Knowing the Limits for Successful Medtech Design Transfer

A common pitfall coming out of early product development phases is that while we might know what a good product is, we might not know exactly what a bad product looks like—and this is key to enabling a successful design transfer. In this blog, I’ll discuss issues that can arise when limits around essential performance are not well defined.

Unfortunately, we can’t just specify the perfect component and have it made time and time again. Each manufacturing process has its own inherent variations. Variations could arise through wear, machine vibrations, environmental changes, etc. All of these can contribute to changes in the output, even though the inputs are unchanged.

As designers, we must take variations into consideration; not only do we need to design parts so they work when integrated into the larger system, we also need to design the parts within the capabilities of the manufacturing processes being used.

Many components come together to make a device. Ideally, manufacturing controls on the individual components ensure the combined assembly meets specifications. However, that’s not always the case, and more importantly, we might not know that we’ve crossed over into a situation where we’re building a bad product.

During initial development, units can be limited due to the low volumes of devices being produced, for a variety of reasons. As a result, we may not be able to explore the full combinations and permutations of all specifications during development.

Why Is This Important?

Part of the danger of not fully exploring variations is that we may unwittingly be fine tuning the device. The essential performance can be defined, but it is only valid in the narrow ranges that we were able to test with the parts available. We might be faced with a new batch of raw material, where the percent of some alloying elements is now larger, or LEDs might be centred around a slightly different wavelength. As a result, the device no longer meets its essential performance requirements.

In these cases, the core technology may not be fully understood. Without knowing what makes a good (within spec) product, the range of acceptable material might be expanded—but shouldn’t be—because there’s a lack of clinical data to support expanding the range. If a similar batch of material (e.g., LED bin, similar raw alloy) can’t be obtained, the likelihood of a full production stop because of a lack of material availability becomes a significant concern.  

For each manufactured device, we’re faced with a stack-up of all the part variations. When we bring all the components (and/or sub-systems) together into a product, the stack together and cumulatively must meet the design output. If we haven’t fully explored the limits of essential performance and part variation, we could be faced with a situation where the constituent parts meet all individual specifications, but the combined product fails.

Mitigating these issues may be costly. We may drive up individual component costs to limit the range of variability. We may have to produce (and discard) a large number of parts or material to get the narrow variability range required. In addition to these costs, several additional runs might be needed to produce the quantity of components we require, resulting in production delays.

In the worst case, if the problem stems from raw material, or a circuit component like an LED or IC that is produced in very high volumes, the range of parts required may simply be unavailable.

Worse still, the solution may require that the essential performance range be increased. If the essential performance window is increased post launch, it may require significant verification efforts or even revalidation of the product, including repeated clinical trials.

How Do We Avoid All of This?

Some issues can be managed through proper tolerancing or other manufacturing controls. In many cases, however, a deeper understanding of the core technology is required. Knowing key performance limits, identifying where variation might impact performance, and testing edge cases will help identify the threshold for a bad product.

The rush to get through development must be resisted in these cases, so the core technology is more fully understood. Design of experiments (DOE) can be instrumental in determining not just the primary variables impacting the outputs, but also any interactions between variables. Characterizing the relationships between variables allows us to choose the range of inputs to the design that have the least impact on the outputs, as the relationships may not be linear. Knowing how inputs change the outputs allows us to control the process to produce the most stable output.

In cases where we cannot manage the materials or range of components (such as LED output), we can design in compensation circuits well ahead of verification or validation activities. This allows us to mitigate potential variations.

In summary, the key to design transfer is not only understanding what makes a good product, but equally, understanding the limits of when we could start producing bad product. If we know where the limits are, we can ensure products being manufactured meet performance specifications and avoid costly scrap, rework, and late-stage troubleshooting.

Dana Trousil is a StarFish Medical Mechanical Engineering Team manager. He has successfully launched many products, with experience in a variety of processes, including NPI for medical devices.