Design Validation – Functional Validation of  Mixed-Signal Devices

Validation ensures that a part will operate correctly in end-user applications.

Design validation is the task of testing a part or a system adequately to ensure that all parts or systems manufactured will function within specifications in end-user applications and environments. The underlying credo that drives design validation is  if anything can go wrong, it will,• especially given the prospect of shipping millions of devices into potentially thousands of applications and environments over the lifetime of a product.

To illustrate the philosophy and principles of design validation, the National Semiconductor DP83865 10/100/1000 Ethernet Physical Layer Interface Integrated Circuit serves as an example throughout this article. Although the DP83865 is a complex mixed-signal semiconductor device, the design-related processes apply equally well to any product.

Many corporations boast ISO 9001 certification, but ISO compliance provides no direct guarantee for adequate validation. It requires that a company document a process and stick to it. If the process is inadequate, the results can be inadequate.

The cost of inadequately validating designs has been experienced by consumers for as long as commerce has existed. Quite simply, products fail.

Product failures are frustrating for customers and can represent significant cost to suppliers due to manufacturing yield reduction, customer returns and complaints, and online product revision turns. Optimistic product business plans can devolve into profit margin nightmares if inadequate attention is given to validation.

Design Validation as Part of The Design Process
Design validation is a subset of the entire spectrum of phases a product undergoes during the development cycle. Figure 1 illustrates how design validation testing fits into an overall product-development cycle.

The design validation phase occurs between product design and product shipment, concurrent with manufacturing test and reliability validation. Design validation serves as one piece within an overall comprehensive quality-assurance strategy.

In some ways, product validation picks up where functional simulation leaves off. Simulation may cover analog, digital, mixed-signal, formal, and even to some extent, system-level verification. Design validation not only proves simulation results on real product, but also extends testing far beyond what simulation can accomplish.

The focus of design validation is different from that of manufacturing or reliability testing although some overlap is possible. Manufacturing testing provides a comprehensive test on a per-part basis but cannot guarantee functionality in all end-user applications and environments. Reliability testing validates a part's viability over a long lifetime of use, but only a subset of functionality can be tested in this fashion. Design validation fills in the gaps that manufacturing testing and reliability testing cannot cover.

Defining the Problem
Validation of any system, no matter how complicated or simple, begins by looking at the task at hand from several perspectives. To get a grasp on all the possible perspectives, start by examining a product from both the outside and the inside.

From the outside, the goal is to validate individually all the features the product is called to support and its conformance to all applicable engineering standards. You also must validate using the various operating environments that the product is called to function in, which for an Ethernet physical-layer part include parameters such as voltage, temperature, cable length, and operational data rate. Finally, look at how the product might be incorporated into OEM customer product applications and ultimately used and abused by end customers.

From the inside, examine the functionality of individual blocks from I/Os to ESD structures; from individual state machines to the operation of PLLs, ADCs, and DACs; and a host of other analog and DSP functions. An important aspect of individual block testing is the stressing of micro-environment stimuli for individual blocks from ppm frequency variations in system clocks, to external/receive signal variability as passed from block to block, to common-mode and differential-mode noise injection across blocks.

In a device such as the DP83865, a cursory look at all of the operational variables reveals how daunting the task of design validation can be. The device has dozens of individual features, some calling for compliance with established engineering standards and some oriented toward nonstandard customer ease-of-use functionality and programmability.

The part is designed to autonegotiate and operate in three different operational modes/wire speeds, using up to three distinct power supplies, an input clock source, and two external digital bus interfaces. The part must operate across cable configurations that meet minimum to maximum attenuation and crosstalk specifications as well as various end-user/patch-panel configurations, all over minimum to maximum temperature ranges.

Just in terms of basic parametric variation, validation might be performed over 27 possible power supply variation scenarios, times 3 variations in operating temperature, and times 3 variations in input clock, resulting in 243 parametric variation scenarios. Multiplying this by three wire speeds and 50 cable configurations results in 36,450 basic operating scenarios, and these combinations account only for validating basic product functionality.

Add all of the associated engineering standards to the scenarios, nonstandard customer-oriented features, and individual analog and digital cells, all of which need to be validated over variances in external environment. Then, you find that almost 100,000 tests must be performed to assure the quality of the design, with each test having multiple steps and measurement requirements.

Furthermore, to prove functionality over a range of possible manufacturing process variations, these tests must be performed on several parts from several silicon wafer lots, which can bring the grand total to over a million tests to be performed.

Reducing the Problem
How do you approach a problem of such a large magnitude? Given the sheer volume of testing needed to ensure a high confidence of total system quality, it is a wonder that any product ever ships.

Planning the validation of any product must start with a statement of the scope of the problem. While examining the overall scope of validation, the magnitude of the problem can be reduced by choosing test scenarios that cover worst-case operating conditions. Choosing scenarios that run the part at its fastest operating condition (high voltage and low temperature), slowest condition (low voltage and high temperature), and conditions that stress I/O interaction against internal core logic (fast I/O and slow core and slow I/O and fast core) can greatly reduce the overall test-scenario burden.

Another key to reducing the overall burden of design validation is establishing an infrastructure of generic and specialized test equipment. Commercial test equipment, including bus analyzers, protocol analyzers, and oscilloscopes that come equipped with industry-standard conformance test software, can help ease the design-validation burden immensely.

The schedule burden can be reduced even further by pulling several individual pieces of test equipment into IEEE 488 bus-controlled validation systems. A fully automated validation infrastructure can help accelerate the design validation schedule of an Ethernet physical-layer product from literally months or even years to mere weeks.

Forming a Design Validation Plan
Having developed a basic understanding of the scope of the problem and the tools available for performing the tests, it is necessary to document the steps used for performing validation testing. A comprehensive plan includes a statement of the scope of the problem, a statement regarding features and applicable industry standards, a statement of overall resource requirements, and perhaps multiple separate sections describing in detail the actual validation testing to be performed.

A section may be added to describe any customer/development partner testing to be performed. Finally, a results summary section must be included to summarize validation testing results for every revision of the product created over the lifetime of the product.

The resource summary details the schedule and budgetary impact that validation testing has on the overall product-development process, including lab equipment and personnel requirements.

Several detailed sections may be required for documenting actual validation test procedures. These sections may include a plan for validating individual analog and digital cells within the product, a plan for validating the overall functionality of the part, and a customer-oriented usability and interoperability plan. A plan also may be necessary for validating any software drivers and/or customer-oriented software applications developed in conjunction with the product.

Finally, and perhaps most importantly, the product design validation results section provides a comprehensive summary of all test results.

Conclusion
Design validation testing is an integral part of an overall quality-assurance strategy for any product. Design validation ensures the functionality and manufacturability of a product and helps prevent costly product revisions due to issues discovered by customers in the field.

About the Author
David Miller is the validation manager for the Communications Interface Division at National Semiconductor. Mr. Miller has more than 22 years of experience in the areas of networking products and systems design, including 10 at National. Previously, he spent five years as a system/motherboard designer at Zenith Data and seven years at GTE Automatic Electric. Mr. Miller received a B.S. in electrical engineering from Milwaukee School of Engineering. National Semiconductor, Communications Interface Division, 500 Pinnacle Ct., Suite 525, Norcross, GA 30071, 770-903-1867, e-mail: [email protected]