Page 48 - ChipScale_Jul-Aug_2020-Digital
P. 48

for high-performing SiP – rather, it is cost.  SiP commercialized applications  Tracker; Automotive ADAS advanced
          Right now, chiplet technology is growing   Based on our System Plus Consulting   computing unit - Mobileye’s EyeQ4, and
        exceptionally fast because advanced node   tear downs of recently commercialized   Nvidia Tesla V100 – GPU + HBM.
        yield is not good (in short). For 7nm, 5nm,   SiP products, the underlying common key   In 2019, the latest AirPods Pro designed
        and beyond, the fabrication cost for SoC is   factor is essentially the ability to create   and manufactured by Apple was the only
        simply too high because of low yield, and it   highly-integrated products in a smaller   earphone in the market with an active
        is not a wise business move. Some smaller   package with increased functionality at a   noise cancelling feature. Two SiP modules
        players pushing node-scaling cannot   lower cost. As examples, Apple’s AirPods   are found inside the AirPods Pro (Figure
        make any more chips right now. So cost   Pro - SiP Audio Module; Fitbit Charge 3   5). The module has a special shape that
        is the key and SiP cost is generally more   (acquired by Google) - Microcontroller   is designed in the shape of the opening of
        attractive than that of SoCs, especially in   + Bluetooth; Motorola 5G Moto Mod   the human ear to maximize the comfort
        the advanced nodes.                - Qualcomm QET 5100 M Envelope     of the AirPods Pro Earbuds. The SiP
                                                                              Audio Module comprises two assembled
                                                                              modules. The Audio Codec Module, H1
                                                                              Module and MEMS Sensors are included
                                                                              in the audio device. The modules contain
                                                                              an audio amplifier that eliminates
                                                                              background noise and produces clear
                                                                              sound. The H1 Module contains the H1
                                                                                           ®
                                                                              chip/Bluetooth  processor that enables
                                                                              wireless connection and drives voice-
                                                                              enabled Siri and enforces real-time noise
                                                                              cancellation. The H1 chip has better
                                                                              power management and increased talk
                                                                              time compared to Apple’s previous W1
                                                                              chip and uses a 16nm FinFET technology.
                                                                                For the Fitbit Charge 3, Cypress
                                                                              has combined two dies of different
                                                                              functionality (a microcontroller with
                                                                                      ®
                                                                              Bluetooth ). This FO SiP is packaged by
                                                                              JCET (STATS ChipPAC’s Singapore fab).
                                                                              Die 1 - MCU is a CapSense (capacitive-
                                                                              sensing) technology that measures changes
                                                                              in the capacitance between a plate (the
                                                                              sensor) and its environment, to detect the
                                                                              presence of a finger on or near a touch
                                                                              surface. Die 2 - BLE is a form of wireless
                                                                              communication designed especially for
                                                                              short-range communication.
                                                                                In 2019, Samsung Electro Mechanic
                                                                              (SEMCO) started the next wave with the
                                                                              introduction of ED packaging technology
                                                                              in the Qualcomm QET 5100 M for
                                                                              Motorola’s Moto Mod 5G hardware and
                                                                              Samsung’s S10 5G US-version smartphone.
                                                                              The envelope tracker IC die is embedded
                                                                              into the 6-layer PCB substrate. All the
                                                                              passives are mounted on the surface of the
                                                                              PCB substrate and overmoulded.
                                                                                Within automotive, Mobileye’s EyeQ4
                                                            P                 is a computing unit used in level 3 ADAS
                                                              RoHS
                                                                              vehicles  that can process data from several
                                                                              cameras as well as light detection and
                                                                              ranging (LiDAR), and radar. The EyeQ4
                                                                              unit is a type of simple SiP where the SoC
                                                                              is packaged in the same package with a
                                                                              few passives. The FC BGA SiP package is
                                                                              done by STMicroelectronics in Malta. It is
                                                                              noteworthy that EyeQ5 is in development
                                                                              and is meant for ADAS level >3.


        46   Chip Scale Review   July  •  August  •  2020   [ChipScaleReview.com]
        46
   43   44   45   46   47   48   49   50   51   52   53