The short answer
Specify a leak or pressure test on a CNC housing RFQ when the part is a pressure or fluid boundary, you already have a written pass/fail method, and you want the shop or a named lab to prove the closed envelope before shipment. Keep the test in-house when you will install your own seals and covers, when the real check is a system or IP test on the finished product, or when you can only write “watertight.”
A CMM report of an O-ring groove is not a leak test, and a shop cannot invent a pressure, hold time, or leak rate from the STEP. Write the medium, pressure or vacuum, hold time, acceptance rule, port and plug map, who supplies seals and test covers, and whether the test runs before or after finishing. Put that paragraph on the PDF you upload to quote-new.php so the quote band covers a real test, not a guessed checkbox.
What a leak test is — and what it is not
On a housing RFQ, “leak test” means the shop closes the cavity the way your note describes, applies a gas or liquid, and records pass or fail against a rule you wrote.
| Method | What it usually shows | Typical use on a CNC RFQ |
|---|---|---|
| Pressure or vacuum decay | A pressure change over a hold time | Closed housings with a documented drop or leak-rate limit |
| Bubble immersion | Visible bubbles at a joint or port | Simple air tests when a visual rule is enough |
| Hydrostatic (water or oil) | Weeps at joints under liquid pressure | When you already test that way in the plant |
| Helium sniff or chamber | A smaller leak than a bubble test can show | Specialist work — attach a procedure; not a default CNC-cell bench |
| IP-code chamber | Ingress under IEC 60529 | Usually a finished-product test, not a bare open CNC shell |
A leak test does not replace groove design (see O-ring groove design and O-ring groove callouts on CNC housings) or first article inspection and CMM vs optical measurement. Those answer “does the geometry match the drawing?” The leak test answers “does this closed envelope hold under the method you named?” An IP code applies to a finished enclosure in the condition the standard describes; open connector bores or a temporary shop lid are not that product. CNC parts from typical wrought plate or billet rarely leak through the metal — leaks almost always start at a joint, a port thread, a scratched seal land, a missed plug, or a vent you forgot to close. If a casting or metal 3D-printed shell may be porous, say so (see SLM vs CNC hybrid part strategies).
When a shop test is worth specifying — and when it is not
Write a shop test when several of these are true at once: the housing is the pressure or fluid boundary and ports will be closed in service; you are buying an assembled unit rather than an open kit; a late leak would scrap a finished, marked housing; you already run a plant procedure you can attach or copy onto the PDF; quantity is a production lot and you want the sampling rule in writing — first article, every piece, or a lot sample (see CMM sampling plans for CNC lots); or the envelope is a split housing whose joint you need to prove (see when should I split a CNC part into two pieces?). Assembly service is a different package from “machine the halves and ship them loose.” Prototype vs production CNC quotes still applies: a development housing whose seal stack is still changing is often better tested on your bench.
Keep the test at your site when you will install production seals and covers yourself; when acceptance is a system check that a CNC cell cannot reproduce from a housing alone; when the only note you have is “watertight” or “IP67” with no medium, pressure, or pass/fail; or when you use a unique fixture or cover you do not want to consign. Still specify the seal-land geometry, finish, and mask notes so the housing can pass *your* test. The RFQ can say “no shop leak test; buyer tests after assembly.” Silence leaves the estimator guessing whether a bench is in scope.
Choosing a method you can actually quote
Pressure or vacuum decay
Name the medium (typically dry air or nitrogen), the test pressure or vacuum, the hold time after stabilization, and the acceptance rule — a maximum pressure change, or a leak rate if you already use one. Write the same values you use in the plant, or the proof step your own standard requires, in the same units as the service rating. Do not leave the shop to pick a “typical” pressure.
Bubble, hydrostatic, helium, or IP
For a bubble test, say that a pass means no visible bubbles at named joints and ports during the hold. If the plant test is water or oil, write the liquid, pressure, hold, and whether a dry exterior is the rule; wet tests add draining and drying, so say whether the test is before or after pack-out or marking. Treat helium and IP-code tests as specialist routes: attach the procedure and say whether the shop may subcontract a named lab.
What must be on the drawing and the RFQ
A useful leak-test note is a short procedure, not a slogan:
- Medium, and whether it is gas or liquid
- Pressure or vacuum, units, and whether that value is the service condition or a documented proof step above it
- Stabilization and hold time
- Acceptance: no visible bubbles, a maximum pressure change, or a leak-rate limit you already use — if you do not have a number, do not ask the shop to invent one
- Port map: which openings are plugged, which is the fill or gauge port, and which face needs a test cover. Vent, drain and weep holes for sealed CNC housings must be on that map or they will vent the test
- Hardware: whether the shop supplies plugs, O-rings, gaskets, and a test cover, or you consign the production parts. Name the seal spec if the shop must install it
- Sequence: before or after anodize, plate, paint, or laser mark (see as-machined before Type II anodize, paint mask and plug notes, and laser marking before vs after anodize). A rack mark or an unmasked land can fail a test the raw machining would have passed
- Sampling, report, and fail path: first article, 100%, or a lot sample; a pass/fail log or pressure trace; and whether a fail is re-sealed, reworked, or isolated
What files to send for a CNC quote and the DFM checklist before an online CNC quote both start from a PDF that matches the STEP. If the test uses a cover that is not in the model, add a sketch or a second body so the shop is not inventing a lid.
Cost, lead time, and a housing that can actually pass
No two benches move the quote the same way, and a shop cannot price a test it cannot see. In typical practice, a written leak test adds a fixture or test cover and a plug-and-seal set unless you consign them; bench time to fill, stabilize, hold, vent, and record; and, for wet or specialist methods, dry time or a subcontracted ship loop. Testing after finish protects the land you ship, but a fail then costs a finished piece; testing before finish is faster to retry and weaker as a shipping proof. What drives CNC machining cost and what affects CNC machining lead time are the general maps; the leak test is a line item, not a free inspection tick.
A leak test only checks the envelope you close. Call out groove size and which face carries the seal; seal-land roughness and edge breaks that will not nick the O-ring (see surface roughness Ra for machined parts and edge-break notes); and mask notes so finish does not fill a groove or scar a land. Plating thickness can also eat a close tolerance on a seal diameter (see does plating thickness eat CNC tolerances?). An unplugged weep hole will fail any decay test. On split housings, design locators, fasteners, and a seal that is actually in the joint. For printed or cast shells, say if you need a sealing skim, impregnation, or a test because the wall itself may leak (see escape-hole notes for SLM hollow parts).
RFQ checklist for a leak test
Before you submit on quote-new.php, confirm:
- Shop test, in-house test after delivery, or no leak test on this revision — written, not implied
- Method: decay, bubble, hydrostatic, helium, or an attached procedure
- Medium, pressure or vacuum with units, stabilization, and hold time
- An acceptance rule you already use, not “as tight as possible”
- Port and vent map, including the fill port and any test cover
- Who supplies plugs, seals, grease, and covers
- Test sequence versus anodize, plate, paint, and marking
- Sampling plan, report contents, and the fail/retest path
- STEP and PDF that match, with the leak-test paragraph visible on the drawing
Common buyer mistakes
- Writing “IP67” or “watertight” with no method, medium, or pass/fail
- Asking the shop to test an open housing with no test cover or plug map
- Forgetting vent, drain, or weep holes, so the cavity cannot hold pressure
- Treating a CMM or first-article packet as proof that the housing will seal
- Testing only a raw machining, then racking, anodizing, or tumbling a seal land
- Leaving a blanket edge-break note that nicks the groove corner
- Assuming every CNC shop has a helium booth or an IP chamber on site
- Not saying who supplies the production O-rings
- Mixing a product test with a housing test in one vague line
- Reusing an old test note after the cover or groove changed on a new revision
Bottom line
Specify a shop leak test when the CNC housing is a closed pressure or fluid boundary, you can write a real method, and you want that check in the shipping packet. Keep the test in-house when you assemble and qualify the product yourself, or when you cannot yet name a medium, pressure, and acceptance rule. Design the grooves, lands, vents, and masks so either bench has a chance, and put the procedure — or an explicit “buyer tests after assembly” — on the PDF. Upload both files on quote-new.php so the quote covers the test you intend to buy.