CHAPTER 1 INTRODUCTION
1.1 Motivation
In recent years, flat panel display has been widely used and become the main stream of future display devices. The most popular display material is liquid crystal.
For LCD (liquid crystal display), its advantages compared with traditional CRT (cathode ray tube) display devices are lower power consumption, lighter weight, smaller volume, and less radiation.
As the display technology advances, the future system requirements for LCD devices are listed below:
Higher color depth. (10 bits/color).
High resolution. (more than 45 dots per inch) Larger panel size. (LCD-TV applications)
Wide view-angle. ( >± 60 in both horizontal and vertical directions) Ultra high contrast ratio. (> 600:1)
Low power consumption.
The market of LCD devices is growing larger and larger because of the electronic products of next generation are all using flat panel displays, such as cell phone, notebook, digital camera, portable digital assistant, LCD-TV, etc. Actually, LCD has already replaced the traditional CRT.
The key considerations of future specification of LCD are higher color depth and less response time. At present days, the response time of the LCD-TV is not good
enough for catching objects with high velocity in display. Also, as the panel size of LCD-TV become larger and larger, driving large panel is another challenge. But since LCD-TVs are superior in many ways compared with traditional display methods, they have already become the main stream of future display devices. As a result, design of driver circuits on LCD display is certainly worth of future research.
1.2 Organization
In this thesis, we focus on the design of high resolution, high accuracy, low response time, and low power consumption data drivers for TFT-LCD (thin film transistor-liquid crystal display).
In Chapter 2, some background knowledge of LCD, such as passive/active matrix LCD, pixel structure, panel structure, frame driving method and periphery circuits, is described.
In Chapter 3, we discuss the output buffers for data driver. The characteristic of low offset voltage is especially emphasized. Also, low power dissipation and high slew rate are certainly important considerations.
In Chapter 4, digital to analog converter for data driver is discussed. The main considerations for the design of DAC are small area and high resolution. Charge recycling circuits are also introduced.
In Chapter 5, the circuit layouts, measurement results are discussed.
In Chapter 6, conclusions and future works have been made.
CHAPTER 2
LIQUID CRYSTAL DISPLAY
2.1 Liquid Crystal Display Structure
2.1.1 Introduction to Liquid Crystal
Liquid crystal was first found by F. Reinizer in Australia in 1888, but it was not applied for modern display until 1960’s [1]. There are many kinds of liquid crystal materials. Distinguished by the arrangement of liquid crystal molecules, they can be divided into three groups, Smectic liquid crystal, Nematic liquid crystal, and Cholesteric liquid crystal. Different kinds of materials are usually blend for different applications.
Differing by the temperature, one important characteristic of liquid crystal materials is called “twice melting”. Below the melting point Tm they are solid crystalline, where above the clearing point Tc they are clear liquid. Between Tm and Tc, the materials look milky liquid but still exhibit the order phases, called mesophase.
Fig.2.1 illustrates the temperature versus phases. For TFT-LCD applications, it is always used in mesophase.
Fig.2.1 Liquid crystal materials phases versus temperature [1].
Fig.2.2 Basic theory of liquid crystal display. (a) twisted liquid crystal (b) non-twisted liquid crystal [2].
Fig.2.2 shows the basic theory of LCD. The basic structure of liquid crystal display is upper and lower polarizers with orientation layers. For upper and lower polarizers with 90° phase difference, we call it “Normally White” where both polarizers with the same phase are called “Normally Black”. In Fig.2.2 (a), without applying any external voltage, the liquid crystal twists 90° phase and guide the light to pass both the polarizers in “Normally White” case. But in Fig.2.2 (b), applying a large voltage supplied by transparent electrodes outside the orientation layers causes all liquid crystal molecules turn into one direction and the light cannot pass. If we apply a smaller voltage in between, the panel would look between black and white. By controlling the applied voltage, LCD can display different gray levels.
2.1.2 Passive/Active Matrix LCD
For dynamic drive, it can be divided into two different methods, passive matrix LCD (PMLCD) and active matrix LCD (AMLCD). Fig.2.3 and Fig.2.4 illustrate the two methods. The PMLCD uses row electrode (X) and column electrode (Y) to determine the gray scales of each pixel. The drawback of PMLCD is that pixels on the same row or column would influence one another. To solve this problem, in AMLCD we use a TFT (thin film transistor) or a diode as a switch for each pixel. Another advantage of AMLCD is its higher operating frequency compared to PMLCD. Since that AMLCD has been commonly used in large and high-resolution panel products.
Fig.2.3 Passive matrix LCD [2].
Fig.2.4 Active matrix LCD [2].
2.1.3 Pixel Structure of TFT-LCD
One pixel is the basic unit of LCD panel. Pixel structures and its layouts are shown in Fig.2.5 and Fig.2.6. There are two kinds of pixel structure, Cs on common and Cs on gate. Comparing Cs on common and Cs on gate, Cs on gate has the advantage of compensating the unstableness of voltage level caused by feed-through effect from Cgd, but Cs on gate need more complicated scan-line signals than Cs on common. Fig.2.5 shows the effective circuit of each pixel. Cls stands for the effective capacitor of liquid crystals, Cgd is the parasitic capacitor between scan-lines and effective liquid crystal capacitor, and Cs is the storage capacitor that stores the voltage between frame transitions. The transistor in each pixel is a TFT (thin film transistor)
used as a switch. Fig.2.6 illustrates the layout of each pixel. The layout area exclusive of the dash-line squared region is called aperture region, where the light can pass from the backlight source. Of course, the larger aperture region, the higher panel brightness it is.
Fig.2.5 Effective circuit of pixel [3].
Fig.2.6 Pixel layout of TFT-LCD [3].
2.1.4 Structure of LCD Panel
The structure of LCD panel is shown in Fig.2.7. As described in 2.1.1, there are two polarizers, backlight, and liquid crystal layer. Between the lower polarizer and liquid crystal is the TFT substrate, which is used to control the applied voltage of each pixel. TFT substrate contains a glass substrate, TFT switches, transparent electrodes, and alignment layers. Transparent electrodes are made by ITO (Indium Thin Oxide), and by voltage supplied from TFT on the glass substrate they can be used to control the directions of liquid crystal molecules in each pixel. There are also color filters, which contain three original colors, red, green, and blue (RGB). For color filter substrate, we also need an alignment layer, a transparent electrode, color filters, a glass substrate and a polarizer. By controlling the amount of light passing through color filter, i.e., different kinds of color intensities, million kinds of colors can be realized.
Fig.2.7 Structure of AMLCD system [4].
2.2 TFT-LCD Driving Method
2.2.1 Inversion Driving Method
Liquid crystal materials contain ionic impurities that drift to electrode under a DC field. If sufficient impurities collect to an electrode, they nullify the charge on the electrode. This would cause a permanent damage of LCD and thus abnormal operations. Since keeping a net zero DC field across the LC materials is the basic driving method of LCD panels. Each pixel should be driven with alternating polarity signal to keep a net zero DC field. This is called “Inversion Driving”.
There are four types of inversion driving: frame inversion, row inversion, column inversion, and dot inversion. These are illustrated in Fig.2.8. The best display quality is the dot inversion, but this method needs more complicated driving signal.
(a) Frame inversion (b) Row inversion
(c) Column inversion (d) Dot inversion
Fig.2.8. Inversion driving methods of TFT-LCD.
2.2.2 Direct Driving and AC Modulation Driving
As described in 2.2.1, an effective zero DC field across the liquid materials should always be kept. There are two kinds of data driving method, “Direct Driving”
and “AC Modulation Driving” as shown in Fig.2.9 and Fig.2.10. In Direct Driving the common voltage of each pixel is fixed where in AC Modulation Driving is not. For better display quality and simpler design, Direct Driving is better. But the voltage swing of AC Modulation Driving is smaller and it is more suitable for low power consumption cases.
Fig.2.9 Direct Driving [5].
Fig.2.10 AC Modulation Driving [5].
2.3 Periphery Driver Circuits
The periphery driver circuits of TFT-LCD panel are shown in Fig.2.11. There are three major parts of the driver circuits, Timing Controller, Data Driver (also called Source Driver), and Scan Driver (also called Gate Driver). Timing Controller receives the input RGB signals and clock from the previous digital circuits and translates these signals to proper signals for Data Driver and Scan Driver.
The Scan Driver would rise gate voltage of each scan line and turn on the transistors sequentially. Meanwhile, the Data Driver sends the display data to each pixel. Same process will repeat again and again during refresh cycles. Detail descriptions of Scan Driver and Data Driver will be discussed below.
Clock
Fig.2.11 Periphery driver circuits of TFT-LCD Panel.
2.3.1 Scan Driver
The block diagram of scan driver is illustrated in Fig.2.12. Scan driver contains shift register, level shifter, and output buffer. The scan driver is just to turn on the TFT switches of a single row in order. Shifter registers can storage input digital signals and pass these to level shifters according to clock timing. Since the voltage to rotate liquid crystal molecules is usually higher than 10V, we need level shifters to get higher voltage. Finally, because each row line can be modeled as a RC-ladder, it is necessary to use some digital buffers that lower the delay time of gate pulse to drive the panel.
The numbers of channels of scan driver depend on the TFT-LCD panel size. Table.I also shows the timing specification of standard video signal.
Source Driver
Fig.2.12 Block diagram of scan driver.
Mode
VGA SVGA XGA SXGA UXGA
800x525 Table.I Timing specification of standard video signal.
2.3.2 Data Driver
There are two kinds of TFT-LCD data driver, analog and digital type. Since analog data driver is only suitable for small panels due to its sample and hold architecture, in this thesis we only discuss digital data driver which is used to drive large panels. Digital data driver contains five parts: shift registers, data latches, level shifter, DAC with gamma correction and analog output buffer. Fig.2.13 illustrates the block diagram of data driver. Input data signals are serially read and stored by shift registers and data latches. Digital to analog converter converts digital display data into analog voltage signals, and gamma correction is used to compensate the sense of sight for human eyes. Finally, to drive the effective RC-ladder liquid crystal, it is necessary to design output analog buffer with high slew-rate.
As the accuracy bits of digital display data become higher, the design of low offset voltage output analog buffer become a challenge. Since in this thesis, low offset voltage buffer will be mainly discussed.
Output Analog Buffer
G at e D ri ve r
pixelpixel
pixel
pixel
pixel
pixel
pixel
pixel
pixel
TFT-LCD Panel
DAC with Gamma correction Level shifter
Second data latches First data latches
Shift register
So ur ce D ri ve r
Fig.2.13 Block diagram of data driver.
CHAPTER 3
OUTPUT BUFFERS FOR DATA DRIVER
3.1 Design Consideration of OP-AMP for AMLCD
For TFT-LCD data driver, it contains shift registers, data latches, level shifter, DAC, and output buffer as described in chapter 2. To discuss the design consideration of OP-AMP for AMLCD, let us take NEC-µPD16721 as an example [6]. The NEC-µPD16721 is a source driver for TFT-LCDs capable of dealing with displays with 256-gray scales. Data input is based on digital input configured as 8 bits by 6 dots (2 pixels), which can realize a full-color display of 16,777,216 colors by output of 256 values γ −corrected by an internal D/A converter. The block diagram of NEC-µPD16721 is as shown in Fig.3.1. Shift register and data register are used to store the input digital data in order, and due to inversion driving method we need a latch to control the signal polarities. In order to drive liquid crystal, it is necessary to use high voltage around 12V, and the level shifter is used to shift low voltage digital signals to high voltage ones. Also, since the NEC-µPD16721 is designed to display 256-gray scales, its D/A converter consists of a string of resistance which is divided into 256 segments, and 8-to-256 multiplexer and some voltage buffers. The relationship between output circuit and D/A converter is illustrated in Fig.3.2. When the voltage supplied to resister string is between 9.8V to 5.5V or between 4.5V and 0.2V, according to the data sheet the standard output deviation of output buffer is
±10mV and the allowed maximum value is 20± mV.
Fig.3.1 The block diagram of NEC-µPD16721 [6].
Fig.3.2 Relationship between output circuit and D/A converter [6].
For output voltage between 9.8V to 5.5V or between 4.5V and 0.2V with 8-bit resolution, i.e., 256 gray levels, the maximum allowed deviation of output buffers are calculated as followed:
Voltage Range V
R= 9 . 8 V − 5 . 5 V = 4 . 3 V
(3-1)
Since the calculated result is 8.4mV, the 10± mV on data sheet is a reasonable range.
The POL signal is used to control the polarity of output voltage due to the inversion driving method. When POL is high, the voltage across the resistor string is between 9.8V and 5.5V. And if POL is low, the voltage across the resistor string is between 0.2V and 4.5V. In Fig.3.3, the testing model of column driver is illustrated. In actual practice, the resistance and capacitance values depend on the panel size. There is only about 10 microseconds to settle the display analog voltage, or the TFT switch will turn off and the pixel will get incorrect display value. The timing specification of NEC-µPD16721 is listed in Table.II. Since the output buffer should be designed as input/output rail-to-rail, high slew-rate, and low output voltage deviation for different voltage applications.
Fig.3.3 Test model for TFT-LCD source driver [6].
Time
Arrival time through digital block to analog output Settling time from output
voltage to target voltage
Specification Output-target voltage
within 10% in 5us Settled voltage within
10mV in another 5us
Table. II Timing specification of NEC-µPD16721 [6].
3.2 Unity-Gain Voltage Buffers
3.2.1 Unity-Gain Voltage Buffers
From the above discussions, the design considerations of voltage buffer for LCD data driver could be concluded. First, we need input/output rail-to-rail operational amplifier due to inversion driving method. Second, we need high slew-rate unity-gain voltage buffer due to timing constrains. Also, layout area and power consumption are important issues. For example, to drive a LCD panel such as XGA standard, it is necessary to use eight 384-pin data driver ICs to drive 3072 columns. Since there are 384 output buffers in a single IC, any wasted area or unnecessary power consumption
in a single buffer would result in a huge penalty. We can conclude that input/output rail-to-rail, speed, layout area, and power consumption are main design considerations.
In this thesis, all will be considered.
3.2.2 Input/Output Rail-to-Rail Operational Amplifier
One traditional input/output rail-to-rail constant- g class-AB operational m amplifier is illustrated in Fig.3.4 [7]. In order to achieve input rail-to-rail, two differential amplifiers including PMOS and NMOS type are used. Class-AB output stage is used because of output rail-to-rail. Left part of this circuit is constant-g m design, since we wish the current conductance g be the same between different m operational regions. There are several drawbacks in this traditional input/output rail-to-rail operational amplifier. First, it is necessary to use many transistors in the circuit, and this would cause unnecessary layout area. Second, the current consumption of this circuit is huge since there are many stages, and class-AB output stage consumes power. Finally, slew-rate and current consumption are trade-off in this design.
3.2.3 High-Speed Operational Amplifier
From the above discussions, this unity-gain voltage buffer is basically a slew-rate limited buffer. There are several reasons for this. First, the input display signals are always unit step functions due to different digital display data. Second, since it is necessary to use inversion driving method, i.e., even with the same gray level during two scan period, the input voltage for this buffer would swing between two different
Gnd
Vi- Vi+
Vb3
Vdd
Out Vb1
Vb2 Vb4
Isrc1
Isrc2
Vb5 Vb6
Fig.3.4 Traditional input/output rail-to-rail constant-gm class-AB operational amplifier [7].
polarities. The worst operation condition of this buffer occurs when the LCD panel displays normally white/black, where the output voltage buffer will be sent during 9.8V and 0.2V between frames to maintain inversion driving. Obviously, the slew rate of this voltage buffer must be enhanced. Several methods to achieve high slew-rate have been proposed [8]-[9].
3.2.4 Class-B Buffer
For traditional operational amplifier with class-AB output stage, the sizes of the output stage MOSs are usually made large for driving large capacitance and resistance loading. But the drawback of this kind of design is that the output stage would consume large current. When designing TFT-LCD driver circuit, there are 384 output voltage buffers and each voltage buffer must consume only a little current, or the total current consumption would be way too huge. Also, class-AB buffer consumes large power even when input signal is static, which means that even when there is no transition, it still wastes power.
Class-B buffer design is a good solution for solving above problems. The main characteristics of class-B are large driving capability, which means high-slew rate, extremely low static current consumption, and low output voltage deviation. Several class-B buffer designs for driving TFT-LCD panel had been proposed [10]-[11].
3.3 High-Speed Class-B Buffer
The class-B amplifier had been presented as a LCD column driver [12].
Although class-B amplifier is limited to some specific applications due to its inherent crossover distortion, it can be used as an output buffer for a stepwise signal as long as the crossover distortion is smaller than the smallest required resolution. Thus, it is suitable for a flat-panel-display column driver because the input to the driver is always a step function. Fig.3.5 shows the block diagram of the class-B amplifier [12].
It contains one pre-amplifier (AMP), two inverter-type comparators (INV_N, INV_P), and two output transistors (Mn, Mp).
+ Fig.3.5 Block diagram of the class-B amplifier [12].
This circuit is connected as a negative feedback type. Basic operation principle of the class-B amplifier is described as followed. At the pre-amplifier, AMP amplifies the difference between Vout and Vin, and the output of this pre-amplifier is connected to two inverters with designed decision level of the comparator. If the voltage of Vin is higher than that of Vout, it would cause node P logically low and turn on Mp to charge Vout. On the other hand, if Vin is lower than Vout, node N is logically high and turn on Mn to discharge Vout. When Vin is close enough to Vout, i.e., in the dead-zone of this class-B amplifier, it would cause node P logically high and node N logically low, thus both of the output transistors are cutoff. This is the reason why class-B has very low static current, but is capable of charging/discharging loading capacitance quickly enough. The complete circuit schematic of p-input class-B buffer is shown in Fig.3.6. The circuits of pre-amplifier and comparator are modified from the current mirror amplifier. C1 and C2 are compensation capacitors.
M2 M1
VSS VSS VSS VSS VSS
P
Fig.3.6 Circuit schematic of the p-input class-B output buffer [12].
In order to eliminate quiescent current under class-B operation when Vout is equal to Vin, both output transistors Mn and Mp must be completely turned off. Thus,
In order to eliminate quiescent current under class-B operation when Vout is equal to Vin, both output transistors Mn and Mp must be completely turned off. Thus,