Industrial leak test machine selection is often reduced to the instrument’s published resolution. For a sealed component on a production line, the result depends on the whole sequence: loading, locating, sealing, charging, stabilizing, detecting, venting, unloading and recording the decision. This guide helps buyers of end-of-line airtightness test cells in Thailand specify the method, fixture, RFP and FAT/SAT evidence. It addresses leaks from a manufactured component. Ambient ammonia gas monitoring and surveys for leaks in a factory compressed-air distribution system require different equipment and decisions.
The decision in brief: buy a validated workpiece-and-cell process
Start with the product’s permitted leakage under defined conditions, owned by product engineering and quality. A tester with an impressive minimum indication cannot compensate for a warm part, an unstable fixture seal or a poor reject mechanism. The acceptance specification must cover the measuring instrument, master reference, pneumatic circuit, fixture, handling, exhaust, safety circuit, PLC and inspection history.
Document seven inputs before requesting quotations: (1) drawings and variants, material, internal volume, ports and sealing surfaces; (2) the functional or customer reason for the permitted leak; (3) medium, pressure and temperature conditions; (4) total cell cycle and upstream/downstream limits; (5) actual good, borderline and defective samples; (6) signals and traceability that distinguish fail from equipment fault; and (7) the conditions to be repeated in FAT and SAT. There is no universal acceptable leak number. It depends on the application, fluid, pressure, life, customer specification and applicable requirements.
COSMO Instruments’ method selection guide organizes choices by test pressure and workpiece shape. It is useful supplier guidance, not proof that one model meets your product’s requirement. Verify the proposed process with your own parts.
Define the workpiece: fully sealed product or accessible port?
A “sealed component” can mean a finished item with no port through which to pressurize it, or a housing that can be connected during inspection. An internal-pressure proposal for the first type may not reproduce the finished state. COSMO lists sealed-chamber pressure-change testing for closed items without an air inlet, such as watches and electronic parts. General workpieces with accessible openings may be candidates for internal differential-pressure testing.
Show the state to be tested with a photograph and section drawing in the RFP. If an O-ring or weld is still open upstream, distinguish an in-process test from a test after final closure. Draw the possible leak path: to ambient air, or between two chambers inside the same component. Do not assume an external chamber can identify an internal cross-leak without a specific demonstration.
Record the incoming part temperature and the presence of wash liquid, oil, dust or residual moisture. Specify orientation and transfer method. A recipe proven with a clean, room-temperature laboratory sample may fail on a hot production part. Burrs and surface damage can wear the fixture gasket and introduce variation before the sensor itself is challenged. Maintenance and gasket replacement are part of the process conditions.
Comparing air leak testers under the same conditions
Compare access to the part, detection requirement, total cycle, test medium and maintenance needs. A method name alone does not decide suitability.
| Candidate method | Suitable starting condition | Main comparison points |
|---|---|---|
| Direct pressure decay | The internal volume is accessible and its pressure change can be observed | Temperature, deformation, line volume, adequate stabilization |
| Differential pressure leak test | The workpiece can be compared with a non-leaking master | Master thermal behavior, fixture symmetry, mastering and calibration |
| Sealed-chamber method | The finished sealed item must be tested from outside | Chamber gap volume, gross-leak detection, fixture cost |
| Tracer-gas method | Air-pressure methods do not meet the defined requirement, among other cases | Gas supply/recovery, detector, background concentration, work procedure and safety |
COSMO’s selection guide notes that pressure decay is affected by temperature, deformation and other pressure changes; reducing internal volume may help a differential-pressure method; and sealed-chamber arrangements need a mechanism for gross leaks. None of this makes differential pressure universally superior. Select using the actual shape, fixture, pressure and thermal conditions.
The published scope of ISO 20485:2017 concerns leak detection using a tracer gas and a detector specific to that gas. It is not a blanket qualification for air-based differential-pressure or pressure-decay testing. If a contract cites ISO 20485, confirm which tracer-gas activities it requires and which edition applies. Do not copy a standard number into an RFP without checking its scope.
Plan for gross leaks as well as small leaks
A detector optimized for a small leak does not necessarily recognize a large opening by the same mechanism. In a sealed chamber, a large opening can allow external pressure to reach the inside of a workpiece rapidly, creating a blind spot in a differential signal. COSMO describes a separate gross-leak detection mechanism for this configuration. Test a missing cap or torn seal separately from a calibrated small leak. If one threshold cannot cover both, specify a two-stage judgment and the intended reject response.
What the master does in a differential-pressure test
Differential pressure measures the pressure difference between workpiece and master. The master cannot simply be treated as “a good part put on a shelf.” COSMO’s March 2026 technical explanation describes how compression initially heats the air, then heat exchange changes pressure during stabilization. This can disturb small-leak decisions. It describes two approaches: make the master thermally similar to the workpiece to offset the change, or use a stable master chamber and prioritize repeatability.
Request the master’s material, volume, location, temperature conditions, leak verification, storage and replacement plan. If a production part is used as master, verify that it stays leak-free over time and that repeated charging and exhausting do not change its thermal response. If a dedicated chamber is chosen, ask how any workpiece-to-master thermal difference is compensated and who controls correction changes. Both approaches need evidence on real workpieces.
Do not confuse the displayed leak unit with a property of the tester alone. COSMO’s leak-calibration guidance discusses K(Ve) calibration using a leak master. The equivalent volume and pressure response of the part, tubing and fixture belong to the measurement system. Define when a fixture replacement, hose extension or product revision requires renewed verification.
Total cycle time is more than detection time
Measure loading, clamping, sealing, charging, equalizing, detection, venting, unclamping and unloading. A supplier’s “measurement time” may cover only the detection interval. COSMO’s timer guidance says that charging and stabilization depend on the actual part, fixture and machine. It advises beginning with enough time to obtain stable baseline data, then shortening intervals experimentally. Its illustrative seconds are not a guaranteed cycle time for your line.
A sound sequence is: establish a baseline with a non-leaking part and fixture; characterize good and known-defect distributions; shorten charging and balancing separately; repeat across part temperatures, batches, operators and shifts; and set the cycle with appropriate margin. Compare a parallel-station concept with an accelerated single station on full cost, maintenance and changeover as well as output.
COSMO’s July 2026 high-speed-charge article explains that preliminary pressurization may help when a narrow inlet or internal passage restricts filling. A large volume alone does not guarantee an improvement. Compression-induced temperature change can still require stabilization before small-leak measurement. Verify the pressure resistance of the actual workpiece and fixture. Treat high-speed charge as an option to test, not a universal mandatory feature or cycle-time promise.
Set pressure and permissible leak from the product requirement
“Higher pressure always means better sensitivity” is an unsafe shortcut. Excess pressure may deform the seal and open a path unlike normal use; a low pressure that does not represent service may miss defects. Define what the test is intended to simulate with product engineering, quality and the customer. Separate proof-pressure testing from leak testing. When changing the gas medium, do not assume that the same numeric leak limit has the same physical meaning: state the conditions, units and any conversion assumptions.
Fixture, air circuit and environment determine usable performance
Important causes of failure are often missing from the tester’s specification sheet: gasket location, hose length, fittings, valves, regulator, air quality, part temperature and vibration. A sound part can show a different result if clamping force changes the seal contact. When the fixture wears, the tester may be unable to distinguish product leakage from fixture leakage. This is why factory results can differ from laboratory demonstrations.
Request fixture drawings, consumables, replacement procedure, positioning tolerances, gasket-life evaluation method, fixture-leak checks, pneumatic diagram, exhaust and purge design. Verify the spare-parts source, Thai service contact and plan while an instrument is being calibrated. Do not invent a gasket lifetime or maintenance interval without supplier and part evidence; make them items for quotation and acceptance.
Watch for baseline drift and distinguish a product change from system change. A master or blank fixture can support a start-of-shift check. Define when abnormal baseline or supply pressure stops production. Record air moisture, temperature, regulator output and ambient conditions in FAT and SAT. These are engineering checks, not claims that every product contains a particular diagnostic function.
A leak-test RFP that suppliers can answer consistently
Ask for evidence tied to each claimed value, not merely a yes/no answer.
| Item | What to specify | What the supplier should show |
|---|---|---|
| Workpiece | Drawing, variants, state, material, ports, seal and tolerances | Match between representative samples and fixture concept |
| Quality | Origin of permitted leak, units, test conditions and decision classes | Borderline-sample results and decision rule |
| Method | Allowed approach and alternatives | Reason for selection and failure modes it cannot detect |
| Cycle | Whole-cell cycle, production conditions and changeover | Time by stage and distribution in production simulation |
| Fixture and air | Seal, pressure, supply quality, tubing and venting | Drawings, leak checks and maintenance plan |
| Data | Serial, part number, recipe, value, decision and time | Sample output, protocol and behavior during data loss |
| Safety | Guard, interlock, residual pressure and restart | Circuit diagram, abnormal-state test and risk assessment |
| Acceptance | FAT/SAT, training, calibration, spares and Thai support | Test protocol, responsibility split and service terms |
Attach sample type, repeats, environment and pass criterion to every numeric claim. A detection limit, repeatability, false-decision rate or cycle shown only on a single ideal sample is not a comparable proposal. If real borderline defects are unavailable, ask how a calibrated leak or test coupon will be used and how it relates to a defective production part.
Treat blanks in supplier comparisons as unresolved work
A price-only offer and a fully scoped cell offer cannot be compared as though they have identical coverage. Mark every requirement “answered,” “conditional” or “unanswered,” and record the condition. Include tester, fixture, master, valves, handling, reject mechanism, PLC interface, data retention, calibration, training, local service and spares in the cost comparison. Ask what a future part or recipe variant will cost to add.
Give shortlisted suppliers the same sample set and desired end result. If suppliers choose different pressure or stabilization settings, request the engineering explanation. Different methods can be compared by whether they meet the same quality decision and total cell cycle, not by matching the same displayed leak value without common conditions.
Airtightness FAT and SAT: what each must prove
FAT checks the proposed configuration, wiring, operation, decisions and data at the supplier’s site. SAT repeats the relevant proof after installation in Thailand with actual air supply, power, handling, temperature, PLC/MES interface and operators. A pass at FAT does not authorize production if the decision cannot be reproduced at SAT.
The FAT protocol should include good, borderline and known-defect parts; a gross leak; fixture leak; unconnected port; wrong variant; low pressure; power loss/restart; emergency stop; communication loss; and failed data storage. For each case specify the steps, expected signal and display, whether data is saved, reject path and recovery. Running only good parts cannot validate reject handling or fault stops.
At SAT, use samples representing production temperature and shifts. A different compressed-air source from the FAT site can change setup. Record every parameter changed since FAT and obtain quality approval. Acceptance is valid for the documented combination of part, recipe, fixture, software and environment, with change control afterward.
Decisions near the limit and measurement uncertainty
A displayed decimal place is not certainty. COSMO’s February 2026 article on measurement uncertainty discusses calibration uncertainty and acceptance decisions. Quality should define calibration evidence, calibrated leak reference, repeatability, day-to-day variation and the decision rule near the limit. Where the product risk or customer specification calls for a guard band, agree its width and treatment before buying. This guide does not prescribe one statistical threshold for every product.
Data and traceability: can the decision be explained later?
An OK/NG contact by itself may not identify which product, recipe and fixture produced a result. Consider recording serial or lot, part number, time, equipment ID, recipe version, measured value, judgment, pressure, major timers, master-check state and fault code. The retained fields and period should follow customer contracts, the quality system and local data policy.
Define what happens during MES disconnection: stop the cell, store records locally for later transfer, or isolate parts under a controlled manual procedure. Keep the original failed result when the same serial is retested, along with reason and authorizer. Align the physical release of a part with finalization of its digital record.
Our broader inspection equipment selection guide covers visual inspection and other equipment decisions. This article focuses on an end-of-line airtightness cell for sealed parts. For wider site-startup planning, see inspection automation FAT/SAT in Thailand.
A staged purchasing and deployment decision
First, establish the quality requirement. Product engineering and the customer-specification owner define what leakage fails and under what conditions. “Perfectly airtight” without units and test conditions does not let a supplier design a method. If the requirement is uncertain, a sample evaluation or test-development phase may be more useful than an immediate purchase order for a production cell.
Second, evaluate methods and prototype the fixture. Secure good, defective and borderline parts, including thermal and batch variation. Compare methods and measure the complete cycle. Do not treat a single successful run after extensive tuning as proof. Look for a reproducible recipe and design margin.
Third, fix the contract boundary. Include roles, drawings, software, recipe ownership, data format, acceptance procedure, spares, Thai service and change control. Distinguish the responsibilities of instrument maker, fixture maker and system integrator, including first response when a result cannot be trusted.
Fourth, perform FAT and SAT. Acceptance means the specified sample and fault-injection cases produce the required part decision, record, routing, stop and recovery—not merely that the machine starts or agrees with a brochure. Transfer start-of-shift checks, periodic calibration, master management, fixture replacement and recipe approval into standard work.
After acceptance: manage recipe, calibration and fixture together
Keep a configuration record capable of reproducing the accepted condition. A tester recipe name is insufficient if gasket material or tubing length changes. Link part number, fixture drawing revision, master ID, pneumatic drawing, tester settings, PLC software, limit and approver to each change. If stabilization time changes on site, record the difference from FAT/SAT evidence.
A start-of-shift check should do more than produce one OK from one good part. A calibrated leak or known-defect coupon checks the NG path; fixture-leak and low-supply-pressure tests show how system faults appear. Define which previously released parts are held if a check fails. Set frequency and containment window from product risk and customer requirements, not a generic count.
Calibration is broader than the sensor alone. Periodically check the response of instrument, leak reference, master, air circuit and fixture as a system. After replacing a master, a zero adjustment alone does not prove the borderline decision is unchanged. After replacing a fixture, check sealing force, position, connections, clamp movement and safety circuit. Agree before purchase what changes require partial or full re-acceptance.
If the NG rate shifts, do not immediately conclude that the product has changed. Compare material and machining lots, shifts, temperature, supply air, fixture use and master checks. Report which condition changed and when. Relaxing a limit to reduce NGs without identifying the cause can hide both product and equipment problems.
Materials that speed up the first technical discussion
You do not need to disclose every confidential product detail at the first meeting. However, missing information needed to choose a test method tends to make quotations conditional. Prepare a drawing showing the external shape and the boundary that must remain sealed, positions of openings and seals, a photograph of the final assembled state, the current acceptance method, required total cycle and layout of adjacent processes. If there are several part numbers, tabulate what is common and what differs. In particular, distinguish variants that retain a test port from those that become completely sealed before inspection.
Even if the permissible leak has not yet been set, the retained fluid, service pressure and temperature, expected service life, consequence of leakage and customer-specified test conditions let you start the discussion with quality. Do not adopt a supplier’s smallest measurable value as the product specification before the design requirement is defined. Conversely, a clear limit without borderline samples cannot establish the decision margin in production. If no defect samples are available, include the design of test coupons and their comparison with real defects in the evaluation plan.
Ask candidate suppliers not only why their method is suitable but which defects it might miss. Examples include a defect that equalizes the product’s inside and outside pressure before detection, a leak hidden at the fixture contact surface, production temperature variation and isolated internal cavities. The ability to explain these limitations at the proposal stage is often a more useful purchasing comparison than a standalone instrument specification.
Frequently asked questions
What number should be set first when selecting an industrial leak test machine?
Define the permissible product leakage and the conditions under which it applies: pressure, medium, temperature, finished state and units. The tester’s smallest display digit and nominal cycle time are not substitutes for a product requirement.
Is differential-pressure testing always the best air leak tester choice?
No. Its comparison with a master can help, but results depend on thermal behavior, sealing, fixture and available ports. An externally tested sealed item may suit a chamber method; a different requirement may call for tracer gas. Compare methods on actual good and defective parts.
Does high-speed charging always shorten the cycle?
No. It may help when a narrow inlet or internal passage restricts filling, but large volume alone is insufficient. Longer thermal stabilization may offset faster charging. Test preliminary pressure only within the verified limits of the workpiece and fixture.
Does ISO 20485 qualify an air-pressure airtightness test?
Its published scope is tracer-gas testing. The standard number alone does not certify an air differential-pressure or pressure-decay process. Confirm the contractual method, edition and scope with quality.
Which samples are needed for airtightness FAT and SAT?
Include good, borderline, known small-leak and gross-leak conditions, plus a way to simulate fixture leakage. Where actual defect samples are unavailable, use a calibrated leak or coupon and document how its result relates to a real part. Set sample counts and acceptance criteria from the quality requirement.
Who controls settings in a Thai plant?
A possible split is quality owning limits and change approval, maintenance owning fixtures, pneumatics and calibration, and production owning shift checks and containment. Document the actual roles under the plant’s organization and customer-audit needs. If only the vendor can edit recipes, examine change history and emergency response.
Conclusion
Select an industrial leak test machine as part of a complete process. Define the product leak requirement and test state, then compare internal-pressure, differential-pressure, sealed-chamber and tracer-gas options with real parts. Put master and thermal behavior, seals, charge and stabilization, gross leaks, data interface and fault response into the RFP. FAT and SAT should prove the same acceptance criteria under documented conditions. Evidence from the installed Thai production cell ultimately decides whether the process can be used for release.
TOMAS TECH can help define the method, draft the RFP, divide responsibility across tester, fixture and data integration, and plan FAT/SAT in a Thai factory. You can contact us while the permitted leak or test method is still being evaluated; a workpiece drawing and process description are enough to begin the discussion.
Sources
- COSMO Instruments: methods and selection — choice by part and pressure.
- COSMO Instruments: role of the master, March 2026 — differential pressure and thermal behavior.
- COSMO Instruments: high-speed charge, July 2026 — conditions, benefits and limits.
- COSMO Instruments: timer settings — charging, stabilization and detection.
- COSMO Instruments: leak calibration — leak-master calibration.
- COSMO Instruments: measurement uncertainty, February 2026 — calibration and limit decisions.
- ISO: ISO 20485:2017 published scope — tracer-gas method.
