It has become an increasingly complex task to introduce the analysis of semiconductor components. Today's analysts can be said to be looking for a needle in a haystack. Every year, the needles are getting smaller and smaller, and the haystack is getting bigger and bigger. Customers hope you can find them faster. In this form, how should we train analysts to accomplish this arduous task? If we want to successfully analyze modern semiconductor devices, our training must make efforts in several main aspects. These aspects include: craft, technology, skills and cross-training. I will discuss the history of product/fault analysis and describe the problems faced in this field. Historical review 1966, Fairchild Semiconductor first introduced a four-gate dual-input NAND gate device. Soon after, Texas Instruments, Motorola, National Semiconductor and other companies also introduced their own production lines to produce standard logic devices. In the late 1960s, many integrated circuits were developed to meet the needs of the US military. At the same time, the military also began to publicize the reliability of its own system. This brings the demand for product analysis ability-understanding the failure mechanism related to integrated circuits. As these products become more and more popular, the demand for fault analysis is also increasing. At the end of 1960s, many companies began to take the first step to establish the failure analysis laboratory as we know it today, so failure analysts were born. In the late 1960 s, most of the failure analysis work was the return of analyzing application sites. The system company returns the failed devices to the integrated circuit manufacturer for analysis. This means that the analyst's job is to respond to the problems that customers may encounter. At that time, the number of products produced by semiconductor manufacturers was relatively low; They made the same equipment for quite a long time. This means that analysts can feed back the information returned from the site to the production line in time, which will affect production. Process engineers improve the production line and solve the problems encountered by customers. Generally, failure analysis can be completed by IV curve analyzer, some unsealing tools and optical microscope. More troublesome problems require a scanning microscope and a possible energy spectrum analyzer. The investment in fault analysis on this scale is not large. A quarter of a million dollars can buy all the tools you need. More importantly, the investment in skills and training is moderate. A qualified electronic technician can complete and master all aspects of fault analysis. In the late 1960s, there was no training program for fault analysts. Public training courses such as those provided by Bud trapp, John Devaney and Howard Dickens still need ten years. Training is completed at work. Analysts talk to wafer process engineers to understand the process, designers understand the layout, schematic diagram and test process, and packaging process engineers understand the packaging process. Other training comes from his or her experience. Excellent analysts usually have many years of work experience and good records, which enable them to identify trends and recurring problems.