5G移动通信的3D IC与RF SiPs技术

需积分: 10 83 浏览量更新于2024-07-18 收藏 11.49MB PDF 举报

"3D IC and RF SiPs是现代无线通信技术中的关键组成部分，特别是在5G移动通信系统中。3D集成电路（3D IC）和射频系统级封装（RF SiP）是实现高性能、小型化及能效优化的关键技术。这本书由Lih-Tyng Hwang教授和Tzyy-Sheng Jason Horng教授撰写，探讨了这些先进技术在5G移动通信领域的应用和解决方案。" 3D集成电路（3D IC）是一种创新的半导体制造技术，它通过垂直堆叠多个芯片层并直接通过硅通孔（Through-Silicon Via, TSV）进行互连，显著提高了集成度和性能。这种技术允许不同功能的芯片在同一封装内紧密集成，减少了延迟和功耗，同时提高了系统的整体效率。在5G通信中，3D IC有助于处理复杂的射频信号处理任务，如多载波调制解调、高速数据传输和高效的功率管理。射频系统级封装（RF SiP）则是将多种射频组件，如功率放大器、混频器、滤波器和天线等，集成在一个单一的封装内。这种方法降低了系统尺寸，简化了设计流程，并提高了系统的性能。在5G移动设备中，RF SiP技术对于满足严格的带宽、频率范围和功率要求至关重要。它还支持多个频段和标准，适应5G网络的频谱多样性和动态分配。 5G移动通信的先进堆叠和平面解决方案主要集中在如何通过3D IC和RF SiP技术提高数据速率、减少延迟、增强连接稳定性和节省能源。书中可能涵盖了这些技术的最新进展，包括新型材料、微电子制造工艺、热管理和信号完整性优化等主题。此外，作者可能会讨论如何通过这些技术解决射频前端的挑战，例如干扰抑制、射频带宽扩展以及与毫米波通信的兼容性问题。 "3D IC and RF SiPs" 是对现代无线通信领域的重要贡献，它为工程师、研究人员和行业专业人员提供了深入理解这些关键技术及其在5G移动通信应用的宝贵资源。通过学习和应用书中的理论和实践，读者可以更好地应对日益复杂和需求增长的无线通信环境。

Preface

xvii

In IBM’s MLC packaging of semiconductor ICs, even though the manufacturing tem-

peratures had been lowered from 1,600 °C in HTCC to 900 °C in LTCC, it was obviously

not enough. The high manufacturing temperatures meant 1) high operating costs, due

to the thermal cycles, 2) the materials set are more expensive, and 3) most importantly,

the manufacturing cycle times are longer, due to the temperature heating and cooling.

That is, the high manufacturing temperature meant high costs to the vendors. In main-

frame market where margins were high, the high manufacturing temperature technol-

ogy may be able to satisfy the business needs; but, in the long run, a lower cost

replacement must be found for high volume products. At the same time in the early

1990s, being in wireless and mobile businesses, Motorola needed manufacturing tech-

nology that could do volume production. Licensing C4 technology from IBM was one

of their strategies.

From the form factor point of view, adopting C4 technology was correct: space saving

from using C4, compared to that using long wirebonds, was significant. Motorola

wanted to apply the C4 on a lower manufacturing temperature organic boards (multi‐

layer organic, MLO). Almost simultaneously, Motorola was working on low‐cost

organic ball grid array (BGA) package that used wirebond as the Level 1 interconnec-

tion. From cost point of view, adopting MLO was logical (lower temperature meant cost

saving), but, not technical. There were at least two reasons: the electrical characteristics

of MLC was superior to that of MLO. For example, the dissipation factor of FR4 (a

MLO) was in percentages, whereas that of ceramic was at least an order of magnitude

lower. The CTE of MLC can be tailored close to that of silicon (2.6–3 ppm/°C), whereas

that of FR4 was about 17 ppm/°C.

FigureP.2 shows a sketch of Motorola’s ball grid array (BGA) package. It was origi-

nally called OMPAC (over‐molded plastic pad array carrier).[14–18] In OMPAC, FR4

MLO was used as the IC substrate. The boldest line in FigureP.2 indicates the active

side of the IC. The overmold was used to protect the IC from the CTE mismatch in

between the silicon and FR4.

OMPAC was under development in the 1980s, with patents issued in 1991–1993, and

originally, they used wirebonds as the Level 1 interconnections, called PBGA (P for

plastic). Licensing C4 from IBM, Motorola was able to develop Level 1 interconnection

using eutectic flip chip bumps in their BGA, called FC‐BGA. Motorola may be first in

PBGA, but, it may not be the first to produce FC‐BGA packages

1418

40 40 51 50 40 32 40

16 16

Overmold

IC substrate (MLO)

BGA balls

FigureP.2 Motorola’s BGA (OMPAC) illustration from US 5,239,198 A.[18] Source: United States Patent

and Trademark Office.

Prefacexviii

C4 technology by IBM, organic BGA invention by Motorola, and later many BGA

variations, such FC‐BGA and CSP that use low temperature (200 °C) eutectic flip chip

bumps, have turned MTM (IC packaging) a mass‐production capable technology. This

benefits handheld market tremendously, since handset production by a major company

could reach several hundreds of million annually. A 2012 report from ITU claims that

the registered phones have reached 6 billion, and in the same year the world population

was 7 billion.[19] In the first quarter of 2013, the mobile phone sales volume was 426

million.[20] Most of the IC packages and components are organic. The decision to

license C4 technology was correct.

P.1.3 The Breakup andtheBrick Phone

Mobility represents a new dimension in human experience. It became reality because of

two events: the breakup of the landline (copper wire based) giant AT&T (legislature

took place between 1982 and 1984) and the development of first analog mobile (a brick

phone demonstrated in 1973, first analog phone in 1983). From the old and new names,

AT&T (the landline Ma Bell) and AT&T wireless, the hand‐over from landline to wire-

less appeared to be logical and direct. In fact, the handover was indirect, unexpected

twist of fate.

Let’s examine a few facts.

The first handheld cell phone was demonstrated by John F. Mitchell and Dr. Martin

Cooper of Motorola in 1973,[1] using a handset weighing around 4.4 pounds (2 kg), a

size of a brick; therefore, the nickname, the brick phone, 9 inches long, 5 inches deep,

and 1.75 inches wide.[2] Ten years later in 1983, the DynaTAC 8000x was the first analog

personal phone to be commercially available. The commercial portable cellular phone

received approval from the U.S. FCC on September 21, 1983. Motorola earned the

National Medal of Technology[21] in 2004 for the innovation that has “enabled portable

and mobile communications to become the standard across society.”

The breakup of the Bell System was mandated on January 8, 1982; effective January 1,

1984, the Bell System’s many member‐companies were variously transformed into

seven independent “Regional Holding Companies,” also known as “Baby Bells.” AT&T

would continue to be a provider of long distance service, while the then independent

Regional Bell Operating Companies (RBOCs) would provide local service, and would

no longer be directly supplied with equipment from AT&T’s manufacturing subsidiary

Western Electric.

AT&T (a 100‐year‐old company then, formed as Bell Telephone Company by Alexander

Graham Bell in 1877, and it became AT&T in 1885) agreed to break up because it has

longed to get into computers and information services beside its core business of long

distance landline service. By giving up the local phone business, AT&T was free to com-

pete in non‐long‐distance areas such as computers. AT&T did not realize the potential

of mobile (wireless) business at the time. That was actually quite reasonable; the mobile

phones in the early 1980s did NOT have roaming capability. AT&T failed the Picturephone

venture, a visual communications device introduced commercially in Pittsburgh in 1970.

AT&T invented the UNIX operating system, and was eager to enter the computer

business by selling the UNIX‐based computers. The 3B2 was the first desktop super‐

microcomputer (1983) with a 32‐bit microprocessor and UNIX. Later, AT&T computer

Preface

xix

systems also bought NCR on September 19, 1991. Today, we did not see any computers,

desktops, servers, or notebooks that bear an AT&T name or logo.

Wireless was not in AT&T corporate’s vision during and even after the break‐up.

However, the mandated breakup may have unintended consequences on the wireless

business; it prepared the Baby Bells for the future wireless business. Two out of three

major wireless companies in the United States today were Baby Bells: Verizon (from

Bell Atlantic) and AT&T Wireless (from SBC and Bell South). SBC Communications

was formed by Southwestern Bell Corporation in 1995. The holding company for T‐

Mobile, the third one, is based in Bonn, Germany, whose subsidiaries operated GSM,

UMTS, and LTE‐based cellular networks in Europe, the United States, Puerto Rico, and

the U.S. Virgin Islands. Baby Bells were free to speed up the wireless development; and

together, they quickly made roaming in the United States a reality. No wonder Motorola’s

Iridium project had failed miserably, a side note.

P.2 Hardware forMobility—The Focus

Let’s use the following functional diagram, FigureP.3, to illustrate the key devices and

components in the mobility enabler, a cellular phone. In one block, there are application

processor (AP) and its main mobile memory. AP, of course, is the brain of a cellular phone.

It is a system on a chip (SoC [22]) based on ARM platform (in Chapter1 computer plat-

forms ARM and ×86 are explained). The brain, consisting of ALU and control units, is

physically close to its memory, as required by the von Neumann architecture. In an Apple

iPhone, the brain and its memory are two chips packaged in a combo unit called “PoP

(packaged on package).” The first block mainly consists of high‐density and logic circuits.

Another block, RF front end (FE), consists of two major parts: one is the trans-

ceiver, an RF IC (TxRx), the other is the section made of PA (power amplifier), LNA

(low noise amplifier), switches, bandpass filters, duplexers, and antennas, packaged

in RF modules. The FE is full of RF‐passives integrated systems realizing important

RF functionalities, such as, balun, filtering, duplexing, and of course, RF signal trans-

mitting and receiving (through antennas).

Accelerometer

Gyro

Touch

Ant

WiFi/

Flash

Mobile

memory

PA/ASW,

filter, and

duplexer

Ant

TxRx

RF FE

Modem

(BB)

A PoP

FigureP.3 Key components in a modern handheld mobile. AP and mobile memory forms a PoP

(packaged on a package).

Prefacexx

Around the AP, there are peripheral devices and wireless communication sections.

The devices include accelerometer, gyro, audio and video, camera, and touch screen,

and so on. The wireless communication sections include connectivity and wide area

network (WAN, i.e., cellular phone) services. Connectivity services are those for shorter

range communications, such as bluetooth and wifi. In today’s technology, a connectivity

chip, for example, wifi, already has a built‐in baseband (BB) processor. Since the phone

services (3G or 4G WAN) involve multi‐mode and multi‐band, layers of protocols, and

database, it requires a large and separate baseband (BB) processor, the one between AP

and RF FE.

P.2.1 ARM (SoC) Processor andits Mobile Memory

TI’s OMAP (Open Multimedia Applications Platform), the father of application proces-

sor, is a SoC (system‐on‐a‐chip) based on ARM platform. OMAP is a series of image/

video processors developed by Texas Instruments, in collaboration with then dominant

wireless company, Motorola. They are a category of proprietary SoC for portable,

mobile multimedia, and wireless applications. OMAP devices generally include an

ARM architecture processor core plus one or more specialized co‐processors. Earlier

OMAP variants commonly featured a variant of the Texas Instruments TMS320 series

digital signal processor (DSP). New‐generation ARM application processors now usu-

ally contain GPU along with CPU on the same platform.

Apple integrates the application processor with its main mobile memory in a PoP

(packaged on a package) format. An AP needs a main mobile memory (low‐power

DRAM), 1 GB, or 2 GB in size. The main mobile memory packaged along with the appli-

cation process is a volatile one. The main memory on a smartphone has to be small

sized and consumes low power. JEDEC has standards for mobile memory, such as

LPDDR3, LPDDR4, and wide I/O and wide I/O 2 memories,[23] and see Chapters 1, 3,

5, and 9.

AP employs the most advanced MM (more Moore, the wafer foundry technology)

volume manufacturing technology; for example, A9, which is used in the new iPhone

6S, employs 16/14 nm technologies (announced in September 2015). Currently, LPDDR

vendors are employing 20‐plus (>20) nm technology, and are trying aggressively to

move to sub‐20 nm technology. AP/LPDDR combo employs the most advanced MTM

(more than Moore, that is, integration technology beyond the technology employed in

wafer foundries) volume production technology, PoP (packaged on package).

On September 26, 2012, Texas Instruments announced that they would wind down

their operations in smartphone‐ and tablet‐oriented OMAP chips and instead focus on

embedded platforms. The fate of OMAP therefore remains uncertain.

P.2.2 RF Front End (FE)

The modern RF front end (FE) processors are based on narrow‐band RF signal process-

ing technique. Digital signals are transmitted and received through air by an RF CW

(continuous wave) carrier. Faster digital signals require wider frequency bandwidths

(1.25 MHz, 5 MHz, 20 MHz, etc.). Due to the narrow band requirement, a broader

bandwidth would require a higher RF CW carrier frequency. In WAN (up to 4G),

700 MHz, 850 MHz, 900 MHz, 1,800 MHz, 1,900 MHz, 2,600 MHz, 3,500 MHz, and

Preface

xxi

5 GHz bands are employed as the CW carrier frequencies. Another way to boost higher

data rates is through the physical (PHY) layer modulation and de‐modulation tech-

niques, such as 256 QAM, multi‐carrier OFDM, and multi‐code CDMA.

In high‐density and logic (HD&L) domain, PoP was used for AP and its main mobile

memory; RF FE consists of many RF‐passives integrated systems, it is where MMMB

(multi‐mode multi‐band) functions is fulfilled. First, WLCSP (wafer‐level CSP) was

used for transceiver IC, WTR1625L; RF modules were used for other area of RF FE,

including PA and switches, filters, duplexers, antenna matching, tuning, and antennas.

In the future, InFO‐PoP and 3D IC may be applied to integrate the AP (e.g., A‐series)

and its main mobile memory (from LPDDR3, LPDDR4 to wide I/O 2). WLCSP, may still

be employed for RF CMOS transceiver IC; on the other hand, RF SiPs, will be used to

package for the rest of RF‐passives integrated systems in RF FE. RF SiPs, including

forms of IPD (WL‐IPD) and IPD‐SiP (defined in Chapter4), utilizing passive substrates

such as glass or hi‐resistance silicon for better RF performance at a lower cost, is a form

expanded from RF modules, to be explained in more details in FigureP.6. RF CMOS

represents a homogenous approach, and RF SiPs are a heterogeneous one for RF-passives

systems; together, they, RF CMOS and RF SiPs, provide total solution for future RF FE.

P.3 About this Book

P.3.1 Guide totheChapters

FigureP.4, below, shows the main topics that are covered in this book: 1) background on

mobility and state‐of‐the‐art hardware in Preface and Chapter 1, 2) state‐of‐the‐art

Mobility and state-of-the-art hardwareState-of-the-art

MTM technologies

Chapter 2:

Interconnects

(L1&L0.5)

Preface: Electrical brain,

transistor, and IC; history

of another planar

technology (MTM);

mobility. Focus on

AP/Memory and RF FE

(including RFIC).

Chapter 1: Conv. of

comm. & 5G;

hardware for wired

and wireless

communications; MM

and MTM, including a

MTM-PKG

Chapter 3: State-of-

the-art packages

(L1.5) and substrates;

MTM-substrate

Chapter 4: State-of-

the-art RF-passives

technologies;

MTM-passives

Electrical design for 5G

Packaging solutions and a new business

model for 5G mobility hardware

Chapter 9: Advanced

stacking and planar

technologies for 5G

mobility; MTM-5G and

new business model

Chapter 5:

Design: Digital

focuses, wide I/O

mobile memory

and TSV

Chapter 6: Design:

RF focuses, carrier

aggregate and

beamforming using

multi-antenna

Manufacturability, Q and R

Chapter 7:

Manufacturability

and quality (six

sigma with hidden

risks removed)

Chapter 8: Product

life and reliability

(theory and

practices)

FigureP.4 A guide to the chapters of the book.

剩余449页未读，继续阅读

ddalecn

粉丝: 8
资源: 31

5G移动通信的3D IC与RF SiPs技术

基于SIPS模型的知识付费类APP营销策略研究——以“流利说·阅读”为例.pdf

oOOFDMA_RAf.rar_SIPS04main

SIPS-lib:SIPS任务使用的库

SIPs:SIREN改进建议

php_sips-开源

Sips Horn-crx插件

sips-sdk-nodejs:允许与Worldline SIPS交互的Node JS库

SIPS04Matlab.rar_OFDMA MATLAB_SIPS04Matlab-v2_ofdma_protocol mat

SIPS en Español-开源

SIPs:Synthetix改进提案存储库

最新资源