Advice for IC design engineers who have been working for 1~2 years (zt)

小猪

Advice for IC design engineers who have been working for 1~2 years (zt) [Copy link]

The impact of the economic crisis on the IC industry is huge. For every IC engineer, the past 1~2 years have not been very comfortable, and the most direct impact is the loss of money. Therefore, when we see the signs of economic recovery, we can't wait to make a new assessment of ourselves: "Should I make more money?", "Should I still develop in the IC field?", "Should I do projects other than Power?". Here, I would like to share my views with you. I have been engaged in IC design for a little more than 4 years. My product line has a certain breadth and depth. At the same time, I have also exchanged views on this with several big turtles. I would like to throw out some ideas here, and I hope you will bear with me. First of all, please make sure that you are doing forward design. If not, switch to forward design as soon as possible. I have seen many engineers with various projects on their resumes, but when asked about the specific details, they can't tell. Even some engineers are still asking very basic questions to others after they have reverse engineered very high-end products. So these projects are of no value to the interviewer, and they are also of no value to any forward IC company. The 1~3 years of IC design experience I mentioned here also refers to the experience of forward design. I don't mean to be disrespectful, but just remind you to pay more attention to your career planning. If you are doing forward design, you should be defined as a rookie by most people after only two years of IC design. At this stage, in addition to doing more design, you also need to start to care about production and the market. Why is your IC accepted (or not accepted) by the market? What market share does your product occupy? Where is your customer base? What do they care about? If you are the CTO, how would you define your chip so that your chip will be accepted by more people? How would you define the application system of your chip? Similarly, there are many considerations in production. How to evaluate a process line? If you are the CTO, how would you choose a process line? How to perform WT and FT tests? What is the entire production process of IC? How to determine the various indicators of your chip's datasheet? How to describe the uniqueness of your chip in the datasheet? How to solve the yield problem? Maybe many people sneer at these questions, but when the sales volume of your chip reaches 100K/M, similar problems may occur at any time. Any small problem, if not handled well, may be fatal. Of course, at this stage, most of these issues are left to senior engineers, project managers, and even CTOs to think about. But if you want to add value to yourself, whether in terms of ability or money, you should start to learn about these issues and even try to face them. Of course, as a design engineer, your job is still design. I believe that some capable engineers may have designed 4 to 6 chips after two years of work under the guidance of experienced senior engineers. At this point, you need to ask yourself, what are the factors that limit your development? Here are some examples: Some engineers are very good at doing chip block-level design, but they lack attention to system-level circuits. So that they cannot describe what the key parameters of their chips should have in the entire application circuit, and they are also unable to describe the frequency response, transient response, and other characteristics of system-level circuits. But if you are a senior engineer, you should start to consider system-level design, and if you are a project manager, you should be responsible for the system-level design of your projects. Some engineers are very happy to take out existing blocks from the company's IP library and add them to their own chips. Of course, this is a very safe way for the company, but for the engineers themselves, this means that you are in an ambiguous stage for any design details and have no obvious contribution to the company; at the same time, some engineers like to design new blocks very much and are fanatically pursuing new technologies, which is also very risky. The correct approach should be to maintain a ratio, with part of the chip being the existing technology and part being new technology. Many people suggest that I keep this ratio at around 7:3. There is a similar situation in the pursuit of chip performance. The chips of the reverse design company mentioned above can almost be said to not pursue performance at all, or to say that they cannot make substantial suggestions on improving chip performance. It is hard to imagine that such chips will have a place in the market. At the same time, some engineers are almost fanatically pursuing every performance of the chip, and every performance must be higher than the performance of existing chips on the market. The design cycle of such chips will be greatly increased, and many opportunities to enter the market will be missed. When your high-performance IC is designed, others have already led this market to another more reasonable definition. Therefore, the general approach is to have a deep understanding of the chip, and then focus on the design of several chip performances that you think are important, and it is enough to keep the others up to date with the market. By the way, I was once an ardent pursuer of new technology and high performance, so I encountered many setbacks but I also enjoyed myself. Chip design is an industrialized product, or an engineered product. After Rookie, who had just graduated, joined the company, he didn't know what he was going to do, and he had no idea where the knowledge he learned in school would be applied. But after two years of practical experience, they found that they had opened up and found that the so-called chip design and building blocks were not very different. This is because you work in a team, and the leader of this team is not only considering chip design, but also considering how to distribute work to all members of the entire team so that they can complete the task perfectly and feel that they have gained something. Therefore, leaders divide a project into many small steps, from simple to complex. At the same time, in order to ensure the perfect completion of the project, leaders will make a summary after completing the project. They will lead the engineers they think have a promising future to a higher level, give them more difficult tasks and more authority, which will also bring more instability to the team, but leaders think it is worth it; they will also define those members who are willing to build blocks as engineers of a certain level. From then on, these engineers will not be assigned more difficult tasks, but will be allowed to engage in block-level design to improve the efficiency of the entire team. Therefore, most engineers in this period will have a feeling like "chip design is just like this". In fact, the most creative work is given to the most capable people. If you feel that what you do is not challenging, it only means that you have been defined at the level you are at. The same is true for learning things. Your forward trajectory is spirally upward. When you reach a certain height, you will find that your previous ideas were very naive, and you will also feel that the circuits discussed by others are "just like that". I suggest that you don't trap yourself, learn more, practice more, and have the courage to take on new projects and more serious responsibilities. Do more things that can help others, and try not to have your leader assign people to help you specifically. In this way, you can see more opportunities. Many experts in the forum say that chip design relies on intuition because they touch the core of analog circuit design. Some people say that system design only needs to be considered when making ADC, DAC or PLL. In fact, it is just the opposite. The requirements of customers facing our analog circuits are usually very simple: charging, lighting, voltage regulation, audio amplification, etc., but what if we meet customer requirements in various complex environments? The solution is the most important, and ADC and DAC can only be a module. Chip design is generally divided into two categories: forward and reverse. In forward design, there are two types: 1st source and 2ndsource. 1stsource is a company that independently defines a solution required by a customer, and then designs a chip's pins, functions, etc.; 2ndsource is a company that sees that big companies such as national have a mature and stable chip, and there will be stable demand in the next 5 or even 20 years, so we design a chip according to others' definitions, following behind others, not ahead, but not behind. However, the starting points of the two designs are different: 1stsource pursues technological advancement and advantages in system-level solutions, so when designing chips, we usually do not consider cost factors, but use the best process and the fastest time to bring the chip to the market; 2ndsource pursues universality. Even if we make the performance the best in the world, it is very difficult to snatch customers from others. Then we need a process line with advantages, which is simply cheap. What is cheap? Pure BJT or pure CMOS process line. This process line may not have an accurate model, or a complete PDK, but it has advantages. I saw friends on the forum say that it is easy to design DC-DC. I guess they use BCD process and have a complete PDK, and even the process line will provide high-performance IP, etc. Then they complained as expected: "What's the difficulty in DC-DC design? But now DC-DC is very cheap. No one wants it even if it is designed." My colleague said that DC-DC is very difficult. How do others achieve it? I told him that we are a forward design, with advantages in forward design. So please think about it from another perspective. Why can't you lower the price of DC-DC? When you are facing such a pure CMOS line, do you have the confidence to design a chip? What is needed here is not only circuit knowledge, but also Model knowledge and certain debugging ability. In summary, you need to gain enough experience by designing a large number of chips and making a lot of mistakes. There are also fields in analog circuits: Power, Audio, LED, etc., but the boundaries are very vague. Only those very large companies will distinguish them when they manage projects with engineering management methods. In fact, from a technical point of view, they are all similar, and for engineers with 1~2 years of work experience, they should focus their energy on the field they are good at. Whether it is LDO, PWM, or ClassD, they all use Op to build circuits, and the principles are the same, all of which are automatic control. Why can you find a job doing PWM after you get tired of working on LDO? Because leaders think that if you have the ability to master the basics of one thing, you will definitely be able to understand the basics of other things, but the value of these basics is really not high, so even if you master the basics of all analog circuits, it is not of much value. You need to be proficient in a certain field, so that you can show your value. Finally, my design experience is only a little more than 4 years, and I am lucky to have a job managing a small team. The above opinions are not mature, and I would like to give some suggestions to new engineers and discuss with senior people. Bricks are welcome, but personal attacks are not welcome.

lixiaohai8211

IC Design...................................