(The author solemnly recommends this article)
The following content may not have the color and organization of grammatical language structure, but it should be more valuable and help you solve problems with limited financial resources and lack of understanding of reliability. In the absence of problems and the lack of accumulation of reliability experience and knowledge, the company's product reliability has been improved, which has greatly reduced the product complaint rate and brought the level of product failure analysis and improvement to a new level.
1. About test instruments
You can buy some cheap test instruments yourself without spending any money. If you want to achieve good reliability, you are a "white wolf with empty gloves". You have to be kind and don't go too far. Chasing a girlfriend who is a bit ugly also costs money on phone bills.
If you just want to spend as much as a phone bill, then choose these, I/V curve tester, you can buy a simple one for 1-3k, or you can buy one for over 10,000, which can be connected to a computer. , can record IV waveforms to compare the differences in IV curves of failed devices and non-failed devices. To analyze the possible faults and causes from the curves, we cannot rely on machines, but on the free human brain of the experimenter. Yes, this instrument is very helpful for internal fault analysis of devices, especially ICs. There were some reliability trainings to promote this instrument and method, which left a deep impression on some trainees. They felt like they had found a treasure. Who knows, this instrument has been used for many years in scientific research institutes of aerospace, weapons, electronics and other ministries. It is recommended that everyone study the characteristics of the IV curve and think about it for yourself, it is quite interesting.
The other is a multimeter, which is a cheap pointer type that costs dozens of yuan, one or two hundred yuan, or more than a thousand yuan. There is no need to mention its function, right? If this is not clear, you should not be the readers of this article. I do not mean to ridicule or disdain, it is just a matter of specialization in the field. However, it is recommended to buy one with a temperature measurement function, so that you don’t have to worry about the temperature measurement method separately when doing thermal design.
The other one is the voltage regulator. You can also buy a few. It is not expensive anyway and its effect is huge. If your product is exported, it will definitely be available in the 110V area and 60Hz area. Yes, at least these can simulate the voltage conditions at the user's site. In addition, when applying electrical stress to the instrument, the upper and lower limits of the voltage can be easily realized with this thing. When doing highly accelerated life experiments on products, there is an experimental test. Let the power supply of the instrument under test fluctuate back and forth between the upper and lower limits to test the machine's ability to withstand electrical stress. At this time, using two voltage regulators and a time relay, you can perform part of the HALT test for less than a few hundred yuan.
An oscilloscope is necessary. Who makes circuits without an oscilloscope? If you are starting a new company, it doesn't matter if you really don't have any money. It doesn't matter if you don't buy it for the time being. It will cost more brain cells to analyze.
If EMC work is carried out, a spectrum analyzer, a near-field probe, and a low-noise field amplifier are generally required. I once discussed with a retired expert from an aerospace laboratory what is the simplest EMC test equipment solution. The old expert talked about spectrum analyzers and near-field probes. The veteran military workers back then had no instruments. Relying on these two, half of China's military industry EMC has been established. But according to my practice, it is best to equip a low-noise field amplifier. After all, the near-field probe is a radiation induction. If the polarization direction is wrong and the antenna resonant frequency deviates from the actual radiation interference frequency, the amplitude of the collected interference signal will be affected. It will be lower. If you add an amplifier and amplify it by more than 10 dB, you will still have a good visual effect. Spectrum analyzers are quite expensive. HP's second-hand ones cost tens of thousands. However, if the frequency range of the instrument produced by your company is not very high, you may consider buying one with a narrower bandwidth. Taiwan's Goodwill has 150k-1GHz. It's a little over 10,000, which is a bargain price among spectrum analyzers. I'm not a shill for Goodwill. In order to prove my innocence, I would like to disclose that the model I provided is now discontinued. You can search in many places. There are near-field probes imported from Canada, and some made in China. Mr. F is also making them, and they are pretty good. The amplifier is very easy, and can be equipped by any probe manufacturer.
The above are all equipment that can be solved within 30,000 yuan, among which spectrum analyzers, oscilloscopes, and near-field probes (several thousand yuan) are the three most expensive ones.
Other equipment is a bit expensive. If you have money, you can come to me. I am good at making expensive laboratory solutions. If you don't even know how to spend money, you are still messing around in the world!
If you don’t want to spend money to purchase the remaining instruments, you can also use your own wisdom and technical skills to make some simple substitutes yourself. It is okay to do your own experiments to feel the performance of the product. , do not use its results to write a report to the military, foreign customers, certification agencies, etc. Although dog meat can satisfy hunger, it cannot be eaten at a banquet.
High and low temperature test chambers cost tens of thousands, and I can’t afford them. Get a pressure cooker, modify it, add water to it, put a waterproof box on the water that is not afraid of boiling, put a circuit board in the box, and connect a temperature measurement Sensor, make a circuit yourself, measure the temperature as a feedback measurement point, and control the power supply of the pressure cooker. In this way, the temperature in the control box can be controlled to be relatively constant through the circuit board made by yourself. The circuit board can simulate any temperature within 100 degrees. , although the temperature is not very stable, it can basically reproduce the fault as thermal stress. A few hundred yuan is enough, right?
The same is true for the salt spray test chamber. It is blown with high-pressure air. Now there is a compressed atomizer sold in pharmacies. Use an infusion bottle to drip salt water, blow it with compressed air, and form a salt band along the small tube. Put the atomized gas in the circuit board into a anti-theft plastic bucket, cover it with plastic wrap, and blow it for 2 days to see whose circuit board can withstand corrosion. It cost more than 300 to get it done. I thought I was the only one who could think of this method. Later, I visited some companies and found that everyone had similar methods. I was actually frustrated by this for a few days.
Vibration test benches are also relatively expensive. I haven’t found a good way to save money. I’m sorry to everyone. If necessary, just find a professional laboratory to do the experiment. If you are well prepared, the time will be shorter. , and it doesn’t cost much.
The last thing worthy of special mention is the HALT comprehensive test bed. This guy is really good. It comes from Italian ACS. It costs more than one million U.S. dollars. The most valuable thing is that it is embargoed to the mainland. , is always suspected of being used for military or military product testing or design verification. There are similar products in the United States. The manufacturer can’t remember clearly. Their temperature change rate can reach more than 60 degrees/minute, or even 100 degrees/minute. The instrument During the test, a large liquid nitrogen tank should be placed next to it. Without liquid nitrogen, I don’t know if there is any other way to achieve such a rapid temperature change rate. Domestic Suzhou Testing Instrument Company has similar instruments, but the indicators are inferior, but the price is as low as 1/6-1/8, more than 1 million yuan. The best way for small and medium-sized enterprises is to find a laboratory and spend money to rent equipment. The price of about 1,000 yuan per hour is still affordable.
As for others, X-ray machines for ICs (if you can’t afford it, find a professional laboratory to do it), you can clearly see the faults inside the chip, which are limited to faults in metal wiring connectors, similar to medical The X-ray machine on the machine can only see bones but not soft tissue), IC test instruments (newer chips cannot be measured without accumulated test data in the test machine, conventional mature chips are OK), microscopes (within 600 times, Used to observe local pins and mechanical failures of the chip).
2. About analysis technology
In reliability work, people are the most advanced instruments. Managers who only shout to buy experimental equipment are stupid managers. Southwest United The teaching conditions of the university were poor enough, but how many talents were produced. The conditions of the military industry enterprises in the 1960s were poor enough, and how many cutting-edge results were produced. Yu Min led a young team at the East China Computing Institute in Jiading, Shanghai, to gnaw out They didn't spend a lot of money and didn't have many people to understand the principles of the hydrogen bomb. They were just a theoretical physics team of 20 or 30 people. Therefore, how to play the role of people is the most important in carrying out reliability work. There is a misunderstanding here. A boss once asked me a question, "If I transfer a product expert to work on reliability, won't the product side collapse? It's better to recruit reliability technicians from the society!" Chapter 1 First, if product experts work on reliability, the product will not collapse. Experts can intervene in product development from testing, review, and control, and the intensity will be greater; second, the basis of product reliability is familiarity with the product, and then reliability Reliability experts recruited from outside need to be familiar with the product first, and the most fatal thing is that if the newcomers want to start to play a role, how can everyone accept it in the short term from a management and psychological perspective? Reliability is a matter of finding faults. . Many companies are also carrying out reliability technology work, but the results are not good. This is the biggest problem.
Chairman Mao said, "After the route is determined, cadres are the only decisive factor." If the boss does not solve the only decisive factor, who will be blamed?
I have summarized a phenomenon. There are more than 20 R&D personnel. In a human company, the strength of product designers is inversely proportional to product reliability. The more powerful the product development engineer is, the worse the product reliability will be, and the more obvious the downward trend will be. Because they are becoming less and less constrained, losing the power to supervise leads to corruption, and losing the technology to supervise leads to unreliability. The reliability technology department monitors product development through analysis, testing, review, summary, and management. Experts fight against experts. The more arguments are made, the clearer the reasoning is, and the products are getting better and better. This is the first rule of analyzing technology: "Let technology be effectively supervised."
The next one is the I/V curve comparison of components. This method is relatively simple. It is to measure the I/V curves of the device pins to the power supply, ground, and adjacent pins. Compare several devices. Come down, there must be a difference between bad and good. Once you determine which one is really bad, go deeper and check the process and internal wiring of the IC. Analyze the possible failure mechanism based on the theoretical derivation and analysis of the characteristic curve, and then find the preventive measures. technical measures or management measures.
Another analysis method is visualization, which makes fault phenomena and fault mechanisms visible. Visualizing principles is the eternal pursuit of experimental science. In terms of reliability, visual fault analysis methods for devices include microscopes and X-ray machines. , transistor/integrated circuit testers, oscilloscopes, and on-site simulations. Remember, rack your brains to visualize the failure. By doing this, more than half of the problem has been solved. I once did an experiment on the interference of an electrosurgical instrument. I went to the site continuously for a week. I was able to visualize the fault and the solution, but it was too inconvenient. Later I bought an electrosurgical device and made rapid progress. In fact, everyone is doing this. Yes, I just proposed the concept of "visualization".
To do some test equipment, I visited a company. They were not sure about the reliability of the button, so they made a reciprocating motor drive mechanism, and a driving rod went back and forth to realize the button pressing/ejection. Each cycle mechanism counts one number for this action. After completing the number of times, you can check the condition of the buttons. It is a very good tool. You can do a lot of similar tooling. You can measure friction, springs, relays, knobs, etc. Facial masks, etc. Use machines to replace human labor to verify.
Others have equipment and can do some experiments. Based on the analysis of the experimental results, a brief summary is: standard compliance test, limit condition test, edge limit condition combination test, saturation operation test, and reliability enhancement. testing, highly accelerated life testing.
The standard compliance test is easy to understand, just find a standard, safety standard, performance standard, and test against the standard;
The extreme condition test is also easy to understand, see what the expected design meets What are the limit index conditions? Just test them. Be careful not to forget the limit conditions outside the environment. For example, if input signals from other external devices are used, the input fluctuation of the other device is also one of the limit conditions;
I need to explain the edge limit condition combination test, because this term cannot be found in textbooks. I created it myself. It selects edge limit conditions, combines them together, analyzes the worst combination conditions, and then simulates Among the tests, such as the combination of low-temperature operation + the machine's simplest function and minimum heat dissipation, the lowest air source pressure input + the highest ambient air pressure, etc.; the condition indicators of this experiment all meet the design requirements, but it is indeed worse than the individual test conditions. ;
The saturation operation test also needs to be explained, because it is also created by me. This test method is to save time. For example, for a relay, the actual working status is once an hour, 10 minutes each time, give it Continuously conductive for several hours, running for one hour is equivalent to actual running for 6 hours or more, because this kind of test is continuous fatigue;
The reliability enhancement test is the official test method. It is to increase the stress profile and see the performance of the equipment under higher stress; check it yourself, there is a lot of information, but it is a pity that there are too many reviews. It is a practical method. Hehe, can you please ask me too and pay appropriately? For a little fee, my company needs to be run!
Highly Accelerated Life Testing (HALT) is a test method that I have been advocating recently, but the equipment is very expensive and imported ones are embargoed. However, if the module-level individual tests are fully conducted during the development stage, HALT will basically not find too many problems, and some will have almost no problems.
The only difference between it and the single item is the dual failure mode of random vibration under high temperature variability.
3. About management methods
Now that the equipment and technology are available, don’t think that it will be fine. There are many companies with very good equipment, but their work is not carried out well. The equipment of large domestic hospitals is not necessarily worse than that of foreign hospitals, but their medical skills may not be higher than those of their foreign counterparts. What is the difference - management.
The first is for senior personnel to conduct a "substantial" review of the product developers' designs. There are almost no companies that don’t do reviews. What’s the result? One is substantive and the other is symbolic, the effect is very different. The solution is the same as the first one in the technical analysis chapter above. Let the technology be supervised by senior personnel. Another thing to note is that the reviewers are full-time. As long as one person has two responsibilities, especially related responsibilities, he will balance it himself. , not completely. Administrative monopoly companies are not just like this. They are both athletes and referees. The quality of their governance results is self-evident. The same is true for the technical review management within the company. This is also a common mistake. One of them is the system engineer and test engineer in the technical department who is concurrently served by the project manager.
The second is to use complaint records, which is not suitable for new companies. The difference between a century-old store and a new company is accumulation. The analysis of complaint records and improvement methods is very helpful for new product development. When I do technical management consulting projects for enterprises, I always need complaint data and records. I take the complaint records to participate in the review meeting. It is actually quite simple. A layman can comment on a lot of problems. Why do you hold a golden bowl and beg for food? , this kind of approach is not what one family or two families do.
The third is that the problem was discovered and the solution was found, but after tracking, it was found that the result was not ideal. Grassroots technical personnel and technical managers often have limited professional experience and do not fully understand the company structure, Regarding processes, strategies, etc., their solutions are technical and not systematic. I gave an analogy to engineers, and I said it this way,
"Everyone wants a salary increase,"
"Salary must be based on the value you create for the company, and then from the value created Is it your own share of the value?"
"The most direct way to get a salary increase is to create more value." "There are two ways to create more value. , I can do one kind of work, but others can't. The value of my work will increase the value of everyone, and my value will be greater; the other kind of work is something that everyone can do, but I did it once and solved the problem 100 times. It's equivalent to creating 99 times more value than others. If I do it this time, it's worth 100 yuan, and someone else's is worth 1 yuan, so I can get more."
"If a machine breaks down, You have fixed it, and it is a one-time thing; but if you find that it was caused by the customer's misoperation, and then place a change order and add a warning mark on the machine to prevent misoperation, will subsequent hidden dangers be avoided? It avoids repairing the machine 100 times in the future. ”
There are many solutions in terms of process. Reliability problems can be prevented from every aspect of the process as long as they are discovered once or twice. Why worry about reliability not being improved? , in fact, look at how many product quality problems are repetitive, and the proportion is not low.
Everyone will look for it separately and analyze it. It turns out that similar problems abound in real-life management. Chinese-style management is all about strategy and blue ocean, but Ballmer in the United States is immersed in management details.
The above methods are less than 20% of the reliability work, but they can solve a large part of the general problems of product reliability. There is a 20/80 rule in nature, and it can also be done here. borrow. If our expectations are not very high, if we can achieve these things for the time being, we can establish ourselves as a good enterprise. When our enterprise has hardly done anything about reliability, it is still running and developing rapidly. , but the executives are worried that they cannot continue to expand rapidly, cannot strengthen brand reputation, and repairs and services are eating away at previous profits. If we want to become a truly great company and build a brand company, these are just the first steps. To go deeper, we can find a teahouse, choose a good time, and sit down and talk.