Bert Uschold

When developing medical device products and the parts that comprise them, engineers must usually decide what elements of the design are critical to function and, therefore, what dimensions require closer attention during manufacturing. There are many names for such dimensions, including critical dimensions, critical-to-function dimensions, and inspection dimensions.

Regardless of what you call them, it is prudent to have a strategy in place to determine what dimensions are critical and communicate to the manufacturer exactly what critical means. If the engineer identifies too many dimensions, the part will require additional inspections, increasing costs unnecessarily. If the engineer identifies too few, quality could be compromised. This article reviews a set of strategies for identifying critical dimensions early in the development process and for communicating their meaning and importance to the manufacturer.

Defining Critical Dimensions

Before describing how to communicate critical dimensions to the manufacturer, medical device engineers should understand how they are defined. Critical dimensions are those that must meet the print requirement to avoid compromising one or more of the product’s critical functions. The critical ‘function’ could be proper assembly, or it could be used to identify dimensions that are important to the interchangeability of parts or assemblies. Legos are a good example of interchangeability. A critical dimension can also be defined as a dimension that must comply with regulatory or industry requirements, such as the size of a USB connector.

In order to indicate how critical the dimension is and how it will be inspected, a critical dimension can be divided into subtypes. The first subtype is called a statistical process control (SPC) dimension. The most important dimension in the design of a product, the SPC dimension is part of an in-process inspection plan throughout the production process. A noncompliant critical dimension will likely cause an important feature to be defective.

Another type of critical dimension is the validation dimension. While the validation dimension is critical to the function of the product, it has also been shown to or is expected to correlate well with another dimension inspected during the process. Therefore, it needs to be inspected only during a process capability study and in the first production article.

A target dimension is another type of critical dimension. While a target dimension is critical to product function or assembly, it is not as critical as an SPC or validation dimension. Thus, it is only inspected in the first production article, whereby it must be within 66% of the drawing tolerance. Why not just make the drawing tolerance tighter? Because a thorough tolerance analysis does not require it. The 66% requirement ensures a high confidence level throughout the production process that the dimension will be in spec. If a first article is measured just within the specified limits, it is probable that it will frequently be out of spec during production. A tighter first-article requirement is like a minivalidation or process capability study.

A process-sensitive dimension (PSD) is a special type of critical dimension. While it might not be important to the function of a part, it is easy to measure and is a good indicator of whether other dimensions are in spec or not. For example, a validation study may show that three part dimensions vary similarly above and below spec. If one of these dimensions is easy to measure but not critical to function, it could be a proxy for the other two if they are important to the function of the part and are difficult to measure. Overall length is a typical PSD.

Applying Critical Dimensions in Practice

How should engineers apply critical dimensions in a typical medical device manufacturing program? In the early stages of a development project, the first ‘manufactured’ parts are usually machined breadboards or those made using a variety of additive manufacturing processes. Often, such parts can be made without drawings. The critical dimensions, if any, are those that the engineer measures when the parts arrive and are probably not identified in any formal manner. In some cases, it is appropriate to sketch a development drawing delineating a few dimensions in order to communicate some design intent to the personnel responsible for making the part. Designating the handful of dimensions on a development print as critical is probably superfluous because most such dimensions are critical.

The next stage in the development process is the creation of prototype parts. While a wide range of manufacturing and design intent options can be used to produce prototypes, they are not final production parts. Prototyping is a production-like process that creates parts that closely resemble final designs. While an accompanying drawing is not necessarily production-ready, it must be more than a glorified hand sketch suitable for an initial evaluation piece. A drawing suitable for creating a prototyped part should show most of the dimensions and tolerances that will appear on the final production print.

At this point, the engineer should seriously consider the critical dimensions of the production part. Because all critical dimensions are treated as target dimensions at this stage, it is not necessary to specify other subtypes—a step that will be performed later. The critical dimensions need to be specified in the drawing, and ovals are commonly used to designate them.

Now, the engineer has developed a moderately mature but not final design, has decided on a production-like manufacturing process, and has created a drawing that identifies critical dimensions without specifying their subtypes. If an injection mold is required, it is probably necessary to instruct the molder to measure the critical dimensions and perhaps a few other dimensions as well. While a complete first article is not usually produced at this stage, it could be produced if necessary.

Honing Critical Dimensions

When the part is ready to be fabricated using a final production method, the critical dimensions—including their subtypes—should be defined. Since a production process is being used, it is appropriate to identify the PSDs. While there is usually just one process-sensitive dimension, circumstances could demand more.

Next, the medical device engineer should decide which target dimensions should be ‘upgraded’ to validation dimensions for measuring in a process validation study. To accomplish this task, the engineer should consider how critical the dimension is to the design and determine whether it is a tight tolerance or a highly variable dimension. For example, if a dimension in an injection-molded part is susceptible to warp, it should probably be validated.

When two target dimensions are close to each other and the engineer is confident that they will vary with each other, one of them can be validated while the other remains a target dimension. While identifying critical dimensions as SPC dimensions at this stage is optional, the engineer may decide to do so if the dimension in question is crucial to functionality, impacts a regulatory requirement, or is known to have been an SPC dimension in the past.

In order to proceed confidently with a process capability study, engineers should determine that all dimensions slated for such a study fall within 66% of the drawing tolerance at first-article inspection. However, the 66% standard is not a hard and fast value. If the tolerance is on the large side or the process is known to be tight, a higher value might be acceptable. On the other hand, the engineer may have reasons to set it at a lower value. Clearly, if a critical dimension were measured at the limit of the drawing tolerance, few engineers would proceed with a process capability study because the chance of passing would be low.

Finalizing Critical Dimension Studies

After the process validation study has been completed, the engineer knows which dimensions easily meet the process capability (Cpk) requirement and which do not. Validation dimensions that barely pass or that do not correlate well with a process-sensitive dimension are good candidates for ongoing evaluation and for being identified as SPC dimensions. In such cases, ongoing inspection is required because the PSD is not a good proxy for the validation dimension. Alternatively, the engineer could identify a different process control dimension as a better indicator. However, especially if a PSD is present, it is not necessary to define any dimensions as statistical process control dimensions. Since fewer inspections result in lower costs, engineers should decide which validation dimension is harder to measure and choose the easier one.

What if a dimension is both an SPC and a PSD? In such cases, the dimension should be designated as both on the drawing. If the SPC designation is left off, the manufacturer could inappropriately decide to change it. If the PSD designation is left off, the drawing will appear to be missing information, especially if other parts made by the manufacturer call out a PSD. Thus, the PSD designation is not critical and can be left off a drawing for the sake of simplicity, while the SPC designation should always appear because it is important to the designer.

Conclusion

The system described in this article is not perfect. For example, what happens if a vendor is unaware that part dimensions must be inspected in process and a post-validation drawing subsequently shows the existence of one or more SPC dimensions? In such cases, tensions between the designer and the manufacturer can arise over the part’s manufacturing scope, cost, or even delivery time.

The solution to this dilemma is early communication, allowing the parties to estimate a final inspection plan and the costs involved. However, depending on a program’s schedule and the time required to create inspection fixtures and gauges, SPC dimensions may need to be identified before the process validation study is performed. Thus, while designers and manufacturers try to create simple procedures, they must follow them with an eye toward accommodating necessary exceptions.
 

Bert Uschold is a principal mechanical engineer with 20 years of experience in product development. His experience includes medical device development at Ethicon Endo-Surgery and Becton Dickinson. At Radius Product Development, he also developed consumer products and worked on consumer electronics projects. The founder of Dexterity Engineering Solutions, which created the Powerstaxx tolerance analysis spreadsheet, he provides contract product development and tolerance analysis services. His skills include plastic part design, geometric dimensioning and tolerancing, tolerance analysis, Pro/Engineer and Solidworks, and failure mode and effects analysis. Uschold received a bachelor’s degree in mechanical engineering from the University of Detroit. Reach him at [email protected].