Functional testing is more and more used in the post-production process, even in the middle of the process, but its system and implementation method are almost completely different from the previous testing. Today's test systems are faster and more compact in most cases. Functional testing is indispensable for verifying the overall function of products, maintaining calibration information, providing data for ISO9000 project and ensuring the quality of high-risk products such as medical equipment. The implementation method of the test is influenced by factors such as budget, output and UUT design, and it is the last one that has the greatest influence on what can be measured. Budget and output will limit the test. In order to get the highest fault coverage in the test, we must pay attention to the selection of components and PCB layout in the design stage. Unfortunately, this is not always the case. Rushing into the market and intense development will often upset your wishful thinking.
The following is a preliminary analysis of how to deal with these restrictions. Some concessions that have to be made for testing (especially in the early stage of design) may affect the design, but it makes the testing work easier and improves the test fault coverage. Please note that the following questions and suggestions are not faced by every test engineer, nor do they need to be solved by every test engineer. Many of these problems will affect each other, so each problem should be evaluated and used flexibly when necessary.
What are the test requirements for the product under test?
Before discussing the design, test system, software and test methods, we must first understand the "object"-the product under test, which not only refers to the PCB or the assembly itself, but also needs to know how much will be produced, expected failures and so on, including: product type.
Structure (single PCB/ prefabricated PCB/ final product)
Test specification
Planned test point
Expected output (per production line/per day/shift, etc.). )
Predicting fault type
Obviously, the "budget" is ignored above, but only after understanding the above items can we determine how much it costs to test a product, and then we can start to discuss the funding problem after figuring out what is needed for fully testing UUT, and only in this way can we know how to compromise to complete the work. After the preliminary report is completed, the company may give you a budget and wish you "good luck"-think about what you can do. You really need "good luck" at this time, but there are others. Here are some.
High density problem
On the surface, component density does not seem to be a problem in functional testing. After all, the main consideration here is to "give an input and get the correct output". It's a bit too simple, but this is the actual situation. When a given excitation signal is applied to the input of UUT, UUT will output a specific series of data after a certain period of time, and connecting with I/O connector should be the only access problem.
However, the density of components also has some influence. Look at the PCB sample in figure 1 (or your own design). You have to answer the following questions first.
Do I need to access the calibration circuit?
Is it important to diagnose specific components or specific areas of UUT?
If the answer to the above question is yes, will the exploration be completed by people or by some automatic mechanical device?
Do you want to use automated test equipment?
Is the I/O connector used easy to touch or connect? If not, can the connector be installed through the through hole contacted by the needle bed?
Let's discuss these problems one by one.
Check circuit
Functional testing is often used for calibration or verification of analog circuits, including checking the interior of UUT (such as the intermediate frequency part of RF circuit) to verify its working condition. To do this, test points or test pads may be needed. One problem in high frequency design is the relative impedance of the test point (path length, test pad size, etc.). However, the impedance of the probe will affect the performance of the circuit, which should be kept in mind when setting the test area. However, automatic mechanical detection and needle bed fixture (discussed later in this paper) can alleviate this contradiction only by requiring a smaller test area, mainly because the accuracy of the automatic machine itself can make the tester detect a smaller area compared with manual operation.
fault diagnosis
If you only use the functional test as the pass/fail filter and do not measure the calibration point, you can skip this part, because the probe may not be needed for the application at this time. In most cases, functional testing is a pass/fail test, because functional testing is very slow in diagnosing faults, especially in the case of multiple faults. However, in some industries, functional testing is going deep into the manufacturing process, such as mobile phone manufacturing. Some manufacturers have to make some key measurements at the PCB level, that is, in the assembly process before the final assembly, which is determined by the nature that mobile phones are easily eliminated. In other words, the mobile phone is designed for low-cost assembly and is not easy to disassemble, so verifying the function before the final test can save rework costs and reduce possible waste products (because the mobile phone will be damaged when disassembled).
Therefore, it is necessary to have enough test points to detect PCB. For example, it is not very convenient to check the J-shaped leads of surface mount devices with a spacing of 20mil, let alone BGA. According to SMTA's suggestion, the minimum spacing between test points is 0.040 inch, and the spacing between pads depends on the component height around the test area, probe size, etc. , but the spacing of 0.200 inch should be the minimum requirement, especially for manual exploration areas. Obviously, the test fixture and automatic mechanical probe are more accurate. There is no doubt that an easily testable design is easier to handle in production than a random design. But engineers usually want to load more technologies in the smallest volume at the lowest cost, which increases the limit of contact with circuit boards in online testing and functional testing.
The market has also responded to this kind of problem. The existing software tools can analyze the design, review it according to the rules stipulated by the assembly and test equipment, and make suggestions to make PCB easier to produce. If these tools are suitable for your product, it is recommended to analyze each design, at least it can quickly point out where the test contact problem is, and the ultimate goal is to make the product easier to manufacture.
Structural configuration meeting the requirements of high density
High density can be a small PCB size, a large number of circuits on UUT, or both. The above topics show that the mechanical and electrical structure of the system must meet the test requirements. The mechanical problems to be considered are:
How to support UUT
test section
Multilayer board test (can the tester do parallel test? )
Input-output connector
In terms of electricity consumption, if it is a multilayer board, which is more economical? Is it a multi-instrument method or a switch converter with fewer instruments? Depending on the UUT structure or the type of instrument required, the answer may not be easy to get.
Automatic test or manual test?
With the increase of the output and speed of each production line (one of the main ways to achieve economies of scale is to improve the productivity of each test equipment), it should be considered whether the test process can be automated. Automated functional testing actually saves loading/unloading time, and there is no need to add other test systems. When considering increasing production, the increase of transportation equipment cost is usually not considered.
The disadvantages of test automation include the initial hardware investment, the integration time with the production line, whether the test system can keep up with the speed of the production line, and the problems that will be brought to production if the equipment fails. Off-line tester will not directly affect the assembly line. If the tester fails, the product can be taken out of the production line and put aside to continue production, so that the production line will not be affected, but the processing time and labor are also problems.
It should be remembered that manual testing usually involves connecting UUT with several cables and connectors. Compared with the probes on the needle bed fixture, the service life of these cables is generally shorter, so they should be included in the maintenance plan, which can reduce intermittent failures.
Fixture problem
Due to the differences in production line, workshop space and labor rate, fixtures can range from simple plywood with pins and connecting cables to complex automatic needle bed test fixtures connected to the assembly line through conveyor belts. Obviously, these factors indicate that there is no fixed plan.
Manually load the double-sided fixture, connect the ribbon cable to the main I/O connector, and the probe installed at the top can touch the key test points on the UUT. This is an ideal design scheme for a medium-sized factory. The operator must connect the ribbon cable, close the top plate, and then start the test. There is no need for manual exploration for calibration and diagnosis here, because the top plate can touch all relevant areas. The connection between the ribbon cable and the roof probe should be designed to be easy to replace, because these cables are often bent and worn.
When dealing with fixture suppliers, we should keep these problems in mind and think about where the products will be made, which is something that many test engineers will ignore. For example, let's assume that the test engineer is in California, but the product is made in Thailand. Test engineers will think that products need expensive automatic fixtures, because factories in California are expensive, and the fewer testers are needed, the better. Using automatic fixtures can reduce the employment of high-tech and high-paying operators. But in Thailand, these two problems do not exist, so it is cheaper to solve these problems manually, because the labor cost here is very low and the land price is very cheap, and large factories are not a problem. So sometimes first-class equipment may not be popular in some countries.
Operator technical level
In high-density UUT, if calibration or diagnosis is needed, manual exploration is likely to be needed. This is because the needle bed contact is limited and the test speed is faster (testing UUT with probe can collect data quickly instead of feeding information back to the edge connector), so operators need to explore the test points on UUT. No matter where you are, make sure that the test sites are clearly marked.
Probe types and ordinary operators should also pay attention to the following issues:
Is the probe larger than the test point?
Is the probe in danger of short-circuiting several test points and damaging UUT?
Is there any danger of electric shock to the operator?
Can each operator quickly find out the test points and check them? Is the test center big and easy to identify?
How long does it take the operator to press the probe on the test point to get an accurate reading? If the time is too long, there will be some troubles in the small test area, such as the operator's hand slipping because of the long test time. It is suggested to expand the test area to avoid this problem.
After considering the above problems, the test engineer should re-evaluate the types of test probes, modify the test documents to better determine the location of test points, and even change the requirements for operators. In some cases, automatic detection will be needed, for example, when it is difficult to detect PCB manually, or when the test speed is greatly reduced due to the technical level of operators, the automatic method should be considered.
Automatic detection can eliminate human error, reduce the possibility of short circuit at several test points and speed up the test operation. However, we should be aware that automatic detection may have some limitations, which vary according to the design of the supplier, including:
Area of UUT
Number of synchronous detectors
How close are the two test sites?
Test the positioning accuracy of the probe
Can the system detect UUT on both sides?
How fast does the probe move to the next test point?
What is the actual spacing required by the probe system? Generally speaking, it is larger than the off-line functional test system.
Automatic inspection usually does not use the needle bed fixture to touch other test points, which is generally slower than the production line, so it may take two steps: if the detector is only used for diagnosis, the traditional function test system can be considered on the production line, and the detector can be placed next to the production line as a diagnosis system; If the purpose of the detector is UUT calibration, then the only real solution is to use multiple systems, which is much faster than manual operation.
How to integrate into the production line is also a key issue that must be studied. Is there room on the production line? Can the system be connected to the conveyor belt? Fortunately, many new detection systems are compatible with the SMEMA standard, so they can work in an online environment. This technology should be discussed as early as the product design stage because it requires special components to perform this task. In UUT with digital circuit as the main body, devices supporting IEEE 1 194 (Boundary Scan) can be purchased, so that most diagnosis problems can be solved with little or no detection. Boundary scan will reduce the overall function of UUT, because it will increase the area of each compatible device (4 ~ 5 pins and some wires per chip), so the principle of choosing this technology is to improve the diagnosis result at cost. It should be remembered that boundary scan can be used to program flash memory and PLD devices on UUT, which further increases the reason for choosing this test method.
How to deal with a limited design?
If the UUT design has been completed and finalized, the options at this time are limited. Of course, you can also ask for changes in the next revision or new product, but process improvement always takes some time, and you still have to deal with it.
The main guiding ideology here is how many tests you can do. It may be enough according to the expected fault type, but if it is not enough, it is usually necessary to strike a delicate balance between more expensive test systems, and choose a more accurate detection method after weighing the COGS and marginal profit of UUT. So, the answer is that there is no simple answer.
The best reference for future design is to complete functional testing when it is restricted. When faced with these limitations, we should write down the tests that can be completed within the time range specified by the speed of the production line and the number of testers on the production line. Time limit is very important, because the output can't give in to you, so your job is to sacrifice test coverage for time, so you will ask for improvement in order to lift these restrictions in the future!