This is how a cool man learns analog circuits.

This is an article by Qi dongjie, an intern in the Internet Data Management Group of Microsoft Asia Research Institute. Article , Copied from the "Microsoft Asia Research Institute blog ". Although many of us will not do research for application development in the future, many of Dr. Ma's views are still very helpful to us, so I would like to repost and recommend it to you. In fact, Microsoft's Asia Research Institute blog has many excellent ideas that are very unique. It is recommended that you take some time to browse the blog, and the gains will certainly be huge.

Fudan University has been studying the design of analog chips for Microelectronics for more than five years. It has been around for eight years and has listened to the guidance of many domestic and foreign experts. Recently, at the invitation of a friend, I would like to share my experiences with you. I remember that when I graduated from my undergraduate course, I was planning to study sensors, and then I entered the State Key Laboratory of integrated circuits and systems for Fudan Yifu Building as a graduate student. Now I want to come to the lab with a deep name, but I was surprised. The circuit and system seem to be two concepts, two layers. I was a graduate student in electronics and information systems. At that time, I knew they were "systems". However, we designed analog "circuits" and naturally had to be biased towards circuits. Beginners of analog chip design are always excited about the circuit of qisiqiqiao, especially the most authoritative magazine jssc (IEEE journal of Solid State circuits) in this field, at that time, I was determined to read articles from the past two decades and get through all sorts of twists and turns. I always wanted to write an article about it. At that time, the articles published in this magazine in China were rare, that is, reading a doctor abroad, being able to publish an article above is also excellent.

At the time of study, my mentor was Professor Zheng zengyu, who retired at the time. yifulou invited Mr. Li to give guidance every week. Instructor Zheng is rigorous in his studies and is a hero of female. Mr. Li is a pioneer in analog circuits in China and is now hired as an expert or consultant in many companies. Mr. Li wrote a book (Operational Amplifier Design) in 87 years. Even now, it seems to be a classic. Teacher Li and Mr. Zheng are classmates, so they are very good. Naturally, I am lucky enough to get guidance from my classmates. Instructor Li and instructor Zheng gave me the training solution: learn from the operational amplifier first. So I remember I was just designing a small current source. At that time, we felt that the design was based on simulation to adjust parameters. However, I always remember the words of Teacher Li's strong speech: the op amp is the foundation, and the op amp design is well done, so it is easy to do anything else. I didn't quite understand it at the time. My classmates focused on "high-end" things such as AD/DA and PLL, while instructor Li and instructor Zheng asked me to create "original" modules, the only article I have published in (solid electronics) (domestic spam magazine) is the Rail-to-rail amplifier. I am very depressed during the process and envy my classmates for their projects, but I feel that there is always a truth in what Mr. Li and Mr. Zheng say, so I will read articles on jssc op-ops, I have read it all over the past 20 years. At that time, I thought I knew this very well. Later I found out that I didn't understand it. The so-called understanding is to be truly integrated, otherwise there will be more knowledge in your head, and it will also die. However, the operational amplifier is the cornerstone of analog circuits. Only with a solid foundation can the Operational Amplifiers flourish, and the two teachers will understand it after their painstaking efforts. In general, in Fudan, what I feel most deeply is the words of teacher Zheng's rigorous study style and Mr. Li's words.

I graduated from the master's degree and went to work for a job. There were several offers at that time. My Senior Brother Sun Liping, a close disciple of Mr. Li, recommended me to go to xintao technology. He said there was Chang Zhongyuan, a doctor from luwen's Catholic University. I followed my advice and went there. At that time, Xin Tao was acquired by IDT for $85 million and became the first successful chip company in China. I interviewed Howard C. Yang, general manager of one of the company's founders ). Howard is a PhD at Oregon State University and an expert on the Phase-Locked Loop. During the interview, he asked me to draw a two-level amplifier with Miller compensation. I am very skilled. He said that there was a zero point on your face. I was very surprised. I never heard of it before. It was in the fog of the cloud that Howard first proposed internationally. The equivalent model had a resistance, he named the young resistance himself. At that time, he nodded his head out of courtesy. But they are very satisfied. They just went in like this. What about me? The only pity for the interview was that I didn't see Chang Zhongyuan. Maybe he was on a business trip.

After entering xintao, I made up my mind to prepare for the specialized skill industry. Because undergraduate and graduate students love physics, mathematics, and philosophy, they spend some energy on it. After work, you have to take a real shot. Every day after work simulation and after work, I read the original English version of books. The first is the popular Razavi book. Read it three times. I feel very rewarding. At that time, in xintao, I was not afraid of tigers. For example, I still did a great job, so I was often appreciated. He was rated as the most talented person in the company. Sometimes, I always give some advice, and others envy me. As a matter of fact, I remember the experiences I often share with me during a chat. The general idea is that there are three realms in analog circuit design: the first is to do it manually, it means that the pensile-to-paper circuit should be manually calculated, and the simulation only proves the result of the manual calculation. The second is to think about it and turn the circuit into an intuitive thing. The third is to create a circuit. I basically follow this trilogy. I have carefully considered the exercises behind Razavi's book. In the company's project, I also tried to give priority to manual computation. The parameters of the amplifier were calculated first and then compared with the simulation results. Over time, my hand's computing capability has been greatly improved, and some small-Signal Analysis and Computation have been very good. Here is an episode. Once in a project, the AC simulation of a protection loop is always unstable, but it cannot be adjusted. The capacitor is added here and the resistor is added here, if you have tried a few times, you will find it. Because this loop is very large, it feels like a blind image. He always draws out three times, five times, and two times. He looks at it carefully and then exports a formula to find out the primary pole and bandwidth expression. Through this incident, I often admire each other and understand the intuitive power. Therefore, when reading a book, I will carefully deduce the formula in the book, and then think about the signal stream intuitively. After more than a year, I finally had a thorough understanding of the amplifier, and I felt that I was able to learn it. After that, I found that I was able to understand it. Conclusion: The amplifier has two difficulties: Frequency Response and feedback. In fact, the so-called circuit is intuitive, that is, to think about the circuit from the perspective of feedback. Every time I analyze some books or the "weird" circuit on jssc, I will sigh: feedback, feedback! Then write the analysis experience on paper.

After learning one field, learning other related fields will have a certain "acceleration" effect. The general approach is to let the following people study each time they create a new project. Before leaving xintao, I made a phase-locked loop. I have never done it before, and I have studied my classmates' master's thesis, books, and many paper documents for a month and a half. I often come to ask me: do I understand the bandwidth of the PLL? I have long understood this. I use the frequency response knowledge of the powerful Op Amplifier on the Phase-Locked Loop. Now I have studied the advanced phase noise and jitter. Soon afterwards, a 30-plus-page English Study Report was published, which is always appreciated !. Later, at commit, a project was to modify an RF transceiver chip from WCDMA to TD-SCDMA. It contains a baseband analog filter. I have never touched the filter before. It took me two months to read three original English versions. The first one was more than 900 pages, and N more paper. All of a sudden, I switched the capacitor to the entire Filter field, GMC and active RC are all understood. When I proposed the modification scheme, because I had a solid foundation in the op-amp and I was easy to understand the signal stream of the filter when I read the article, I was able to propose the Principle Analysis and Modification Scheme of the chip circuit in a short time. Finally, the report is written to Ti. ti was very respectful to us. During the conference call, they first said that the report was "great job! ", I didn't understand English. Julian gave me a thumbs up and said, "They rate you very well ". Later I went to Dallas and Ti was very respectful to us. Many people came to listen when I made a report. In short, we now know that everything is very important. It is easy to learn other things with solid foundations, and the faster you learn.

I went to commit in November, And I was interviewed by my current company boss, Julian. Julian asked me: Where do you think is the process of SOC (system on chip) design? I said: it should be a analog circuit. This is more difficult. Julian said wrong. It's the system. I did not agree at the time. I felt that analog circuit engineers should spend their energy analyzing and designing circuits. Julian then ran the company on-bright and brought me back. He also pulled two from Ti, one of which was Dr. Fang. I recommend Dr. Zhu to Julian. In the past one or two years, Dr. Zhu and I have admired each other. Dr. Fang is a top expert in TI's Chinese community and has strong product capabilities. On-Bright is now working on a power chip. Dr. Zhu and I have been working on it for nearly two years and have learned about the importance of the system. Chip Design must eventually go to the system, which is the fourth realm of chip design. The circuit is like a brick, and the system is like a building. Chip Design Engineers must consider the problem from the system perspective. Otherwise, they will only see the trees, not the forest. In power chips, amplifiers and comparator are the most common, and the difficulty lies in a thorough understanding of the system. At on-bright, I really saw products, from definition to design, to debug, chip testing and system testing, and finally to RTP (release to production ). Julian introduced TI's advanced product development process and project management methods into on-bright. Dr. Zhu and I were very eye-catching and knew the hardships of making products.

 

(Original address:Http://hi.baidu.com/danaldyang/blog/item/86bbcd1d0872a28787d6b694.html)