A Quality Inspector's 6-Step Checklist for Medical-Grade Manufacturing

Who this checklist is for

Every part that leaves our floor gets reviewed before it reaches a customer. Medical parts especially. I'm a quality compliance manager at a contract manufacturer—I review roughly 200 unique items a year, and in 2024 I rejected about 12% of first deliveries. Mostly tolerance drift, wrong surface finish, missing documentation.

If you're specifying medical device components, you have more options than ever: CNC machining on machines like Makino's, injection molding with engineered resins, 3D printing, laser welding. Each one has a place. The problem is figuring out which one will actually pass inspection. Here's the 6-step checklist I use internally. It's practical, not theoretical.

Step 1: Lock down critical dimensions before you get quotes

The first thing I look for on any drawing is which dimensions are marked critical. Marked. Not implied. If a tolerance is under ±0.005" or a feature affects fit, function, or safety, it needs to be flagged on the drawing.

Here's why that matters: machinists default to general shop tolerances when you don't specify. And general tolerances are usually ±0.010" or looser. Makino CNC machines—the horizontal and vertical machining centers I've seen in medical device plants—hold much tighter tolerances with the right program. But even the best equipment can't fix a drawing that never flagged the critical features.

So: circle the critical dimensions, assign realistic tolerances, and get written acknowledgment from your vendor. (Should mention: I've seen prints with tolerances that were impossible for the material specified. A ±0.002" tolerance on a thin-wall plastic part needs a lot of engineering behind it.)

Step 2: Match the material to the process, not the other way around

Sounds obvious, but it's the second most common reason I reject first deliveries. People pick a process first, then try to force a material through it.

For the two processes people get confused about most:

• What kind of material do 3D printers use? For medical parts: Ti-6Al-4V titanium, cobalt-chrome, PEEK, ULTEM, and medical-grade stainless like 316L. Powder bed fusion is the dominant metal process, and the powder comes with its own handling requirements. What I mean is: don't assume the powder supplier's certificate is enough—you need lot-level traceability through the build.
• Common plastic resins for injection molding: polycarbonate, ABS, POM (acetal), nylon (PA66), and PEEK. The resin's moisture sensitivity, glass transition temperature, and shrinkage all affect final dimensions. A resin that works great for a consumer snap-fit part may not hold tolerances for a surgical instrument handle.

A customer once saved $1,200 by switching to a cheaper resin on a 10,000-unit order. The parts failed tensile testing—every single one. The rework cost $9,000 and delayed the launch by three weeks. (Which, honestly, was painful to watch.)

Put another way: material choice drives process choice. CNC machining gives you the widest material selection. 3D printing gives you geometry freedom but a narrower material list. Injection molding gives you low per-unit cost but high upfront tooling.

Step 3: Verify the process, not just the part

Here's the step most people skip: don't just measure the part. Verify the process that made it.

In our Q1 2024 audit, we caught a machine that was producing parts to print—but spindle load was 18% above baseline and the tool wear log showed the end mill was past its replacement limit. The parts measured fine. They would have passed any dimensional inspection. But the process was unstable, and within another 50 parts, the tolerance would have drifted.

What I check: calibration records, tool wear logs, in-process data (spindle load, temperature, vibration), and environmental conditions—temperature swings affect metal machining and resin molding.

If you're in a regulated space, this is where process validation (IQ/OQ/PQ) comes in. Under FDA 21 CFR Part 820—which now incorporates ISO 13485 as of February 2024—you don't just prove the part works; you prove the process can consistently produce it. Makino machines for medical applications are popular in part because their control systems log the kind of process data this step depends on.

Step 4: Check the joining method before you commit

Medical assemblies often involve joining metal: instrument handles, cannulas, implant components. Laser welding is one of the cleanest options—no filler material, localized heat, high weld strength.

If you're evaluating laser welders, the X1 Pro laser welder price is what typically catches the eye. But the price is the starting point, not the decision. Full cost of ownership includes consumables (shielding gas, lenses, nozzles), calibration and maintenance intervals, operator training, and rejection rate. A welder is only as good as the person running it.

I can only speak to our experience with mid-size production runs. If you're doing high-volume automated welding, the calculus might be different. And if you're doing low-volume work, outsourcing to a validated contract shop might beat owning the equipment—buying a laser welder that sits idle half the year is hard to justify. (Not that I have an agenda toward contract manufacturing. Maybe I do.)

Step 5: Inspect surface finish and cleanliness like your product depends on it

Because it does. For implants, surface roughness affects how cells attach—too smooth and tissue doesn't integrate; too rough and bacteria colonize. For surgical instruments, surface finish affects sterilization.

I always verify: Ra (roughness average) with a profilometer, not a guess; passivation for stainless steel (this creates the corrosion-resistant layer); residue from coolant, machining oil, or release agents; and burrs—a burr can cut a glove, a surgeon, or a patient.

The surprise wasn't the dimensional results. It was how much contamination showed up on parts that looked clean under a shop light. A handheld UV lamp revealed residue the eye couldn't see. Add that to your checklist.

Step 6: Run a pilot batch before full production

Whatever process you choose, I insist on a pilot batch. Not a first article—a batch of 10 to 50 units that mirrors the exact full-production process. CNC machines drift. Resins age. Operators change shifts. A lot happens between the first article and the thousandth piece.

When I implemented our verification protocol in 2022, we started requiring three consecutive pilot batches before approving a new production line. Customer satisfaction scores rose 34% in the following year's survey.

Pilot batches add lead time. But catching a tolerance drift on 20 parts beats catching it on 2,000. One drift we caught in pilot saved us an $18,000 redo. Missing it would have been a $22,000 write-off.

Common mistakes I see

Buying on unit price alone. The cheapest resin, the cheapest welder, the cheapest machine shop. I've never seen it work for medical parts. The hidden costs—rejection, rework, delays—always surface. (Surprise, surprise.)

Assuming certifications mean standards match. ISO 13485 tells you the vendor has a quality system. It doesn't tell you their internal tolerance standards match yours. We once received a batch where a dimension was 0.008" off our 0.005" spec. The vendor claimed it was within industry standard. We rejected it, and they redid it at their cost. Now every contract states the spec explicitly.

Ignoring documentation. Part traceability is the law in medical manufacturing—21 CFR Part 820 requires it. If your vendor can't provide material certifications, process logs, and inspection data, the part might as well not exist.

Bottom line

The industry is evolving—what was best practice in 2020 may not apply in 2025. CNC machining remains the workhorse for precision medical parts, and Makino machines for medical applications are everywhere for a reason. But injection molding and 3D printing have earned their place, and laser welding is more accessible than it used to be. The fundamentals haven't changed: lock down specs, verify the process, document everything, and don't let short-term savings drive long-term decisions.