productronica Daily: Test equipment capability is a concept that is becoming increasingly important. For the most part, calculations from mechanical test equipment capabilities for sliding calipers and the like are advanced in this context. But these are neither applicable nor meaningful for an electronic test system. Even if they were, though, what would be the advantage?
Hans Baka, Digitaltest: Well, then you’d know that the test system is capable of testing within defined limits and tolerances. No more. But whether the adapter and the test program yield the desired test depth and defect coverage consistently remains open to question. You see, test systems can be parameterized. This means that various settings can be modified in order to achieve the test result together with the desired test depth and defect coverage. Normally, a program generator uses a bill of materials and a circuit diagram to carry out a circuit analysis and create a test program with all the necessary parameters like stimulus, guard points, integration times, delays, and Kelvin measurements. But because the result doesn’t always yield the desired success, the test is modified during debugging, meaning that the parameters of the automatically generated test program are altered and adjusted for as long as it takes to show the desired test result consistently.
Are the test results gained in this way completely reliable?
Manipulating items like this makes it perfectly possible to force a test result without actually measuring anything useful. Such testing is known in tester jargon as a “ticket.” So you can force a test result by setting various parameters without this being immediately evident.
What impact would this have on defect coverage?
One fatal consequence of this could be that a component is tested even if it is incorrectly embedded or not embedded at all. This means that the expected defect coverage is only theoretically correct and the test program leads the user to believe something that is false.
How could that be avoided?
This could be avoided by manipulating–or “desoldering”–each component to be tested individually or by replacing it with other components with other values and verifying every change made using the test program, in order to determine what is actually being recognized. But even with a relatively small board with around a hundred components, this turns into a time-consuming and error-prone affair.
What solution would be conceivable instead?
Digitaltest has come up with a technique called FailSim that makes it possible to connect further components with the component under test in parallel or in series during testing and hence to modify the nominal value of the device under test. So if an additional resistor is connected in parallel to the resistor being tested, the test result should be smaller. Or the other way round, if you use a capacitor. If a series of components is now connected additionally, and a measurement is taken and analyzed every time, this should also be reflected in a change in the test result. If this is not the case, we again have the infamous ticket and have to assume that a defect in this component will not be identified. This in turn makes it possible to modify the parameters of this test in such a way that defects are spotted. Alternatively, you can take the measurement right out the test and replace it with other measures.
What does this mean in practice?
The FailSim technique can be employed in our in-circuit test systems right now. To do so, a new board is loaded on the new AMU05 Analog Measurement Unit with a series of resistors and capacitors. This can be hooked up to the test bus either in series or in parallel during the in-circuit testing. The result is that these components can be connected in parallel or in series with the components under test for every measurement, enabling defects to be simulated. An analysis program is used to compare and evaluate the test results recorded. The outcome is a clear statement about stable and reliable tests that are also capable of locating both genuine defects and to evaluate those that are not accordingly.
To what extent can this solution be transferred to Industry 4.0?
Industry 4.0 centers on connected, software-based production machinery. This automatic method fits in here seamlessly, as such mechanisms are both helpful and purposeful not just for execution in production but also for the manufacture of production tools.
Digitaltest, Hall A1, Booth 365