Creating uniform LED intensity in RGB billboard and video displays LED billboard video displays - Part one: Equal brightness

Michael Day and Tarek Saab Texas Instruments

Video Imaging DesignLine (10/17/2005 12:31 AM ET)

The proliferation of light emitting diodes (LEDs) in various end equipments has surpassed even the most aggressive expectations. From present-day technologies, such as car headlamps, traffic lights, alphanumeric displays, billboards and large form-factor video displays, to the newest applications, including general and architectural lighting and LCD backlighting, the burgeoning acceptance of this light source has fueled the re-design of even the most common devices. As the efficiency and brightness of LEDs improve and the cost decreases, it is anticipated that LED usage will eventually replace conventional lighting methods in consumer applications. This article addresses some design challenges with using LEDs by comparing techniques used in large form-factor video displays with those of LED-based LCD backlighting.

A stadium or advertising display integrates dozens of display panels and thousands of LEDs. Inside each array, LEDs (also referred to as pixels) will yield significant variations in brightness, such that the delta in lumens between the brightest and dimmest LED can regularly approach 15 to 20 percent, if not more. As a result, companies continually must solve difficult quality and maintenance issues. While this problem is endemic to all LED applications, it is especially critical in higher-quality systems requiring pixel uniformity such as full-motion video. To compensate for these variations, manufacturers often employ two techniques: first, they purchase “matched LEDs” from a supplier (also known as “binning”); and second, they utilize a high-quality LED driver with “dot correction” functionality.

LED suppliers offer the benefit of matched LEDs for an incremental charge. They measure and bundle these RGB (red, green, blue) diodes together with LEDs that generate similar lumens at a specified current. Using this method can provide the desired uniformity with minimal design considerations for low-end lighting systems. However, the variance in decay rate, or degradation in brightness, per pixel over time makes this method a short-lived solution. In other words, in a year or two the picture becomes “blotchy.” Furthermore, should a defective panel need replacement, the lumen output of the “new” panel will be visually dissimilar to the others.

In high-end display systems, the brightness matching requirements are far more rigid, rendering LED binning to be insufficient. In order to achieve pixel and panel uniformity over the lifespan of a display unit, manufacturers use advanced LED drivers with “dot correction” capability. Dot correction is a method for managing pixel brightness by adjusting the current supplied through each individual LED in the array. By using the dot correction feature, the processor can control full current to a panel of LEDs while the LED driver scales the current to each LED and creates uniform brightness. As a result, the processor power is free to perform other tasks by eliminating the need to check a look-up table, and by removing complex multiplication tasks for each LED in every refresh cycle. To implement dot correction, manufacturers measure the brightness of individual LEDs through photo capture. The dimmest LED in the system is designated as the “base” LED to which every other pixel is matched. To accomplish this calibration, the current supplied to each pixel is multiplied by a fractional value proportional to the LED’s lumen output. In a device like Texas Instruments’ TLC5940, the dot correction value for each LED can be dynamically changed every refresh cycle, or stored inside an integrated EEPROM. This dual-dot correction method offers the flexibility to update overall panel brightness as external lighting conditions change, and provides long term, non-volatile dot correction information that ensures panel uniformity. The EEPROM data can be rewritten as lumen measurements vary over time or as panels fail, requiring correction and replacement, respectively. The following example illustrates this concept.

For simplicity, the example considers only 16 LEDs of a single color from the many panels and thousands of total LEDs constituting a full-working system. The brightness requirement for a green pixel in a video panel may require the pixel’s green LED to have a luminous intensity of 80 mcd (millicandela). The chosen LED (Osram LP E675) is available in four different luminosity bins: 45-56, 56-71, 71-90, and 90-112 mcd, each measured at a normalized current of 50mA. Selecting the highest bin guarantees at least 80 mcd per LED. For an IC like the TLC5940, a single resistor is used to set the maximum current for each IC, which can drive up to 16 LEDs. This resistor value must set the current high enough to allow even the dimmest LED to produce 80 mcd. Therefore, according to the datasheet for the LP E675, it is necessary to drive 43 mA in order to achieve 80 mcd. By measuring the LED’s full-current (43mA) brightness at installation, one might produce an LED histogram of luminous intensity resembling Figure 1. The data in the foreground is the LED current in mA, while the data in the background is the LED brightness in mcd. As Figure 1 illustrates, the brightness variation measured between each LED in the panel may vary as much as +/-10 percent without dot correction. A measurement with such a wide deviation is unacceptable in higher-end displays. The histogram depicts the relevant data to individually adjust, or “dot correct”, each LED to a uniform brightness. When programmed to full brightness, the IC must dot correct the luminous intensity of LED1 from 83 mcd to 80 mcd. The TLC5940 has six-bit dot correction (64 steps), which corresponds to a full-scale resolution of 1.56 percent per step.

The following formula calculates the correct dot correction level for each LED:

where DC_production is the required dot correction value at production, L_baseline is the desired brightness level, and L_initial is the measured brightness at maximum current.

View full size
Figure 1 - LED Brightness and Forward Current Histogram Before Dot Correction

By rounding the calculated dot correction value to the closest fractional number, and then multiplying the original luminosity by the new dot correction ratio, one can produce the updated LED brightness.

Continued on next page

Next Page