An auditor walked a device plant in Minnesota this spring, opened the work environment section of their quality manual, and asked a question nobody had a good answer to: where does compressed air live in your quality system?
The plant had a beautiful cleanroom monitoring program. Particle counts, pressure cascades, gowning, HEPA integrity scans, all of it documented. And then there was a quarter inch line blowing air directly onto a catheter hub right before it got bonded, fed from a thirty year old lubricated screw compressor in a shed behind the building, with no record that anyone had ever tested what came out of it.
That gap is the whole subject of compressed air for medical device manufacturing. The air is a process input. It touches product, it carries your leak test, it feeds your cleanroom tooling, and since February 2 of this year the regulation that governs all of it changed shape.
What Changed in February and Why It Matters Here
FDA's Quality Management System Regulation took effect on February 2, 2026, replacing the old Quality System Regulation. The rule still lives at 21 CFR Part 820, but it now incorporates ISO 13485:2016 by reference, so most of Part 820 points at an ISO clause instead of spelling out the requirement itself. The old QSIT inspection technique was withdrawn the same day.
For your air system the practical effect is this. ISO 13485 clause 6.4 covers the work environment and requires you to document requirements for conditions that can affect product quality. Clause 7.5.1 covers production controls. If compressed air contacts product, contacts a product contact surface, or performs a measurement, it falls inside both, and an inspector can now ask for the documented requirement, the monitoring, and the records.
We are a compressed air supplier, not a regulatory consultant. We will not tell you your plant is compliant and nobody selling you a compressor should. What we can tell you is what the equipment has to do and what evidence it has to be able to produce.
Where the Air Goes in a Device Plant
Device manufacturing has a different load profile from drug manufacturing. There are fewer big continuous users and many small, fussy, high value ones.
| Use | Typical demand | What the air has to be |
|---|---|---|
| Injection molding of housings, hubs, connectors | Largest single load, often 50 to 300 CFM | Dry, clean, general plant air unless it contacts the part |
| Pressure decay and burst leak testing | Small volume, extremely demanding | Dry, stable pressure, stable temperature. This is a measurement |
| Pick and place, grippers, indexers, rotary tables | 10 to 60 CFM, cyclic | Dry and clean, instrument grade |
| Adhesive and UV dispensing, needle purge | Small, continuous | Product contact in practice. Oil free |
| Tray sealers, pouch sealers, form fill seal | 10 to 40 CFM | Product contact at the seal. Oil free, filtered at the machine |
| Cleanroom tooling, pass-throughs, air showers | Varies | Oil free, sterile filtered at point of use |
| Vision system lens purge and part blow-off | Small, wasteful if unmanaged | Oil free, and use an engineered nozzle |
| Laser and ultrasonic welding assist | Small | Often nitrogen rather than air |
Look at that list and notice something. The biggest consumer, molding, usually has the loosest requirement. The smallest consumers set the air quality spec for the whole plant unless you split the system. That inversion is the single most useful thing to understand before you buy anything.
Leak Testing Is Not a Utility, It Is an Instrument
This is the part that separates device plants from everything else we work with, and it is where we take the most calls.
Pressure decay is the dominant leak test method in the industry, and for catheters and fluid path assemblies it is often what you correlate back to a standard like ISO 10555. Modern testers resolve pressure drops into the hundred thousandths of a PSI. At that resolution the air is a measurement variable, and three things wreck your repeatability.
- Temperature. Air temperature change inside the part during the test looks exactly like a leak, because pressure follows temperature. If your supply air arrives at a different temperature than the part and the fixture, you will chase false rejects forever. This is the number one cause, and it is almost always the air, not the tester.
- Moisture. Liquid water in the test circuit changes the effective volume and fouls the precision regulator and the tester's internal valves. Dry air is not optional here.
- Supply pressure stability. If a molding machine cycles and your header sags, the tester's fill phase changes and your cycle time and result drift with it.
The fix is boring and it works. Give the test cells their own dry receiver downstream of the dryer with a check valve isolating them from the plant header, let the air equilibrate to room temperature by running the last run of pipe inside the conditioned space, and put a precision regulator at the fixture. We have seen plants cut a nine percent false reject rate to under one percent without touching the tester.
Air Quality, With the Numbers
There is no USP monograph for device air the way there is in pharma, so your spec comes from your own risk assessment. In practice almost every device plant we work with lands in the same place.
| Where | Reasonable ISO 8573-1 target | Equipment |
|---|---|---|
| Product contact, cleanroom, packaging seal | Class 0 for oil, with particulate and water classes stated | Oil free compressor, desiccant dryer, coalescing plus point of use sterile filtration |
| Leak and burst test air | Dry and particle free. Oil free strongly preferred | Desiccant dryer, dedicated receiver, precision regulation |
| Automation and actuation outside the clean space | General plant quality, dry | Refrigerated dryer, particulate and coalescing filters |
| Molding machine air | General plant quality, dry | Refrigerated dryer, filtration |
A word on Class 0, because it gets sold badly. Class 0 means cleaner than Class 1, with the measured values stated by the manufacturer. It is not a badge, and it is not what an oil free compressor necessarily delivers at the point of use after four hundred feet of pipe. If your spec says Class 0, your evidence is a test at the point of use, not a brochure. Our breakdown of ISO 8573-1 air quality classes explains what each digit means.
In a cleanroom, sterile grade filtration at 0.01 micron or better is the normal final stage, mounted at the machine rather than in the utility room. Everything downstream of your last filter is a potential contamination source, so put that filter as close to the product as you can.
Validation and the Evidence Trail
Here is the part that costs people a finding.
ISO 8573 is a multi part standard. Part 1 is the classification you write in your spec. Parts 2 through 9 are the test methods for oil aerosol, oil vapor, particles, water, and microbiological content. If your quality system says the air meets a class, you need a test performed by those methods, by a lab that runs them, and a record of the result.
- Sample at the point of use. Testing at the dryer outlet tells you the dryer works. It tells your auditor nothing about the air coming out of the drop in the clean room.
- Set a frequency and justify it. Annual is common for device plants, semiannual where the risk assessment pushes harder. Pick one and keep to it.
- Qualify the equipment like equipment. Compressor, dryer and filtration go through IQ, OQ and PQ like the rest of your production equipment, and filter change intervals belong in your PM system.
- Change control applies. Swapping a filter element brand, moving the intake, or adding a drop is a change to a validated utility.
Our piece on compressed air for pharmaceutical manufacturing covers the drug side, and much of the filtration logic transfers.
The maintenance program that keeps all of this defensible, including why your PM tasks should name specific part numbers rather than descriptions, is in preventive maintenance for a validated compressed air system.
What We Can and Cannot Sell You
Being straight about this saves everyone time.
We can supply the oil free compressors, dryers, filtration trains, nitrogen generators, receivers, and piping that make up a device plant's process air and instrument air system, and we will size it with you.
We do not supply certified medical gas systems. Piped medical air for patient care under NFPA 99, with the verification that goes with it, is a different scope with a different contractor. Device companies with an attached clinical facility do ask us, so it is worth saying. Our hospitals guide covers where that line sits.
We do not write validation protocols. We will give you equipment documentation, performance data, and test port locations that make yours easier to write.
Sizing a Device Plant
| Plant | Typical demand | Common setup |
|---|---|---|
| Single cleanroom assembly cell, no molding | 15 to 40 CFM | 10 to 15 HP oil free scroll, desiccant dryer, 120 gal, point of use filtration |
| Assembly plus packaging plus leak test, molding outsourced | 50 to 150 CFM | 25 to 50 HP oil free screw, duplex desiccant, 240 to 400 gal, dedicated test cell receiver |
| Integrated plant with in house molding | 200 to 600 CFM | Split system: lubricated screws for molding and shop air, oil free for clean and product contact, 660 gal and up |
The split system is what we quote most often above about 150 CFM. Paying the oil free premium on the molding load to protect a 20 CFM packaging line is a bad trade. Run two systems, keep them physically separate so nobody can cross connect them with a hose on a Friday afternoon, and label every drop.
Redundancy deserves a sentence. A plant that cannot ship because the compressor is down has a supply problem that ends up in a corrective action, so most go to two machines rather than one larger one.
What Device Plants Get Wrong
- One system for everything, specced to the fussiest user. Expensive, and the cost usually gets cut somewhere worse.
- Testing at the compressor instead of the point of use. The most common finding we hear about.
- Chasing leak test false rejects at the tester. Check the air temperature and the dew point first, every time.
- No check valve between the test cells and the plant header. Your test results now depend on what the molding floor is doing.
- Treating a filter change as maintenance rather than change control. Fine until an auditor asks which element is in there and why.
- Blow-off with an open tube in a clean space. It wastes air and moves particles around a room you spent money making clean.
- Intake drawing from the loading dock. Truck exhaust into the compressor is a contamination path you did not write down.
Frequently Asked Questions
Do I need an oil free compressor for medical device manufacturing?
For any air that contacts product, a product contact surface, or the inside of a cleanroom, yes in practice, and that is what an auditor will expect to see. For molding machine air, shop air, and automation outside the clean space, a lubricated screw with proper filtration is normal and defensible. Most plants above about 150 CFM split the system and pay the oil free premium only where it buys something.
What ISO 8573-1 class should a device plant specify?
There is no regulation that hands you a number, so it comes out of your risk assessment. In practice, product contact and cleanroom air lands at Class 0 for oil with stated particle and water classes, and general plant air sits well below that. Whatever you write, you have to be able to prove it by the ISO 8573 part 2 through 9 test methods, measured at the point of use.
Why does our pressure decay leak test keep giving false rejects?
Start with air temperature, not the tester. A temperature difference between the supply air, the part, and the fixture produces a pressure change that is indistinguishable from a leak. Then check dew point, then check whether the header sags when other equipment cycles. Give the test cells a dedicated receiver with a check valve and run the final pipe inside the conditioned space so the air arrives at room temperature.
Did the QMSR change what I have to do about compressed air?
It changed the framing more than the substance. Since February 2, 2026, 21 CFR Part 820 incorporates ISO 13485:2016 by reference, so the relevant requirements now read through clause 6.4 on work environment and clause 7.5 on production controls. If air affects product quality, you document the requirement, control it, and keep records. Many plants that were already ISO 13485 certified had this covered. Plants that were running on the old QSR alone are the ones finding gaps.
Can you supply medical air for a hospital or clinic?
No. Certified piped medical gas systems under NFPA 99 are a separate scope with their own installers and verifiers, and we do not do that work. We supply process air and instrument air for manufacturing. If your facility has both, keep them completely separate systems and do not let anyone tie them together.
Where to Start
Two questions will tell you where you stand. Has anyone tested the air at the point of use, by the ISO 8573 methods, within the last year? And is your leak test air on its own receiver? If the answer to either is no, that is your next project, and both are cheap compared to a finding or a scrap rate nobody can explain.
Browse oil-free air compressors, or the desiccant air dryers and filtration that go with them. Send us your clean space classification, your leak test equipment, and whether you mold in house, and we will size the system with you.
