Access Official Technical PDF

Share Summary:LinkedInEmail

Automotive semiconductor testing must address more than whether a chip works at the point of manufacture. The ART 2018 program placed in-field diagnostics, test coverage, functional safety and reliability alongside manufacturing test, reflecting the range of concerns facing engineers developing electronics for vehicles. Its sessions offer a useful map of the technical questions involved: how to test complex ICs efficiently, how to assess safety mechanisms, and how to align software development practices with automotive standards.

The agenda covered design-for-test architectures, safety analysis, connected-vehicle security, artificial intelligence, software process compliance and component reliability. Taken together, these subjects show why test planning needs to consider the device, its intended safety role and the conditions in which testing will take place.

Automotive semiconductor testing across the IC lifecycle

The opening technical session focused on design for automotive testing. Presentations addressed on-line memory built-in self-test (MBIST), error-correcting code (ECC) logic testing, system-test interfaces for heterogeneous IP, in-system test architectures and scalable test infrastructure. Other topics included compression and logic built-in self-test (LBIST).

These approaches serve different purposes. Manufacturing test is used to identify defects before a component is deployed. In-system and in-field test techniques, by contrast, enable checks after integration into a larger system or during operation. On-chip test resources can provide ways to assess selected logic and memory without relying solely on external test equipment.

The program’s emphasis on scalable architectures also points to a practical design concern: test structures must be planned for increasingly complex automotive ICs. A strategy that works for one block or device may need to be organized differently when it spans heterogeneous IP or a large system-on-chip. The listed work addressed infrastructure and test-time considerations, but the agenda does not prescribe one universal architecture.

Balancing test coverage, diagnostic mechanisms and overhead

A central question in the program’s “Smarter test for Automotive” panel was how to avoid both insufficient testing and unnecessary testing. The panel description raised the relationship between continuous ECC checking and periodic on-line MBIST, including whether a power-on self-test might be sufficient in some situations. It also questioned whether an overall test-coverage target can hide weaker coverage in individual blocks.

For engineering teams, the implication is that a high-level coverage figure should not be treated as the complete picture. Test planning also needs to consider what is being tested, which logic is critical, and whether the selected diagnostics are appropriate to the system’s safety goals. The panel was framed around these trade-offs and the use of functional-safety methodologies to meet ISO 26262 demands; it did not present a single answer applicable to every design.

Related sessions covered reliability-aware safety analysis, failure modes, effects and diagnostic analysis (FMEDA), residual error probability in safety-critical data communication, and failure-in-time (FIT) rate calculation and mitigation. These topics connect test design to broader safety and reliability assessments. Test mechanisms are one part of a safety case, alongside analysis of failure behavior and the effectiveness of detection and mitigation measures.

Functional safety includes software and system context

ART 2018 also included a tutorial on ASPICE- and ISO 26262-compliant software development methodologies, with IEC 61508 included in the described process context. The tutorial focused on establishing a development process for reliable software and outlined a case study concerning its implementation at Texas Instruments.

The tutorial abstract highlighted a particular challenge for semiconductor suppliers: software such as device drivers and software development kits may be developed “out of context,” with limited knowledge of the final application. At the same time, automotive customers may expect standards-based development practices even for foundational software components. The abstract cautioned that standards have broad scope, so teams need to interpret their recommendations in a way that matches the complexity and role of the software rather than adding process overhead without a clear fit.

For integrators and technical decision-makers, this highlights the importance of defining software scope, development evidence and relevant safety expectations early. Compliance is not only a documentation exercise; the development and testing methods need to be selected for the component and its intended use.

Reliability questions extend to AI and advanced devices

The program included sessions on deep neural network reliability, fault injection for assessing autonomous-vehicle safety and reliability, and the use of deep learning to improve stereo reconstruction dependability. These topics bring fault and error analysis into systems that rely on data-driven processing. The listed work treated AI reliability as a subject for assessment, rather than assuming that the use of a learning method guarantees dependable behavior.

Separate automotive reliability presentations addressed functional degradation in 14 nm FinFET SRAM, hybrid verification for automotive radar and a technique identified as the “Scratch Algorithm.” The program also listed a poster on single-event-rate qualification and radiation effects in relation to functional safety. Together, these topics span device-level behavior and the verification of larger automotive functions.

That range matters because reliability concerns can arise at different levels of the stack. Memory behavior, radar verification, software, and system responses to faults are not interchangeable problems. Each requires methods suited to the component or function being assessed.

Practical implications for engineering teams

The ART 2018 agenda suggests several useful considerations for organizations developing or integrating automotive electronics:

  • Plan testability early: Consider how manufacturing, in-system and in-field test requirements affect IC architecture.
  • Look below aggregate coverage: Examine coverage and diagnostic needs at the block level, especially where safety relevance differs.
  • Connect test to safety analysis: Relate test mechanisms to failure analysis, detection capability and the system’s safety objectives.
  • Fit processes to software scope: Interpret standards-based methods in light of component complexity and the information available about the end application.
  • Assess emerging functions explicitly: Treat AI-based processing and complex functions such as radar as subjects for reliability and fault analysis.

Conclusion

ART 2018 brought together automotive semiconductor test, functional safety, software development and reliability as connected engineering concerns. Its program emphasized the need to make test strategies both sufficiently thorough and appropriately targeted, while considering the device’s role in a broader automotive system. For engineering teams, the practical takeaway is to align test architecture, coverage analysis, safety processes and reliability assessment with the specific component and application—not to rely on a single test metric or method.

Access Official Technical PDF

About The Author