Huawei R & D reading-19 (chip design, from scratch)

Starting from scratch
Many of Huawei's early chip design engineers did not design chips or have no practical experience in chip production.
Huawei has no conditions to introduce high-paying engineers from abroad. It can only rely on young chip design engineers to learn from theory.
First, and then combined with practice-step by step.
Sun hongjun, a graduate student from Nanjing University in 1996, came to the Huawei Basic Research Department. At the beginning, the chip design was simply
-No connection. He started from the circuit anatomy analysis of existing chips, one side facing the chip, the other side facing the schematic
Point-to-point analysis. Various original foreign books related to chip design, such as analysis and design by P. R. Gray
Of analog integrated circuits, a variety of non-professional AWB (analog Workbench) simulation software, he has
I learned it carefully. He also used various opportunities to consult, for example, from hardware engineers in Huawei's customer circuit Department.
Solve the design principle of the User Board circuit system of the C & c08 switch, and learn the peripheral features of the prepared chip. However
Then we can understand the working principle of the thick film circuit from a micro perspective. In three months
The user's circuit Department left and right asked, soaking in the Data heap, reading all the relevant information, and working with the user's Circuit
The System Department started to improve the design and simulation of Thick Film circuits and learned a lot of practical knowledge. Soon, he
The circuit designed by the chip can be connected through the simulation software, and all the work conditions of the circuit can be simulated again.
Sun hongjun's first chip involved in the projection is a Speech Codec Chip. He searched a large number of devices in Huawei
He has learned some basic knowledge about English materials, digital signal processing and other teaching materials. Learn from competitors Siemens
The paper published by IEEE (World Journal of electronics) is his preliminary understanding of the working principle of chips; chip field
AMD, the leading us chip company, elaborated on the working principle of chips in depth.
This gives him a better understanding of the chip principles. In a lot of learning, he realized that if he doesn't have to understand the market
There will be a lot of detours if you want to do things on your own. In-depth exploration of market leader Technologies
The incoming analysis and learning are not just acceptable, but also an important reason why Huawei has the opportunity to surpass them. Ongoing
On the basis of studying and digesting the latest information in the industry, Sun hongjun and his colleagues turned to the schematic diagram and went to the calculator,
Shift operation, Alu structure, and definition of Rom lattice. In just three months
He and his colleagues broke through the algorithm.
After the design of the chip system was completed, they found a professional drawing company to draw a circuit diagram of the chip, and then
It is sent to foreign chip processing plants for projection. After the chip is complete, they have not only conducted a comprehensive test in the laboratory,
We also took the chips to the company's internal users. Huawei's System Department repeatedly conducted tests in various situations for a long time.
-Year. At that time, they repeatedly tested the call loss rate in a high temperature and high humidity environment. The call fails.
In a day or two, a large number of simulated phone calls may only appear, but they are not easy to put
Problem. After a long period of continuous capture of failure cases in the laboratory, and then carefully analyzed, finally they finally
Find out the cause of the problem. It turns out that the serial CLK line does not have a Schmidt structure, and the PCB Board has a signal integrity problem when the strip is too long,
The result is that the tick of the clock edge is misunderstood. Although the probability of trigger by mistake is extremely low, this core will not be affected.
The implementation of the system device functions of the slice will affect the system performance indicators.
After a year-long "torture", the chip, Sun hongjun and his colleagues
They have developed a rigorous attitude towards improving the chip, and have a profound understanding of it. Knowing that a technology is, what is going on?
It is another thing to commercialize technology. The problem of chip-like is not important,
It may be rare to say that the chip does not work, but it is also rare to say that the chip does not work. Annual Light Industry of Huawei Basic Research Department
Programmers are not satisfied with the fact that they only master the technology, industrialize the chip technology, and refine the chip. To this end,
They put forward the internal slogan "Do not be a warrior with flaws", and "do fly with beauty"-cheaper, more technical
Simple: as long as it is perfect, acceptable in the market, and profitable, it must be well done meticulously. Starting from childhood
From scratch, there will be unlimited development possibilities in the future.
Sun hongjun and his colleagues are like this. They have almost no idea about Huawei's basic research department, and they have learned from themselves
Having some practical experience, familiar with R & D processes and system applications, from reverse design to positive design of chips,
Challenge the communication giants again and again. Through the cooperation of the team, a digital 32-way 64 ms response from the basic research department in 1997
The wave elimination chip is officially put into operation. The chip is only half the size of the Communication giant Lucent, and its business functions and other
The performance indicators are the same as those of Lucent. This chip with excellent performance also greatly improves the circuit board integration of Huawei users.
Saves a lot of costs, and performance indicators are superior to other competitors.
Sun hongjun participated in the in-depth R & D of the five chips, including high-voltage bipolar process and BCD process.
There are mixed mode Low-Voltage CMOS, and pure digital circuit positive algorithm implementation. This is not a breeze.
Not all of them are successful. Some projects failed and made "Bricks" (failed products in the engineer's mouth),
This "brick" allows everyone to learn more: You can't select partners at will, you need to do more on the Process
To learn and understand, you need to carefully study the chip design principles and so on. Although some projects have succeeded, they have not been able to produce data,
We all feel sorry. However, among the many regrets and failures, the chip is also very successful and scaled.
Huawei has created a lot of value and made engineers feel the realization of their own value.
A chip investment requires millions of dollars, and a subtle error will make this chip
However, Huawei did not dare to let young engineers take responsibility for the failure. Huawei
The courage to let go, also enables young engineers to grow rapidly, so that the basic research department can make the chip in a short time
The design has been greatly improved. Once, St, a world-famous chip manufacturer, cooperated with Huawei to design, St
The company's senior engineers have more than 20 years of experience in chip design. When they see that Huawei's young engineers are very
I was surprised when I completed the circuit design. I think Huawei's chip design level has reached a level not far from that of St.
Ping. All these achievements are the result of Huawei's courage to let go and the hard work of engineers. Huawei Zhongyan Department passed
The new project management architecture brought about by matrix management, such as "making big and small" and "making difficult. Relying on the general office and general office
Groups and other departments that provide layer-by-layer technical support, as well as a large number of experienced overall design and planning talents, break through the complex
The management of technical projects allows new users to do projects. Through effective management of cross-department matrices, Huawei's new product R & D can
Quickly achieve breakthroughs. At Huawei, as long as a new employee is brave enough to learn from and learn from various experiences, and Huawei is brave enough to help
New employees can develop rapidly, even if it is difficult to design chips.
Matrix management makes the management of Huawei R & D department further, reducing the difficulty of organizing new technological breakthroughs. Following Huawei
In, ZTE decided to implement matrix management in its R & D system.