LED string.
Many LED driver ICs have a PWM dimming input pin that accepts a PWM dimming input signal generated by a microcontroller. Typically, the driver IC turns off the MOSFET driver only when the PWM dimming signal is low. The MOSFET driver turns back on when the PWM dimming signal is high. The internal circuitry is fully operational during the PWM dimming off cycle. This prevents the IC from restarting, which causes a delay in the PWM dimming rising edge.
For switch-mode LED drivers, a capacitor is typically placed across the LED string to filter out the high-frequency switching noise. This capacitor can slow the rising and falling edges of the PWM-dimmed LED current. So for high-frequency, low-duty-cycle PWM dimming applications, this capacitor needs to be removed. Fig. 3 shows a LED buck regulator PWM dimming waveform without an output capacitor.
Some LED driver ICs provide a function called analog-to-PWM dimming. The IC's dimming pin accepts the analog signal and converts it to a PWM dimming signal. The PWM dimming frequency is fixed, and the PWM dimming duty cycle is proportional to an input analog signal level.
Analog-to-PWM dimming is very useful in those lighting applications where microcontrollers are not available. It can also be used to implement the thermal fold back function where the LED current is reduced by PWM dimming when the temperature of the LED board rises above a set point.
Shunt FET PWM dimming
Shunt FET (field effect transistor) PWM dimming often is used for very high-frequency LED PWM dimming. Fig. 4 shows a buck regulator-based shunt FET PWM dimming circuit. An external shunt FET is placed in parallel with the LED string to quickly bypass (short) the converter's output current. The LED string is shut off when the shunt FET is on and is turned on when the shunt FET is off. So the LED string is effectively PWM dimmed by this shunt FET. Some LED driver ICs integrate a MOSFET driver for shunt FET PWM dimming and eliminate the need for an external MOSFET driver.
The switching regulator's inductor current stays continuous during shunt FET PWM dimming. There is no delay caused by inductor current ramping up or down. With a strong MOSFET driver, the shunt FET can be turned on and off at very high speeds. Consequently, the PWM-dimmed LED current has very sharp rising and falling edges. The shunt FET PWM dimming is ideal for high-frequency PWM dimming applications.
Dimming is critical to achieving the desired color and brightness in LED color-mixing applications. There are many approaches to dimming LEDs. The two main categories are analog dimming and PWM dimming. Analog dimming usually can be achieved using a relatively simpler circuit. Analog dimming is generally lower in cost and good for a system where a microcontroller is not available. However, it may not be appropriate for applications that require a constant color temperature. PWM dimming, alternatively, can achieve very accurate color temperature by reducing the color change associated with the LED current level. PWM dimming usually requires an input digital signal generated by a microcontroller — and it increases system cost.