Yesterday our group of 10bit-40m ADC measured come, oneself finally can relieved, can unencumbered graduation.
In the evening boss BG Whole group of graduates, drank a lot, I said to the boss: "This 3 years in the group, I have it as a career to do!" "
Yes, although in these 3 years, setbacks may be more than success, joy is always intertwined with pain, but whether it is technical or work, even in the capacity of liquor, I have been honed, up to 3 years.
On the verge of graduation, I can say: come here to read 3 years, I have no regrets ...
I woke up this morning with a sudden impulse to write this 3-year master's career.
One to leave some memories, to commemorate the lost time;
Secondly, I am grateful to my mentor, Professor Yanghuaching.
Sunline by the way propaganda + marketing our group, hope he can successors, the better (so hair in this version, everyone less pat ah ha: P)
(1) Reasons
I am a telecommunications professional, sophomore summer vacation also in Lu Jianhua teacher there finished a SRT project, very cool.
But in the end, the route was chosen, and the transition was made during the SARS period.
At that time, the domestic IC design industry just up, the star micro-Xinfei force has not been listed, but from many news I feel that China's IT industry bottleneck lies in the chip.
In particular, it is reported that the Shenzhou spacecraft on the day after the foreign countries began to blockade of our electronic components. Our manufacturing added value is low, but also because we can not master the design and production of chips, only a large number of imported chips, become someone else's fitter.
When I was doing SRT, I thought, our system can be comparable with 802.11, but can't sell, why? Because a chassis-sized system was not made into a chip.
Later, it was found that chip design is a very high threshold of the industry.
Not only does this mean fewer people to compete with later, but it also reduces the opportunity cost of working 3 years later than the undergraduate.
My undergraduate time to find internships can be found to write FPGA work, if Master's degree or only will write FPGA, then and in the industry wrote 3 years of FPGA undergraduate ratio, it is not necessarily dominant.
and chip design is not the same, because the undergraduate knowledge reserves is not enough, the company is difficult for undergraduates to do the core research and development, often do application engineers such peripheral development, and the threshold of undergraduate students formed.
Compared with other majors, the IC also has a training threshold, a single stream of tens of thousands of to hundreds of thousands of, the cost is not to adjust a few circuit boards, do a few simulation can be burned.
Take me for example, Master 3 years altogether 3 times, the conservative calculation of the cost of the flow sheet more than 60,000, plus other costs, a master down can burn 1,000,080,000 than the tuition is too value ...
Another secondary product of the high threshold is higher paid water.
I have not done detailed statistics, but from these two years of technical class salary, IC company's average salary is higher than do communications, software and other professional.
A proof is that every year a lot of communications, images, letters and other professional graduates choose to work in IC companies.
Of course, at the end of the feeling of the circuit is better than doing the algorithm.
AEW's project was a tough one, but for ZHANGZQ's negative feedback amp test, I was willing to start thinking about it one months in advance ...
With these 3 points, I decided to devote myself to this field ...
If the choice of line is considered, then the choice of my boss is very hasty, or even reckless.
Originally pushed the research is planned is another group, did not have any contact and understanding of teacher Yang here, but after the completion of a volunteer after a cause I hurt self-esteem, the provisional decision to change the group.
Once wanted to past the Chinese Academy of Sciences to do godson, finally in Huahua Education + flicker (:P), chose the current boss ...
Instead of quirks the boss's famous "audition interview" Every year, he was able to go straight into the group Heihei
BTW: The boss will interview a large number of students every year before pushing the research, this has become a tradition.
And as far as I can see, almost all of the students in the interview were very badly filmed, many people because of self-feeling is not afraid to come.
In fact, this and job-hunting technical interview is similar, pat you do not represent not satisfied, the reason is very simple: shoot to death, understand you do not know what to assess the total know what;
1+1=2 will answer, this person can not evaluate what level, in turn, if the interviewer does not shoot you, stating that his level may not be as good as you, to go there is likely not to learn things ...
(2) Graduation Design
The first direct contact with the boss is the assignment of graduation designs.
As the saying goes, women afraid of marrying the wrong Lang, men afraid into the wrong line. After receiving the notice from the boss, I was so nervous that I was afraid to give my task no fun, or to give others a good life.
Remember that day 7, 8 of the students sat around the table. The boss first task is to give me, after listening to a description, I said: I do not want to do this .... So the boss described a new task, and I said, "I don't want to do it."
......
This process finally lasted 6, 7 rounds, still did not find the desired job. See is going to noon, I have to say, you first assign to others!
Remember that day will be open from 10 o'clock to 1:30 before the end, at least set up a topic.
Think of me is too impetuous, capricious and picky, in fact, the main reason is that I do not know what they want. At that time only know to do IC, but do the system or the number? Is it analog or RF? There is no concept.
For fear of leaving a bad impression on the boss, after the meeting hurriedly fill up a letter of apology, fortunately he did not blame me what.
After the 2nd semester, my bi set up the topic quirks to a wireless transceiver project for the system simulation, but there is only one of the United States standards and senior discussion of the system structure of the prototype. Soon, found that my work can not only be limited to simulation, but also with the senior members of the system to discuss the structure of the allocation of indicators.
Transciever, the most complex system in the analog circuit, needs to compromise the linearity, noise, gain, power consumption and even cost of the circuit by more than 10 factors. and the standard requirements of more than 10 system indicators and these circuit indicators are inextricably linked, no experience.
At first, the system structure is my day to write a program for everyone to discuss, often this indicator is pressed down, the indicator is not satisfied with the requirements, and other indicators are almost, a module of the brother said to his module requirements outrageous, do not, on a vote veto ...
But this job is really interesting, one is to find the undergraduate course of almost all knowledge to use, from analog circuit, signal system to communication circuit, DSP, communication principle and even senior microwave circuit, antenna principle and so on, the basic concept can not be wrong, otherwise it may be a wrong indicator to ruin the whole program.
The other is the discovery of the system structure this thing is really changeable, endless, although everywhere is landmines, but never found themselves to the cliff, always see hope, people will not because of temporary difficulties and discouraged.
System Solutions we basically feel that the clearance, began to carry out the indicator allocation and system simulation, finally realized what is called "the key to get the girl, the boss does not master".
Although I have to assign, but no circuit experience, do not know who the module can do what indicators, and do the modules are senior, also dare not offend, can only ask once we can do how much ah, come back to see the simulation to meet the system indicators. Can, then happy, not, you have to see who to say good words, let him try to do the indicators up ...
The advantage of the system is that it does not hang on a tree, and a module cannot be improved, and can be adjusted to meet the overall requirements by adjusting the indicators of other modules.
Of course, the disadvantage is here: we all think of their own circuit is not moving, others to adjust, the result is often "tough" brother won the victory, "weak" brother was forced to modify the circuit.
The most deadly is, some seniors on their own circuit is also flaky, today said the indicator can be achieved, two days and say no
So everyone was angry, fortunately, the boss as a fire fighters, around to appease, the project can be difficult to move forward in the bumps.
Using a rookie like me to make the system architecture, like an architect who does not draw a straight line, that is, the use of better reinforced concrete, the building also has a collapsed day.
A year later, when everyone was on the system, a fatal bug in the structure was discovered, and the building collapsed, and PT failed in the first project in the Graduate stage ...
Of course, this bug also allows PT to have a new understanding of the circuit: simulation is always the abstraction and simplification of the physical system, do not know what is to be ignored when modeling, which can not be ignored, blindly believe that the simulation results are meaningless.
PT is due to the failure to take into account the mixer of the local oscillator leakage, so a majority of simulation model is ignored in the physical quantities, resulting in the entire project failed ...
(3) adc,3 years of grinding a sword
For why Master chooses to do ad, is in the process of doing transceiver gradually clear.
Bishi, the accumulation of each module circuit in the group is still very few, in the system design fully realized by the module design ability everywhere constraints of taste.
In terms of scale, ad in the entire transceiver, and even a variety of analog circuits, is the largest, the most complex of a module, the group of students have complained, do the students to do radio frequency map is the "day", and the ad is according to the "month" meter. Have Berkeley doctor 3 years to do an ad out can graduate.
In terms of difficulty, ad is also one of the most difficult modules. Intuitive understanding, in the transmitter-channel-receiver This communication loop, the ad is at the end of the analog loop, the signal through the transmitter, channel, receiver Front end, has introduced a variety of interference and distortion, and ad also need to stand the last shift, the various signals faithful quantification, to avoid those "troublemakers" mixed " Revolutionary troop ", otherwise the next digital filter will also be difficult to distinguish between them.
It has been calculated that a 14bit pipeline ADC, every time in the circuit amplification, the error is less than "270 electrons" line ...
Especially after the RF receiver software, if the increase in frequency, dynamic range, high-speed high-precision ad often become the bottleneck of the system, but also become the focus of our blockade abroad.
At present, the United States to 12bit 50MHz above, 14bit 5MHz above the ad limit exports to China, and the students who have contacted the software radio know that at present, radar, base station and other receivers, often need to speed in 80MHz above, the accuracy of more than 12bit ad line.
Difficulty, fatigue is the disadvantage of factors, in turn, it is advantageous, indicating that his threshold is higher, fewer competitors.
Especially for high-speed and high-precision ADCS on radar and base stations, there are few applications in civil field, and few domestic industry do
This also gives the university and the Institute an opportunity, what we do is no longer just for hair paper, although some distance from the actual mass production, but at least it is possible through our efforts to force the foreign countries to release the blockade ...
In this year, through the literature reading, the domestic academia in the field of analog circuits and foreign gaps have a more sober understanding.
IC design field is a focus on the field of experimentation, top journals need test results to testify, if the lack of basic realization ability, even if there is good idea, can not measure the world-class performance, or can not publish high-level paper.
Due to the difference in process conditions and other aspects, and the algorithm, the other person's circuit plan is difficult to repeat the verification of the paper, not to mention digestion and absorption.
and idea-circuit Design-flow sheet test Verification-published paper this academic cycle is generally more than 1 years, compared to make the theoretical results of the algorithm is much slower.
Especially in the current academia more impetuous, value the number of papers under the environment, more need to calm down to be able to withstand loneliness, otherwise it may repeat the Chen of the mistakes ...
Want me to say, the circuit is to have some ah Q spirit ...
These objective factors have increased the difficulty of our work, but the boss is well aware of the truth and provides a very relaxed environment.
Hardware, we from the beginning of 1 servers, gradually developed to 7 servers 18 CPU core of the cluster day and night operation to ensure adequate simulation validation.
Although the flow of the opportunity to not reach the United States like how to flow, but through the fine management, routine review mechanism, to ensure that there are reasonable requirements of the people can have the opportunity to flow.
On the soft indexes, he never has a thesis on the graduate student who makes the circuit, as long as the test results meet the requirements, he will always find a way to let you graduate.
As a student, in this environment also slowly sank the heart down, at the beginning of the top not to judge the outstanding paper also to be born belly gas, to later also can calm.
Write Leigh words into the SMD to Ming Zhi:
Solve the actual problems in the project is the real mission of our electronic engineering people, send two articles that do not contribute to human knowledge, it seems really a waste of time.
-----Leigh
After all, when the winner of the country is the King loser, compared to the number of papers, idea new and old, the decision will be the missile radar components of the accuracy and speed ...
(4) dedicated + process
If someone asked me 3 years ago what I needed to do the circuit?
I will answer, the concept of physics should be clear.
Ask me again today, the answer is only two words: dedication!
Research 2 started, the group began to do a codec project, St China Company sent an engineer to help us review, to do technical guidance.
The boss gave me 7 people, I press the module to look, found that they do not work, Hoho,8cuo
St is a large company that has strict specifications for the process of development work.
Only from the schematic, big to the module division and naming, small to a switch symbol drawing, even the circuit signal line can turn a few bends, there are strict rules, and every pipe in the circuit size, every simulation, we have to tell the reason to pass review. So many requirements, we are not professionally trained bandits for a moment difficult to adapt, and at first I was not very important.
The first time to submit the circuit was St beat back to scold a getting yelled at, they look very carefully, each team members have received nearly 10 serious errors.
I found not to work the most miserable, was scolded 7*10=70 sentence ...
ST's engineers are very dedicated, speak very straight, a "Use your head!" It always makes me remember ...
Some rules are quantified, such as the signal line can only turn a bend, very good judgment, but some rules are principled, such as the circuit diagram to draw the intuitive, good-looking.
What kind of a good-looking? This is hard to define.
Sometimes see the team submitted to my schematic, I think it is really ugly but everyone's aesthetic is not the same, the team members are trying, I also do not tamper with the work of others, the results of the past only a word back: "ugly!"
Those months, often the morning received email, not to be angry in the laboratory to scold, is wronged want to cry.
Can only find the boss to complain, and then back to a tone of moderation, humbly consult the email, the afternoon to little gentle to appease those same angry and wronged members continue to work.
ST Engineers, although the tone is tough, but their dedication and the pursuit of perfection of the Spirit really left me a deep impression.
From the example of the circuit they gave us, we can see that the size of the design indicator, as small as a capacitor up and down the plate this is difficult to affect the overall performance of the details, are carefully considered, to ensure that every point on the circuit is optimal.
In fact, analog circuit is so, large structure who know, a lot of details added together, decided the quality of the product.
On the contrary, some of our team members usually careless, always feel "almost on the line," the completion of the circuit diagram of a big problem, a bunch of small problems, see let people feel "live rough" ...
Once wrote the report, the structure of their own circuits can be mistaken, let me furious, wrote to them an email:
I always thought that the circuit should be treated as if it were my own flesh and blood. You are familiar with their smell, familiar with the location of every birthmark, willing to give them the best looking clothes to wear, sick will be anxious to give him a doctor, injections of medicine, will prepare for a rainy day, think of which kindergarten they want to go ... And some of them? Please put the child aside to think, looking forward to the number 23rd can lead the wages home ... Usually, if you give the child to breast-feeding, want to do not want the child is not need, take up a bottle of bubbling to the children, choking all regardless, finished hurriedly to rest, see not a look, not to mention the temperament of the child of the temperament at hand, estimated by others switched, do not know ... Do you think this is an attitude towards your own son? Don't you think it's more like a nanny?
It's not a question of love for work, it's a matter of responsibility ...
Later to St debriefing, with another team leader chat up, he said, see which student company is willing to.
Mainly to be dedicated, engineers will have the pursuit of perfection of the spirit ...
I remember correctly, the latter half of the sentence Leigh also to his students said.
Although the project ended in the design drawing phase, the group established a relatively standardized process by working with large companies.
And the importance of the process, there is a profound lesson.
Just in the middle of the project, my first Sigma-delta ADC was back, and there was no movement on the oscilloscope after powering on ...
Later after inspection, all the way bias current is not connected, this mentally retarded mistake let me regret straight to jump, because a successful stream film experience, job-hunting means thousands of dollars of salary premium.
But the boss did not stay on the wrong, he told me, in the IC design field, only hope that through careful to avoid mistakes is not feasible, must be from the process of checks.
After careful examination, it is really my process of the non-standard caused by the bug ...
After 21 years of research, I did not do a one-day master's project except for the question, but it let me know two important things:
Dedication and process ...
(5) Finally
He has a lot of things these days, the last one for me, anyway, is the group of people, who wrote the same.
At the end of the semester, our 13bit200k Sigma-delta ADC measured the noise floor of the <-80db. In the afternoon when the boss invited the graduates to dinner, our 10bit40m pipeline ADC measured the DNL 0.38, INL 0.51, ENOB 9.14b,sfdr 72dB. is a shot of the indicators and test results, said not outstanding, but for us, has been very happy.
Our group of analog circuits started 5 years ago, and officially started streaming films just two years ago. Is the gray brother with us, starting from Sigma-delta. To this day, we have been able to put the ad in the most cicatrization of the pipeline to do the domestic leading, it can be said that 5 years of time to do what others do not do in 10 years.
From the completion of the calculation, came to the line set is 2.5. This 2.5 of the time, I really touch the spirit of Tsinghua two years. In order to rush the film, Sleepless night is a regular thing. Pipeline ADC who has not done, with our design of the senior also no flow film experience, from the circuit to the layout, and then to the long and arduous test, are we 1.1 points to explore out. Sometimes the psychological imbalance, want to foreign students, conditions do not know how much better than us. But want to think, finally still have to do, sometimes really helpless, encounter thorny problem, a long time also have no clue. Fortunately later, in our unremitting insistence, the problem has been solved.
Tsinghua has been 6 years, but read Tsinghua, but it is these two years. I am very glad that I entered the line group, there is a solid boss, in the entire academic community to crush the skull to think idea, save paper time, can do things with ease. AD research and development cycle is long, it takes several generations of effort to be successful, in the time of not seeing the results, money or time, really is in vain to cast in. Fortunately, the boss has never been a quick buck, has given us a lot of space and support. Even when the previous seniors did not have idea and paper, he also managed to put people to graduate.
Many people are talking, Tsinghua people are steadfast. Commendatory to say is dedicated, derogatory said is too stupid, do not understand flexibility. I often think, what is stupid, what is smart? Big article with small thing is clever, then do work is stupid? Remember a brother told me that a university, a certain laboratory, a few years ago had a big project, its significance and godson, however, after the project to get, the entire laboratory is diffuse impetuous wind, just want to save paper, do not want to do things, the results of several years passed, hand holding a little small results, large projects but nothing. I was thinking, if at the beginning, they can do things, do not engage in such a clever, is it possible to see the results today? Even if you don't see the results, you can at least see hope.
Making a chip is not an easy thing, it takes you a long time to endure loneliness. But once a day, the results come out, then the prospects will be very broad. Like the vulgar words: 10 years bided no one, once fame world known. A wise man will not be tempted by the vanity of the present, and he will abandon his sight for a higher goal. In such a impetuous academic, the boss with us sank down to the bottom of our hearts to do, the result is the fastest results.
Career version of the day, met quite a CEO demeanor Hellsing classmate, he very familiar to tell a lot of Baidu anecdotal. Very happy to see him, listen to him say because not to be able to sign to do fighter's place and regret so far, because the body in Baidu for Chinese search engine work and feel full. I think he must be at ease and happy.
I was sorry that I had no chance to go abroad for a degree because I stayed at home and read a blog. But now, I'm happy to stay here. These two years, all of us in the group for the same dream efforts, humiliation, from scratch, sweat and tears, step by step firmly forward. In the wilderness of yellow sand, pouring bricks earth to build our own manor.
I also once did not understand, even pretentious ridicule those dengjiaxian, why give up so good condition, return to this backwater, catch a lifetime. However, if you do not experience the process of pouring bricks from the earth to the base of the building, you will never be able to experience such a sense of accomplishment and a deep sense of mission. To do things for others and to do things for themselves, the two feelings can never be compared.
On the day of the Graduate dinner, the boss was very happy and spoke a lot. Just like the annual graduation dinner, he said to every graduating brother and sister, no matter where to go, you must not forget your younger sisters.
Posts here, there must be the future of the younger brothers and sisters can see. As the future of the elder sister, say a few words to you: P
Originally chose Line Group Edge from the sophomore class meeting, Dickson a sentence. He said that China is missing two technical talents, large system architecture and basic hardware design and manufacture. China's technical bottleneck is also in these two areas, the lack of a system structure, so far without its own large aircraft, the lack of basic hardware, high-performance ad is also embargoed by the United States. In the sense of national defense, these two kinds of things are essential, so talent is also indispensable. From the other side, the group has a flow of experience in the elder sister, graduated from the salary 15k*14 is quite normal.
The line group has always been in the long belt, for analog circuit design with special emphasis on the heritage of the professional, this is very important. Elder sister not only in school when the good teaching, and even out of school to work, but also keep in touch with us, hands-on teaching good sister. The line group is a large family. In such a large environment, I and the same group of two non-micro-electric undergraduate background, the first time design circuit and flow film, research two first Test, it can be pipeline ADC so, really is not easy, no elder sister's accumulation and help, we do not come.
I am happy to see the dawn of years of accumulation. Although we have not built a splendid pagoda yet, we have built a solid foundation and we are building it with great vigour. Do you want to fill the beautiful pagoda with brick by brick, you want to be the tower of the dazzling pearl? Let's join our group now: P
Write my master three years "ZZ"