There will be 107.6 million smartphone AP (application processor) shipped in the China market in the third quarter of 2014 and 113.5 million units in the fourth one, increasing on quarter by 4.5% and 5.5% respectively mainly due to increasing demand for LTE (4G) smartphones, according to Digitimes Research.Taiwan-based MediaTek will be the largest vendor, followed by Qualcomm, Spreadtrum Communications, HiSilicon Technologies, Leadcore Technology, Marvell and Nvidia, Digitimes Research indicated.

Taiwan-based makers shipped 67.518 million large-size (9-inch and above) TFT-LCD panels in the second quarter of 2014, increasing by 8.5% on quarter and by 0.9% on year, according to Digitimes Research.Most of the major application saw double-digit growth in shipmnts in the second-quarter, Digitimes Research indicated.Global Ultra HD TV panel shipments grew 154% sequentially in the second quarter. The top-two suppliers, Taiwan-based Innolux and Korea-based LG Display, both shipped about 1.5 million UHD TV panels during the quarter.

Japan's Ministry of Economy, Trade and Industry has been boosting the development of service robots, with a target production value of JPY2.65 trillion (US$26.1 billion) and target production of 9.4 million robots specifically for use in health and medical care in 2025, according to Digitimes Research.The robots will meet demand in Japan where the birth rate remains low and the elderly population is increasing, which is expected to result in a lack of personnel in the nursing segment.Robots are currently used mainly in Japan's industrial segment but by 2025 the service market will surpass it to represent a JPY2.65 trillion market compared to the JPY1.58 trillion of the industrial segment, according to statistics from Japan's New Energy and Industrial Technology Development Organization.The robots will primarily be used for for logistics, security, health and medical care, added Digitimes Research.

Since the implementation of China's 12th Five Year Plan in 2011, the country's semiconductor industry has shown a steady yearly revenue growth from CNY179.61 billion (US$29 billion) in 2010 to CNY240.8 billion in 2013, representing a compound annual growth rate (CAGR) of 10.2%, thanks to a steep rise in sales of entry-level and mid-range smartphones, in addition to the continued expansion in foundry and assembly house production capacity.This growth rate compares favorably with the global semiconductor industry annual growth rate of less than 5% each year from 2011 to 2013, with 2012 even registering a negative growth rate of -2.7%, due to lower-than-expected end market demand and inventory adjustments.Based on the assumption that 2014 will see better global economic prospects than 2013 and the shipment of mobile devices including entry-level and mid-range smartphones and tablets will be able to maintain a momentum of double-digit percentage growth, Digitimes Research estimates that revenues of the China semiconductor industry are likely to increase to CNY267.37 billion in 2014, an annual growth rate of 11% from 2013 and a compound annual growth rate of 10.5% from 2010, which far surpasses the CAGR of 1.7% of the global semiconductor industry in the same period.However, the goal of CNY350 billion for the China semiconductor industry in 2015, laid out by China's State Council in the "National Semiconductor Industry Development Guidelines" promulgated on June 24, 2014, may be out of reach.In response to this challenge, the State Council published guidelines to strengthen government support for China's semiconductor industry, including expanding tax benefits mentioned in the State Council Document 4 (2011) for IC design houses and foundries to testing firms. The central government also plans to set up a national industry investment fund of CNY120 billion. The fund will mainly focus on investing in the construction of advanced process capacity, semiconductor firm reorganization, and mergers.About DIGITIMES ResearchDIGITIMES Research is the research arm of DIGITIMES Inc., Taiwan's leading high-tech media outlet. Operating as an independent business unit, DIGITIMES Research focuses on monitoring key high-tech industries, while also guiding clients toward suitable new businesses. Digitimes provides market intelligence and analysis to more than 1000 corporate customers worldwide. Research and consulting services including a full range of products, from in-depth Special Reports on industry trends in the flat panel display (FPD), LED, and semiconductor industries, to Tracker services that monitor the global mobile device supply chain on a quarterly basis.

Xiaomi Technology ranked first in smartphone shipments in the China market among China-based vendors in the first half of 2014, but will see growth in shipments gradually slow down mainly because other vendors started using online marketing in the second quarter and have launched or will launch models with high performance-price ratios for sale at CNY799-999 (US$129-161), according to Digitimes Research.Hongmi and Xiaomi smartphones both use MIUI operating system, so consumers feel no difference in terms of usage experience and prefer Hongmi, which is cheaper, as a result. This will cause an imbalance in Xiaomi's high-end and entry-level smartphone shipments.With other China-based smartphone vendors also joining the online channel competition in the second quarter and pricing their models at CNY799-999, Xiaomi will maintain its shipments in the short terms because of its strong brand recognition, but will see its shipments weakening in the long term as their products all share similar specificitons, added Digitimes Research.

Following its aggressive investments in 8.5G production lines, China Star Optoelectronics Technology (CSOT) has become the largest supplier of 32-inch panels worldwide. The company has also established a joint venture with the government of Wuhan, China, through which CSOT will officially enter the small-to-medium-size panel application market using the joint venture's 6G LTPS/AMOLED production lines.CSOT has already begun trial production of AMOLED panels, and by 2016 is expected to start mass producing the technology. The company's AMOLED production facilities will contribute to a total of 10 overall AMOLED production lines in China that will go into production or have already begun production from 2014-2016.China panel makers continue to receive financial assistance both from the central government in China as well as from many local governments to develop AMOLED panel facilities. The makers are also receiving various bank loans, subsidies and plentiful access to state property in China, and are making efforts to use talent from Taiwan and Korea to develop the AMOLED panel industry.Despite China makers' efforts to establish AMOLED facilities through 2016, Digitimes Research believes they will face high production costs and low yields, making the technology less competitive in terms of pricing compared to TFT LCD technology, and limiting the influence China makers will have in the segment in the short term.

There will be an estimated demand for 12.44 million mm of 2-inch-equivalent sapphires for optical purposes in 2014, mainly consisting of 54.3% for use in smartphone button covers, 24.1% iWatch covers and 17.4% smartphone camera lens covers, according to Digitimes Research.The demand will increase in 2015 to 26.2 million mm, with iWatch holding the biggest proportion at 57.2% followed by smartphone button covers at 30.1%.Sapphire for Apple's iWatch is expected to primarily come from GTAT while 2-inch units for use in optical purposes will come from VHGF technology. Additionally, demand for advanced sapphire furnaces needed to produce iWatch applications is expected to increase from 67 units in 2014 to 333 in 2015, added Digitimes Research.

There were 55.06 million tablets shipped globally in the second quarter of 2014, decreasing 4.5% on quarter but increasing 17.9% on year, according to Digitimes Research.The shipments consisted of 14.1 million iPads, down 10% on quarter, and 18.96 million units launched by vendors other than Apple, down 12.7% on quarter. Additionaly, 22.3 million white-box units were shipped in the second quarter.Shipments of small-size Wi-Fi-enabled units in particular slowed down in the second quarter and the time period was also a slow season for shipments. Supply chains also faced yield issues and Samsung saw less-than-expected shipments for its 8-inch tablets. Tablets sized 10-inch and above have seen shipment increases since fourth-quarter 2014.Taiwan tablet makers meanwhile surpassed 20 million in shipments for brand tablets during the second quarter, which made up 60% of overall brand tablet shipments during the time period, added Digitimes Research.

Connecting an application processor to a DRAM chip through a 3200 Mbps LPDDR4 interface is not any easier than routing a 2600 MHz 4G LTE antenna. While RF front ends enjoy ceramic packages and careful routing in electromagnetic proof modules, digital signals flow through ball grid array packages and small dense printed circuit boards (PCB) making them more liable to high frequency effects.As data rates increase in the gigabit range and beyond, PCB traces can no longer be treated as simple conductors. The parasitic resistance, capacitance and inductance of the copper trace make it act as a transmission line producing all kinds of high frequency effects not commonly considered in digital design. For example, high frequency components of the signal suffer more attenuation due to skin effect than the lower frequency components causing signal deformation. The inductance and capacitance between parallel copper traces result in crosstalk and the high switching currents lead to ground bounces. Bit error rate (BER) increases as more ones become interpreted as zeros and vice versa. Hence, transmission line effects of PCB traces as well as the frequency response of packages, connecters and cables need to be thoroughly analyzed to ensure signal integrity. Accurate SPICE analysis at the PCB level can save time and cost by reducing numerous iterations of PCB prototyping and measurement.Figure 1: A typical system configuration for signal integrity analysisFigure 1 shows a memory interface, a typical example of multi-Gigabit inter-chip communication. The same concept applies to high speed serial IO such as USB 3.0 and HDMI as well as multi-Gigabit Ethernet devices. The communication channel consists of IO models for the chips, Scattering Parameters (S-parameters) models for the packages, connectors and cables, and lossy coupled transmission line models for the PCB traces. IO models are provided by the chip vendors. Simple IO buffers can be accurately represented by IBIS models. More complex IO circuitry with active pre-emphasis and equalization are usually available in the form of encrypted transistor-level HSPICE netlists, or as IBIS-AMI models derived from the transistor-level representation. As the golden reference for transistor-level simulation, HSPICE uses foundry-certified transistor models to deliver the most accurate behavior of IO circuits. Moreover, most chip vendors use HSPICE to validate their IBIS and IBIS-AMI models. Hence using HSPICE at the board-level gives the best correlation with the chip vendor's intent. When it comes to IBIS-AMI, HSPICE has the unique capability of simulating these models in true transient mode in addition to bit by bit and statistical eye diagram modes.Lossy coupled transmission lines for PCB traces can be extracted in different ways, the simplest of which is using HSPICE W-elements. The W-element reads in the PCB properties and the dimensions of the parallel traces, and then uses a built-in 2D solver to extract the transmission line response. The model accurately represents frequency-dependent loss and coupling, has no limit on the number of coupled lines and ensures the passivity and causality of the system. Most PCB layout tools can extract the trace geometries and automatically generate W-element models in the HSPICE netlist. Third party quasi-static 2.5D field solvers can also be used to generate broadband models of PCB traces. Depending on the field solver, these models can be inserted into the W-element in the form of RLGC tables. For critical layouts, full-wave solvers can be used to extract the frequency response of the PCB traces in the form of S-parameters which can also be used as input to the W-element.The S-parameter models for the packages, connectors and cables are provided by the component vendor, measured by a network analyzer or extracted using a 3D electromagnetic field solver. In either case, the S-parameter model gives a robust linear representation of the component taking into account its distributed nature and any arbitrary frequency dependent behavior. Examining S-parameters on a Smith Chart gives deep insight of such distributed systems beyond what can be obtained using lumped elements in a circuit schematic. There is one challenge though. S-parameters are frequency domain models originally invented for RF and microwave devices, while signal integrity analysis of digital multi-Gigabit systems is predominantly performed in the time domain. The HSPICE S-elements overcomes this challenge using state-of-the-art automatic rational function model generation. Furthermore, HSPICE deploys multi-delay enhanced rational function models to capture the complex high frequency behavior of long data cables with several meters in length (e.g. HDMI cables). HSPICE uses parallel computing technology available in modern processors to simulate large - over 500 ports - S-parameter models with superior speed and accuracy. In addition to the use of s-parameters in circuit simulations, HSPICE also supports multiport linear network analysis (.LIN) which allows S-parameter extraction from arbitrary circuit types.Figure 2: Working with S-parameters is more than running transient analysisThe flexibility and robustness of s-parameter modeling are sometimes compromised by the lack of quality of some s-parameter models. Low-quality S-parameter models are likely to end up with poor simulation results. Quality issues include - among other causes - passivity violation, coarse frequency sampling and narrow frequency bandwidth. For example, the starting frequency of an S-parameter model can be too high to capture low-frequency transient behavior, or the end frequency can be too low to capture the high frequency components of digital transitions. HSPICE includes a standalone S-parameter utility (SPUTIL) which can manipulate S-parameters in different ways to ensure the quality of S-parameter modeling. Figure 2 shows how HSPICE combines S-parameter extraction using multi-port linear network analysis (.LIN), S-parameter simulation using transient analysis (.TRAN) and S-parameter quality assurance using the S-parameter utility (SPUTIL). Figure 3: Checking impedance matching using HSPICE S-parameter utility (SPUTIL)SPUTIL provides a number of convenient s-parameter manipulation techniques such as merging multiple data files, passivity checking or enforcement, re-sampling with flexible frequency point specifications, and file format conversion. For example, impedance matching is an important requirement of high speed channel design. The quickest way to test impedance matching is to observe S11 with different reference impedances. SPUTIL offers a convenient way to convert the reference impedance of a given s-parameter set using a simple script, as shown in Figure 3. Then observing the resulting S11 plot on a Smith chart and finding the smallest S11 value will give a good starting point for designing the channel termination impedance.This concludes our discussion of the different components and models used in signal integrity analysis. Now we move to the discussion of the different analysis techniques. Eye diagram analysis is widely used for high speed communication channel evaluation. The eye diagram overlays the unit intervals of a long digital bit sequence into a compact form that readily gives picosecond-level observations of the system. Generating an eye diagram of the target system under test, examining the eye opening and measuring BER as accumulated probability are the key aspects of channel compliance testing.HSPICE offers different techniques to analyze eye diagrams at different levels of simulation speed and accuracy. BER evaluation of multi-Gigabit systems requires the analysis of millions of unit intervals. Transient analysis of such long bit streams takes hours and produces huge data files. Thousands of simulation may be needed to cover the design space for channel optimization. Bit-by-bit and statistical eye diagram generation in HSPICE significantly improves the productivity of channel design by reducing simulation time from hours to seconds.Figure 4: HSPICE statistical eye diagram analysisHSPICE state-of-the-art statistical eye diagram analysis is applicable to all types of channels and models. HSPICE uses accurate transient analysis to calculate the impulse response of a number of small bit patterns, and then uses statistical methods to quickly generate the eye diagram as a probability density function (PDF) map that takes all the possible bit patterns into account as shown in Figure 4. HSPICE automatically extracts the jitter profiles by observing the vertical and horizontal cross sections of the eye diagram. HSPICE also generates the bit error rate (BER) map based on the PDF eye diagram. Then similarly taking cross sectional views of the BER, bathtub curves can be examined. HSPICE also captures the shortest necessary bit pattern which reproduces the inner-most - or in other terms the worst-case - eye fragment at a given time point. Using this worst bit pattern in subsequent short transient analysis, one can analyze the causes of eye closure and improve the design. HSPICE can extract time domain waveforms at particular bit positions, a useful technique for qualifying adaptive equalizer designs.Bit-by-bit eye diagram analysis uses a fast transient technique to generate eye diagrams for specific bit patterns at a fraction of the time needed for transient analysis. Bit-by-bit and statistical eye diagram techniques are very useful for channel design. For signoff, transient analysis offers the most accurate eye diagrams. HSPICE supports all types of IO models - including algorithmic ones (IBIS-AMI) - and frequency-dependent elements in time domain transient analysis. Parallel computing technology is used to speed up the simulation of extremely long bit sequences without compromising accuracy.Figure 5: combining AMI and transistor-level models in statistical, bit-by-bit and transient eye diagram analysisHSPICE has the unique ability to mix and match analysis technique and model types to best fit every stage of channel design and compliance testing. As shown in Figure 5, HSPICE treats the channel as a complete black box giving the user the option to include any combination of active and passive devices. HSPICE also allows for different transmitter and receiver representations. For example the transmitter can be a transistor level IO circuit while the receiver is an IBIS-AMI model and vice versa. In early design stages, one can use fast statistical eye diagram analysis to evaluate pattern independent transmitter emphasis. For such analysis, IBIS-AMI models are used in the transmitter side only keeping the receiver as an idealized receiver termination. Then as design evolves, one can use bit-by-bit simulation replacing the idealized receiver with an algorithmic model in order to test how well the adaptive equalizer adjusts its parameters to achieve maximum eye openings. Then by switching the analysis from bit-by-bit mode to full transient, one can capture any nonlinear effects that may happen to be in the channel. At the final verification stage, most likely full transistor representations of both transmitter and receiver buffers will be used. HSPICE has the flexibility to run all these analysis in the same testbench.About the authors:Hany Elhak, Product Marketing Manager, Synopsys, Inc. Elhak has more than 10 years of EDA experience spanning both technical and marketing responsibilities. Prior to EDA, Hany worked as RF designer, designing RF ICs for cellular and wireless networking standards. Hany holds B.S. and M.S. degrees in Electrical Engineering from Ain Shams University, Cairo and MBA from UC Berkeley, Haas School of Business.Ted Mido, R&D Engineer, Synopsys, Inc. Mido is a senior staff member of HSPICE R&D group. He received the BS, MS and PhD. degrees in electrical and electronic engineering from the University of Tokyo, Japan. His researches and developments have focused on analyzing high speed signal propagation systems in analog/mixed-signal ICs, high speed chip packages and printed circuit boards. He has published numerous technical papers on interconnect analysis and parasitic extraction.Hany ElhakProduct Marketing Manager, Synopsys, Inc.Ted MidoR&D Engineer, Synopsys, Inc.

There were 178 million handsets shipped in the China market during January-May 2014, decreasing over 70 million units on year. The decrease was mainly because China Mobile shifted marketing resources to promote 4G TD-LTE smartphones and reduced subsidies for 2G and TD-SCDMA (3G) handset purchases, according to Digitimes Research.The reduced subsidies were a factor in the 77% decline of 2G handsets in China during the time period. Additionally, China Telecom and China Unicom had not yet started 4G commercial operation due to licensing issues, which further pushed down shipments.However, China's Ministry of Industry and Information Technology issued two FD-LTE pilot operation licenses to China Unicom and China Telecom on June 27, 2014 for several cities in China, which will push 4G developments in China as well as increase handset shipments, added Digitimes Research.Meanwhile, the China Internet Network Information Center (CNNIC) has announced that mobile Internet users in China reached 527 million as of June 2014.