Today's consumers expect to view crisp, responsive displays on their handheld devices. This presses product developers to select high-quality displays that are low in cost, consume minimal power, and can mount on a flexible substrate. However, the all-but-ubiquitous SuperTwist-Nematic (STN) liquid-crystal display (LCD) is coming up short. Its unquenchable thirst for backlighting shortens battery life and causes luminance to diminish dramatically in direct sunlight. What's more, this commonplace LCD must also be sealed in glass, significantly adding to overall weight and cost. Also, slow refresh rates make scrolling text difficult to read, let alone video—a big drawback as broadband becomes ever more pervasive.
There are two new flat-panel display technologies that exhibit great promise. They're the organic light-emitting diode (OLED) and the cholesteric LCD (CLCD). Both consume considerably less power, are visible in bright sunlight, offer a wide viewing angle, and can be manufactured on flexible materials (such as plastic) to slash weight and cost. OLEDs are a lot faster than CLCDs and support the high refresh rates needed for high-quality video. CLCDs, on the other hand, support refresh rates needed for smooth text scrolling but will not be capable of video speed for a while (if ever). (See "OLEDs And CLCDs Each Seek Their Niche," p. 84.)
While the OLED and CLCD technologies are both progressing rapidly, the OLED is further along the development curve, with prototype display subsystems already being devised and tested. But if these displays are to reach consumers anytime soon, new driver-IC designs must be perfected. Voltage drivers borrowed from LCDs provide only an interim answer. OLEDs and CLCDs must have driver-IC solutions distinctly tailored to their unique characteristics. Only then can these displays be fully evaluated so that they can go into mass manufacturing.
Being passive-matrix displays, OLEDs are the most cost-effective to manufacture. But since they are row-multiplexed, they place significant demands on the edge drive electronics. With row multiplexing, each row in the panel display is activated sequentially. While the row is activated, its individual pixels may be illuminated. Because an OLED is a variant of a light-emitting diode, it's a luminescent semiconductor and needs a current-driver solution. At every instant, the brightness of each pixel is directly proportional to the current flowing through it. Typically, an OLED achieves optimum luminance efficiency and power efficiency at less than 3 V, as shown by its representative luminescent-current-voltage (LIV) characteristics (Fig. 1).
OLED column drivers re-quire a selectable current magnitude, ranging from 10 µA to more than 1 mA and a 6-bit duty-cycle control. Each column connects to a column of the OLED to adjust its brightness, or luminance. Each current pulse delivered by the column driver is programmed for both magnitude and length. There are four current-magnitude settings: monochrome, red (R), green (G), and blue (B). The duty cycle sets the width of the current pulse from zero to 100% of the row scan time.
OLED row drivers are much less complex. The function of each row driver is simply to complete the path from the OLED to ground, sinking the current from the active columns in a row. Essentially, it must sink the total current through all of the active pixels in any given row. So, the challenge is to make sure that the row driver has sufficiently low impedance, typically 20 Ω.
Depending upon the resolution and the size of the OLED display, a row length may range from a few hundred to over a thousand pixels long, driven by a cascade of column drivers. For the largest arrays, split-row scanning can be employed, with cascaded column drivers on the top and bottom, each driving half the pixels. In small displays, a single row driver is attached to the end of the matrix (Fig. 2).
For large displays, row drivers may be configured in parallel on opposite sides of the array. This ensures the row driver behaves as a switch and doesn't develop a high IR drop that would have to be contained by the column-driver voltage headroom. Drivers such as the Clare MXED102 column driver and the Clare MXED202 row driver are built on 30-V silicon processes. So, they can accommodate a large range of OLED ON voltages; the IR drops that accompany high-magnitude, pulsed-current operation; and the cascading of array panels.
An OLED driver IC is quite a mixed-signal challenge. Its digital-logic circuits convert the pixel data into magnitude- and width-controlled current pulses. But analog circuitry actually delivers current to the display. The OLED's relatively high parasitic capacitance, on the order of 20 to 30 pF, must be charged and discharged rapidly for the driver to sustain a reasonable frame-refresh rate.
OLED displays are fabricated by first depositing a transparent, indium-tin-oxide anode electrode on a plastic substrate. A very thin layer of organic material, typically less than 0.1 µm thick, is then added. Next comes a cathode electrode made of matching OLED material. It aligns with the anode electrode to define the OLED's typical active area of approximately 250 by 250 µm. This display structure also behaves as a parallel-plate capacitor with 250- by 250-µm plates, spaced less than 0.1 µm apart.
To raise the selected row pixel to its conductive voltage (i.e., the ON threshold of the OLED), the entire capacitive load must be charged. If this entire capacitance had to be precharged by the precision ON current, which may be less than 10 µA, the time to reach the conduction voltage would be too long. The response time would be slow and depend on the specific size of the array. Plus, in large arrays, the charge time might even approach the entire row time interval.
Therefore, a portion of the scan cycle is reserved to pre-charge the columns to a voltage threshold that's just below the onset of conduction. When the precision ON current is applied, it flows instantaneously through the OLED rather than the capacitance of the array. The trick is to have the driver IC precharge the array to just the right forward voltage before allowing the current source to take over. If this occurs properly, the OLED light outputs are are directly proportional to the current magnitude and duty cycle.
Precharging is more complex for color OLED displays than monochromatic versions. Red, green, and blue OLEDs have different intrinsic forward voltages as well as different photometric efficiencies (drive current required for a human to perceive the light). To achieve the desired relative brightness and hue, each color (R, G, B) commonly requires its own drive current range and individual precharge voltages. With today's materials, current levels for the different-color OLEDs may have to differ by several multiples for best appearance.
Figure 3 shows the row-voltage and column-current waveforms for a multiplexed OLED display. Relative to the Active, Reset, Precharge cycle at the top of the figure, each active cycle produces different column pulse widths. Wider current pulses mean brighter pixels.
OLED column drivers must perform within 3% of their programmed current output with respect to their adjacent drivers. Doing so guarantees uniform brightness among pixels. Otherwise, the quality of the image will suffer.
The traditional approach for achieving tight matching is to place the individual transistors close to each other. But this approach is unrealistic for an OLED column-driver IC supporting several hundred or more column drivers, because the current-source transistors are spread across the entire chip.
Techniques such as degeneration offer the best option for guaranteeing all the transistors perform within the same tight, 3% tolerance. Resistive degeneration in the transistor circuitry reduces the effects of small VGS changes across the chip, thereby improving the match of each column output. It's also effective when resistor matching is tighter than device thresholds.
To further enhance matching, other methods are em-ployed to monitor current-source differences and to dynamically balance currents. For example, Clare's MXED102 column-driver IC offers parameters (accessible by the display user) that may be used in algorithms to modify current magnitude and duty cycle.
With the CLCD, the transmissive (or transparent) mode is a property of the focal conic state of its material. While in this phase, the CLCD passes a broad, white-light spectrum. In the reflective mode, the CLCD reflects a certain wavelength but transmits all others, a property of its planar state. Individual pixel material is prepared to reflect red, green, or blue light.
The typical structure of a full-color CLCD panel comprises three sandwiched layers of the cholesteric liquid crystal material—each one reflective of red, green, or blue. Each layer has a number of transparent row electrodes on one side and transparent column electrodes on the other. They access each pixel either in a matrix, row by column—or in an x-y array. The backing is an absorbent black material.
In operation, the information to be displayed determines, for each pixel, a particular degree of color reflectivity versus white-light transmissivity. Light incident on the CLCD panel is either reflected by a preselected degree at each layer or passed through to be absorbed by the backing material.
Like conventional twisted-nematic LCDs, CLCDs are voltage-driven, electric-field-controlled devices. When they are driven with low-cost, passive-matrix (edge-driven) electronics, specific waveforms must be generated at their row and column electrodes. Doing so ensures the selected pixel is written accurately, while unselected pixels remain unaffected. A CLCD driver must deliver upwards of seven distinct, high-ranging voltage levels in a precisely timed (selectively transient) sequence. This distinguishes it from commercial STN-LCD drivers, which may supply just two or three low-magnitude voltages while scanning the array continuously.
Figure 4 illustrates a CLCD panel as an array, along with its parasitic capacitance. The net differential voltage across an element equals the difference between the row and column electrode voltages at any given instant. The array is written selectively by accessing a row at a time. Rows need not be accessed if the image information hasn't changed.
Selected row pixels are written to be either transmissive or reflective, or varying degrees in between. Unselected row pixels must not be written or altered. The cholesteric writing phase's voltage/time trajectory determines the stable state.
The CLCD driver must provide a four-step, time-voltage sequence to write a cholesteric pixel into a new re-flective/transparent condition (Fig. 5):
- To clear/prepare or reset the pixel, the highest voltage is applied first. Upwards of 40 V must be supplied symmetrically by the row driver for 10 to 20 ms.
- In the preselect phase, the pixel is conditioned to assume its programmed characteristic. The magnitude and duration of the voltage applied determine the degree of reflectivity versus transmissivity. Preselect programs the differential pixel-voltage waveform characteristic to determine final appearance. So, exact synchronization and timing control is necessary between the column driver and the row driver during preselect.
- The instantiation phase involves the application of an intermediate-level pixel voltage that drives the progression through the preselected state transitions in the CLCD material.
- Finally, the voltage is removed from the pixel, and its preselected optical characteristic becomes visible.
Note that the various row driver voltages develop the net pixel voltages (VCLEAR, VREFLECT, VTRANSMIT, VFIX) differentially, in conjunction with the application of the column driver voltage (VCOL). The voltages must be impressed with time certainty. Such a task is complicated by the parasitics of the array and the concomitant time constants encountered at remote pixel locations.
The total voltage differential can approach 100 V, which must be applied to all pixels in the array with the same time integrity. This requires compensating for the CLCD array's parasitic capacitance and interconnect resistances. Not only must the driver swing to a high voltage, it must provide sufficient current to charge all the pixel parasitic capacitance in a given row or column, in the prescribed time. The CLCD driver must also overcome the resistances found in the display itself.
Each state of CLCD material is sensitive to the magnitude of the electric fields imposed. As a result, the voltage applied to a pixel must be controlled within a few percent. Care must also be given to symmetry and the magnitudes of the less critical voltages. Then, no accidental writing or crosstalk will occur in the other pixels.
Liquid crystals are sensitive to the magnitude of the electric field, the duration of the field, and its rate of change. A thicker layer of material requires a higher driver voltage to produce a given field strength (in units of V/µm). Higher fields (voltages) may be used for shorter intervals. This induces similar effects as lower intensity fields are applied for a longer period or switched at a higher rate. Drivers must provide a degree of programmability so the driver patterns can be optimized for a particular display application.
Power is an issue for all displays, especially for portable, battery-powered devices and nonvolatile, static display applications. Since CLCDs don't call for any keep-alive power, their companion display electronics must be capable of complete power-down.
Another concern is cost. CLCD material promises to be inexpensive to manufacture, making it well-suited for inexpensive (plastic) substrates and encapsulation. However, the driver requirements exceed those of today's high-volume LCD drivers. For this reason, the drivers account for a higher percentage of the display module's cost.
Packaging can appreciably affect the cost and size of the finished driver solution. To reduce costs, driver manufacturers try to make the die as small as possible and the pixel pitch as tight as 60 µm. Tape-carrier packages (TCPs), which can accommodate an output pitch of 100 to 110 µm, allow driver-IC manufacturers to straddle the difference between the smaller-pitch ICs and the larger-pitch OLED or CLCD panels themselves. The TCP acts as a lead spreader, aligning the driver channels with the display pixels.
With a working set of current-driven row and column driver chips in hand, OLED and CLCD display manufacturers will be able to fabricate highly realistic prototype systems. In turn, this will promote a more productive prototype and verification environment. Therefore, the response and quality of OLED and CLCD display subsystems can be tested and evaluated against a variety of conditions and environments.