This Product How-To article discusses use of antifuse nonvolatile memory in display driver ICs (DDI) and touch sensor controllers and how proprietary technology from Kilopass can be used to make NVM an integral part of a system-on-chip design.
Antifuse NVM IPs have been used in display driver ICs (DDI) to store critical parameters for analog and image processing circuitry to calibrate out chip-to-chip variations and to meet different requirements of the system integrators.
Uniform display quality depends on the characteristics of panels and accuracy of driving voltage or current levels, and any tolerance must be compensated for. Antifuse NVMs are perfectly suited for the task of permanently recording the following:
* reference parameters for power management and analog circuits,
* calibration information for timing elements and analog-digital converters,
* brightness and color correction tables,
* initialization data for organic materials, as well as
* configuration selection for different features.
Once antifuse NVM programming has been performed in the final stage of system testing, the display module becomes a tuned unit, eliminating the need for additional software configurations. This allows an automated system assembly and ensures reduced production costs.
In addition to pervasive uses in trimming and yield enhancements, another popular use of antifuse NVM is for boot and firmware code storage. Traditionally, microprocessors for display and touch sensor controllers have to employ externally attached EEPROM chips or embedded flash blocks to store control code with the anticipation of frequent updates.
For some designs, however, as development matures, the number of expected code upgrades has decreased to around 5 to 10 times in the life span of the end product. High-density embedded antifuse NVM offering multi-time programmability is a cost-effective alternative to expensive flash-based counterparts.
To satisfy the industry demand for simplified system designs, future trends point to the consolidation of touch screen and display panels and their respective controller ICs. Embedded antifuse NVM plays an integral role in supporting on-chip storage of calibration data to better match the characteristics of individual panel and high-density software code for the touch screen controller.
The basics of antifuse NVM
Antifuse NVM bit cell contains a program transistor and relies on its gate oxide breakdown to store a logic 1. The oxide breakdown is achieved by applying a precise voltage to the gate of the program transistor that creates a low resistive path between the gate and source, as opposed to a capacitor before the breakdown.
This irreversible physical procedure guarantees outstanding reliability in data retention compared to other technologies that depend on trapped charges. Aside from a small footprint, both bit cells and peripheral circuits are in standard logic process, requiring no additional manufacturing and backend process steps; they are also portable to different foundries and advanced process nodes.
An optional built-in charge pump to generate the necessary voltage required for programming adds flexibility to logistics management because programming can take place at the foundry, in test houses, or in end user systems.
Future directions of DDI
Conventional touch-display system shown in Figure 1 below consists of a display panel, an overlay touch panel, a polarizer and a glass layer. Discrete touch sensor controller and DDI are individually mounted to their respective panels.
Figure 1: Comparison between Conventional & Integrated Touch Display Modules
Lately, as touch screen technology becomes mainstream in the handheld communications market, many display module makers are finding ways to further compact the module profile and lower material and assembly costs by merging the touch-screen panel with the display panel.
As a result, it calls for a unification of touch sensor controller and DDI chips. The integrated solution promises to offer improved display quality, faster time-to-market, lower power consumption, reduced component count, and greater inventory savings.
As system functions are being actively consolidated on single chips, embedded antifuse NVM technology is expanding in density to serve different purposes in SoCs.
In addition to optimizing existing designs to keep driving down costs and improving profit margins, designers are also aggressively exploring advanced deep-submicron fabrication processes to remain competitive in the global market.
Antifuse NVM offers excellent coverage in foundries and process nodes, and with its capacity and reliability, it is a future proof embedded NVM solution for DDI manufacturers.
Table 1: XPM vs. Gusto in 130nm Process
Two antifuse NVM from Kilopass’ IP product portfolio are particularly well suited to the Display Driver IC application and its future implementation: XPM and Gusto.
The XPM family with capacities ranging from 1kb to 128kb is used to store calibration data and security codes.
The Gusto family with capacities ranging from 256kb to 4Mb is idea for containing program code. Gusto has improved peripheral designs that greatly reduce area, access time, and power consumption. Most display system ICs are in 180 to 90nm process nodes. Table 1 above compares XPM and Gusto in a baseline 130nm process.
XPM for high voltages designs
DDI users with a diverse product line for different panel technologies, resolutions and sizes have used NVM primarily for Extended Display Identification Data (EDID), chip configuration, and display adjustment.
Common densities range from a few kilobits for trimming and EDID to one megabit for large display calibration. Figure 2 below shows a typical block diagram of a DDI SoC. Many foundries offer different types of high voltage devices for TFT-LCD, LTPS, CSTN, and OLED display drivers.
XPM can take advantages of the availability of high voltage sources for programming and be custom ported to these specialty processes with uncompromised reliability.
Figure 2: Use of NVM in DDI
DDI manufacturers are migrating to more advanced process nodes in order to integrate denser frame buffers on chip to support larger displays. At the same time, the NVM has to be able to address the following three critical concerns:
– High in-field programming yield due to high panel cost
– Minimal metal layer stacking and macro height for better routability
– Low standby power consumption for mobile applications
XPM can address these key factors by offering built-in error correction algorithms to improve production yield, custom macro layout to meet the special aspect ratio requirement, and ultra-low standby current suitable for power sensitive applications.
Gusto for touch screens
Touch screen has become the preferred user interface for a wide range of mobile applications such as cellular phones, tablet PCs, and handheld game consoles.
There is a variety of touch sensing technologies with different performance and cost profiles. Though analog resistive technology has demonstrated high resolution and low cost, therefore achieving the widest market acceptance, projected capacitive touch screens have been rapidly adopted in high-end mobile devices due to its improved transmissivity, sensitivity and durability.
In addition to a diversity of touch screen technologies, different configurations of touch sensors like in-cell, out-cell, and on-cell placements also add complexity to the design of touch sensor controllers, which, with the assistance of built-in Microcontroller Unit (MCU) and advanced power management system, detect and filter touch coordinates, compensate for parasitic capacitances, and communicate with the host controller. Complex control software, or touch screen driver, compiled on specific MCU, is stored in NVM.
Figure 3: Use of NVM in touch screen controllers
To fulfill the requirement of on-chip firmware code storage, an embedded NVM needs to support the following high-end features:
– Large density and code efficiency: Typical code size ranges from 16KB to 64KB depending on algorithms and configurations.
– Execute-in-Place (XIP) capability: To internalize code storage and remove shadow DRAMs, the software can run directly from NVM.
– Few-times-programmability: This allows the opportunity to upgrade the software in system should manufacturing or operating conditions change.
Gusto offers a maximum storage capacity of 512KB and an array of built-in features to ensure small macro area, fast access speed, and efficient error corrections.
For common firmware code sizes, a properly configured Gusto can support 8 to 30 updates in a single instance. As production ramps up and the firmware becomes fixed, Kilopass’ ROM technology can easily convert Gusto to ROM to help customers save on testing cost.
From storing calibration data and cryptographic keys to high-density firmware code for microprocessors, antifuse NVM products are not limited by fabrication technology scaling and deliver to SoC designers the promise of faster turnaround, larger density, smaller area, and higher performance.
Edward Cheng is the Field Marketing and Applications Manager at Kilopass, Inc. Prior to Kilopass, Edward was the Senior Staff Applications Engineer at Telegent Systems, supporting customers and teams on areas of software integration, RF hardware design, and system bring-up and debug.