Surgical planning

My $11,000 Quality Lesson: Why 'Within Spec' Wasn't Good Enough for ResMed AirMini Batteries

Posted on 2026-11-17 by Samuel Okafor
Surgical article header

The Tuesday Batch That Changed How We Verify Everything

In our Q1 2024 quality audit, I was standing in a receiving bay with a clipboard and a battery tester, staring at 340 units of battery for ResMed AirMini that had just arrived from a new supplier. The paperwork said everything was within spec. The paperwork, as it turned out, was missing a few things.

I've been a brand compliance manager for a mid-size DME (durable medical equipment) distributor for about six years. I review every medical device deliverable before it reaches customers—roughly 200 unique items annually, spanning everything from ResMed CPAP consumables to surgical energy devices, mechanical ventilators, and even different types of MRI coils. And yeah, I've become the person teams don't love seeing in the receiving bay. I've made my peace with that.

That Tuesday's shipment was supposed to be simple. ResMed AirMini batteries, some ResMed Air Mini muffler units (the little inline noise reducers), and a handful of replacement masks. Routine.

It wasn't routine.

What 'Within Tolerance' Actually Looked Like

We test every lithium-ion battery pack that comes through. The protocol is straightforward: charge cycle, discharge cycle, capacity check at 37°C (body temperature, because these devices run against a patient's face all night).

The first unit read fine. The second read 8% below rated capacity. The third was within 1%. By the twelfth unit, I had a pattern: roughly one in ten packs was coming in at 7-9% below the manufacturer's rated mAh, which—critically—sits inside the "acceptable industry tolerance" the supplier kept citing in their emails.

Here's the thing about tolerance. The vendor's spec sheet said ±10% capacity variance was "industry standard." Fine. But our patients aren't flying with these devices in a lab. They're using them on 14-hour international flights, or overnight in places where a power outlet isn't guaranteed. A 10% shortfall on a 90-minute rated battery isn't a lab curiosity—it's a patient stranded without therapy.

We rejected the batch. The vendor redid it at their cost. That conversation went about as well as you'd expect.

The Blind Comparison That Made Me Rethink Specs

Here's where my thinking shifted. After the AirMini battery mess, I ran a blind test with our team: same device, same therapy settings, two batches—one from the original supplier, one from the "within tolerance" batch. Three of my six colleagues couldn't tell the difference in a 30-minute session. Two noticed the battery life. One—our most experienced tech—picked up on it in about twelve minutes.

I'd love to tell you there was a dramatic aha moment. There wasn't. The real insight came when I compared our Q1 and Q2 rejection logs side by side—same vendor, different spec requirements. In Q1, before we tightened the battery spec, we were rejecting 22% of shipments. After the spec change, that dropped to 4%. The specification, not the vendor, was doing the work.

So now every contract we sign for ResMed-compatible accessories includes the tighter number. Not because I'm a perfectionist. Because a spec that lets 22% of bad units through isn't a spec—it's a suggestion.

I still kick myself for not running that blind test in 2022. If I had, we'd have caught the muffler issue (the ResMed AirMini muffler housing wasn't seating cleanly on about 6% of units, causing a rattle the vendor called "acoustic variance") about eight months earlier than we did.

The Parts Nobody Warns You About

Quality work in medical devices is strange because the categories don't talk to each other. A surgical energy device failure and a CPAP muffler failure share almost no physical characteristics. But the failure patterns? Weirdly similar.

Let me give you the pattern, because once you see it you can't unsee it:

  • Packaging passes before the product does. Every single time. The carton is perfect. The insert is perfect. The unit inside is the one that fails. I've stopped trusting a clean box.
  • "Within tolerance" is a negotiating position, not a measurement. When a vendor leads with tolerance language in the first email, tighten the spec.
  • Small-unit failures hide in small batches. A 2% defect rate in a 340-unit shipment means six or seven devices that will fail in the field. On a 50,000-unit annual order, 2% is a thousand devices.

That last one isn't hypothetical. In 2023, we ran a batch of 8,000 CPAP filter assemblies where the defect rate was "only" 1.8%. That's 144 filters that didn't meet spec. We caught most of them in QC. We didn't catch all of them. Two made it to a sleep center in Ohio, and I spent three weeks on that phone call.

That quality issue cost us about $11,000 in returns and a very uncomfortable conversation with a respiratory therapist who had every right to be frustrated. The containment process we built afterward—full-batch quarantine on any sub-2% defect finding—has held up across every device category since.

Where the Honest Limitation Comes In

I'm going to say something that'll sound strange coming from someone in my seat: Not every ResMed product is the right fit for every facility, and I don't recommend ResMed for every scenario.

Here's what I mean. If you're a large sleep center running 40+ patients a night and you've got the infrastructure for centralized humidification, cloud-based compliance tracking, and dedicated equipment staff—the full ResMed ecosystem is genuinely excellent. The AirSense line, the AirMini for travel, the AirFit mask range—the whole stack works together, and the myAir app data is actually useful for adherence monitoring.

But if you're a small two-bed sleep lab, or a home-care distributor serving patients with limited tech literacy, that same ecosystem can be overkill. You're paying for connectivity features nobody uses, and your support burden goes up because patients can't figure out the app. I've seen distributors buy into the full platform and then eat the cost of unused licenses because their patient population just wanted a mask and a machine that worked.

The question isn't "which brand is best." It's "which product fits your actual operating reality." Those are different questions. I'd rather tell a small lab to look at simpler alternatives and keep them as a customer for ten years than sell them a platform they'll abandon in six months.

What I Do Now That I Didn't Do Before

Three things. In order:

First, I document verbal promises. Every single one. If a vendor says "this batch is functionally identical to the previous one," that goes in an email within the hour. I still kick myself for the 2022 AirMini muffler situation—if I'd had the "functionally identical" claim in writing, we'd have had grounds to recover the delayed launch costs. Instead we ate them.

Second, I standardized the spec document. One document, one process, across every device category we touch—surgical energy devices, mechanical ventilators, MRI coils, all of it. Before 2023 we had different spec sheets for different suppliers. That was the actual source of most of our problems, not the suppliers themselves.

Third, I do the blind test before I need it. Not during a crisis. Before. The eight months we lost on the muffler issue would have been four if I'd just run the test when the vendor relationship started.

Per FTC guidelines (ftc.gov), advertising claims for medical devices need to be substantiated and not misleading. That applies to what we tell patients, what we tell facilities, and—in my view—what we tell ourselves about our own supplier relationships. "Within tolerance" is a claim. It needs evidence. The evidence has to hold up in the field, not just on the spec sheet.

None of this is glamorous work. It's clipboards and battery testers and uncomfortable phone calls. But it's the difference between a device that reaches a patient and one that shows up in a return bin.

One Last Thing

If you're specifying ResMed equipment—or any medical device, really—the honest question isn't "is this the best product on the market." It's "have I written the spec tightly enough that the product I receive is the product I paid for."

Those two questions have very different answers. One of them will save you money. The other will save you a phone call you really don't want to make.

Permalink Ask a Specialist
Samuel Okafor

Samuel Okafor

Samuel Okafor is an in-vitro diagnostics and clinical laboratory equipment analyst covering chemistry and hematology analyzers, immunoassay systems, PCR platforms, centrifuges, and electronic pipettes. He applies IEC 61010-2-101 and examines analytical sensitivity, precision, carryover, calibration traceability, sample throughput, reagent stability, reportable range, and operator maintenance requirements. His guides help laboratory directors, clinical scientists, quality teams, and buyers evaluate assay performance, biosafety, automation capacity, consumable dependence, and total workflow cost.