From June 29th to 30th, the “2021 China Automotive semiconductor Industry Conference” hosted by Gasgoo was grandly held. This conference mainly focuses on the current situation of core shortage of Chinese auto companies, the safety construction of supply chain localization, the construction and design of vehicle chip platform, the chip requirements and application cases in the field of autonomous driving, intelligent cockpit, the application of power semiconductors in the three powers and chip testing Discuss topics such as functional safety, and seek the future development of the industry. The following is the speech of Chen Dawei, deputy chief engineer of China Electronics Standardization Research Institute, at this conference.
Chen Dawei, deputy chief engineer of China Electronics Standardization Institute
Good morning everyone, today I will share with you the topic of automotive chip standards.
I believe everyone is very aware of the AEC-Q100 standard. So far, there are 6 standards. The AEC-Q100 standard is basically divided into four grades. The lowest is -40 degrees and the highest is 150 degrees, which was a division five years ago. As a car-regulated chip, from design to application, a series of standards must be followed at all stages.
At the design stage, everyone knows the ISO26262 active functional safety standard. This standard only has clear requirements for chips in 2018. Before that, automotive-grade chips followed high-reliability design requirements. This may not be mentioned too much in consumer electronics and industrial control chips, including redundancy, special Reliability design technology, of course, will also bring increased area and cost.
In the manufacturing stage, first, tape out on the IATF16949 production line. All auto parts are required to meet the requirements of IATF16949, so they must be made on the IATF16949 production line. According to AEC-Q100 requirements, all chips around the failed chip in the CP test must be picked out, which increases the cost of automotive chips. There’s also statistical analysis, a zero-defect guide, and more. The so-called zero-defect is pursued through a series of management controls, and it is not really possible to design a zero-defect chip. Reliability certification, these are the six standards I just mentioned. Today, I mainly discuss the AEC-Q100 standard with you.
In 2010, there were very few domestic automotive chips, maybe just a few. In 2015, there were about 10, but they were mainly in the aftermarket. In 2020, there were about 40. As of yesterday, the statistics 70 companies, 250 quasi-vehicle chips.
In fact, since 2012, the National Nuclear High-Based Project has supported the research on the key technologies of two domestically produced automotive Electronic chips, and made on-board information system chips. There is no industrialization index assessment. By 2014, the R&D and industrial application of the body controller chip and the R&D and industrialization of the SOC chip of the body electronic controller were arranged. 2019 gave Horizon another project.
In the case of domestic chips, the design of high-reliability chips is still a bit lacking. Vehicle-specific chips are designed, not tested. In addition, most domestic chip companies have just begun to contact the ISO 26262 standard, and maybe only a very few companies have just passed the ISO 26262 system certification, but the chip certification may still be in progress.
The control of the production process, at this stage, the process and line width of a chip may be determined, and then the production line to be selected just meets the requirements of TS16949, so there is less room for choice. Sometimes there is no TS16949 certification, and it has to be done.
The testing and certification agency mainly verifies the compliance of the chip to the AEC-Q100 standard through testing. It does not mean that the chip is turned into a vehicle-grade chip through testing and verification. In fact, it is a process of verifying whether the chip meets the standard. Test institutions should reflect their respective strengths and work together to cover all the contents of AEC-Q100. For now, our testing facility still has a lot to improve.
There is no national standard and no national standard, which makes everyone’s understanding of the AEC-Q100 standard confusing, because AEC-Q100 is a series of standards, including the following test methods, which are a series of nested standards.
This is the general structure of the AEC-Q100 standard. There are scopes, reference documents, terms, requirements, appraisals, appraisal inspections, and appendices in front of them, which are basically similar to the existing standards.
These are other supporting standards of the AEC-Q100 standard. The following standards are all supporting the AEC-Q100 main standard. If implemented according to the AEC-Q100 standard, both supporting and auxiliary standards need to be used.
This is the AEC-Q100 test grouping chart, which is divided into seven categories and 41 specific test items. Group A is the accelerated environmental stress test (6 items), and group B is the accelerated life cycle simulation test. Group C is the package assembly integrity test. Group D is the chip manufacturing reliability test. Group E is the electrical verification test. Panel F is the defect screening test analysis. Group G is the cavity package integrity test. The red color can be deleted here. The last deletion is the G group. If it is a plastic sealing test, then the G group can not be done.
There is a power temperature cycle PTC in group A. If the power of the sample is greater than 1W, this test should be done, and if it is lower than 1W, this test can be omitted. Group B volatile memory durability, data retention, working life, that is to say, there is no memory in the chip itself, then this project can not be done, if there is memory in the chip, and it is greater than 1 megabyte, the memory must do this test. For example, we sometimes see MCU samples. Generally, there is memory in the MCU. If this project is not done, then for the MCU, the AEC-Q100 experiment is flawed. C group solder ball cutting, if it has no solder ball chips, it can not be done. Group E has short-circuit and hot-spot effects, which are required for power devices, which is probably the case.
The more time-consuming ones are group A and group B, all of which are experiments with a cycle of 1000 hours, and each group is basically done at the same time for 1000 hours, and some need to be powered on. For AEC-Q100, in some cases, all kinds of tests are deleted, and at least 28 tests are required, that is to say, 28 tests are done, and the AEC-Q100 standard is considered to be fully completed. As mentioned on the first page of AEC-Q100, only when all items pass the test of this standard can we claim that this material has passed the AEC-Q100 certification.
Here are a few places where the existing industrial regulations or consumer chips are not the same. Group A is similar in that the stress is slightly stronger. The difference is that of group D. At the beginning of the design and production process of the car gauge chip, it must be considered that this chip is a car gauge chip. Otherwise, you cannot get this data. Before this data, AEC-Q100 lacks Group D. The first item of group E is the test before and after the stress test experiment, which is very critical. According to the AEC-Q100 standard, the test of group E is related to all the tests of group A and group B above. That is to say, the first step is to do the test of group E1, and then to do group A and group B after the test, and then calculate its failure rate after the test, and calculate the value of Cpk, if it is greater than 1.67, it is considered acceptable. Because for a long time before, consumer products basically used normal high temperature, and industrial control chips also used normal high temperature.
Another special thing is the EMC problem. The EMCs made in our country are all EBCs of the whole machine and system. Now there are many EMC laboratories in various places, but there are still relatively few EMC laboratories that actually do chips. Our unit has an integrated circuit EMC working group, which has been working for three or four years, and everyone has been working together to set standards.
In addition, the most different is the soft error rate of group E, which basically does not have a large number of experimental conditions in China at present. There is only one company that makes single particles in China, and there is no way for car chips to line up. This is the only particle accelerator base in China. This project is cumulative. After the cosmic rays hit the car, if the time is long enough, it may cause a certain soft failure.
Group F is the average statistics of AEC-Q100 in the production process, these two are also to be pushed back to the data analysis in the production process.
This is a group A accelerated environmental stress test. How to identify whether the criterion is 0 or 1 requires passing the electrical parameter test before and after the test.
I would like to report to you here that I made a catalogue of innovative products for automotive electronics last year, and in fact there is a catalogue of automotive chips. This was done by the Ministry of Science and Technology last year under the leadership of the China Integrated Circuit Innovation Design Alliance, and the second edition was made this year. So far, a total of 293 product registrations have been received from 73 companies, which is a lot better than last year. Last year, there were 40 companies with 130 products, which has basically doubled this year. Last year, we came here to register. This year, we proposed that because the AEC-Q100 standard is the most basic front-loading entry standard in foreign countries, this time we asked everyone to submit these 293 certification reports. In the end, 25 companies provided 38 AEC-Q100 standards. Q100 report. After reading these reports, it is not optimistic. It seems that the most concentrated is that the E1 group does not have it, and most of them do not. There are many reasons for this. Some of the entrusting chip companies said that they did not want to do it or it was difficult to do it, and it took a long time to develop programs and debug. Long time is more troublesome. There are also test institutions that do not have test equipment, try not to do this, and so on. In the end, 25 companies provided 38 reports. The common problem is that the three thermoelectric test was not done, and the group D did not do it.
The automotive electronic components standard working committee will make some standards in the next step. If you are interested, you can scan the QR code, and there is a form in it to join the organization. The 2021 product catalog just mentioned, if you need to add it, there is still a chance, and it will be released at the Suzhou conference on July 15th.
For chip reliability design technology, we will carry out research on the ISO 26262 standard in domestic chip verification to help chip companies design chips that meet the standard from the source. The tape-out and packaging and testing plant implemented TS16949, accelerated the formulation and implementation of relevant basic reliability standards, and established a national automotive chip research and development project.
Finally, there are two suggestions. One is for chip companies. AEC-Q100 can actually be done by itself, not necessarily by a third party. All three types of laboratories can do AEC-Q100 experiments. The key is to have a correct and deep understanding of the AEC-Q100 standard, otherwise it will be problematic for anyone to do it.
For Teir1, how to judge that the material given to you meets the AEC-Q100 standard, Bosch has a team, which includes microelectronics experts, electronic circuit application experts, vehicle machine experts, vehicle experts, statistics experts, In this way, all the data of each report can be reviewed strictly, carefully and comprehensively according to the standard. If there is a problem in the review, it is not that you go, it does not meet my requirements, but will patiently help the other party to solve the remaining problems. Completed tests are completed and progress is made together, and finally this chip will enter its supply chain system.
For tier 1 interpretation test reports, here are a few suggestions: 1. Is the content comprehensive? 2. Are there any design parameters? Some reports seem to be quite high-level, but in fact the key content in it is absent or ambiguous, which will be misleading, because we are not familiar with the requirements of this standard, so there will be problems. 3. Are the test conditions in compliance with the regulations?
Finally, to sum up, the understanding of AEC-Q100 is very important. Only when all of them are met can they claim to have passed the AEC-Q100 standard. You don’t need a third-party test, you can test it yourself. The other 7 categories of 41 experimental items can be deleted, but there must be reasons to explain. The same series of products can use general data to reduce test items, but need to provide proof. Before and after the reliability test, a three-temperature electrical test must be carried out, which is the only basis for judging whether it is qualified or not. Some test item criteria are determined by the user, such as EMC. AEC-Q100 test data is to prove that the chip meets the standard.
Source: Gasgoo
The Links: 2MBI200NB-120-01 6DI75MB-050 IGBTCOMPANY