Progress is gaining momentum and along with new wonderful inventions, a number of problems associated with their implementation in production accumulate. Want to know what's going on on the production front?
Anyone who studies the hardware of modern smartphones sooner or later subscribes to corporate mailing lists of electronics manufacturers and ends up on developer forums. It is in such places that you can see the outlines of the future digital world, track trends and understand where everything is going. Information from such places gets into the specialized media, and you can track its appearance through electronics standardization sites. In such a tricky way, news and information were chosen that excite the manufacturers themselves. Let's take a look at this slice of reality together.
AI as an accelerator for factories and chip factories.
Chip manufacturing has a fixed cost. These include rejects, incorrect calculation of the number of chips produced due to an incorrect forecast for demand, and so on. The biggest cost is marriage. It is not a problem to build a new production line, the problem is in setting up a clear line of quality control, because we are talking about potential defects invisible to the eye. And the longer the scrap of the blank remains undetected, the more production operations it goes through, the greater losses the manufacturer will ultimately incur. The number of decommissioned 'near-normal' chips is often measured in tons. Buddy Nicoson is the silicon wafer manager at Micron and shares how his company is fighting this scourge. To understand the scale of Micron, 13,000 people are involved in the production of silicon wafers alone. The company is actively engaged in the development of production planners – this is specialized software based on AI (artificial intelligence). If earlier Micron's plans for the release of chips were drawn up several days in advance, now they change every minute, sensitively reacting to external and internal events. The introduction of quality control systems based on neural networks and AI makes it possible to find defects 35% faster than the classical combination of a person + a control device. Buddy does not hesitate to admit (and openly speaks about it) that the quality control system of silicon wafers works on the same principle as Face ID in iPhone X. Snapshots of each wafer and other parameters are analyzed by the neural network, after which it is taken out verdict on its compliance with the standard.
Burton 'Buddy' Nicoson, Samsung Vice President of Fabrication Engineering, at a chip manufacturing campus in Austin, Texas on April 11, 2011.
So, we see the further development of production automation, where a machine will replace a person as a quality controller. There is no doubt that other companies are following a similar course. This will affect the end buyer only in the form of increased product quality, there is no need to dream of a significant reduction in prices for electronics, because the investment must be recouped.
09/15/2017 Original material.
Qualcomm has announced its 10nm qualification at SJ Semiconductor.
It would seem, what's the news, because SJ Semiconductor is already producing chips using the 10nm process technology? However, information about a large number of defects in this Chinese production passed by the consumer. In this light, Qualcomm's statement can be regarded as an objective reality, or as a protection of their investments in production. However, there are also reports of upgrading existing production to produce 10nm chips with a denser layout.
This news suggests that the massive influx of smartphones with 10nm Qualcomm chipsets, if it happens, will not be in the near future. They simply do not report the qualifications of an already operating production.
September 18, 2017 Original material.
News from the front 5G.
It is expected that by the end of 2017 MediaTek a prototype of the modem 5G will be presented, and it will be tested in 2018. Thus, MediaTek will not be in the role of catching up (Qualcomm), as happened with 4G LTE, and will release its SoC simultaneously with the launch of providers 5G – networks.
Earlier MediaTek announced their merger with Nokia to develop mobile communication systems 5G. Apparently, the news about the creation of its prototype 5G – a modem and is the result of a fruitful cooperation MediaTek and Nokia. In the future, it is planned to cooperate with NTT DoCoMo, for joint deployment 5G – networks in Japan.
Spreadtrum Communications (a company that makes chipsets for budget smartphones) plans to release its prototype modems 5G in 2018. Telecommunications equipment suppliers Huawei, Ericsson and ZTE will act as a support group.
In light of the above, Qualcomm CEO Stephen Mollenkopf's optimistic forecast of a commercial launch 5G of networks in 2019 instead of 2020 does not seem unfounded. Qualcomm previously announced their prototype system 5G NR mmWave.
Prototype 5G NR mmWave Qualcomm.
As you can see, at this stage, Asian manufacturers are not far behind the leader (by half a year?) And may well release ready-made solutions at the same time.
September 18, 2017 Original material.
Friends, how do you like this format? If you were interested, do not forget to write about it in the comments.