[Original] work summary

This is the first month of a new job. I always wanted to write something every week, but I found that I was too busy. Aside from completing some scheduled tasks during the first weekend, the rest of the time was allocated to sleep, which was a bit lacking. But in general, this month's work in the M community gave me a new understanding of the embedded industry, and as a R & D personnel (to be honest, in the real sense of the Chinese mainland, it cannot be called R & D. It can only be called secondary development orCodeMaintenance Personnel) have a brand new understanding of the way of working, and I will share with you the following;

Work:

In the M community (currently only known as M community, because it has not been finalized, although you can get a badge next week, but it still takes a month to become a positive, after the change can be used throughout the process) there is no way to compare the previous work. I used to work in a very small design house. Now I work in an original factory and work in two places. I have also been in touch with some original chip manufacturers before, my experience is that collaborative work and corporate culture are really impressive to everyone, although I am tired (this is basically the case for companies like tw, JP, and kr ...), I usually leave the company at every day, but the old guys are working together until the end. Although there are still some complaints, it seems very normal for me. Overtime actually accelerates my personal progress to some extent. Although I work in shifts on Saturday, many people work overtime together. In addition, I have indeed learned a lot during overtime, after receiving the advice from the experts, we will talk about it later. In addition, the company's boss will seize a very effective opportunity to express its plans for the company's development direction and objectives in employee communication, make the employee's work direction and work methods very clear.

This month, I think the biggest difference in shipping is that I have a very clear understanding of the sales time model of the entire consumer electronics. From development to sales of a product, we usually talk about the low peak season. In the past, the company only felt that Christmas was a peak season, while others were dominated by R & D, of course, this is about overseas markets, but this is a narrow view to some extent. A product is a foreshadowing in the early stage. The real need lies in a certain key point. However, the grasp of this key point often requires a one-year period of preparation, in this one-year period, products are generally divided into two to three cycles. For example, a product is sold in China. According to the current domestic habits, the period of tradeselling is generally on the National Day and Spring Festival. How can the National Day and Spring Festival holidays really grasp the best sales practices, the preparations for the first phase of R & D in the last few months are crucial. The first phase of R & D must begin in March and enter the sprint phase in March, at this stage, the company generally performs a sprint-type development to create machines that can be produced in batch before May 1, and conduct small-scale promotion in the market, the months from January to August are a trial period. These months are quite painful because of the continuous feedback on different problems due to machine production and sales, with the upgrade and update of this version, the development will be closed in the past few months, and the existing problems will be revised and upgraded. Everything is for one purpose, massive sales during the National Day. By the National Day, a stable and high-performance machine will be sold in an overwhelming manner, and the work from October to December will turn to supplement and assist in production, if there is an external ticket, the two months are also very busy. In China, it is to continue to sell in large quantities during the Spring Festival of New Year's Day. in foreign countries, it is necessary to ensure that the production and testing of the product will be completed before January 1, November, you know, if the product goes to Europe and America to catch up with the Christmas schedule, it means it will take more than a month at sea. If there is no way to complete the production test in November, the Christmas schedule will be basically hopeless.

As mentioned above, it is basically fun for masters. In fact, it is a basic principle. I have only a general understanding of the annual plan of this industry. For those who are engaged in consumer electronics, especially those who are engaged in R & D projects, there is no idle time in one year, no off-season, and only the peak season;

Technology:

In terms of technology, it is getting farther and farther away from wince. However, I often read the blogs of Tom and Tom. Tom has been fully conversion to wince and has been awarded the MVP of Microsoft. Congratulations. However, a few days ago, I chatted with Brother h about what the OS is. What is important is the customer experience and actual needs, as a product, end customers do not care about what OS you use, WinCE? Android? And so on ..... What an end customer needs is a stable and somewhat novel consumer product. If there are other demands, let alone.

When I was resting, I heard a conversation, but I still had some questions about it. However, this conversation made me suddenly clear, and a new colleague kept asking me, why don't we use the so-called mainstream OS? For example, W and a or others mentioned above, a very senior engineer said that W and a are both others, in particular, although a is free now, we may not be able to use it any day. We have our own practical experience. Indeed, you can continue using your own things with confidence. You don't have to worry about copyright, whether you can continue to be free, as long as you are stable and functional. After all, we are the original manufacturer. Of course, W and a will also involve some.

After a month of study, we can look back at the entire architecture of wince and feel more clear. We often cannot get rid of a familiar traditional Microsoft shell, I think there will be no OS when I see a command line or a piece of black call. In fact, shell is just a representation of OS. Although it is also a very important part, at the entire operating system level, the operating system is at the top level, and at the customer level, we often feel that the operating system is on this layer. Below we do not want to carefully examine it, because, some talented people from Microsoft or Google experts helped us solve the problem. Now, let's look at the shell of wince, which is really superb. In order for shell to fully respond to the control of underlying devices and functions, the current hardware status can be quickly reflected. The shell contains a wide range of API interfaces connected to the hardware driver, and these driver interfaces are only one layer, at this layer, we can see that Win32 APIs are represented by the kernel, GWES, device management, file system management, and network. This layer is the core of the entire OS, the basic OS does not need to manage devices and files without coordination Communication and work collaboration between the same device, that is, the interface between the driver and the driver. After this is done, it is the underlying actual driver interface, oal, stream device interfaces and local driver interfaces, which are managed by the kernel layer. After these interfaces are completed, they are of course the underlying bootloader. Here we will look at the actual chip architecture, whether it is arm or MIPS. Here we divide it into another boot layer. Of course, sometimes we will directly merge it with the underlying driver layer.

The above is so long-winded that we want to illustrate one thing. We just use wince as an example. Most of the OS processes should be in a hierarchical structure, which is roughly divided into the shell app layer, kernel kernel layer and driver Driver boot layer, but we may actually have only one driver layer, however, you still have to understand each layer to understand the actual working mechanism. Alternatively, you don't have to worry so much about it, because it is actually better, and it doesn't feel difficult to adapt to what operating system it is. What layer does it mean, you can simply follow the standard format. Although it is not recommended, you can do this in the case of urgent development. Afterwards, you will learn the principle and then become enlightened.

The last step is to give a simple idea to some recently developed things. Recently, we have developed things mainly around keypad and remote control lever, and occasionally touched the display part, however, because the display format is relatively fixed, it is just a simple driver porting process, coupled with the original factory cool... The supplier quickly gave accurate screen information, and the actual debugging was useless for a few days. It spent a lot of effort on a touch capacitive button and easy point. The touch capacitive button is generally used in white appliances, LG's chocolate series is also said to use that stuff, but to tell the truth, I don't know why or why it is difficult to debug it. In summary, it can also be used on other devices, check from power supply,

1. Whether the power supply is normal, whether the ripple is normal, whether the power supply is clean, and whether the grounding method is clean;
2. Whether there is a reset action;
3. Communication Mode: whether the communication is normal. Generally, the waveform is captured;
4. It is important to verify whether the correct device ID is read by the initialization value and register;
5. Check the working status of the actual operation, perform the interface at the upper layer, and try to debug the actual operation;

Capacitive buttons do not have an accurate value, which may be due to the voltage. The voltage of 3.3v on the board may lead to inaccurate sampling, however, for the first time in this development, the actual pressing and lifting actions of the buttons need to be done here.

When a capacitive button is lifted, it does not send an interrupt message indicating that the master device can read the key. This means that we can only judge whether it is pressed and cannot judge whether it is lifted. Here we discuss a solution with our colleagues, when a timer is pressed, the system first checks whether the timer is enabled when there is an interruption. If it is enabled, the system closes the timer. After the process is completed, the system opens the timer. When the timer times out, the master device reads a key independently. If the key is kept pressed, no action is taken. If it is lifted, the key value will be restored to the normal state. This is the release action, these processes are used in combination with AP control. This example illustrates the connection between the upstream and downstream drivers and the AP. In addition, there is much easier to do with easy point. easy point has a dead zone, once the device joystick enters this zone, the interruption will not occur. Once there is no interruption, we think that the easy point will return to the center point, and we will immediately perform the lifting action corresponding to the previous action, compared with capacitive buttons, buttons may have the concept of combining keys, while pressing two at the same time. However, these two functions can be solved in the same way. After all, keys are used.

The above are my personal thoughts and experiences over the past month. We also need to continue the revision. If you have any mistakes, please kindly advise.