Medical Device Design Basics: ISO 13485 and the Extra Rigor Mechanical Parts Require

The first time I was pulled onto a medical device project after years of designing production equipment for automotive and electronics lines, I assumed the mechanical design skills would transfer almost directly. A bracket is a bracket, a linkage is a linkage. What I wasn’t prepared for was how much of the actual engineering effort shifts away from the geometry itself and into documentation, traceability, and risk analysis that has to exist before you’re even allowed to cut a single part. Working under ISO 13485 changed how I think about mechanical design permanently, even on projects that have nothing to do with medical devices. This article covers what surprised me most, and what I now treat as non-negotiable whenever a mechanical part might end up inside or near a patient.

Design Controls Come Before the CAD Model, Not After

In most industrial equipment work, you can sketch a concept, model it, and refine requirements as you go — the requirements and the design co-evolve. Under ISO 13485’s design control process, that order is inverted. User needs and design inputs have to be documented and reviewed before design output begins in earnest, and every design output has to trace back to a specific input. The first device project I worked on, a component of a drug-delivery mechanism, had a design input document nearly as long as the eventual drawing package, specifying things I would never have written down for an industrial machine: maximum actuation force a patient’s hand could realistically apply, expected number of use cycles over the device’s shelf life, and tolerable levels of particulate generation from sliding surfaces.

I learned to treat that input document as the actual specification, not the informal conversations with the product manager. On one occasion I redesigned a latch mechanism based on a verbal request to “make it easier to open,” without updating the formal design input for actuation force. The change looked fine in testing, but during a design review the quality team correctly flagged that we had no documented basis for the new force value, and I had to go back and formally justify it against usability data before the change could proceed. That taught me to route every mechanical change, however small, back through the input document first.

Risk Analysis Drives Geometry Decisions You’d Otherwise Skip

ISO 14971 risk management runs alongside 13485 on nearly every device project, and it changes mechanical decisions in ways that aren’t intuitive if you’re used to designing for manufacturability and cost alone. On a handheld surgical instrument housing I worked on, the risk analysis flagged pinch points between a moving trigger and the fixed housing as a hazard, even though the gap was well within what I’d consider normal and safe on industrial equipment. The mitigation wasn’t a warning label — it was a geometry change, adding a radiused guard that physically prevented a gloved finger from reaching the pinch point at all.

I now build a habit of running a rough hazard walk-through on any moving mechanical interface before I finalize geometry, asking specifically about pinch points, sharp edges, entrapment gaps sized for fingers, and any surface that could shed particles or fragments during use. It’s a different mindset from optimizing for assembly time or material cost, and it has to happen early, because retrofitting a guard or a radius after tooling is committed is far more expensive than designing it in from the start.

Material Selection Means Biocompatibility, Not Just Mechanical Properties

Choosing a material for a bracket in a factory machine is mostly a question of strength, cost, and availability. For anything that contacts a patient, even indirectly, biocompatibility testing requirements under ISO 10993 become part of the material decision, and they eliminate options that would otherwise be perfectly reasonable. I once specified a common engineering plastic for a housing component based purely on its mechanical properties and machinability, only to have the material rejected because it hadn’t been characterized for the relevant biocompatibility endpoints and getting it tested would have added months to the schedule.

Since then, I keep a running list of pre-qualified, commonly used medical-grade materials — certain grades of stainless steel, specific medical PC and ABS blends, USP Class VI silicones — and I default to those unless there’s a strong reason not to. Switching to an already-characterized material, even if it’s slightly more expensive or slightly harder to machine, is almost always faster overall than introducing a new material and absorbing the biocompatibility testing timeline.

Tolerances and Process Validation Go Far Beyond Drawing Callouts

On industrial equipment, a tight tolerance on a drawing is usually verified by inspecting a sample of parts and trusting the process to hold. For a device component that affects patient safety, a tight tolerance often triggers a requirement for process validation — installation qualification, operational qualification, and performance qualification, commonly shortened to IQ/OQ/PQ — which proves statistically that the manufacturing process reliably produces parts within spec, not just that a handful of samples happened to measure in range.

I ran into this directly on a fluid-path component with a critical inner diameter tolerance. What would have been a straightforward CNC machining callout on an industrial part turned into a multi-week validation study involving dozens of parts, capability index calculations, and a formal report before the process was approved for production use. It changed how I write tolerances now — I ask myself, for every tight tolerance on a device drawing, whether I actually need that tolerance for the function to work, because every unnecessarily tight callout multiplies the validation burden downstream. Loosening a tolerance from what “feels right” to what’s actually functionally required has saved real time and cost on more than one project.

Traceability Means Every Part Needs a Paper Trail Back to You

In device manufacturing, if a failure shows up in the field, the manufacturer needs to be able to trace that specific unit back through its manufacturing lot, its component suppliers, and ultimately back to the design decisions and risk analysis that justified the design in the first place. As the mechanical designer, that means my drawings, my design change records, and my rationale documents all become part of a permanent Design History File that has to hold up to scrutiny years later, potentially during a regulatory audit or a field investigation.

This changes how carefully I document rationale, even for decisions that feel minor. Early on I’d change a fillet radius or a wall thickness based on a quick DFM conversation with a machinist and just update the drawing. On device projects I now write a short rationale note for every dimensional change, even trivial-seeming ones, because six months later — after the DHF is closed, and a batch of parts is already in the field — nobody remembers the informal conversation, but the paper trail is what an auditor or an investigator will actually see.

Sterilization Compatibility Constrains Geometry and Material in Ways You Don’t Expect

If a device or component needs to be sterilized — by autoclave, ethylene oxide, or gamma radiation — that requirement reaches back into both material selection and geometry decisions that would otherwise seem unrelated to sterility at all. I designed a housing early in my device work using a snap-fit assembly method I’d used successfully on dozens of industrial enclosures, only to learn during design review that autoclave sterilization cycles subject the polymer to repeated thermal cycling that fatigues snap features over time, and that the sterilization validation would need to demonstrate the snaps still held after the device’s full expected number of reprocessing cycles.

Geometry also matters for sterilant penetration and residue removal — blind holes, tight crevices, and lap joints can trap moisture or sterilant residue, so I now actively design surfaces to be self-draining and avoid unnecessary crevices wherever the part will be exposed to a sterilization process. It’s a design consideration that simply doesn’t exist on most industrial equipment, and missing it early means expensive redesign after a failed sterilization validation.

Documentation Discipline Is the Real Skill, Not the Mechanical Design Itself

If there’s one lesson from device work that has permanently changed how I operate as a mechanical designer, it’s that the mechanical engineering itself — the stress calculations, the tolerance stack-ups, the material selection — is often the easier half of the job. The harder half is maintaining a documentation trail rigorous enough that someone with no memory of the project could reconstruct exactly why every dimension, every material, and every design change exists, years after the fact.

I now apply a lighter version of that discipline even on non-medical projects, because it consistently pays off: a one-line rationale note attached to every non-obvious design decision, a clear traceability chain from requirement to feature, and risk considerations written down rather than just discussed in a meeting and forgotten. Medical device work is demanding in ways that can feel bureaucratic when you’re used to moving fast on industrial equipment, but the underlying discipline — know why every feature exists, and be able to prove it later — makes for better engineering everywhere, not just where a regulator is watching.