Building a Software Automation Framework to Drive Standardization and Increase Efficiency in Semiconductor Post-Silicon Validation Engineering

Vijay Krishna Guru, Soliton Technologies Pvt. Ltd. 

"We truly believe that the framework today is ready to scale from upstream design verification teams to downstream production test floors or application engineering teams and provide seamless data correlation to enhance the debug experience throughout the entire product development cycle."

- Vijay Krishna Guru, Soliton Technologies Pvt. Ltd.

The Challenge:

Establishing an effective mechanism to share measurement expertise and device, and instrumentation knowledge across validation teams to facilitate accelerated product development and validation cycles driven by growing time-to-market pressure.

The Solution:

We developed a novel semiconductor validation framework based on LabVIEW and TestStand, which allows validation teams to increase automaton, increase reuse of their existing software assets and code bases, and easily onboard new members to the validation activity.

Author(s):

Vijay Krishna Guru - Soliton Technologies Pvt. Ltd. 
Anand Krishnan - Soliton Technologies Pvt. Ltd. 
Sandeep Achari - Soliton Technologies Pvt. Ltd.

 

Accelerating product development lifecycles by resolving the toolchain fragmentation in the post silicon validation engineering.

As the semiconductor industry moves toward shortened product development cycles with higher levels of complexity and integration, tools that empower designers to mix and match IP are becoming standard in the design and verification processes. However, the tools used in post-silicon validation are typically fragmented across groups because different teams and teams located at different geographic sites often custom develop and maintain software catering to their needs. While some of these software interfaces are built from scratch, others leverage a variety of commercial off-the-shelf (COTS) test sequencers. This disparity and non-uniformity in the software interfaces and toolchain limits the data collection and correlation, and often results in wasted effort in large organizations in which multiple groups develop similar functionality independently. With more of the available IP being integrated in a single chip, previously distinct validation teams must work with each other and explore means for sharing and correlating data across different engineering sites and across stages of the product development lifecycle to get the product out in the market faster.

 

 

About a decade ago, we partnered with a leading global semiconductor company on a successful pioneer initiative to standardize their post-silicon validation software framework. In subsequent years, we have specialized in designing, developing, implementing, and supporting such frameworks in the world’s leading semiconductor companies including Analog Devices, Cirrus Logic, Cypress Semiconductor, NXP, Qualcomm, and Texas Instruments. Leveraging this unique and unmatched experience and expertise, we have developed a framework called Soliton’s Integrated Validation Architecture (SIVA) that we can deploy in a semiconductor company, integrate with existing systems, and demonstrate benefits within six months. This automation framework provides the benefits of:

 

  • Increasing automation while enabling an intuitive debug environment for designers and validation engineers
  • Facilitating higher reuse of software assets and code across teams, projects, and programs
  • Supporting easy onboarding of new engineers onto the validation activity

 

 

A Modular and Scalable Framework to Adopt and Build On

To ensure that the standardized software framework was successfully adopted in an organization, we had to meet the following list of key objectives:

 

  • An easy-to-operate environment for the user that addresses existing pain points and improves productivity  
  • The adoption (the one-time effort) should not be painful or disruptive  
  • The organization should be able to validate the benefits and return on investment through measurable metrics  

 

To meet these objectives, we had to understand the use cases, pain points, and future needs of the organization and users. We accomplished this through detailed interviews with a representative cross section of potential users. The interview process led us to refine the choice of development platform, architecture, delivery mechanism, and the long-term support and maintenance.  

 

A key decision we had to make was the choice of the underlying platform on which to build our framework. We decided on the NI products LabVIEW and TestStand. LabVIEW, as the programming language, delivers a high-productivity GUI development environment and powerful yet intuitive object-oriented programming capabilities that enabled abstraction, along with extensive instrument driver support (third-party instruments included) through the LabVIEW Instrument Driver Network (IDNet). TestStand, as the test sequencing engine, is flexible out of the box and features scalable and reliable test sequence management capabilities.

 

The resultant framework - the SIVA automation framework expands from the first silicon bring-up to device PVT characterization to system-level testing. We also developed custom tools like data logging APIs, automated report generation, and more for different workflows using the common underlying base components. The base components like instrument handles, device under test communication mechanisms, test parameters, test logic, data logging, and visualization are required to implement the measurement in the lab and in the test engineering workflows. The framework also establishes templates, guidelines, and processes for individual component development, which the team maintaining this standardized framework within the global organization can use to develop and easily roll out updates to different users to meet their unique needs. We use custom ‘glue logic’ between base components of the framework to help balance performance and flexibility.

 

Various components and features of the framework have evolved iteratively over time with interactions and feedback from the users and customers. Some of these feature groups include:

 

  • Custom Test Automation GUI that supports iterative testing for a product or support testing of different variants within the same product line
  • Automated Report Generation and APIs that export and consolidate results in .pdf, .ppt, .xls, and more
  • Connectivity to Databases for centralized data management, correlation, and more

 

These readily reusable components offer customers and users a high-level starting point, which can help prevent repetitive investment in terms of time and dollars spent to meet project-specific needs. 

 

 

Framework Adoption Leading to Business Impact

The SIVA framework is a ready-to-deploy enterprise automation framework for semiconductor post-silicon validation and characterization. SIVA users can focus on validation tasks instead of having to focus on developing their automation software. We have heard firsthand from our customers that adopting SIVA has helped them realize quantifiable productivity gains of at least 30 percent within six months of deployment. The table below shows the ROI computation and time and cost savings for the same three feature groups that we discussed in the previous section.

 

In addition to increasing productivity in the post-silicon validation phase, we see a large potential for significant efficiency improvement across the board through the entire software standardization exercise. The use of a singular framework across several product teams and geographies can:

 

 

  • Reduce time spent on repeatedly developing, testing, and debugging automation software and increase the time spent debugging the device
  • Facilitate a better tie-in for tracking and reporting requirements and specifications; thereby, unifying data analysis and reporting flow across the entire product development lifecycle
  • Deliver an enhanced debug experience through better interoperability and interactivity with upstream (design verification and emulation) and downstream (applications engineering, high-volume production testing) tools  
  • Provide well-defined structures, documentation, and processes within the framework, which enable quick recreation of the automation software originally used for design evaluation for failure analysis after the product release  
  • Support easy onboarding of new engineers and reducing the learning curve in moving the engineers between different product lines  

 

 

We truly believe that the framework today is ready to scale from upstream design verification teams to downstream production test floors or application engineering teams and provide seamless data correlation to enhance the debug experience throughout the entire product development cycle. As the framework usage scales across multiple product lines within large semiconductor organizations, the future potential for increasing productivity, and thereby the test coverage, is significant.

 

 

 

Author Information:

Vijay Krishna Guru
Soliton Technologies Pvt. Ltd. 
#683, 15th Cross Road, J.P. Nagar 2nd Phase
Bangalore 560078
India
vijaykrishna@solitontech.com

Figure 1. Resolving the Toolchain Fragmentation Using the SIVA Framework
Figure 2. Soliton’s Integrated Validation Architecture
Figure 3. Silicon Bring-Up and Test Automation Development GUI
Figure 4. Test Automation Run-Time GUI
Figure 5. – Tabulation of ROI computation for a sample set of feature groups in SIVA
Figure 6. How the SIVA Framework Fits Into the Product Development Lifecycle