Design Verification to Production Test Why Early Test Input Matters

Simon Payne, CEO, XJTAG | Subscribe to LinkedIn newsletter

XJTAG Testing Times cartoon - Just a few thousand connections

There is an important transition in almost every hardware development project.

The design has progressed through schematic capture, PCB layout, prototype manufacture, bring-up and verification.

The engineering team has worked to establish that the design behaves correctly across the operating conditions, component tolerances and variations the product is expected to encounter.

As the product moves towards production, another engineering challenge becomes increasingly important:

How are we going to test every board we manufacture?

That is where the requirements of production test begin to look rather different from those of design verification.

A Proven Design Isn’t the Same as a Production Test

The development team and production test team are ultimately working towards the same objective: confidence in the product.

But they approach that objective from different perspectives.

During development, engineers need to demonstrate that the design itself is correct.

They investigate interfaces, validate performance, exercise operating conditions and establish that the product meets its requirements.

Production test has a different task.

Once the design moves into manufacture, engineers need a repeatable way of determining whether each individual PCB has been assembled correctly.

And when one hasn’t, they need enough information to determine where the problem is likely to be.

Both are important, but they answer different questions.

The Design Influences the Test Long Before Production Begins

By the time a design approaches production, many of the decisions that influence how it can be tested have already been made.

The PCB determines which signals are physically accessible.

The circuitry determines which devices can be controlled and observed.

The JTAG architecture determines how Boundary Scan devices can be accessed.

Reset, enable and mode circuitry affects which parts of the board can be exercised independently.

Power architecture, buffers, multiplexers and connectors can all influence the eventual production test strategy.

Test engineers may already have had some involvement during development—and in many organisations they will.

The important point is that the earlier test requirements are considered, the more opportunity there is to make useful changes.

As the design progresses towards release, those options naturally become more constrained.

Give the Test Engineer More Options

Consider a peripheral connected to a Boundary Scan-enabled FPGA.

The FPGA may provide excellent access to the interface.

But suppose the peripheral’s reset can only be released once the processor is running application software.

Or perhaps a buffer sits between the FPGA and the peripheral, but its enable cannot be controlled independently during test.

The potential test path exists, but making practical use of it has become more difficult.

Identified during schematic development, the answer might be as simple as a resistor option, an alternative control path or access to an additional signal.

Identified after PCB layout or during an early production build—the available solutions are likely to be more limited.

This is why relatively small design decisions can have a disproportionate effect on testability later.

Production Testing Asks Different Questions

During development, one of the central questions is:

“Does this design operate correctly across the conditions it needs to support?”

Production test asks:

“Has this particular board been manufactured correctly?”

And if it hasn’t:

“Can we identify the likely fault quickly enough to diagnose, repair or reject it?”

In a production environment, those questions need to be answered repeatedly and efficiently.

A diagnostic process that is perfectly reasonable when investigating an occasional prototype may become impractical if it has to be repeated across every failed production board.

The test therefore needs to be repeatable.

The results need to be meaningful.

And ideally, a failure should provide useful information about where further investigation should begin.

Boundary Scan Can Help Bridge the Gap

This is one of the reasons Boundary Scan can be particularly valuable as a product moves from development towards manufacturing.

The structural tests created during prototype bring-up and verification don’t necessarily have to disappear once development progresses.

They can potentially become part of the production test strategy.

Tests that verify PCB connectivity during development can continue checking those same connections on manufactured boards.

Tests developed for memories, peripherals and interfaces may also be reusable.

That creates an opportunity to carry both engineering knowledge and test capability forward rather than starting again when the product reaches production.

Develop Once, Reuse Many Times

This can significantly increase the value of the original test development effort.

A Boundary Scan test might initially be created to provide confidence during prototype bring-up or design verification.

That same test capability can potentially support:

Prototype bring-up
Design verification
Production test
Repair and rework
Failure analysis
Service and support

The test doesn’t have to belong to one phase of the product lifecycle.

It can become an engineering asset that continues to provide value as the product moves from development into manufacture and beyond.

Test Engineers Need More Than a Working Design

As a product approaches production, the test team benefits from understanding not only how the product works, but how it was designed to be tested.

Which areas can Boundary Scan verify?

Which devices can be controlled independently?

Which tests proved particularly useful during development?

Which areas require application software?

Where are the known gaps in coverage?

What diagnostic information will be available when a test fails?

Answering those questions earlier makes the transition into production test much easier.

It also gives the test engineer an opportunity to identify areas where a relatively small design change could provide substantially greater test access.

Bring Development and Test Together Earlier

This doesn’t mean every hardware designer needs to become a production test engineer.

Nor does it mean the complete manufacturing test needs to exist before the first prototype is built.

It means giving the people responsible for production test an opportunity to contribute while their input can still influence the design.

The hardware engineer understands how the circuit is intended to operate and what must be demonstrated during design verification.

The test engineer brings a different perspective:

How can this be tested repeatedly?

How can this section be isolated?

Can this device be put into the required state without relying on the complete application?

If a board fails, what information will the test provide?

Bringing those perspectives together can produce a product that is not only well designed, but considerably easier to manufacture and support.

Avoid Making the Test Engineer Reverse-Engineer the Test Strategy

When testability has been considered during development, the production test engineer starts with a much stronger foundation.

They don’t have to discover late in the process whether an enable can be overridden.

They don’t have to find an unexpected way of disabling a watchdog.

They don’t have to discover that one inaccessible control signal prevents an otherwise testable interface from being exercised.

Instead, many of those questions have already been considered as part of the design.

The test engineer can then concentrate on developing the most effective production test rather than finding ways around avoidable hardware constraints.

A Better Handover

Moving a product towards manufacturing shouldn’t be a moment where one team simply finishes and another begins.

Ideally, knowledge moves with the product.

The design team understands what has been verified and where the difficult areas are.

The test team understands what must be proved repeatedly on each manufactured assembly.

Boundary Scan provides one way of carrying structural test capability between those stages.

And when test requirements have influenced the design early enough, the production test engineer has substantially more options available.

Closing Thought

Design verification and production test perform different jobs.

One establishes confidence that the design is correct.

The other helps establish that each manufactured board has been assembled correctly.

Thinking about both while the PCB can still be changed makes that transition considerably easier.

The objective isn’t to ask the test team to design the product.

It is to make sure that, when they need to develop an effective production test, the design hasn’t unnecessarily restricted the tools available to them.

Board Design Tip of the Week

Before your next design is released for layout, involve someone who understands how the product will eventually be tested in production.

Don’t ask them to develop the complete production test.

Ask: “Is there anything we could change now that would give you better control, observation or diagnostic access later?”

It might be a resistor option, an accessible enable, a controllable reset, a way of disabling a watchdog or a small change to the JTAG architecture.

These changes may have little impact on the product’s normal operation.

But they can give the test engineer considerably more freedom when each manufactured board needs to be tested quickly, consistently and with useful diagnostics.