The design for test (DFT, design for test) is not a single person's work, but a group of representatives from the design engineering department, the test engineering department, the manufacturing department, and the purchasing department. Design engineering must specify functional products and their error requirements. Test engineering must provide a strategy that achieves the highest possible first pass-through yield (FPY) with the lowest cost and minimal rework. The manufacturing and quality departments must provide input on production costs, what has been done in similar products in the past, what has not been done, and the help of DFV (design for volume) to increase production. The Purchasing Department must provide information on the availability of components, especially reliability. The test and procurement departments must work together to purchase components on the on-board test hardware to ensure that they are available and easy to implement. The test system is often used as a sensor to collect historical data to achieve process improvement, which should be the goal of the quality team. So these functions should be done before placing/removing any node selection.
Preparation and understanding are key before developing a policy for the test environment. The parameters that affect the test strategy include:
Full access and large test pads are always the goal of manufacturing a printed circuit board. There are four reasons why you cannot usually provide full access:
1.1. The size of the board.
The design is smaller; the problem is the "extra" footprint of the test pad. Unfortunately, most design engineers believe that the accessibility of test soldering is less important on printed circuit boards (PCBs). When the product has to be debugged by the design engineer because of the inability to use the simple diagnostics of the in-circuit tester (ICT), the situation is quite another. Test options are limited if full access is not available.
The performance lost in high-speed designs affects the board, but can gradually reduce the impact on product testability.
1.3. Board size / number of nodes.
This is when the physical board size cannot be tested on any existing device. Fortunately, this problem can be solved by adding a budget on new test equipment or using an external test facility. When the number of nodes is larger than the existing ICT, the problem is more difficult to solve. The DFT team must understand the test methods that will allow the manufacturing department to output good products with minimal time and money. Embedded self-test, boundary scan (BS) and function block tests can do this. The diagnosis must support the unit under test (UUT); this can only be achieved by an in-depth understanding of the test methods used, the existing test equipment and capabilities, and the fault spectrum of the manufacturing environment.
2. The DFT rules are not used, observed or understood.
Historically, DFT rules have been enforced by an engineer or group of engineers who understand manufacturing environments, process and functional test requirements, and component technology. In the real world, the process is lengthy and requires communication between design, computer-aided design (CAD) and testing. This ubiquitous repetitive work is prone to human error and often rushes through time-to-market pressure. Nowadays, the industry has begun to use automatic "productivity analyzers" to evaluate CAD files using DFT rules. When a contract manufacturer (CM, contract manufacturer) is used, multiple sets of rules can be classified. Rule continuity and error-free product evaluation are the advantages of this approach.
Test methods and defect coverage
It is important to understand existing test methods and defect coverage before developing a test strategy. There are many electrical test methods with different defect coverage.
1. Short circuit and open circuit.
MDA and ICT are good at finding short circuits - they have a needle bed to reach each electrical node, and the resistance between the dots can be measured to confirm the short circuit.
The empty board tester uses a capacitance-to-ground technique, which is high in efficiency and speed if it is limited to empty boards.
The flying probe test uses a capacitor technique and a proximity shorts technique; the former is not repeatable enough for most manufacturing facilities and lacks a good diagnosis.
The best approximation test uses raw CAD data to confirm trace position, allowing the programmer to choose the maximum distance between test points. This provides a degree of control over the speed of the test;
However, it should be recommended that the functional test equipment has a current-clamping or fold-back power supply to prevent damage to the board or tester because the low impedance short circuit through the component may only be during the short circuit test. Can't find out.
2. Passive analog
Acceptable process quality is ensured by confirming that the UUT has been soldered to the board and components with the correct parameters are installed. This test is often performed with only a small number of WIPs, so the problem can be corrected before a large number of problematic products appear. The board is not powered, with selective passive or active guards to zero the current in the parallel current path. For the UUT and the surrounding guard position, an entrance to the needle bed is required.
3. Visual system
Provide device-level diagnostics. They use a golden board to compare it to a UUT without electrical testing. MDA provides electrical testing and component-level diagnostics, again compared to known good boards. ICT performs electrical testing, providing equipment-level diagnostics, comparing values and errors with BOM. The functional tester is tested according to the designer's specifications (usually called the sample golden board). If the functional test is thorough, it guarantees that the product can be shipped out. However, if the FPY is not particularly high, the manufacturer's price will be poor product, waste and expensive manual diagnosis and repair costs.
4. Active analog.
ICTs, functional testers, and non-needle tests that power the board are good at finding bad active analog components. ICT and flying probe tests, while providing pin-level diagnostics, cannot measure some key manufacturer specifications (eg, bandwidth, input bias current, etc.). The functional tester measures the output characteristics without providing pin-level diagnostics. With the help of no vector technology, the visual system only confirms the existence of components. X-ray provides a diagnosis of weld quality.
5. Testing of digital and mixed signal components.
Visual, X-ray and MDA only diagnose open and short circuits. ICT uses a variety of methods, depending on components, circuitry, and accessibility. It can only use vectorless techniques for continuity, and when there are all entries, BS is used for continuity and component validation. Modeling a particular component by manual vector generation can be time consuming and may not be sufficient to cover the defect to determine the effect. A combination of continuous vectorless techniques and finite vector testing that guarantees component operation can be used to maximize coverage while limiting development time.
The functional system tests the circuit/module according to design specifications, but lacks foot/component level diagnostics that will reduce the cost of rework. In most cases, functional testing does not provide deep data that needs to be used for process improvement. Both function and ICT are programmed on-board flash, in-system programmable, and on-board memory components
Manufacturing test strategy
There is no one strategy that will or should be suitable for the manufacturer. When developing a test and process improvement strategy, countless variables must be considered.
The confirmation of the manufacturing defect spectrum should be factory specific and product specific. These data, if relevant and reliable, will reduce staff and scrap costs and increase customer confidence. Defect data should be collected, edited, and discussed at a meeting of the normal quality team. This data should also be used to develop a test strategy to find common preventable defects. These data should include factory and site failures, marked with dates. New product defects should be monitored, while mature products should be monitored to improve FPY and supplier quality. Defect data should be compared internally to short-term, along with other locations to improve overall quality. Data on weather conditions, personnel, suppliers, and line changes should be tracked as these are often potential quality factors.
Two important quality factors are related data collection and distribution experiments. A sensor's ability to collect data that will improve quality, and the ability of data managers to communicate data to the right group, affects current and future products. The definition of the correct data is determined by the facilities and products. The test machine acts as a sensor to monitor the process. An effective distributed test strategy finds process problems as close as possible to the root cause and reduces the amount of bad goods produced.