Quality in electronic design is not a final inspection step; it is a set of considerations that shape how circuits and systems are conceived, verified and prepared for manufacture. As electronics become more complex and serve increasingly demanding applications, design teams must balance reliability, security, performance and energy use from the outset. These concerns also have growing relevance to industrial automation, where dependable electronic hardware underpins control, sensing and connected systems.
The themes highlighted at the 2020 International Symposium on Quality Electronic Design (ISQED) show why a multidisciplinary approach matters. Design for Quality (DfQ) connects disciplines such as circuit design, testing, verification, semiconductor technology and security. Rather than treating each as a separate late-stage concern, teams can use this wider perspective to identify design trade-offs and risks earlier.
Quality in electronic design starts with verification
Simulation is central to evaluating circuits, but verification demands have grown. Safety-critical applications, statistical timing characterization, wider FinFET voltage domains and tighter product margins can require designs to be assessed across many process, voltage and temperature (PVT) conditions. Some intellectual property blocks may also need high-sigma evaluation, which can involve very large numbers of Monte Carlo samples.
Brute-force simulation can become difficult to fit within available production runtimes as the number of cases increases. The ISQED program described active learning as an approach to accelerating SPICE-level verification while preserving its accuracy. In general terms, active learning uses results from earlier evaluations to guide which additional cases should be examined, rather than relying only on exhaustive sampling. This makes it relevant to verification planning when simulation coverage and compute budgets must be managed together.
For engineering teams, the broader point is that verification strategy should reflect the intended operating conditions and risk profile of the design. Decisions about corners, statistical coverage and simulation methods affect confidence in circuit behavior, not just the time required to complete a run.
Hardware security belongs in the quality conversation
Connected electronics create security concerns that extend across hardware and software design. Security is therefore part of product quality: a system that performs its intended function but cannot protect sensitive data or resist relevant attacks may not be suitable for its application.
Two hardware building blocks discussed at ISQED illustrate how security requirements reach down to circuit design:
- Physically unclonable functions (PUFs) use device-level variation to provide a hardware identity or security primitive.
- True random number generators (TRNGs) produce random values used by security systems, including cryptographic applications.
A tutorial described a design approach that uses a shared entropy source for PUF and TRNG operation, with self-calibration and extraction methods to support both functions. The underlying design challenge is that stable behavior is desirable for PUF bits, while randomness is essential for a TRNG. This illustrates how security requirements can create competing circuit objectives that must be addressed deliberately.
Cryptographic hardware also faces side-channel risks. In a correlation power analysis attack, an adversary studies a chip’s supply-current behavior to infer information about embedded keys. Circuit and arithmetic countermeasures can be designed to reduce the relationship between current signatures and secret data. Security analysis, then, is not limited to algorithms: physical implementation and observable hardware behavior also matter.
AI hardware makes energy and architecture first-order concerns
Machine learning is driving demand for computing systems that can process large volumes of data. Yet implementing neural algorithms on conventional von Neumann architectures can carry substantial area and power costs compared with biological computation. The resulting design challenge is broader than simply increasing processing capacity.
Work presented at the symposium explored a range of possible responses, including brain-inspired learning algorithms, specialized network architectures, spintronic devices and hardware that brings computation closer to memory. These approaches illustrate the value of co-design: algorithms, circuits, devices and system architecture need to be considered together. A change at one level may affect storage needs, data movement, power consumption or the kinds of computations a system can perform efficiently.
Machine learning is also being considered within the design process itself. The program included discussion of its role in chip design, validation and testing, alongside active-learning techniques for circuit verification. These applications do not remove the need for engineering judgment. They make it important to understand what is being modeled, what the method can establish and how results fit into a wider verification plan.
Emerging technologies require system-level thinking
Several other topics reinforce the importance of evaluating hardware as a complete system. Spintronic devices, for example, are being investigated for memory, logic and neuromorphic computing. Their non-volatility and switching mechanisms offer different design possibilities from conventional CMOS, while their practical use still depends on how devices, circuits and applications work together.
Advanced integration is another consideration. The N3XT approach described at ISQED combines emerging logic and memory technologies with dense three-dimensional integration. Such architectures aim to address data-intensive workloads, but integration also brings variation and reliability challenges that require specific techniques. Similarly, quantum computing introduces new questions for electronic design automation (EDA), including how design tools can support quantum systems and how their performance should be measured.
These examples are not interchangeable solutions. They represent different paths for addressing constraints in computation, energy, integration and reliability. For technical decision-makers, the practical task is to assess technologies against application requirements and the maturity of the overall design flow.
What Design for Quality means for engineering teams
Design for Quality is most useful when it connects technical decisions across the development lifecycle. For industrial automation engineers, integrators and operations leaders involved in hardware selection or system development, several questions help make that connection concrete:
- Are verification conditions representative of the system’s expected operating range and reliability needs?
- Have security requirements been considered at both the system and hardware levels?
- Do energy and data-movement constraints influence the choice of computing architecture?
- Are emerging devices or integration methods supported by suitable validation and reliability practices?
- Can design, test and verification teams share assumptions and evidence across disciplines?
These questions help prevent quality from being reduced to a single metric. A design may meet a performance target while remaining difficult to verify, vulnerable to a physical attack or challenging to manufacture reliably. Cross-disciplinary review makes such trade-offs visible earlier.
Conclusion
Quality in electronic design depends on integrating reliability, verification, security and system-level performance into the design process. The developments discussed at ISQED—from active learning for SPICE verification to secure circuits, AI hardware and emerging computing technologies—show how tightly linked these concerns have become. For organizations building or deploying electronic systems, the practical implication is clear: evaluate hardware choices across the full lifecycle, align verification with application risks and treat security and reliability as core design requirements.