LED Journal
Online Buyers Guide

dayLED Dimming Techniques Enhance Video Display Image Quality
By Michael Day, Power Management Application Supervisor • Texas Instruments

Video billboards, jumbo-trons and the like require a multitude of smaller subsystems including power supplies, video encoders and decoders, line drivers, digital signal processors (DSPs), etc., all working together in complete harmony to provide a video image. For the audience, the end result is a very large full-color video with stunning high definition image quality. While the stadium vibrates from the crowd’s audible response to a video replay, the fans are completely unaware of the functions performed by the lowly LED driver in this system. If you closely examine the video screen, you see hundreds of individual video panels. Looking closer, you find that each panel contains 16 rows and columns of pixels. Each pixel is made up of three LEDs: one each of red, green and blue. Each LEDs’ anode is electrically connected to the output of an LED driver. Ultimately, these tens of thousands of LED drivers control the forward current through hundreds of thousands of LEDs to generate the video image.

The challenge for the electrical designer is to understand the best way to use the LED drivers to control the LED currents. Most high-end LED drivers provide the designer with several methods for controlling the LED currents in their system. These drivers include several features that assist with dimming LED brightness such as dot correction (DC), pulse width modulation (PWM) dimming and global brightness control (BC). Although these features provide the same basic function; dimming LED brightness, they are intended to be used differently. Understanding how to use these features properly is the key to providing the best quality video.

fig1Analog versus PWM Dimming

Before understanding the advanced features of an LED driver, one must first understand the two methods for reducing LED brightness: analog dimming and pulse width modulation (PWM) dimming. Analog dimming refers to changing the forward current through an LED. Driving the LED with half the current reduces brightness by a factor of two. Although not an exact one-to-one ratio, LED luminosity is fairly linear with forward current. 

PWM dimming refers to turning the current on and off to change the pulse width of current, not the amplitude. LED brightness is controlled by the percentage of time the LED is turned on, or duty cycle. The human eye’s frequency response is limited to approximately 60 Hz and lower. Therefore, turning LEDs on and off at a rate faster than 60 Hz causes the human eye to see only the average LED brightness. Most displays implement PWM dimming at 100 Hz or higher. For a fixed PWM dimming frequency, a lower duty cycle results in a lower LED brightness. Figure 1 shows a comparison between analog and PWM dimming for a 20 mA LED being dimmed to 25 percent brightness. Both LEDs have identical brightness.

Dot Correction

LED manufacturing processing tolerances are difficult to control. The two major LED characteristics important for high-quality displays are luminous intensity and spectral purity. Unfortunately, these characteristics may vary significantly from LED to LED, even when these LEDs come from the same manufacturing lot. In an ideal situation, when the processor tells the LED drivers to turn on all red LEDs with the same current, they turn on with the same brightness. In reality, LEDs with the same current have significant differences in brightness. A typical orderable LED part number may exhibit 2:1 variations in brightness with the same forward current.

Without compensating for this variation, an LED panel has very poor pixel-to-pixel brightness matching. Dot correction (DC) offers a well suited solution to compensate for these variations in luminous intensity. The term DC refers to correcting, or normalizing, the intensity of each pixel, or dot, in an LED display. This is accomplished by adjusting the analog current through each individual LED. The initial DC must take place in the factory after a panel’s production. The test fixture measures LED intensity and generates DC coefficients that are programmed into the IC to adjust LED brightness.

fig2A typical LED driver like the TLC5946 sets the maximum LED current for each output with a single resistor, Rset. Dot correction uses analog dimming to adjust the current to a percentage of the maximum current. The greater the number of DC correction bits, the more closely one can match the final LED brightness. A 6-bit DC provides 64 discrete current levels from 0 mA to the maximum current defined by the Rset resistor. This provides 1.59 percent steps of forward current. A 7-bit DC provides 0.79 percent steps of forward current. Figure 2 shows a production panel’s pre- and post-dot corrected brightness levels.

Before dot correction, all LEDs have the same forward current, which results in a 2:1 variation in brightness. After implementing dot correction, the LED driver drives each LED with a different current, resulting in uniform brightness of all LEDs. It is important to note that DC must be designed to reduce the brightness levels to a value lower than the dimmest LED in the panel. In a production environment where many panels are combined to make a larger display, the brightness level of all LEDs must be set at or below the dimmest LED in the system. The LED’s datasheet should provide the minimum specified brightness range.

Dot correction can also be used to dim an entire display’s brightness. While full brightness may be appropriate in an outdoor setting at high noon on a sunny day, the same brightness may be overwhelming for night time or indoor viewing. Overall panel dimming can be accomplished by reducing all DC values to a percentage of their previous value. A system that uses an IC with DC, but does not take advantage of the DC feature can achieve 50 percent brightness by programming all DC values to half of their maximum value. With a 7-bit dot corrected LED driver, 50 percent brightness is achieved by reducing all DC values from 127 to 63. The following equation sets an LED’s forward current when DC is used:

I_LEDx=Imax*DCx/2^n

Where:

  • Imax is the maximum output current set by Rset
  • DCx is the dot correction value for output x
  • n is the number of bits of dot correction

Using the example above, with Imax equal to 40 mA, DC equal to 63, all LED currents are set to 40 mA * 63 / 127 = 19.84 mA.

If the system already uses the DC feature to achieve uniform panel brightness, DC can still be used for dimming. Fifty percent dimming is achieved by programming the DC value to half of the original value. As an example using a 7-bit DC LED driver, if an LED requires a DC value of 108 to set its current to the correct value for uniform panel brightness, it can be dimmed to 50 percent brightness by programming the new DC value to 54.

The disadvantage of using DC to dim a dot-corrected panel is that it reduces the effective number of available bits for brightness matching.

fig3Brightness Control

To maintain tight brightness matching while still allowing full panel dimming, high-end LED drivers include separate DC from overall panel dimming by providing a separate set of registers for brightness control (BC). Like DC, BC is also implemented through analog dimming. The difference between the two is that DC dims each LED independently, while BC dims many LEDs at the same time. The number and configurations of the DC and BC registers are IC dependent.

As an example, the TLC5951, a 24-channel LED driver designed to drive three separate banks of eight LEDs, contains 24 DC registers (one for each LED). It contains three BC registers, one for each bank of LEDs. This contrasts to the TLC5943, a 16 output LED driver designed to drive 16 LEDs of the same color, which contains 16 internal DC registers. It contains a single BC register to dim all 16 outputs simultaneously. Separating the DC and BC functions allows precise control over LED brightness matching while still allowing for full panel dimming. The LED current in an IC containing both DC and BC is defined by a combination of the two register values:

I_LEDx=Imax*(DCx/2^n)*(BC/2^m)

Where:

  • Imax is the maximum output current set by Rset
  • DCx is the dot correction value for output x
  • n is the number of bits of dot correction
  • BC is the global brightness value
  • m is the number of bits of brightness correction

fig4PWM Dimming

Precise brightness control is only one component required for high-quality video. The second component is accurate color matching. Previous generation displays used analog dimming to set LED brightness levels for color mixing, which negatively affects color quality. Figure 3 shows that an LED’s color shifts with changes in forward current. Figure 3’s “true” green LED defines its color at full-brightness, or 20 mA, for this specific LED. Using analog dimming to achieve 25 percent brightness requires 5 mA of forward current. This shifts the color spectrum from 525 nm to 531 nm, which may become unacceptable in displays requiring a true-color representation.

Pulse width modulation (PWM) dimming, or grayscale PWM dimming, eliminates the color shift associated with dimming LEDs. This type of dimming maintains accurate LED color at reduced brightness levels which is critical for rendering high-quality video. PWM dimming eliminates LED color shift by keeping a constant current through the LED regardless of the LED’s brightness level. Each pixel in a color display is made from three LEDs: red, green and blue. By simultaneously pulsing and mixing the red, green and blue LEDs, the pixel is capable of producing up to 68.7 billion colors. The following example illustrates PWM dimming.

For simplicity, the example assumes only three bits of PWM dimming. Three bits is equivalent to 23 = 8 shades, so each LED can be programmed to stay “on” anywhere from zero to seven PWM grayscale steps. Each video frame starts with all LEDs turned “off”. At the rising edge of the first PWM clock, all LEDs turn “on” except ones that are programmed with grayscale values of zero. The IC increments a grayscale counter at the beginning of each PWM clock-cycle. Each LED stays on until the PWM grayscale counter goes above the LED’s programmed PWM value.

fig5This process is illustrated in Figure 4, which shows waveforms and a block diagram (Figure 5) for a simplified 3-bit PWM dimming controller. Programming the grayscale value of the red, green and blue LEDs to 7, 4, and 1, respectively, produces an orange pixel on the screen. The green LED, which is programmed to a grayscale value of four, turns on at the rising-edge of the first PWM clock cycle and stays on for four full PWM clock cycles. This 3-bit PWM dimming example is capable of producing 23 * 23 * 23 = 512 colors for each pixel. Expanding this math to a 16-bit LED driver, like the TLC5943, results in 216 * 216 * 216 = 281 trillion colors.

When dot correction, brightness control and PWM dimming are properly used together, the image on the big screen is flawless. As a result of this thoughtful design, the crowd reacts to the replay on the big screen, unaware of the care and effort taken to properly match pixel-to-pixel brightness, match the display’s brightness to the ambient light conditions, and mix colors to create the perfect image.

References


Michael Day, Power Management Application Supervisor for Texas Instruments’ Power organization has 16 years design experience in the field of power conversion. Currently, Michael manages the DC/DC Power Applications group at TI. He received his BSEE and MSEE in Pulsed Power from Texas Tech University, Lubbock, Texas. Michael is a member of IEEE and has published more than 60 articles on power, portable power and lighting topics. You can reach Michael at ti_michaelday@list.ti.com.

[Get Copyright Permissions] Click here for reuse options!
Copyright 2010 Webcom Communications Corp.
Webcom Home Directories/CD-ROMs Advertising Subscriptions E-Newsletters Webcom Products Mailing Lists Webcom Events Contact Us Webcom's Publications About Webcom Webcom Home
© 2011 Webcom Communications Corp.