Leak testing machines that give every part a verdict — and a reason
How a pressure-decay leak tester decides
The part is sealed and filled with air to the test pressure. The air is then left to settle, because filling warms it and a part that is still cooling loses pressure without leaking. Then the fill valve closes and the machine watches the pressure for a fixed test time. A sound part holds; a leaking part keeps falling. The drop over the test time, compared with the limit for that part, is the verdict.
A drop on its own is not enough to tell an operator what to do, so every NOT OK carries a reason:
- GROSS LEAK — the part never reached fill pressure: a missing seal, a cracked part, or a part not seated in the fixture.
- LOW FILL / HIGH FILL — pressure outside its window at the end of fill: a supply problem or a blocked port.
- LEAK — filled correctly, then lost more than the limit during the test.
- SENSOR FAULT — the reading itself cannot be trusted, so the part is not passed on a bad signal.
The full method — including how to work out the pressure drop your allowed leak rate produces, and why part temperature matters more than most buyers expect — is in our guide to pressure decay leak testing.
What goes into one of our leak testers
| Stations | 1 to 6, each with its own clamp, fill valve and pressure reading. Any station can be switched off from the HMI. |
|---|---|
| Test method | Pressure decay on a transmitter per station, or a commercial leak test instrument (Ateq, Cosmo or your preferred make) driven by the PLC. |
| Visual check | Optional water dip: the pressurised part goes into a tank so a leak shows as a bubble trail. |
| Start | Two-hand start, both buttons pressed together; guarding to suit the risk assessment. |
| Result | OK / NOT OK per station with a reason, station and total counts, optional marking punch on the part. |
| Pressure signal | 4–20 mA or 0–10 V transmitters into the PLC, or digital indicators over RS-485 Modbus RTU. |
| Control | Your plant's PLC and HMI standard. Our reference build: a Weintek cMT HMI with built-in CODESYS PLC and Modbus TCP remote I/O. |
| Settings | Fill, stabilise and test times, pressure windows and leak limit, per part type, with limits and defaults; kept through power off. |
| Safe stop | Emergency stop runs a defined sequence — exhaust, lift out of the tank, release pressure, unclamp — not just “power off”. |
| Data (option) | Result, value, station and cycle time per part, logged to a database against time or part serial. |
Screens an operator can run without calling anyone
Every page carries the same header: the machine state, one message that says what to do next (“Load parts, then press BOTH start buttons together”, “Cycle complete — NOT OK part(s)! Segregate them”), the first active alarm, and RESET and STOP. Behind it:
- AUTO — the station table: pressure, leak value, result, reason and counts.
- MANUAL — clamp, fill, exhaust and dip by hand, refused when the move would be unsafe.
- SETTINGS — times, pressure windows and leak limit, with allowed ranges.
- MAINTENANCE — which stations are fitted and in use, counters, lamp test.
- CALIBRATION — zero and span per station against a reference gauge.
- I/O — live inputs and outputs, with output force for wiring checks.
- ALARMS and DIAGNOSTICS — what tripped, when, and the likely cause.
Built to be fault-found by your own team
Every wait in the sequence has a timeout and a named alarm, so the machine never just stops with no explanation. The machine ships with its I/O list, alarm list, parameter sheet and HMI page guide, and the PLC and HMI source.
Choosing the test method
| Method | What you get | Good for |
|---|---|---|
| Pressure decay | A measured drop and an automatic OK / NOT OK | Production testing of most pneumatic and hydraulic parts |
| Water dip (bubble) | Where it leaks, seen by the operator | Finding the leak path; rework; a visual check alongside decay |
| Commercial instrument | The instrument maker's measurement and compensation | Tight limits, or where the customer specifies the instrument |
| Tracer gas (helium) | Very small leak rates | Refrigeration and sealed parts — a specialist system; we integrate it rather than build it |
We will tell you which method fits, the cycle time to expect and a budget range — free, as part of the concept.
Leak tester questions
What leak rate can the machine detect?
It depends on the part: its internal volume, the test pressure, how long you can afford to test, and how stable the part's temperature is. A small volume and a longer test make a small leak easier to see. We work out the pressure drop your allowed leak will produce on your part during the concept, and choose the transmitter or instrument to suit — before anything is built.
Pressure decay or water dip — which do we need?
Pressure decay gives a number and a verdict with no operator judgement, which is what a production line needs. A water dip shows where the leak is, which is what rework needs. Many of our machines do both: the verdict comes from pressure decay, and the dip lets the operator see the bubble trail on a rejected part.
Can the machine test several parts at once?
Yes. Each station has its own clamp, fill valve and pressure reading and gets its own OK or NOT OK, so one leaking part does not reject the other five. A station that is out for maintenance can be switched off from the HMI, and the machine runs on the rest.
We already own an Ateq or Cosmo leak tester. Can you build around it?
Yes. The instrument does the measurement and the machine does everything around it — clamping, sealing, start, reading the result, marking and sorting. The PLC starts the instrument's program for the part and reads back the result and the measured value.
Can the results be stored for traceability?
Yes. Each cycle's result, measured value, station and cycle time can be logged against a date and time, or against a part serial or barcode, to a database your quality team can query. Counts per station are kept on the machine and survive a power cut.
How is a leak tester calibrated?
Each station's pressure reading is set against a reference gauge with a two-point zero and span adjustment from the CALIBRATION page, and the settings are protected. To prove the machine catches the limit, you test a master part with a calibrated reference leak fitted: it must fail every time, and a good master must pass every time.
Have a job that a machine should be doing?
Send the part and the target output. You get a concept, a sequence and a budget range back — free, and with no obligation to build it with us.