As a test engineer before you can start building a test system or writing test code you have to decide on a test strategy. You have some sort of electronic product that needs to be tested before it can be sent to the field (sold). The question is, what problems could the product have, where and how are they going to be detected? Most companies already have some idea of what their strategy to test and overall product quality is, but these decisions have to be made or were made at some point.
Three Ways to Fail There are three ways an electronic device can fail 1. Poor quality raw materials 2. Design is incorrect 3. Process variation That’s it, three ways. So, what does each of these entail? 1. Poor quality raw materials An example of this would be defective components or bad PCBs. 2. Design is incorrect This is just what it sounds like, the product as designed does not meet its specifications. These errors should not have to be checked for on every instance of a product that is built in manufacturing. These problems should be caught early in the design validation and verification. However, design problems can show up in the manufacturing test, particularly when high volume manufacturing measurement data is analyzed. Some design problems - Does not meet analog specifications - Digital logic error - Functional failures 3. Process variation This type of failure is basically manufacturing errors. Here is a list of a few manufacturing errors - Shorts - Opens - Missing component - Wrong component - Backwards component - Bent Leads - Functional Failures Failure Mode Effects Analysis There are certainly a lot of failure modes that can happen to electronics in general but how do you know what failure modes to test for in a particular product? One way is to perform an FMEA. An FMEA is a formal method for determining what could go wrong so you can design a test strategy to check for all these failure modes. It is easy to find an FMEA worksheet on the internet to see how it works. The idea is to think of all the possible failures for each function or sub-system (or however you break it down) and then rate those failures by how serious their consequences are, how frequently they occur, and how easily they are detected. Here are some typical fields in the FMEA table - Function - Failure Mode - Effects - Severity Rating – this could be 0 to 10 or 0 to 3 or whatever works. - Causes - Occurrence Rating - Current Controls – what should be preventing this failure from happening in the first place - Detection Rating - Risk Priority Number (RPN) – This is a method to make an overall ranking of the items, this is the three rankings multiplied together. RPN = S * O * D. Test Strategy Complexity In reality, a test strategy comes down to the type of product you are testing and how much money you have to spend to test it. If you are testing a toy fire truck with an electronic siren I’m sure all the testing that takes place is pressing the button to see if the siren goes off on the completed toy, if that. Of course in a mission critical situation the test strategy could be very complex and expensive and would require a lot of work to design. Here are some additional factors to consider when developing the strategy. Cost – How much budget do you really have to work with? How much testing can you do with that money and where are the most important places to apply it? Volume – What volume of this product is being built? Product Complexity – How complex is this product being tested? Will it require a lot or little testing? Will the testing have to be broken up into multiple levels and tactics? Level of reliability needed – How reliable does the product you are testing really need to be? Ideally, it would be 100% for anything, but it’s really more about what testing is cost effective to perform. Environment where the product is used – will you have to duplicate any extreme conditions to verify that the product will work under those conditions? How to handle ASICs and components? – Will you assume that all ASIC and components are good before they are used on a board? Do you need to sample or fully test these components in-house? Outsource – will any of the testing be outsourced to another company? Is there a way to outsource part of the testing or building equipment that makes sense? Environment Stress Screening (ESS) or Reliability testing – Will you have to attempt to predict the reliability of the product or just verify that it is working correctly as it is built? Test Levels – Is there a way to divide up all the necessary testing into different levels that will make the testing simpler or cheaper? Some test levels include: - Incoming inspection, vendor test - Prescreen (auto shorts, opens, manual inspection) - In-circuit test of manufacturing defects analyzer - Board level burn in or ESS - Functional or performance - Hot mockup - System burn-in - Field test Test Tactics – A test strategy is the why and the test tactic is how. What methods will be used to best implement your test strategy? Some test tactics include, - In circuit test - Functional - Emulation - Inspection - Hot mock up - Design for test - Boundary scan - ESS Two Test Strategies There are two general strategies that can be taken when building a product. 1. Build it correctly the first time 2. Build it, test it and repair the failures 1. Build it correctly the first time This may seem like what you would obviously try to do no matter what, but it goes beyond just trying to do a good job. This is really the idea of putting inspection and process monitoring into your manufacturing process. As an example, components may be screed before they go onto a circuit board, then the circuit board undergoes some type of inspection, maybe an IR inspection and process monitors are put on this inspection. Finally, the final product is built with some final functional testing. It may be more elaborate than that but the idea is to end up with nearly 100% yield at the final testing. There is no need for scrap, debug and repair. This is really what Lean manufacturing and Six Sigma type concepts are all about. Judging from the volume of information about these topic on the internet, this strategy is very popular right now. 2. Build it, test is and repair the failures Here the products would be built without worrying too much about inspection or process monitoring then the products would undergo electrical testing. Anything that fails the testing is then debugged and repaired. The advantage here is that you saved a lot of time, cost and effort building the products. So, if the yield is okay and this method is appropriate for the product being tested, it should be fine. The disadvantages are that you have either lots of scrap or lots of products to debug and repair. The testing that was preformed should be an aid, but it still may be pretty difficult to debug and repair the problems. Summary Few engineers really get the chance to develop a test strategy from scratch because it can be tough to change the momentum of a company at such a fundamental level. However, if an engineer knows all the issues that go into a test strategy they can pick spots that and argue a case for changing them. Test strategy can be very simple or very complex, it’s all based on the product you are testing and the goals of your company. Comments are closed.
|
Archives
December 2022
Categories
All
|