LED driving techniques reduce power in LCD TVs
There seems little doubt that LCD technology with LED backlighting is the only viable way to reach the efficiency targets that authorities are proposing. Plasma has the disadvantage that each pixel is an active light emitter, so power consumption is directly proportional to the number of pixels. As a result, HD plasma televisions consume around two to three times the power of an LCD display for the same resolution & brightness. Highly touted OLED technology — as recently reported — may not come any time soon, if ever. The investment required for this “bleeding edge” large panel technology is prohibitive. However, large display panels with current state-of-the-art TFT-LCD technology and “smart” direct LED backlighting with local dimming is far less expensive than OLED but compares well for both power consumption and picture quality.
But today’s LCD TVs, even those with LED backlighting, are still some distance from achieving the efficiency targets they will face in the coming years. However, new design techniques in LED driver circuits promise to deliver significant energy savings that will go a long way to helping TV manufacturers meet the tough requirements for power consumption.
Changing requirements of TV power standards
Standards for TV power consumption, such as Energy Star, came out in 2008 and each year the specification reduces the amount of power the TV can draw. The design challenge gets even tougher for large screen TVs since the current maximum for any size screen is 85 Watts.
Energy Star is voluntary but highly influential but it’s not the only form of regulation. For instance, the State of California’s Energy Commission introduced its own standard. This regulation is a bit tougher than Energy Star and also has real teeth – it prohibits the sale of TVs in California that do not meet its efficiency specifications. In Europe, regulations have for many years allowed direct comparison of the energy consumption of white goods (EU Energy Label) and customers use it as a basis for purchasing decisions. These regulations are now mandatory for TVs, cars etc.
The operation of LED backlighting
Since LED backlighting power ranges from 30 percent to 70 percent of overall system power in LCD TVs, improvements in the efficiency of the backlighting power circuit can make a considerable contribution to system efficiency. As is often the case in power system design, a number of small improvements in efficiency can deliver a large combined saving.
There are two ways to implement LED backlighting (see figure 1). In indirect or edge-lit backlighting, the LEDs are arranged at the edges of the screen. A light guide distributes the light uniformly across the display. This arrangement can be deployed with good optical uniformity in screen sizes up to 40”, and enables backlighting units with thickness of just 5-10 mm.
Figure 1: LCD TVs can adopt one of two arrangements for LED backlighting
In direct backlit systems, the LEDs are located directly behind the LCD, enabling low power, good thermal design and excellent scalability with practically no limit to the screen size. These panels tend to be thicker than edge-lit versions, but with the latest technologies for light distribution, displays as thin as 8 mm can now be found. An important advantage of direct backlighting is that it enables sophisticated local dimming, which lowers power consumption and increases the dynamic contrast ratio, allowing the latest TV designs to compare favorably with OLED.
LED backlit system architecture choices
The choice of architecture for an LED backlit driver system is the decision with the greatest potential to produce power savings and significantly enhanced picture quality. The designer looks for the best balance between local control of strings of LEDs and the lowest possible bill of materials (BOM).
Single string and single DC-DC converter
A switched-mode power supply (SMPS) is used to provide the voltage for backlit LEDs arranged in strings. A current sink is included to regulate the current through the LED string. To minimize power dissipation, the voltage at the ILED sink needs to be a fraction above the voltage necessary to guarantee that the LEDs receive their specified current (see figure 2).
Figure 2: Single-string, single DC-DC converter backlight system architecture
A common design approach is to use a feedback path from the ILED sink to the SMPS to regulate the SMPS’ output voltage. This feedback path is required to allow for variations in forward voltage (Vf) from one LED to another. The typical Vf of a white LED is around 3.2 V, and may vary as much as ±200 mV per LED. So, for example, in a string of 10 LEDs, the total for VLED may range from 30 V to 34 V.
The voltage required at the DC-DC converter can be expressed as:
VSINK is assumed to be 0.5 V, so the ILED sink must regulate VDC-DC in the range of 30.5 V to 34.5 V, depending on the actual LED forward voltages.
Multi-string and multi DC-DC converter
A single string of LEDs is rarely adequate because as the number of LEDs in the string rises the required output voltage also rises. Above a certain VOUT/VIN ratio, the SMPS’s efficiency falls dramatically. LED backlight designers can therefore use several strings in order to avoid an excessively high output voltage required of the SMPS.
The simplest approach is to duplicate the single-string single DC-DC converter topology at each string (see Figure 3). The advantage is efficiency, because each string’s voltage is regulated separately. The disadvantage is the high cost, since each string requires its own DC-DC converter, MOSFET, coil, diode, and output capacitor. In order to save bill of material (BOM) cost, the designer could reduce the number of LED channels, using long strings with many LEDs in each string. But this compromises the system’s ability to implement local dimming, which is another important power saving technique. Therefore, none of the trade-offs of this topology is particularly attractive.
Figure 3: A separate DC-DC converter with each LED string is an expensive option
Multi-string with single DC-DC converter
A more radical approach to reducing BOM cost can be found in the multi-string with a single DC-DC converter topology (see Figure 4). The drawback of this approach is that the SMPS voltage must be regulated higher than the LED string with the highest forward voltage, which means that it operates at a higher voltage than is necessary for those strings with a lower forward voltage. This means that the ILED sink must dissipate the excess power from the strings with lower forward voltage, generating heat that must be conducted away from the circuit board, and resulting in reduced power efficiency.
Figure 4: With one DC-DC converter serving multiple LED strings, SMPS voltage is not optimized
Multi string mixed architecture
The architecture that provides the best balance between efficiency and BOM cost is one that combines elements of the multi-string and multi DC-DC converter architectures previously described. This mixed architecture (see Figure 5) has multiple DC-DC converters supplying groups of LED strings.
Figure 5: A mixed architecture optimizes balance between BOM cost & and power efficiency
This solution offers the best overall power efficiency because it combines the advantage of local dimming in direct backlit systems with good DC-DC output voltage regulation. It also offers a substantial BOM saving over the efficient multi-string, multi DC-DC converter architecture.
Regulating current to match the characteristics of LEDs
The LED manufacturing process causes wide variations in brightness and color temperature from one LED to the next. As a guide to users, white LED manufacturers allocate each manufactured unit to groups or ‘bins’ of LEDs with comparable performance in terms of color, brightness and forward voltage. But the manufacturer’s specification for each brightness and color temperature bin is only valid under specific nominal operating conditions. This means that the LED current must be set to the nominal current stated in the datasheet in order to generate the specified brightness and color.
Consequently, dimming and brightness control can only be implemented by switching the current to any single LED either to ON (nominal current) or OFF (zero current) through a digital PWM control signal. In analog dimming, the LED would be operating outside its specified nominal current, leading to unacceptable changes in color temperature and poor LED-to-LED brightness matching (see Figure 6).
Figure 6: Brightness of LEDs from the same bin is guaranteed to match only at nominal current (in this case, 20 mA)
Current sink characteristics
Since LEDs require a perfectly regulated constant-current power supply, it follows that the primary role of the LED driver is to set the current to the nominal value when ON and to 0 A when OFF. Therefore, the feedback loop controlling the accuracy of regulation requires an extremely precise current sink (see figure 7).
While there are a variety of current sink designs, the precision requirements of TV backlighting (current regulation better than +/- 0.5 percent) require an accurate op amp to set the ILED current independent of the ILED voltage. But in backlighting driver applications, the task is more challenging because the accuracy of current regulation must be maintained even when the voltage at the current sink falls to very low levels.
This is a difficult requirement to meet but 4 generations of very accurate current sink LED drivers from ams – AS369x, AS381x, AS382x, AS385x — have been designed specifically for such applications. These devices also incorporate an offset-compensated op amp. Current sink drivers require a minimum voltage at the drain (VDS(sat)) to ensure the full accuracy and proper operation of the sink transistor inside the saturation region. For the saturated region the gate-source voltage primarily controls the output current.
If the current sink is to operate at high efficiency, it is important that the voltage drop between VSET and VDS is low. LED drivers with op amps that include built-in offset cancellation can maintain VSET at levels as low as 125-250 mV. Allowing an additional margin for VDS above VDS(sat) of 150 mV, a total voltage drop at the current sink of approximately 400 mV is necessary. For a string of eight LEDs (where Vf = 8 x 3.2 = 25.6 V) this results in a power loss of around 1.5 percent in ISINK. Without the offset cancellation included in ams’ backlight LED drivers, the value of VSET would be higher, leading to higher power losses at the current sink.
Figure 7: Current sink designs; a precision current sink requires an accurate op amp with offset compensation
Feedback regulation for power optimization
As has been shown above, a feedback path from the LED driver to the SMPS sets the drain voltage to the minimum required value. The output current sink can be implemented either with a simple, defined current output driver and an external capacitor (see Figure 9, left-hand diagram) or with a digital control circuit which sets attack/release times and controls the current output with a digital-to-analog converter (IDAC) (see Figure 8, right- hand diagram).
Figure 8: Two different methods for implementing a feedback loop to the SMPS
Both of these solutions offer good efficiency, work with every type of SMPS with voltage feedback, and can be implemented by attaching feedback lines from more than one driver to the same SMPS, as is required in mixed- architecture systems.
However, the second, digital implementation provides some special advantages. As well as not requiring an output capacitor, the digital circuit also gives the designer the freedom to define the feedback system’s attack and decay times. By selecting a fast attack time combined with decay latency and relatively slow decay, the display’s performance can be improved. This benefit is particularly noticeable in scenes that require brightness to change rapidly. In this case, a fast attack time eliminates perceptible brightness artifacts as the screen changes from dark to full brightness. The analog solution (from figure 8) dims the LEDs’ output gradually during a short dark frame, resulting in a visible delay in achieving full brightness for the next bright frame.
This is a noticeable distraction for TV viewers because films and other video content create large dynamics from one frame to another. Such artifacts can be eliminated with digital regulation circuits by inserting latencies of several hundred milliseconds into the decay instruction. Thus, when bright frames are interrupted by a short sequence of dark frames, the second bright frame starts at full brightness because the driver has automatically delayed the voltage ramp-down. Digital feedback algorithms implementing decay latency can be found in products from ams.
Another useful feature integrated in LED driver ICs is a fast Serial Peripheral Interface (SPI). In direct backlit TVs, the LEDs are arranged in a large number of relatively short strings, so that small areas of the panel can be dimmed to save energy. Typically, such arrangements contain 256 channels in a matrix of 16×16 fields, each individually configured through pulse width modulation (PWM). But generating 256 PWM signals with variable PWM width and delay is a hugely intensive processing task, even for the fastest microcontroller.
These backlighting systems therefore use local PWM generators integrated into the LED driver ICs. This enables brightness to be configured with simple SPI data transfers. In an architecture with multiple driver ICs (e.g. 256 channels with 16 channels per IC, and 16 ICs), the LED channels can be configured by daisy-chaining SPI signals and transferring the data that are used in a VSYNC frame to the frame before.
In this arrangement, data transfer over an SPI can reach speeds of 20 Mb/s, or 50 kb/frame at a 400 Hz frame rate. This is fast enough to change dimming of each field in sync with the actual frame. So ideal local dimming can be achieved with minimum overhead on the microcontroller.
Smart dimming for edge-lit systems
This local dimming technique is only possible with direct backlit systems. But a certain amount of smart dimming can also be achieved with edge lighting. In particular, PWM dimming can be used to vary brightness without changing the color temperature of the white LEDs. Instead of having the edge-lit LEDs permanently set to a specific brightness value, the brightness can be dynamically altered via changes to the pulse width.
Another technique for saving power is Dynamic Luminance Scaling (DLS). With this technique, the LCD’s white level/ brightness level is increased in certain scenes to allow the backlight LEDs’ power output to be reduced.
Yet another method to reduce power consumption is the use of ambient light sensors. If the room where the TV is being watched is fairly dark, the backlight brightness can be reduced (see figure 9).
Figure 9: Energy saving methodologies using smart LED drivers and smart ALS sensors
Even more sophisticated methods are being explored by TV manufacturers. For instance, cameras are beginning to be designed into displays to enable consumers to use video-telephone services such as Skype on their TVs. These cameras can also be used to detect if someone is actually watching TV; if the TV is on without anyone being in the room the backlight can be reduced to a minimum brightness level.
Even customized energy consumption patterns can be implemented. While you might prefer watching in the energy friendly eco-mode with reduced backlighting, another member of the household might prefer full brightness.
In sum, considerable power savings can be realized by implementing today’s advanced techniques for efficient LED driving. This is important since ever-tougher regulations continue to reduce the maximum power that a new TV can consume.
About the authors:
. Herbert Truppe is a Product Line Manager at ams.
. Peter Rust, Werner Schögler and Manfred Pauritsch are all Sr. Design Engineers at ams
This artice originally appeared on EE Times Smart DesignLine.
If you enjoyed this article, you will like the following ones: don't miss them by subscribing to :
