The Leak Tester That Rejected Every Good Part
We built a leak test station for a plastic housing. The spec was pressure decay at 50 kPa for 10 seconds, with a decay limit of 5 Pa. The tester rejected 30% of parts that the customer’s manual gauge test passed. We checked the seals, the fixture, the instrument — everything was fine. The problem was that the housing was a plastic part that expanded under pressure. When we charged it to 50 kPa, the plastic ballooned. The pressure dropped not because of a leak, but because the part stretched. The volume changed, so the pressure changed. The leak detector was measuring part expansion, not leakage.
Leak testing methods for production look simple on a datasheet — pressurize, wait, measure pressure drop. But the physics of the test depends on the part material, volume, temperature, and the sealing method. A leak tester that works on a rigid metal housing fails on a flexible plastic part. This article is how I design leak test stations that actually measure leakage, not part physics.
The Three Main Leak Test Methods
There are three common production leak test methods. Each has a sensitivity, cycle time, and cost range. Pick based on what the part is and how tight the spec is.
Pressure Decay
The most common. You seal the part, pressurize it to a set pressure, isolate it from the supply, and measure the pressure change over a set time. If the pressure drops more than the limit, the part leaks.
How it works: ΔP = Leak rate × Time / Volume. The smaller the part volume, the more sensitive the test (same leak creates a bigger pressure change in a small volume).
Best for: rigid parts (metal, glass, hard plastic), moderate leak rates (1–100 Pa/s), most production applications.
Mass Flow
Instead of measuring pressure decay, you measure the air flow needed to maintain a constant pressure. A small leak requires a small make-up flow. A large leak requires more flow. The mass flow sensor measures this directly.
Best for: large volume parts (where pressure decay would be insensitive), very small leak rates, and high-speed testing (no settling time needed).
Pressure Rise (For Vacuum Applications)
For parts that operate under vacuum (vacuum chambers, sealed enclosures), you pull a vacuum, isolate, and measure the pressure rise. Air leaking in causes the pressure to climb. This is the vacuum equivalent of pressure decay.
| Method | Sensitivity | Cycle Time | Best For |
|---|---|---|---|
| Pressure decay | Medium (0.1–10 Pa/s) | 10–30 s | Rigid parts, general production |
| Mass flow | High (0.01–1 Pa/s) | 3–10 s | Large volumes, fast cycles |
| Pressure rise (vacuum) | Medium-high | 15–30 s | Vacuum-operated parts |
| Immersion (bubble) | Visual, qualitative | Slow | Finding leak location, not production |
The Variables That Actually Determine the Test
The leak rate number on the spec (e.g., “must not leak more than 5 sccm”) is meaningless without understanding the test conditions. Four variables determine whether the test works.
Part Volume
Pressure decay sensitivity depends on volume. A small part (10 mL) shows a big pressure drop from a small leak. A large part (10 L) shows a tiny pressure drop from the same leak. For large-volume parts, use mass flow instead of pressure decay — or use a differential test (compare the part to a known-good reference volume).
Temperature
Air temperature changes pressure. If the test air is warmer than the part (or vice versa), the temperature equalizes during the test, causing a pressure change that looks like a leak. This is the #1 source of false failures in leak testing.
When you charge a part with compressed air, the air is at compressor temperature (often 40–60°C). The part is at room temperature (20°C). As the air cools inside the part, the pressure drops — and the leak detector reads it as a leak. The fix is a dwell (settling) time after charging, before the measurement starts. Give the air 5–10 seconds to equalize temperature with the part.
Part Material (Rigid vs. Flexible)
Back to the plastic housing that ballooned. Flexible parts expand under pressure. The volume change from expansion causes a pressure change that’s not leakage. For flexible parts (plastic, rubber, thin-walled aluminum):
- Use a lower test pressure (to minimize expansion).
- Use a longer dwell (let the expansion stabilize before measuring).
- Use a differential test (compare against a reference part, which also expands, canceling the expansion effect).
Sealing Method
The part has to be sealed at its openings. The sealing fixture itself can leak — and a leak in the fixture seal is indistinguishable from a leak in the part. Design the seals to be more reliable than the part’s own leak rate.
Fixture Design for Leak Testing
The leak test fixture holds the part, seals the openings, and supplies the test pressure. It’s half the test.
Sealing the Openings
- Pneumatic seal plugs: A rubber plug that’s pushed into the opening and expanded (or compressed) to seal. Standard for round ports. The plug should be sized to the port, not forced into a mismatched hole.
- Gasket face seals: For flat-face openings, a flat gasket clamped against the surface. Use a urethane or silicone gasket that conforms to minor surface irregularities.
- Custom-machined seals: For irregular openings, a custom-machined urethane plug that matches the part geometry. The cost is worth it if the seal leaks.
Fixture Rigidity
The fixture should not flex under test pressure. A flexible fixture that moves when pressurized changes the test volume, causing false decay. Use a rigid aluminum or steel fixture frame, not a thin sheet metal plate.
Reference Part (Master)
Every leak tester needs a reference — a known-good part that doesn’t leak. Run the test on the reference part periodically to verify the instrument, fixture, and seals are working. If the reference part starts failing, the fixture or instrument has drifted. If the reference passes and production parts fail, the parts are actually leaking.
The master part rule: Keep a known-good part in the fixture area. Test it at the start of every shift and after every fixture change. If it fails, stop production and troubleshoot the test setup — don’t start rejecting parts until the master passes.
Test Sequence: Charge, Dwell, Measure, Vent
A production leak test follows a sequence. Each step has a purpose.
- Clamp: The fixture seals the part. Sensors confirm all seals are engaged.
- Charge: Pressurized air fills the part to the test pressure. This takes 1–5 seconds depending on part volume.
- Dwell (settling): Wait for pressure and temperature to stabilize. 5–15 seconds. This is where thermal equilibrium happens. Skipping the dwell is the most common cause of false failures.
- Measure: Isolate the part from the supply and measure pressure change over the test time (5–30 seconds). This is the actual leak measurement.
- Vent: Release the pressure. The part vents to atmosphere. Wait until pressure is zero before unclamping.
- Pass/fail: If the decay is below the limit, the part passes. Above the limit, it fails. The result is logged.
Total cycle time is typically 15–40 seconds per part. The dwell and measure times are the long ones. You can’t rush them — rushing causes false failures.
Setting the Leak Limit: Don’t Set It Too Tight
The leak limit is the threshold between pass and fail. Set it too tight and you reject good parts. Set it too loose and bad parts ship.
How to Set the Limit
- Start with the engineering requirement: What leak rate does the application tolerate? (A medical device needs tighter than an air manifold.)
- Test a sample of good parts: Run 20–50 known-good parts. What’s their decay distribution? Set the limit above the 99th percentile of good parts. If good parts read 1–3 Pa, set the limit at 5 Pa — not at 1 Pa.
- Test a known-bad part: Introduce a deliberate leak (a pinhole, a loose seal) and verify the tester catches it. If it doesn’t, the limit is too loose or the instrument isn’t sensitive enough.
- Consider the test variation: The measurement itself varies ±1–2 Pa due to temperature and fixture repeatability. The limit needs to be above that variation band. Don’t set the limit at the edge of the good-parts distribution.
Common Leak Test Failures
| Symptom | Likely Cause | Fix |
|---|---|---|
| All parts fail | Fixture seal leaking, instrument not calibrated | Test master part; inspect seals |
| Random fails | Temperature not equilibrated, part variation | Increase dwell; check part temp |
| First part fails, rest pass | Cold fixture or cold air charge | Warm up; run a purge cycle first |
| Good parts fail on flexible material | Part expansion under pressure | Lower test pressure; use differential test |
| Leaks found visually but tester passes | Limit too loose, instrument too insensitive | Tighten limit; verify with known leak |
A Leak Test Station Checklist
- What is the required leak rate? (Engineering spec.)
- What test method matches the part volume and leak rate? (Pressure decay, mass flow, vacuum?)
- Is the part material rigid or flexible? (Determines test pressure and dwell.)
- How are the part openings sealed? (Custom fixtures?)
- Is the dwell time sufficient for temperature equilibration?
- Is the leak limit set above the variation of good parts?
- Is there a known-good master part for verification?
- Is the fixture rigid enough (no flex under pressure)?
- Is the cycle time acceptable for production?
- Is the result logged for traceability?
The Bottom Line
Industrial leak test automation isn’t just buying an instrument and connecting it to the part. The test measures pressure change, and pressure changes for reasons other than leaks — temperature, part expansion, fixture flex, and seal wear. Design the fixture to seal reliably, add enough dwell for thermal equilibrium, set the limit based on actual good-part variation, and verify with a master part. The tester that accepts 98% of good parts and catches 100% of bad ones isn’t luck — it’s a test that accounts for the physics of the part, not just the instrument’s datasheet.
