Robustness Analysis

What is Robustness Analysis?

Designing a device is difficult because it must perform reliably over time under various conditions. Your device may need to maintain functionality in a scalding desert or the freezing Arctic. If you are designing a device to work in space, it may require certain anti-gravitational mechanisms. However, it might not be possible to test its performance by going to such places. Another major issue is that devices will inevitably degrade over time.

It is almost impossible to test how a device will degrade over five years in the real world if you are looking at a two-year development lifecycle. In most cases, chip designers only have a few months to design the next groundbreaking chip. Therefore, all the ideation and testing must happen in that short window of time. Outside of environments, you may also want to predict how the device will perform for certain types of users.

As you cannot predict the physical environment, it is crucial to build a robust design to handle everything a user might throw at it. This is where robustness analysis comes into play. Robustness analysis analyzes how your device will perform with various voltages, temperatures, and processes when it ages. This technique gives your team the data to ensure your device will perform well, regardless of the operating environment. It is a complementary methodology to static timing analysis, letting you know how your machine will perform in the various cases. Overall, robustness analysis allows you to maximize your device's power, performance, and area more efficiently than other methodologies.

Why is Robustness Analysis Important?

Chips are now in everything, from your toaster to NASA equipment. They have never been more vital for everyday life. Consider the James Webb space telescope, currently taking some of the best pictures of the universe we have ever seen. Any malfunctions could cost billions of dollars to recover the time lost because it is impossible to service the telescope in its current location and orbit.

The chip must also work for the entire device lifecycle, or it will be useless to the thousands of scientists who depend on it to get groundbreaking insights that could change the world. It also means that the people using your device can rely on it to function properly for many years to come.

Devices are tested extensively before manufacturing; however, certain tests are impossible to run in a reasonable time. Furthermore, other tests are necessary to ensure the device works well before manufacturing. Robustness analysis falls into this category because it gives design engineers accurate and useful models for designing dependable devices that work effectively long-term. Robustness analysis is also a valuable piece of the puzzle, which ensures device reliability for various users.

Models are critical when designing chips. Naturally, more accurate models will lead to better chip designs in the long term. Modeling a chip accurately allows companies to reduce fabrication costs dramatically. A significant advantage of robustness analysis is the creation of more accurate models that can transform the chip design process. By accurately testing the performance of your chip design over its lifecycle, companies can create better designs considering those factors. Furthermore, chips improve in mission-critical applications. This aspect of robustness analysis also allows chip designers to iterate much faster, creating higher quality designs in a much shorter timeframe.

How Does Robustness Analysis Work?

Robustness analysis is more of a process than a tool. It uses other tools and scientific know-how to build chips that will last much longer. Crucially, robustness analysis depends on modeling circuits accurately at various levels. It also depends on complicated statistical analysis, helping designers make more accurate predictions in areas where modeling and simulation do not work. While modeling and simulation are useful tools in chip design, they cannot replace actual testing in the real world because certain areas are too complex to model in software. You might also see a future with more machine learning and artificial intelligence in the robustness analysis process. Chip designers can create even more accurate and reliable circuits for mission-critical areas.

Simulation

Modern chip design would be impossible without accurate simulation. Simulation is the foundation for modern EDA tools and methodologies. Simulation is also at the heart of robustness analysis. Simulation is an important part of the chip design process because building a modern chip with tens of billions of transistors without using accurate circuit simulation software and SPICE is impossible. Simulation allows engineers to model various parameters for a chip before manufacturing.

With the exorbitant cost of chip manufacturing, most companies cannot afford the iterative process necessary to build a working chip without simulation. Various aspects of a chip must be simulated to work as accurately as possible. For example, circuit simulation uses the SPICE language to model circuit performance and variance by voltage and current. It lets engineers know whether a specific design will work and is functionally accurate. However, simulating this aspect of the circuit does not consider its future performance. This is where robustness analysis comes into play because circuits, like all other things in the physical world, degrade over time.

Manufacturing physical products involves too many factors. Understanding and simulating a circuit’s performance and degradation over time is vital. Robustness analysis helps engineers simulate circuit performance without using a pessimistic model that forces them to create a sub-optimal design that costs more. Robustness analysis works with simulation and statistical methods to accurately model a circuit's performance as it degrades through a typical usage cycle.

Statistical Analysis

As mentioned, the world is too complicated to simulate perfectly. Therefore, we must rely on statistical methods to get as close to perfection as possible when building useful models. Robustness analysis involves statistical analysis to compute metrics that describe cell, path, and overall circuit performance as it degrades under various conditions. This is usually done by building a SPICE model of a circuit based on the main device degradation and aging mechanisms the design is likely to encounter in the real world.

The two factors listed enable EDA developers to create tools that accurately model device degradation and performance, allowing chip designers to develop robust products over the lifecycle. This enables better products while also maximizing performance, area, and costs.

Achieve Faster Design Closure from Cadence’s Superior Robustness Analysis

Cadence understands that robustness analysis is a key component of the chip design lifecycle. Therefore, Cadence offers the Tempus Design Robustness Analysis (DRA) Suite of advanced analyses to help with the process. This suite can accurately analyze complex chip designs for reduced pessimism and overall negative timing slack to ensure a dependable design while maximizing the number of working chips when it reaches manufacturing. The Tempus DRA Suite provides accurate analysis capabilities for specific phenomena, such as aging and other effects that can negatively impact PPA. This suite delivers a significant uplift over traditional approaches to robustness analysis. Tempus DRA Suite also works seamlessly with the Innovus Implementation System and Cadence Certus Closure Solution to maximize performance at the block, subsystem, and full-chip level in your design.

Uncover the advantages of integrating Cadence into your design process today.