REQUEST QUOTE
In a groundbreaking achievement for Maketronics, we are thrilled to announce that we have secured a spot among the top 60 electronic sellers, as recognized by the reputable platform Feedspot.com. This prestigious acknowledgment underscores our commitment to excellence in the electronic retail industry and serves as a testament to the trust and satisfaction of our valued customers.
Article Link: https://electronics.feedspot.com/electronic_products_and_components_blogs/
In 2017, The MagPi Magazine – the authority Raspberry Pi magazine – reported that over 12.5 million of the Raspberry Pi single-board PCs (SBCs) had been sold since the 2012 presentation of the primary form. The Pi has not just shown to be a #1 of specialists, producers and designers the same, it has even gotten the interest of AI scientists at Microsoft.
Combined with its blast in fame is the rundown of frill that supplement this SBC. Coming up next is a rundown of suggested extras for you, the Raspberry Pi fan.
If, notwithstanding, you’re not there yet, we can assist you with the beginning. So we should begin with a pack.
The starter pack for the new Raspberry Pi proprietor
We suggest Raspberry Pi Foundation’s own starter pack for an overall investigation of the SBC. The unit incorporates a Raspberry Pi 3 Model B, 8-GB or 16-GB NOOBS SD card, case, a 2.5-A multi-locale power supply, 1-m HDMI link, optical mouse, and a console with top-notch scissor-switch activity.
What we truly like about this pack is that it accompanies a duplicate of “Undertakings in Raspberry Pi.” The writer, Carrie Anne Philbin, guides you through your new SBC obtaining, exhibits how to compose programs in Scratch and Python, utilizes engineer devices like Minecraft Pi and Sonic Pi and examines a few first ventures.
Equipped with a Raspberry Pi, now is the ideal time to take a gander at ways you can get the majority of your SBC – through embellishments and direction.
Raspberry Pi Cookbook
First up is the book by Simon Monk that goes past the manual in the starter pack to give you north of 200 recipes or tasks. It will show you how to set up and deal with the board, associate it to an organization, work with its Linux OS, program in Python, control other equipment including engines, secure data from sensors and much more. The Raspberry Pi Cookbook is simply $39.99.
To dive further into the Python programming language, we additionally suggest Python for Microcontrollers: Getting Started with MicroPython. This book won’t just show you the Pi’s Python local language yet arm you with eight extra model tasks.
Google AIY Voice Kit for Raspberry Pi
To rapidly construct an all set venture – in about 90 minutes level – attempt AIY Projects’ Google AIY Voice Kit. For just $29.99, you can construct yourself your own voice-actuated home right hand.
The pack incorporates a voice Hardware-Attached-on-Top (HAT) embellishment board, voice HAT mouthpiece board, 3-in. speaker, arcade-style pushbutton, edge and plastic stalemates. You’ll should simply supply a Raspberry Pi 3 board, a SD card, size “00” Phillips screwdriver and some Scotch tape.
Happiness Bonnet
In the event that you are wanting to construct a gaming framework, you’ll require a game regulator. Adafruit’s Joy Bonnet sits on top of a Pi Zero to empower arcade-like control. For just $15.61, this deal regulator is a completely collected HAT, and that implies you don’t have to bind anything. All you want to do is to introduce your content on the Pi and transfer your emulator, like the RetroPie or the EmulationStation.
You can likewise utilize the Joy Bonnet with Pi A+ or B+/2/3, however it’s the Pi Zero that it fits the best.
Raspberry Pi Sense HAT
Prepared to utilize sensors to assemble games, control other equipment or lead science tests? The Raspberry Pi Sense HAT appends a few on-board sensors on top of the Raspberry Pi by means of the GPIO pins. It was made for the Astro Pi mission and was sent off to the International Space Station in December 2015 to perform science tests.
This HAT incorporates a gyro, accelerometer, magnetometer, temperature sensor, barometric tension sensor and a mugginess sensor. A 8 × 8 RGB LED grid will show information from the sensors, and you can utilize a five-button joystick to connect with your task here on Earth.
You can purchase the Sense HAT for $70.39 and download Raspberry Pi Foundation’s Python library here to control it.
Adafruit 7-in. LCD TFT touchscreen
Transform the Pi into a tablet with Adafruit’s 7-in. touchscreen, which gives you a 1,024 x 600 goal that is a decent counterpart for inserted processing needs.
You can drive it over USB and contribution over HDMI. Included is a resistive touchscreen-to-mouse regulator (AR1100) – a board that handles A/D change – as well as adjustment programming. A wired PCB with buttons permits you to enter the showcase’s menu for changing splendor, differentiation and variety.
On the off chance that you are fostering a LED divider show, consider adding an Adafruit RGB Matrix HAT to make a looking over show utilizing 16 x 32, 32 x 32, and 32 x 64 RGB LED frameworks with HUB75 associations.
The Pi-Top Laptop
For the people who are considering utilizing the Pi a dainty client or more, the Pi-Top will get you an open-source PC for just $299.99. You’ll get a 14-in. 1080p LCD, complete with a nook that incorporates a console, trackpad and battery, a PCB to deal with the associations between the Pi and the remainder of the framework, 8-GB SD card, links and charger.
What’s best about this Pi-to-Laptop converter is that gathering is pretty much as simple as “clicking” the provided parts into place.
Pimoroni SkyWriter HAT
Our last pick is an extra that will acquaint you with the developing signal control market. With the SkyWriter HAT, you can carry the ability to utilize signals to your Raspberry Pi (Models 3, 2, A+, B+ and Zero) for just $18.99.
The HAT has an electrical close field 3D/motion sensor that can recognize your developments to the extent that 5 cm away and can work from behind a non-conductive material-like texture. It tracks finger developments and signals like tap/twofold tap and finger flicks.
To get more familiar with these frills and investigate extra choices, go to Arrow.com. On the off chance that you pick to make a buy and spend more than $20, you’ll appreciate free one-day transport.
Over the course of the last year, the Center for Advanced Electronics through Machine Learning (CAEML) has acquired four new accomplices. The group of 13 industry individuals and three colleges has extended both the expansiveness and profundity of its work.
“Last year, we zeroed in principally on signal trustworthiness and power uprightness, however this year, we differentiated our portfolio into framework examination, chip format, and believed stage plan — so the variety of the exploration has gained the most headway,” said Christopher Cheng, a recognized technologist at Hewlett-Packard Enterprise and an individual from CAEML.
“Work in Bayesian improvements and convolutional brain networks in plan for-assembling have both high level essentially in the capacities we have demoed, and we’re beginning to contemplate utilization of in-line learning in the plan cycle,” said Paul Franzon, a recognized teacher at North Carolina State University, one of the gathering’s three host schools.
“One of the difficulties we face is gaining admittance to information from organizations,” said Madhavan Swaminathan, a teacher at Georgia Institute of Technology, another CAEML have. “The majority of their information is restrictive, so we’ve thought of a few components to deal with it. The cycles are functioning admirably, yet they are more extensive than we’d like.”
The gathering hosted a kind of coming-out get-together for itself at this occasion a year ago. It began with support from nine sellers including Analog Devices, Cadence, Cisco, IBM, Nvidia, Qualcomm, Samsung, and Xilinx. Its underlying interest regions included fast interconnects, power conveyance, framework level electrostatic release, IP center reuse, and configuration rule checking.
Rhythm portrayed out a guide recommending that the EDA business is entering a subsequent stage in its utilization of AI. Snap to augment. Source: Cadence.
EDA merchants, for example, Cadence Design Systems began following exploration in AI back in the mid 1990s. The procedures initial advanced into its items in 2013 with a variant of Virtuoso that utilized examination and information mining to make AI models for parasitic extraction, said David White, a ranking executive of R&D at Cadence.
Until now, Cadence has sent more than 1.1 million AI models for its instruments to speed extensive estimations. The following period of item improvement is set up and-steering instruments that gain from human creators to suggest advancements that speed completion time. The arrangements might utilize a mix of nearby and cloud-based handling to saddle equal frameworks and huge informational collections, said White.
MIT engineers have fostered a strategy to manufacture adaptable ultrathin semiconducting movies produced using GaAs, GaN, and lithium fluoride – materials that show preferable execution over silicon yet as of not long ago have been restrictively costly to create in utilitarian gadgets.
The new strategy, scientists say, gives a financially savvy technique to create adaptable hardware produced using any blend of semiconducting components, that could perform better compared to current silicon-based gadgets.
“We’ve opened up a method for making adaptable gadgets with such countless different material frameworks, other than silicon,” says Jeehwan Kim, the Class of 1947 Career Development Associate Professor in the divisions of Mechanical Engineering and Materials Science and Engineering. Kim imagines the method can be utilized to produce minimal expense, elite execution gadgets like adaptable sun-powered cells, and wearable PCs and sensors.
Subtleties of the new method are accounted for in Nature Materials. Notwithstanding Kim, the paper’s MIT co-creators incorporate Wei Kong, Huashan Li, Kuan Qiao, Yunjo Kim, Kyusang Lee, Doyoon Lee, Tom Osadchy, Richard Molnar, Yang Yu, Sang-hoon Bae, Yang Shao-Horn, and Jeffrey Grossman, alongside scientists from Sun Yat-Sen University, the University of Virginia, the University of Texas at Dallas, the US Naval Research Laboratory, Ohio State University, and Georgia Tech.
In 2017, Kim and his partners concocted a strategy to create ‘duplicates’ of costly semiconducting materials utilizing graphene. They found that when they stacked graphene on top of an unadulterated, costly wafer of semiconducting material, for example, GaAs, then streamed molecules of gallium and arsenide over the stack, the particles seemed to connect here and there with the fundamental nuclear layer, as though the halfway graphene were undetectable or straightforward. Subsequently, the iotas gathered into the exact, single-translucent example of the basic semiconducting wafer, framing a precise duplicate that could then effectively be stripped away from the graphene layer.
The procedure, which they call ‘far off epitaxy’ gave a reasonable method for creating different movies of GaAs, utilizing only one costly fundamental wafer.
Not long after they announced their most memorable outcomes, the group puzzled over whether their procedure could be utilized to duplicate other semiconducting materials. They had a go at applying distant epitaxy to silicon, and furthermore, germanium yet found that when they streamed these molecules over graphene they neglected to collaborate with their individual basic layers. It was as though graphene, beforehand straightforward, turned out to be out of nowhere obscure, keeping iotas of silicon and germanium from “seeing” the molecules on the opposite side.
As it works out, silicon and germanium are two components that exist inside a similar gatherings of the intermittent table of components. In particular, the two components have a place in bunch four, a class of materials that are ionically unbiased, meaning they have no extremity.
“This gave us a clue,” says Kim.
Maybe, the group contemplated, molecules can interface with one another through graphene assuming that they have some ionic charge. For example, on account of GaAs, gallium has a negative charge at the connection point, contrasted and arsenic’s positive charge. This charge distinction, or extremity, may have assisted the iotas with interfacing through graphene as though it were straightforward, and to duplicate the basic nuclear example.
“We tracked down that the communication through not entirely set in stone by the extremity of the particles. For the most grounded ionically reinforced materials, they cooperate even through three layers of graphene,” Kim says. “It’s like the manner in which two magnets can draw in, even though a dainty piece of paper.”
Opposites are inclined toward one another
The specialists tried their speculation by utilizing far-off epitaxy to duplicate semiconducting materials with different levels of extremity, from unbiased silicon and germanium to somewhat spellbound GaAs, and lastly, exceptionally captivated lithium fluoride – a superior, more costly semiconductor than silicon.
They tracked down that the more prominent the level of extremity, the more grounded the nuclear communication, even, now and again, through numerous sheets of graphene. Each film they had the option to deliver was adaptable and simply tens to many nanometers thick.
The material through which the molecules connect additionally matters, the group found. Notwithstanding graphene, they explored different avenues regarding a halfway layer of hexagonal boron nitride (hBN), a material that looks like graphene’s nuclear example and has a comparative Teflon-like quality, empowering overlying materials to effortlessly strip off whenever they are replicated.
Nonetheless, hBN is made of oppositely charged boron and nitrogen iotas, which create an extremity inside the actual material. In their tests, the scientists found that any iotas streaming over hBN, regardless of whether they were exceptionally energized themselves, couldn’t connect with their fundamental wafers totally, proposing that the extremity of both the molecules of interest and the transitional material decides if the particles will cooperate and frame a duplicate of the first semiconducting wafer.
“Presently we truly comprehend there are rules of nuclear cooperation through graphene,” Kim says.
With this new getting it, he says, specialists can now essentially check out at the intermittent table and pick two components of the inverse charge. When they secure or create a fundamental wafer produced using similar components, they can then apply the group’s far-off epitaxy procedures to manufacture various, precise duplicates of the first wafer.
“Individuals have for the most part utilized silicon wafers since they’re modest,” Kim says.
“Presently our strategy opens up a method for utilizing higher-performing, nonsilicon materials. You can simply buy one costly wafer and duplicate it again and again, and continue to reuse the wafer. Also, presently the material library for this procedure is completely extended.”
Kim imagines that far-off epitaxy can now be utilized to create ultrathin, adaptable movies from a wide assortment of beforehand outlandish, semiconducting materials – as long as the materials are produced using iotas with a level of extremity. Such ultrathin movies might actually be stacked, one on top of the other, to deliver little, adaptable, multifunctional gadgets, like wearable sensors, adaptable sunlight-based cells, and even, in the far off future, “cellphones that connect to your skin.”
“In brilliant urban areas, where we should put little PCs all over the place, we would require low power, profoundly touchy registering and detecting gadgets, produced using better materials,” Kim says. “This [study] opens the pathway to those gadgets.”
The promotion stage is finished. Engineers are focusing on accomplishing the difficult work of building an Internet of Things.
Early projections for IoT are currently viewed as exaggerated. Be that as it may, the market actually has immense potential, so an enlarging circle of organizations and consortia is looking forward to enormous difficulties in regions like interoperability, security, and convenience.
You could follow the finish of the starting to June 2015. That is the point at which one economic specialist and previous designer became weary of the expanded expectations of a huge number of associated gadgets just around the corner. He chose to put out a more clearheaded conjecture.
On his most memorable occasion committed to the Internet of Things, Linley Gwennap assessed 1.9 billion new IoT gadgets would transport in 2020, up from around 200 million he projected were delivered around then, predominantly in the modern space. Gwennap adopted a base’s up a strategy to estimate the market that brought about an undeniably less blushing perspective than the 50 billion gadgets in 2020 that Cisco Systems hopefully anticipated.
From that point forward, the viewpoint has obscured marginally.
On its most recent occasion in late July, the Linley Group said the IoT market won’t hit a run pace of a billion units per year until 2019. The issue is customer frameworks expected to make up the brunt of the market sometime are still excessively costly and too difficult to even think about utilizing, said senior examiner Mike Demler.
“Interfacing things to Wi-Fi is so troublesome and temperamental that it’s been keeping things down,” he said, giving an illustration of his disappointments with an Amazon Echo.
In their authentic minutes, chip chiefs chasing after this nirvana concur. “IoT is in somewhat of a box at this moment,” Tyson Tuttle, CEO of Silicon Labs told EE Times in late June.
The shopper IoT still can’t seem to have its “iPhone second” when a plan fires the public creative mind, Tuttle grumbled. “It will require numerous a long time for IoT to work out,” he said, adding that he actually accepts it addresses “the greatest chance of our lifetime.”
text
Examinations of cell phone-based chips for midrange wearables (top) and MCUs with Wi-Fi. Snap to grow (Tables: Linley Group)
Engineers are feeling some blowback from the IoT bubble exploding, said Gwennap.
“Last year and the year prior to a lot of chips were springing up like MCUs with coordinated radios and chips focusing on shrewd watches and different wearables. This year, I am seeing less interest. We are not seeing follow-ons to last year’s items and more individuals are simply reusing existing cell phone SoCs,” the market watcher said.
Barely any imaginative chips make it harder for OEMs to create the iPhones of the buyer IoT. The uplifting news is in interchanges.
Chips for Bluetooth network and the 802.11ah form of W-Fi for minimal expense, long reach associations north of 900 MHz groups ought to transport this year. Outside the shrewd home a modest bunch of new choices – LoRa, Sigfox, Cat-M1, and NB-IoT- – are extending the still-new low power wide region (LPWA) area.
A startup sent off by a siblings group from Broadcom will show at MWC Americas a millimeter-wave module for 5G organizations. Co-CEO Maryam Rofougaran discussed the innovation and its business sectors as well as the condition of ladies in innovation and techno-governmental issues around her introduction to the world nation of Iran.
Morandi’s BeamX module is an RF front-end including everything from a radio wire cluster to a baseband interface. It targets 28 and 39 GHz 5G frameworks including base stations and collectors along with use in indoor gigabit, fixed-remote access and satellite organizations.
The startup means to convey a 4-5 dB preferable connection spending plan over contending items while consuming 30% less communication power. That could mean items that are twofold territory to 400 meters or view information paces of 4-5 Gbits/second.
Morandi will test 8-and 16-receiving wire modules for 28 GHz nets before the year’s end and a 39 GHz form right on time one year from now. They will be founded on RF ICs taped out around three weeks prior. Its most memorable model was a 64-receiving wire exhibit in light of chips taped out toward the finish of a year ago.
The items target client premises gear expected to utilize 16-32 radio wires and base stations that might utilize around 256 receiving wires. The modules connect to basebands by means of IQ simple, computerized SDI control, and a third anonymous point of interaction.
The first BeamX model was a 64-receiving wire exhibit. (Pictures: Morandi)The unique BeamX model was a 64-receiving wire exhibit. (Pictures: Morandi)
By conveying a superior sign at lower power, Movandi desires to jump-laid out monsters including Intel and Qualcomm which as of now have models in transporter preliminaries.
“At this moment, nobody has great proficiency, particularly on the transmitter. Since there are so many transmitters you can undoubtedly consume 10W… and we’re getting serious responsiveness,” said Maryam Rofougaran, who helped to establish the startup with her sibling Reza.
The couple spearheaded RF CMOS for Bluetooth in startup Innovent procured by Broadcom in 2000. There, they spent numerous years as senior designing directors venturing into Wi-Fi and cell chips.
SAN JOSE, Calif. — An arising organization with an original AI innovation and similarly remarkable monetary design appeared today first equipment item. BrainChip carried out an FPGA-based gas pedal for its spiking brain organization (SNN) programming and desires to convey an ASIC in the span of two years to extend its current business sectors.
SNNs are connected yet unique in relation to the convolutional brain nets (CNNs) presently broadly utilized and, some express, advertised by web monsters for occupations like voice and picture acknowledgment. SNNs utilize an easier, single-shot preparing technique and are appropriate to errands, for example, face acknowledgment in low-goal and loud conditions like observation video. Until this point, BrainChip’s items are utilized essentially by policing security.
The BrainChip gas pedal packs six SNN centers in a Xilinx Kintex chip on a PCI Express barricade handling video at 600 edges/s at around 15 W max. It gives however much six-overlap execution support for the BrainChip Studio programming for x86 PCs that the organization carried out in July at an expense beginning at $4,000 per video channel. The organization originally portrayed its design in late 2015.
The present arising CNN chips basically speed up meager straight polynomial math to abbreviate preparing circles. Paradoxically, BrainChip’s gas pedal velocities processes in advanced pathways are said to imitate brain neurotransmitters, supporting or restraining traffic and setting edges as fitting.
The card will be accessible toward the finish of September at a $10,000 list cost for single units. The organization will offer the card to framework integrators. It additionally may sell incorporated heaps of cards, programming, and servers straightforwardly to end clients.
The PCIe accelerator consumes 15 W at 600 casings/s. (Picture: BrainChip)The PCIe accelerator consumes 15 W at 600 edges/s. (Picture: BrainChip)
The organization guarantees that the card is the main business equipment to speed up SNNs. IBM’s True North is all the more commonly known however has been to a greater degree a universally useful examination vehicle, albeit the U.S. Flying corps said in June that it would involve it in a supercomputer. Stanford and the European Union additionally support research endeavors in SNN gas pedals.
BrainChip authorized innovation for an ASIC to speed up solo advancing in SNNs from an examination bunch in Toulouse, France. The organization is concentrating on the potential in auto, online protection, monetary, and clinical business sectors to decide how to tailor the silicon that could be accessible in 12 to two years.
The organization got its beginning quite a while back as a spinout from the Toulouse University research exertion that was making custom programming for clients in France. BrainChip presently comprises a product group in Toulouse, an equipment bunch in southern California, and keep going year, it welcomed on new administration essentially in Silicon Valley.
A fellow benefactor from Australia shrugged off the monetary terms of customary financial speculators. As another option, the group designed a converse takeover of an underactive mining organization in Australia and raised $15 million from public financial backers on the Australian securities exchange, where it is presently recorded as a little cap stock.
The organization wouldn’t remark on plans for additional gathering pledges, which it will obviously have to finance ASIC advancement while its still-small programming and card incomes gradually extend.
Fab chiefs need to shuffle liquid client interest while at the same time carrying out continually changing cycle advances in various assembling destinations all over the planet. This occurs as chipmakers mean to accomplish yield and quality focus on a corporate level as fast as could be expected, as indicated by Buddy Nicoson, VP of wafer fabs with Micron.
During a profession spreading over 30 years, Nicoson has overseen chip offices for Samsung, AMD, and Cypress Semiconductor. In the wake of joining Micron in 2014, he has assisted the organization with sending off the first-of-its-sort work to bring AI innovation into fab activities.
“Artificial intelligence was 25% quicker to get to the yield target wanted to be contrasted with already without these applications,” he said for this present week in a show at the Semicon Taiwan show. “It was 35% quicker to get to the deformity per million (DPM) levels wanted than it was previously.”
Micron, which has fabs in the U.S., Japan, Taiwan, and Singapore, has been building groups to gain from one another. Assuming somebody gains from huge information in Taiwan, “we need that engendered through our organization quickly,” Nicoson said.
With AI, Micron is moving fab administration from chip offices to controller focuses.
“You never again need individuals out in the field that you once did,” as per Nicoson. “You get a more extensive extent of view in the event that you can set up a remote working community by means of dashboards.”
Consistently, Micron pulls in 5 terabytes of information. The organization has 3,000 dashboards of information. The organization separates the work to workstation gatherings and workstation groups.
“In the event that you have 13,000 individuals in your association, 3,000 isn’t numerous dashboards,” as per Nicoson. “We’re attempting to make an interpretation of something exceptionally muddled to our people out there who are attempting to have an effect. We need to make interface apparatuses that are substantial and that our kin can accomplish something with.”
Request Forecasting
One AI application that shows potential is request estimating, which is turning out to be progressively hard for independent people, as per Bill Wiseman, a senior band together with the board consultancy McKinsey.
“At the point when Apple dispatches another iPhone, you better know the best figure,” he said. “Any other way, you can’t keep up. Assuming you’re one of the horrendously tragic creatures who need to supply to every other person other than Apple, who will win? Who do you transport items to? We see lots of overages and benefits at each point in the worth chain.”
Applying AI can give a superior thought to the number of chips to supply, he said.
Nicoson sounds like a comparable note.
“Planning presently is liquid,” he said. “It changes each and every moment. On the off chance that you don’t have schedulers who can adjust to the continuous elements of the climate where you are running your processing plant, you will be behind. That can be upgraded through virtualization.”
No More Silos
“No joking matter.” Comprehensive perception can uncover stowed away misfortunes and secret waste representation.
“Assuming you go out and converse with engineers in the field, one of the disappointments that they have is what I would call siloed sees,” he said. “They must go to one framework to see something, and afterward they must go to one more framework to take a gander at something different. It’s not exhaustive or durable.”
Unstructured information can be utilized to wipe out a current inclination Micron has in its information, he said.
“You wind up seeing secret misfortunes and secret waste that you were unable to see previously,” as per Nicoson. “Presently they become efficient marks, and you can take care of business.”
Overseeing Human Resources
AI can likewise let the administration know when a worker will stop, as indicated by McKinsey’s Bill Wiseman.
“You can watch long-range interpersonal communication conduct, how often representatives go to LinkedIn and their email conduct. You can anticipate representative turnover with a precision of around 0.95,” he says.
By expecting the outcome, an organization the executives can act and “save” workers before they leave. The ramifications is that there should be new interchanges arrangements among the executives and workers, he said.
Profound Learning
Profound learning is an innovation that is still in its early stages for chipmakers, as indicated by Nicoson. One point is to utilize AI to perceive absconds on silicon wafers from the get-go in the assembling system.
“It is basically the same as facial acknowledgment,” he said. “There are different sides to this: mathematical acknowledgment and photometric acknowledgment. It’s exceptionally successful in advising us about an interaction.”
The slack in the reception of AI is a piece humiliating in the semiconductor business, Nicoson said. The business is simply starting to consolidate AI advancements to get esteem, he noted.
Artificial intelligence will assist with crossing over new ages of laborers with old ones, as per Nicoson.
He gave the case of a worker who worked at Micron for quite a long time as a photograph administrator.
“He’s an extremely sharp person who has a deep understanding of photographs. We expected to change him to keep him feasible,” Nicoson said. “Due as far as anyone is concerned and involved insight, he turned into a significant coach to more youthful ages who see more about IoT, frameworks, and programming yet miss the mark on reasonable applied information.”
TAIPEI — United Microelectronics Corp. (UMC), Taiwan’s second-largest foundry, has started its first production of 14nm products, narrowing a technology gap with its larger competitors such as Taiwan Semiconductor Manufacturing Co. (TSMC) and Samsung.
UMC said that, during the second quarter this year, 14nm accounted for 1 percent of its total revenue for the first time.
In the meantime, the company slashed its 2017 capital expenditure budget to $1.7 billion from the $2 billion it had earmarked earlier this year. UMC’s CAPEX for 2016 was $2.8 billion.
“Operational efficiency will be the focus,” said Jason Wang, the newly appointed co-president of UMC, at an event to announce the company’s second-quarter results. “We are aligning to the current demand outlook.”
The company said it may cede competition at the 10nm and 7nm nodes to larger rivals TSMC and Samsung, where UMC said there will be “less demand”. UMC said it will focus on addressable markets where there is still plenty of room to grow.
While the company declined to forecast its 2018 CAPEX, UMC said it will be “cautious” with capital expenditure plans in the future.
No. 1 foundry TSMC plans to hold steady, with capital expenditures reaching about $10 billion this year. In 2016, the company’s CAPEX came in at $10.4 billion. TSMC, Samsung, and Intel are the top three spenders in the chip industry.
UMC’s pared Capex plan comes at the same time as Samsung reportedly aims to grab a bigger slice of the foundry business. Samsung plans to triple its foundry share as it aims for the No. 2 position after TSMC.
Mainstay Products Solid
Demand for UMC’s mainstay 28nm and 40nm products held firm during the second quarter. UMC’s 28nm brought in 17 percent of the company’s overall revenue during the second quarter, the same percentage as in the first quarter this year. UMC’s 40nm accounted for a 28 percent chunk of overall revenue in the second quarter, dipping from 29 percent in the first quarter.
Due to a weaker outlook for 28nm, the company forecast that its 28nm high-k metal gate (HKMG) business may decline during the next few quarters this year.
Analysts at the event noted that UMC may face increasing competition from Chinese foundries such as Semiconductor Manufacturing International Corp. (SMIC) which are ramping up 28nm poly gate and SiON gate dielectric (poly-SiON) production.
UMC said that while demand for 28nm HKMG is “lagging expectations,” the company expects full loading for 28nm poly-SiON in the foreseeable future.
UMC makes most of its 28nm products at its Tainan fab in southern Taiwan and is ramping up production at its new fab in Xiamen, China, that’s a joint venture with the local government. By the end of this year, the company expects to have 39,500 12-inch wafers per month — including about 5,000 wafers at the Xiamen fab — in 28nm process technology.
While TSMC and Samsung are starting up production of 10nm, UMC has targeted 14nm as its most advanced technology node. In the meantime, TSMC has been filling gaps in its legacy 28nm process to blunt the competition.
TSMC has increased capacity for 28nm, which accounted for more than a quarter of its revenue during the second quarter of this year. The company, which has counted on 28nm as a cash cow for more than five years, is hanging on to a 90 percent share of the market even as rivals such as UMC, Intel, and SMIC are trying to grab a chunk of that business.
UMC said its capacity utilization during the second quarter was 96 percent, unchanged from the first quarter. The company said it expects the utilization rate to fall to the low 90 percent range during the third quarter this year.
Earlier this year, UMC’s board of directors appointed Jason Wang and SC Chien as co-presidents of the company. Chien will be in charge of fab operations and technology development while Wang will be responsible for business management and corporate strategy.
MOUNTAIN VIEW, Calif. — Fans of systems from the beginning of the personal computer era in the 1970s and 80s gathered last weekend at the Vintage Computer Festival West. The event hosted by the Computer History Museum here showcased some minicomputers and workstations in addition to early microcomputers.
A panel session with a number of the early Commodore engineers talked about their once-popular, early home PCs. Commodore International started as a calculator company. Later it bought MOS Technologies, the manufacturer of the iconic 6502 microprocessor used in the Apple II, and entered the home computer market. A no-show for the panel was Chuck Peddle, considered the father of the 6502 and the man who convinced Commodore CEO Jack Tramiel to move beyond calculators and into computers.
Two panels discussed the Xerox PARC Alto workstation, the inspiration for modern personal computers from both Apple and Microsoft. The Alto was made in the mid-1970s, yet almost all the bipolar TTL chips inside systems at the event were still working. Engineers only had to replace one inverter chip.
Engineers who restored the systems said their biggest challenge was cobbling into one working machine piece from multiple salvaged Altos using different disk drives, disk controllers, and other mismatched peripherals.
Clubs and private collectors showed their vintage computers and related peripherals at the event, with one booth specializing in every flavor of the floppy disk. Attendees could buy their own classic products at a mini flea market for computers, peripherals, chips, manuals, and other related odds and ends.
