Issues in High-Volume Manufacturing of Large-Screen LED-Backlit LCDs

Environmental concerns, including initiatives for the reduction of hazardous materials and requirements for reduced energy consumption and longer product life, are driving the need for non-CCFL-based backlight technologies. Cost and technology barriers have historically kept the volumes of LED-based solutions low, although that situation is now changing.

by Evan O'Sullivan and Bob Pantalone

LIGHT-EMITTING DIODES (LEDs) have been used for many years as the backlight for laptops and small desktop flat-screen displays. LEDs allow these displays to be thinner and to dissipate less power than other backlight alternatives. The natural progression for this technology would be an expansion into larger screen sizes, but the adoption has been slow. Early large-screen implementations used a direct-lit array of hundreds to thousands of LEDs to replace the lamps on a standard backlight. More recently, a few large-screen displays have adopted the same method of edge lighting that was previously implemented in smaller displays (Fig. 1). LED-based systems have conventionally faced cost and yield issues, but recent breakthroughs in LED efficiency and lower-cost packaging are making the adoption more possible than ever before. These technical advancements, coupled with the high priority on the part of OEMs of making products more eco-friendly, are paving the way for LED backlights to garner a much larger market share in the next 2–3 years.

Fig. 1: One of the first edge-lit LED-backlit LCD TVs was this 32-in. version shown at the 2005 SID International Symposium in Boston. Today, this sized LCD can be backlit by as few as six RGB LEDs.

Regardless of the implementation (direct-lit or edge-lit) used, there are important factors to be considered when moving from prototype quantities to high-volume manufacturing. This article examines the most important of these challenges and offers some advice on how to address the issues.

LED Issues

It is well understood within the LED industry that a variation of LED color-point and lumen output will be expected during the high-volume manufacture of LEDs. This variation is the driver for binning in the LED industry. In some applications, a variation in lumen output and color point may have little effect on the end-customer's viewing experience; however, display backlighting is not a very tolerant application for LED variations because these variations will affect both brightness and color uniformity. Typical brightness-uniformity specifications are greater than 85%, and color-uniformity specifications are ±0.03 from the desired color-point X,Y setting.

There are many efforts under way by LED manufacturers to achieve tighter and more predictable control over the variation in LEDs. One such effort is Philips's Lumiramic phosphor technology, which is designed to enable tighter control of correlated color temperature (CCT) and to reduce the number of bins for a given CCT by 75% or more for white LEDs. Until the LED industry can tightly control LED variations, a high-volume product must either be able to tolerate these variations or else be able to rely on a very specific bin of LEDs and accept the associated extra cost and possible lack of supply associated with this approach.

A robust LED driver design should account for lumen-output differences in LEDs, whether the backlight consists of white or RGB LEDs. The current through each LED can be controlled in implementations using a small number of high-power LEDs, or the current through a string of LEDs can be adjusted for implementations using a large number of low-power LEDs. These adjustments must be made during the manufacturing process to calibrate the brightness uniformity of the backlight unit (BLU).

For white-LED implementations, there is little that can be done to account for variations in color (white) point, so care must be taken to use pre-determined matched bins of LEDs. For RGB implementations, the duty cycle that each color is driven by can be adjusted to make corrections to obtain the desired color point, sometimes at the cost of brightness.

A secondary issue with LEDs is their degradation over time. Without the benefit of having LED backlights in the marketplace for a number of years, it is up to the individual designer to determine if the variation in the degradation of LEDs over time will be acceptable or if some type of aging correction should be performed at predetermined usage periods. Without aging correction, the white/color point will shift. In a backlight system that incorporates multiple LEDs, these LEDs will degrade at slightly different rates over time. This results in the need to tune the operation of individual LEDs, or perhaps small groups of LEDs, in order to maintain a uniform brightness and color point over time for an entire backlight. This aging correction can be performed in a manner similar to the initial calibration and one or more light/color sensors must be added to the system to provide a feedback loop for recalibrating the color/white point and lumen output of the LEDs.

The following discusses an edge-lit RGB-LED-backlight LCD-TV design and shows how it handles LED aging.

Figure 2 shows an example of an edge-lit LED-backlit LCD TV that incorporates a bladed-light-guide approach (using horizontal sections of a light guide with attached LEDs) with RGB LEDs and color sensors used for aging correction. The system uses a microprocessor-based adaptive control system to progressively control each blade and also maintains a consistent color-point performance. Color sensors installed behind each blade measure the color saturation and tri-stimulus value for a given LED color. This value is used adaptively by the microprocessor to correct the non-linearities that arise due to LED aging and thermal changes.

Fig. 2: Ilustrated is an edge-lit LED-backlight LCD-TV block diagram showing the bladed approach incorporating horizontal sections of the light guide with attached LEDs.

The adaptive algorithm used is a "Self-Tuning-Regulator" mechanism and is depicted in Fig. 3. The control law and model estimator modules are the two critical components of the mechanism, and these modules are invoked at a periodic sampling time. The best-suited model estimator for the given scenario is an nth-order polynomial regression. The reference polynomial regression is constructed one time, based on the LED's subsystem behavioral response. Any performance changes over time are dynamically compensated through new model estimates. Based on these new inputs, the control law drives the system to achieve the performance goal.

u(k) is the activator variable derived from the performance goal, estimated model, and performance measurements. The model estimator estimates the system behavior for a given input and predicts the model. This model estimation is based on a one-time system learning process:

With time, this behavior will change due to several conditions explained above; thus, model prediction, combined with performance goal, and current performance measurement is key in order to achieve the goal.

Aging Control

As LEDs age, their brightness performance begins to degrade. A study of this decay can help the adaptive algorithm to correct for this effect periodically. The adaptive algorithm monitors the LEDs' performance by using a built-in color sensor in order to obtain a precise state of decay and then applies compensation to offset this change. The decay factor is propagated to the dynamic-brightness-control algorithm to adapt this new decay change factor. This will allow accurate white-point preservation over time.

Manufacturing Issues

The electrical to optical efficiency of an LED-based backlight is dependent on a number of parameters. Certainly, the efficiency of the driver scheme is important, but once the driver design is set, the variation from system to system must be minimal. Light leakage and light coupling are two other areas that can drastically affect the efficiency of the system. For edge-lit systems, the light guide is the primary optical component, and its design determines much of this loss. Losses can range from 5% to 10% in well-designed systems, to 15% and more in poorly designed systems. Once the light-guide mold is created, the variations from system to system will be minimal.

Fig. 4: Cross-sectional view of the LED die and light guide (blade).

Fig. 5: This photo shows LED dice placed directly onto a printed-circuit board and wire-bonded directly to the circuit board with no protective package or window. Packaging choices such as these improve the system performance, but also create the need for new handling and manufacturing processes.

The mechanical positioning of the LEDs with respect to the light guide is crucial for capturing the maximum amount of light output from the LEDs. Care must be taken to ensure that the LED is properly aligned to the light collecting surface of the light guide, which in this case requires that the LED is at a 90° angle from the light guide, and that any air gap between the LED die and the light guide is tightly controlled. Control of the air-gap distance is crucial, due to the dispersion angle of the light emitted from the LED, which in most cases is a lambertion pattern. While achieving a direct connection with no air gap would be ideal, it is not currently practical because of the height required for the wire bonds on the LED die. Hence, a very small air gap is required.

A robust design will achieve these close tolerances without over-complicating the manufacturing process. Figure 4 shows a typical implementation and the tight tolerances needed to ensure a robust design.

Large light guides can be manufactured in small quantities using various techniques, but these techniques do not necessarily translate into a high-volume solution. Light-extraction features can be molded into the light guide, printed onto the light guide, or cut/scored onto the surface. Each method – molding, printing, or cut-scoring – has its own merits and drawbacks, but the ability to consistently create the same light-extraction patterns will affect the uniformity of the system.

LED packaging for backlight applications may not meet the robust standards of the LEDs that most manufacturers are familiar with handling. In some cases, the optics or protective glass windows may be removed to improve coupling and LEDs may be wire-bonded directly to a printed-circuit board to maximize heat transfer (Fig. 5). The opportunity for damaging these components can become a real problem in a high-volume environment if proper care is not taken in the handling and placement of these devices.

High-volume manufacturing of LED-backlight systems also requires more attention to rework procedures for repairing failed systems (Fig. 6). Not only is handling an issue during the rework process, but care must be taken to use methods that do not over-stress the LEDs or the LED packaging. Developing a close relationship with the LED supplier will help ensure that the processes used will not cause catastrophic or latent LED issues.

A final word of caution relates to the selection of other system components, most notably the system power supply. The LEDs in an LED-based backlight may have a stated reliability of 50,000 hours or greater, but the system reliability number may be no better than current products if the other components in the system are not carefully selected for high reliability. Also, LED reliability is extremely dependent on the operating temperature of the LED die, so care must be taken when choosing components used for dissipating heat. The junction temperature of the LED has a large effect on its wavelength, brightness, and life. Typically, a junction temperature over 65°C will cause some degradation in these parameters.

The use of LEDs as a light source for backlighting is not new, but their use in high-volume large-screen-sized applications is still in its infancy. What is needed is a robust design that will meet the required brightness and color specifications without the need to specify a very narrow bin of LEDs. Manufacturers of high-volume products that are not familiar with using LEDs for backlight applications will need to educate themselves on the proper handling and repair of these systems. The realization of high-volume LED backlights for large-screen displays is a certainty in the near future and with care and attention to the product design and the assembly process, these products will certainly deliver on their promise of low power, high reliability, and enhanced user experience. •

Fig. 6: Having proper rework procedures in place to repair failed systems is important. Here, a rework technician using approved handling and rework processes repairs a failed LED PCB from a backlight module.