Your Medical OEM Tooling Questions: Quality Standards for Injection Molds, PVC Molds, Silicone Molds, and More

What We'll Cover Here

If you're sourcing OEM tooling for medical devices—think injection molds, PVC components, silicone parts, or assemblies like IV cannulas and stopcocks—you probably have a bunch of specific questions. Not the generic 'how much does a mold cost' kind. The real ones, like 'what tolerances actually matter for a luer lock?' or 'can I use the same mold for PVC and silicone?'

I've been reviewing tooling specs and production deliveries for Medline for about six years now. Before that, I was on the supplier side, managing quality for a contract manufacturer. So I've seen these questions from both ends of the table. Here are the ones I get asked most often—and the answers I wish someone had given me earlier.

1. What's the single most important specification for a medical injection mold?

Honestly? It's not the material or the cycle time. It's the cavity-to-cavity consistency. I rejected a batch of 8,000 stopcocks back in 2019 because one cavity was producing parts with a flash line 0.02 mm over spec. The other seven cavities were fine. The vendor argued it was 'within industry standard.' But for a fluid pathway component, that's a contamination risk and a potential leak path. We rejected the whole lot. That $18,000 project got delayed by three weeks.

So if you're specifying a mold, demand a cavity-to-cavity variation report during first article inspection. Most vendors can do it. Not all will offer it unless you ask.

2. Is a steel mold always better than an aluminum one for medical parts?

That's one of those things that was true 15 years ago when aluminum molds wore out fast. Today, it depends on volume and material. For high-volume PVC parts—like the tubing on an IV set—you'll want hardened steel (usually P20 or H13). Steel holds tolerance for millions of cycles. For shorter runs of silicone components, like respiratory masks or seals, an aluminum mold can be perfectly adequate and a lot cheaper to modify if your design iterates. We've got aluminum molds that have run 100,000+ parts with no issue. But—and this is key—aluminum doesn't handle abrasive fillers well. If your material has glass or mineral fillers, go steel.

3. What tolerances should I specify for an IV cannula or a stopcock?

The short answer: tighter than you think, but not blindly tight. For a luer lock connector, the critical dimensions are the thread form and the taper angle. ISO 594 (now ISO 80369) specifies a 6% taper. If your mold doesn't hold that taper within ±0.05 degrees, you'll get connectors that leak or cross-thread. I've seen this cause field failures that required a recall—not something you want. For a stopcock bore diameter, typical tolerance is ±0.05 mm. For a cannula bevel angle, you're into ±0.5 degrees. The cost of a high-precision mold is real, but the cost of a recall is a lot higher.

4. Can I use the same injection mold for PVC and silicone?

Not recommended, and here's why. PVC and silicone have different shrinkage rates (PVC is around 0.5-2.5%, silicone is higher, often 2-4%). They also process at different temperatures—PVC runs around 160-180°C, while silicone can require 180-220°C (or more for LSR—liquid silicone rubber). If you try to use the same mold, you'll get parts that are dimensionally different because the shrinkage isn't the same. Also, silicone can leave residue that contaminates the next PVC run. I had a vendor try this once to save cost. The silicone parts were fine. The PVC parts that came out next? All rejects.

5. What's the biggest misconception about PVC molds for medical devices?

The assumption that 'PVC is cheap, so the mold should be cheap.' Actually, medical-grade PVC (like USP Class VI) has tighter thermal stability requirements. If your mold isn't designed with proper cooling channels to maintain uniform temperature, you'll get degradation—the PVC yellows, and you get 'fish eyes' (unmelted particles) that clog filters or lines. I've seen a $200,000 annual contract lost over a poorly designed PVC mold that caused intermittent yellowing. The material cost per part was pennies. The rework and lost trust cost a lot more.

Also, people think PVC is always the lowest-cost option. That was true a decade ago. Today, with resin price volatility and stricter REACH and RoHS requirements, silicone or thermoplastic elastomers (TPEs) can sometimes be comparable or even cheaper when you factor in scrap rates and regulatory compliance costs. Worth running a total cost analysis.

6. For silicone molds, how important is the gate location?

Critical. Silicone flows differently than thermoplastics. It's a shear-thinning fluid, so gate location determines flow length and fill pressure. If you put the gate at the wrong spot on a respiratory mask, you get knit lines that can become crack initiation points during sterilization. I've seen parts fail burst testing because of poor gate placement. Good simulation analysis (mold flow analysis) before cutting steel is worth its weight in gold—especially for silicone. You can also look into cold-runner systems for LSR. They reduce material waste significantly.

7. How do I verify my mold meets the spec without a lab?

This was accurate as of early 2025. The FDA's QSR (21 CFR Part 820) and ISO 13485 both require you to have documented acceptance criteria and to verify conformance. You don't need a full metrology lab yourself. You need: (1) a first article inspection report (FAIR) from the mold maker with CMM (coordinate measuring machine) data for critical dimensions; (2) a visual inspection under good lighting for surface finish, flash, and sink marks; (3) a functional test—assemble the cannula into a luer adapter, rotate the stopcock, etc. If it binds or leaks, spec sheets don't matter. Trust the part's behavior. And keep a 'golden sample' from the first good run for comparison on future batches.

That rule about keeping physical samples? I learned it the hard way. A vendor changed their polish compound once and the surface finish changed from 0.4 to 0.8 Ra. The new parts looked matte and felt 'sticky.' We didn't have a reference physical sample, only a print spec. I could argue the numbers, but a physical comparison would have cut the debate in half. Keep the sample.