Industry Solutions
Leak testing for direct-to-chip liquid-cooling hardware in AI data-center production.
Cold plates, rack manifolds, quick-disconnects, hoses and coolant distribution units bring liquid close to high-value computing hardware. Production teams use a controlled, dry leak test to verify the sealed boundary before assembly.
Air leak testing and trace-gas leak testing are peer options. Pressure decay, vacuum decay and mass flow suit gross-leak and production screening; helium or hydrogen trace-gas methods provide a measurable tracer when the required target or localization task calls for it.
Request for QuoteRequest a DemoDirect-to-chip liquid cooling puts a coolant circuit next to GPUs, CPUs and memory. Every joint in that circuit — brazed cold-plate passages, manifold branches, quick-disconnect seal faces, hose crimps and CDU connections — is a potential leak path, and a slow leak in a running rack is an expensive field failure.
Production teams therefore test the component where it is made and again when it is integrated into a rack or loop. Each part needs a method, fixture and acceptance target appropriate to its volume, access and purpose rather than one generic recipe for the whole assembly.
Methods used
Choose a method around the connected volume, temperature stability, required decision and whether the station must localize a leak.
Pressure or vacuum decay measures a controlled change after a sealed part is isolated. Mass flow measures gas through a connected path. These air methods are dry, repeatable production tools when the fixture, settle time and acceptance target are defined.
Compare air leak-test methods →Helium or hydrogen forming gas gives the detector an identifiable tracer. Sniffing can localize accessible seams, accumulation measures tracer collected around a part, and hard vacuum addresses the smallest leak paths.
Explore trace-gas methods →Method selection and acceptance criteria are part-specific. The table identifies common starting points; validate the fixture, sequence and target for the production part.
| Component | Recommended method | Why | Zaxis product |
|---|---|---|---|
| Cold plate or cold-plate loop | Pressure or vacuum decay screen; helium or hydrogen sniff or accumulation for fine leaks | Brazed passages and port seals need a gross screen first; a tracer method resolves the fine-leak target without the volume and temperature limits of a decay test. | Z5 or Z8 air testers; Z-Charge + Z-Ion for trace gas |
| Rack manifold or CDU | Pressure decay or mass flow; helium or hydrogen sniff to localize | Many ports and a large connected volume call for a sealed-port fixture, stabilization time and a sniff probe to find which joint failed. | Z8 multi-port air tester; Z-Charge + Z-Ion |
| Quick-disconnect or immersion-tank fitting | Pressure or vacuum decay, coupled and uncoupled; helium or hydrogen sniff for seal faces | Small volume gives a fast decay test; the seal face is the failure point and sniffing locates it without masking access. | Z1 or Z5 air testers; Z-Charge + Z-Ion |
| Hose, pump or brazed heat exchanger | Pressure or vacuum decay; helium or hydrogen accumulation or hard vacuum for the tightest targets | Flexible or multi-circuit parts add compliance and volume; accumulation and hard vacuum measure total leakage when a decay test cannot reach the target. | Z5 or Z8 air testers; Z-Charge + Z-Ion |
Zaxis air and trace-gas systems are selected with the part, fixture and required method.
Z8Modular air tester with up to four test ports for sequential or concurrent testing.
Z5Compact multi-channel air tester for configured production stations.
Z1Single-channel air tester for machine integration and cascading test cells.
Z-Charge + Z-IonCoordinated part preparation and helium leak detection in a trace-gas test cell.ASHRAE Technical Committee 9.9 provides thermal guidance for data-processing environments, including liquid-cooling context. The manufacturer defines the acceptance criteria and validation plan for its product and process.
A component maker can screen a cold plate, hose or manifold before shipment, then a rack integrator can test incoming hardware or a completed fluid path. This two-stage approach keeps a part-level result separate from a later assembly verification.
Fixture design controls where the test volume begins and ends. Seal unused ports, minimize unnecessary adapters, allow stabilization, and preserve access when a trace-gas sniff is required.
Connected cooling paths can be tested with an air or trace-gas sequence selected around volume, seals and the required production decision.
A component-level test can identify a defect before shipment; an incoming or assembled-rack test evaluates the later connections and assembly boundary.
Pressure decay, vacuum decay and mass flow are air methods; helium or hydrogen trace-gas testing can use sniffing, accumulation or hard vacuum.
A sealed cold plate can use pressure decay when the fixture, volume, stabilization and target are defined.
Trace gas provides a measurable tracer when the target calls for fine-leak detection or a station needs to localize an accessible leak.
Connected volume, part and gas temperature, stabilization time, fixture volume and the selected method all affect a production sequence.
Standards and guidance identify scope; the manufacturer establishes the acceptance target and validation plan.
