3 Displays Figure 3.1. The University of Texas at Austin s Stallion Tiled Display, made up of 75 Dell 3007WPF LCDs with a total resolution of 307 megapixels (38400 8000 pixels) High-resolution screens have become a mainstay on modern smartphones. Initial resolutions like 480 320 pixels have given way to the iphone Retina Display (960 640 pixels or 326 ppi) and Samsung s Super AMOLED displays (800 480 or 217 ppi, used in the Samsung Galaxy S II). The upcoming Samsung Galaxy S II HD will use a Super AMOLED HD screen with a stated resolution of 1280 720 or 319 ppi, approaching laptop resolutions on a 4.65-inch screen. Two standard technologies are used for mobile phone screens: liquid crystal displays and organic light emitting diode displays. Additionally, different subpixel arrangements are used to reproduce colours at each pixel in the display. 3.1 LCD A liquid crystal display (LCD) is normally made up of six individual layers sandwiched together (Fig. 3.2). From the front of the screen to the back, they are: vertical polarizer, to block light that is not polarized in the vertical direction, 9
Chapter 3. Displays Figure 3.2. Layers of a typical LCD screen, with the backlight behind the horizontal polarizer and a glass substrate to protect the screen in front of the vertical polarizer vertical electron grid, to index columns of pixels, liquid crystal (LC) layer, to twist light s polarization as it passes through the LCs, horizontal electron grid, to index rows of pixels, horizontal polarizer, to block light that is not polarized in the horizontal direction, and reflective layer and backlight, to produce and reflect light through the display and out to the viewer. The LC layer is made up of crystalline molecules arranged in a spiral pattern. Normally, LCs rotate the direction of the polarization of light 90. So, as light is produced by a screen s CCFL or LED backlight it is polarized to be horizontal, rotates to vertical as it passes through the LCs, passes through the vertical polarizer, and makes its way out to the viewer. When an electric field is applied to the LCs, they line up and do not change the polarization of transmitted light. This means that light polarized horizontally is not 10
3.1. LCD Figure 3.3. Liquid crystals undergoing a phase transition from nematic (aligned) on the left to smectic (unaligned) on the right rotated, and therefore will not pass through the vertical polarizer on its way to the viewer. This produces a dark spot on the screen where the light is being blocked. Different techniques are used to control the LCs. In a passive matrix array, a grid of control wires is embedded in the LC layer, with wires crossing at each pixel location. To produce a dark spot at pixel (x,y), a negative voltage V is applied to horizontal wire x, and a positive voltage +V is applied to vertical wire y. Neither V nor +V is sufficient to line up the LCs, but their difference is. Because of the indexing method of one wire for an entire row/column, individual points are selected in succession, similar to an electron gun on a CRT. The basic scheme is: for each row x, apply V to x s horizontal grid line, then apply +V to all vertical grid lines that need to be darkened, lighting the row. By varying the amount of +V applied, LCs can be partially lined up, allowing different amounts of light to be rotated to produce different intensities. The LCs will maintain their aligned state for several tens of milliseconds after the voltage difference is withdrawn. Modern LCDs are active matrix. Rather than voltage grid wires, each LC uses a transistor to rapidly change the LC s state, and to control the amount of of alignment or twist in the LC. This allows higher refresh rates, since transistors act as memory units to allow a pixel to maintain a constant intensity until a change is requested. This also allows for brighter displays. Control wires are still needed to change each transistor s state. To support colour output, individual subpixel LCs have red, green, and blue filters placed in front of them. This produces a triad, a group of three subpixels similar to red, green, and blue phosphor dots on a traditional CRT. Because the subpixels are small, the eye fuses them into a composite pixel colour made up of the combination of the red, green, and blue light intensities being produced. 3.1.1 Twisted Nematic vs In-Plane Switching Modern LCDs come in two common types: twisted nematic (TN) and in-plane switching (IPS). TN displays work as described above, with voltage differences being used to twist the alignment of individual LCs to pass or block outgoing light. TN is cheaper, but it has poor off-axis viewing, often causing significant colour shifts. TN is 11
Chapter 3. Displays Figure 3.4. Layers of a typical OLED screen, with current from cathode anode producing positive holes that jump to meet negative electrons, releasing energy as visible light also normally limited to 6-bits per subpixel, or 18-bits per pixel. This means the 24-bit information from the graphics hardware must be interpolated into an 18-bit estimate prior to display. In-plane switching (IPS) was developed by Hitachi in 1996 to improve viewing angle and colour reproduction. The key difference is that, with TN, LCs are stacked vertically perpendicular to the display plane, whereas with IPS, they are stacked horizontally to lie in an arrangement that is parallel to the display plane. This allows for significantly better off-angle viewing and colour reproduction. IPS displays also normally use a full 8-bits per subpixel, or 24-bits per pixel. In order to twist horizontally aligned LCs, two transistors are required for each R, G, and B subpixel, as opposed to the single transistor needed for TN displays. 3.2 OLED Displays built with organic light emitting diodes (OLEDs) use electroluminescent organic polymers to produce lit regions (pixels) in the display. First proposed by scientists from Eastman Kodak in 1987, OLEDs use a dye with excitation states that consist of an electron and a hole that the electron can fall into. When this happens within the dye, energy is released in the form of visible light fluorescence with a colour based on the type of dye participating in the excitation. As with LCDs, OLEDs are made up of a set of layers sandwiched together (Fig. 3.4). From front to back, they are: glass substrate, to protect the screen, 12
3.3. Subpixel Arrangements transparent tin-oxide anode, to generate a positive voltage, hole injection layer, a conductive layer to produce positively charged holes, organic emitters, to emit visible lights through fluorescence when a electron and a hole meet and release energy, electron transport layer, an emissive layer to produce negatively charged electrons, and aluminum cathode, to generate a negative voltage. A current is passed through the cathode to produce excited electrons at the electron transport layer. In areas where illumination is required, the anode is charged, causing the excited electrons to stream off the cathode, through the dye, and into the hole injection layer. This produces visible light. The more electrons that are attracted, the brighter the intensity of the light being produced. OLEDs have a number of claimed advantages, although it is still unclear to what extent these can be achieved: self-luminous, so no backlight is required, suggesting (theoretically) a much lower power requirement, thin, so can be made flexible, wide field-of-view, so good off-axis viewing, good contrast and brightness ratios, so good color reproduction with a wide color gamut, and high refresh rate, so capable of high-speed motion reproduction. One disadvantage of OLEDs is that they can perform poorly under directly illumination, for example, outdoors in bright light. This is because the incoming light washes out the outgoing light from the display. This is less of a problem on LCDs, because incoming light can be reflected (by the reflective layer at the back of the display), and can therefore help augment the brightness of the display. OLEDs have no such reflective layer. Different flavours of OLED displays have been produced. The most common variant seen on phones is active matrix OLED (AMOLED), which uses transistors to drive the conductive and emissive layers. Super AMOLED, produced by Samsung, offers two additional advantages. First, the gaps between subpixels are reduced, resulting in higher overall brightness. Second, touch detection is built directly into the screen rather than being laid overtop of it. Samsung is also producing a new Super AMOLED HD display with higher resolutions, although some of this resolution improvement is achieved through subpixel rendering, rather than through a full increase in the number of subpixel units in the display. 3.3 Subpixel Arrangements Different methods are used to arrange subpixels to produce individual pixels. Common examples currently in use include stripe and PenTile. 13
Chapter 3. Displays (a) (b) Figure 3.5. Subpixel arrangements: (a) LCD subpixels, the dots and lines at the top of each subpixel are transistors connected to a control grid; (b) PenTile G subpixels 3.3.1 Stripe The most common arrangement for subpixels in an LCD is the stripe (Fig. 3.5a). Here, vertical stripes of red (R), green (G), and blue (B) are interspersed as follows: Notice that this arrangement means you need six subpixels to produce white black alternating lines: seen as If you fall below this frequency, you get coloured cyan (C), magenta (M), and yellow (Y) lines or chromatic aliasing: GB GB GB GB BR BR BR BR seen as CC CC CC CC MM MM MM MM 3.3.2 PenTile The PenTile matrix family is a subpixel scheme built on basic properties of our human visual system. The original PenTile Matrix was built on the idea of having five (penta) 14
3.3. Subpixel Arrangements subpixel groups containing two R subpixels, two G subpixels, and one B subpixel in a star-type layout: B B B (three GR PenTile pixels) The justification for two R and G subpixels per one B subpixel is that the visual system has about twice as many medium and long-wavelength cones specialized to respond to light in the green and red wavelengths versus short-wavelength cones specialized to respond to light in the blue wavelengths. This layout requires an intelligent conversion from input data to proper intensities for the GR subpixels to be effective. Since a star-shaped layout is difficult to manufacture in an LCD panel, two PenTile matrix alternatives are normally used. The first is PenTile G (Fig. 3.5b), which interleaves two G subpixels for every R and B subpixel: GGG GGG GGG GGG Again, this is justified through human psychophysics. The human visual system is most sensitive to green, and the G scheme is built on this idea. The claim is that one-third fewer subpixels are needed versus a traditional display, that is, four subpixels BR can display two pixels of information, versus the six subpixels needed in an stripe configuration. This type of resolution counting is controversial. Some claim that resolution is defined as the number of white black lines a display can produce. In this case, stripe needs six subpixels to produce a white black line,..., but so does PenTile, BGR...However, an alternative standard claims that if two successive lines differ in contrast by at least 50%, they count as individual lines. If this measurement is used, PenTile can produce alternating lines with fewer subpixels than stripe,...pentile G displays are used on various Android phones, including the HTC Nexus One and the Samsung i9000 Galaxy S. An alternative matrix is PenTile W. Here, a white subpixel W is added to the traditional arrangement: WWW WWW WWW WWW Again, this is built on models of human vision that show that we are more sensitive to high-resolution luminance information, rather than chroma hue plus saturation information. Special conversions are used to vary the luminance of the W subpixels to present higher resolution luminance, coupled with a combination of the 15
Chapter 3. Displays subpixels to present lower resolution chroma information. Again, manufactures claim that one third fewer subpixels are needed to match the equivalent pixel resolution of an stripe display. The Motorola Atrix 4G uses a PenTile W display. The obvious benefit, if you accept the argument that fewer subpixels are needed to produce pixel-equivalent results, is that PenTile displays can claim higher resolutions for a given number of subpixels. For example, the Nexus One has a stated raster resolution of 480 800 pixels. The underlying AMOLED display has a total of 960 800 subpixels. An stripe display would need 1440 800 subpixels to produce the same 480 800 pixels. Technological arguments aside, the general consensus from users is that PenTile displays are not as good as stripe displays. Common complaints are that the display looks grainy, and that individual subpixels are visible, especially when photos or other continuous-tone material is presented. Many of these issues may be solved as we move to physically smaller subpixels, however. 16