Even if they don’t know it, everyone loves product designers and engineers because they give us the amazing gadgets that we love to use. At companies like Apple (iPhone, iPad, iPods) and Zojirushi (rice cookers), a lot of that engineering time is spent making sure the products will work again and again, for years. Designing for reliability is expensive, but the very high volumes of consumer products makes it feasible to put that much time and money into designing products the right way.

Outside of consumer goods, there is much less focus on designing for reliability … or designing for manufacturability, testability, or serviceability for that matter. Between the pressure to release the product and the fact that the product volumes are just not high enough, manufacturers just may not think it’s worthwhile. But those decisions can have big, negative downstream effects.

Design is Critical

The design of a product sets the stage for everything. It affects how manufacturing builds and tests a product, how service is able to diagnose and repair it, how a user interacts with it, and ultimately what the market will conclude about it. Is there more value if a product is pretty, but it falls apart quickly? Or is it better that a product be just handsome, but long-lived? Obviously, every manufacturer has choices, but when the product is critical in some way, those choices have consequences.

Consider an example in medicine. In 2022 alone, five medical device manufacturers had to recall six different products (including important life-sustaining products like ventilators) because of a range of battery failures[1]. In the worst cases, patients might have been affected, perhaps negatively. Only the manufacturers can determine if changes in the design process could have prevented these recalls (and whatever reportable incidents precipitated the recalls).

Prototyping and Prototype Testing. Before designs even get to production, the engineers often build prototypes. This is a critical way that engineering can work out the details of the design. For complex products, prototyping, including having real customers come in and play with the system, is a great way to figure out if the product will work well for the personas described by marketing.

But often overlooked during this testing is how a customer could potentially use the product incorrectly. Called negative path testing, it is a series of “what if” questions that reflect how a product might be used if an unskilled, untrained, or even unintelligent person began using it. What if they do X before Y? What if they never add oil? What if they never clean the filters? What if they interpret the error message incorrectly and don’t reboot the system?

Negative testing is critical because it represents the real world. Humans make mistakes; they misinterpret things, and sometimes they just do something stupid. When a person uses a product incorrectly, counter to the way the engineers designed it to be used, the results can be anything from excessive wear to total product failure. Often the service department is called, and left to pick up the pieces.

Prototyping and Prototype Testing are best practices … and that means both positive testing, with a user carefully following directions, as well as negative testing.

Design for Manufacturability. Before a device goes into production, the design is locked in place. Then manufacturing has one simple job—to make the same product again and again. If the design team considers manufacturability, then the task should be made easier. Individual parts, components and assemblies are chosen or designed so that they go together more easily, in fewer steps, and ultimately at lower cost. By designing the product this way, the engineers can lessen the impact that an individual manufacturing technician might have on the final product.

Humans do make mistakes in many ways, even in carefully controlled manufacturing facilities. But the effect of those mistakes can be lessened through careful design and good manufacturing practices.

Designing for manufacturability is a best practice … taking the time during design to make sure the manufacturing department will have an easier time making every product the same.

Design for Testability. It should be standard practice that products are designed for testability, but that isn’t always the case. For printed circuit board (PCB) design, it is now standard to include test points on PCB layouts, and that facilitates testing with automated in-circuit test (ICT) equipment. However, that forethought it is not necessarily the case for higher level assemblies. Sometimes there is no easy, practical, cost-effective way to test an assembly. Instead, the testing waits until later in the manufacturing process.

Assemblies and products can be tested automatically or manually. Automated testing is great but has a dramatic limitation – it only tests for the product parameters that have been included in the test. If the developer of the automated test equipment jigs and programming of the tests is uninspired it’s possible that classes of parameter combinations will not be tested.

Manual testing is usually more time consuming but can be equally effective. However, manual testing is prone to human error, errors in documentation, and accidental introduction of faults (such as inadvertently shorting two contacts together, burning out a component). There is a silver lining with manual testing in that a human interacting with the assembly might observe a fault that automated testing might not detect, such as a missing screw, a nicked wire, or another potentially critical issue.

Designing for testability is also best practice … taking the time during design to make sure the test department can validate that every product is fit for purpose.

Design for Serviceability. When the engineering team is designing a new product, especially a low-volume, high-cost product that will have to be repaired at some point, it is critical to design for serviceability. That basically means that the design itself will make it easier for the field engineer to take the system apart and replace key components, or that the product will be more easily diagnosed through remote access.

For the most common faults, companies often have procedures documented, which field engineers can follow to get the product up and running again. Less common faults often require the knowledge and experience of the field engineer for resolution. The engineer will go “off script” and take apart whatever is necessary to get to the root cause and fix the problem.

Product designers can do a lot to support this. They can design in log files that succinctly describe problems. They can specify screws that don’t strip easily. They can ensure clearance for tools when things must be taken apart.

Even Big Companies Get It Wrong

It’s true. Even the largest of companies get design wrong. This was clearly the case with General Motors and the Chevrolet Monza 2‐plus‐2 V‐8. As reported by the New York Times in 1967[2], Chevrolet designed a car in which the spark plugs could only be replaced if the engine was lifted.

Chevrolet could argue that the design for manufacturing was fine … the spark plugs were inserted into the engine at the engine factory. Then the engine was inserted into the car on the assembly line. Not a problem.

They could also argue that the design for testability was fine. The engine could be fired up at the engine factory – clearly the spark plugs were OK. Likewise, the car could be started and driven off the assembly line. Again, not a problem.

But Chevrolet did not have a defensible position for design for serviceability. In those days (1967), replacing the spark plugs was very common during a tune-up, which was a regular occurrence. Having to lift the engine, even a little, to replace the spark plugs was a complete failure in design for serviceability. And in that era, because cars simply were not that reliable, serviceability meant a lot.         

A New Day is Dawning

Believe it or not, even in 2023, negative testing (in prototypes), design for manufacturing and design for serviceability are not universal. Engineers these days are often focused on designing for personas given to them by marketing. Those personas are almost always different kinds of end-users, rarely a service engineer, and almost never a manufacturing or test technician.

In the end, because the service department always winds up being the customer-facing responsible party, they deserve a new technology that is just around the corner. With minimal product impact, design for serviceability can be instantly enhanced by integrating software tools that provide new levels of remote monitoring, access, diagnosis, and repair … and, very soon, service automation.

Even better, it is now possible to retrofit the installed base, those millions of products from thousands of manufacturers that are already out in the field, with remote service tools that add serviceability where before there was little. While software cannot fix mechanical problems, software with a few decent log files can often diagnose the problem, determine the most likely parts and tools required for the job, and only dispatch the field engineer when those parts and tools are available.

maiLink™ SRM software is a service relationship management platform that automates repetitive service tasks and helps you build a rich database about your installed devices. It seamlessly integrates telemetry from your products into a database rich with profound product insights. maiLink has no per-user fee—any employee you authorize can have access to the data. To learn more about maiLink software, visit www.maiData.io or email info@maidata.io.

[1] 2022 Medical Device Recalls, US Food and Drug Administration

[2] Problem is Reported in Replacing Chevrolet Monza Spark Plugs, NY Times, Jan 10 1975