Purchasing and specification guide
Custom Thin-Wall Plastic Housings: How to Specify One You Can Actually Receive
A housing fails in the field for reasons that are rarely on the drawing. The wall is thin enough but the seam does not close, the visible face sinks opposite a screw boss, the first batch fits and the third does not, or the finish that looked acceptable on a sample plaque is rejected on a production face. This page is about the specification decisions that prevent those outcomes — the wall map, the cosmetic grade, the stack-up and the acceptance schedule — because those are the parts a buyer actually controls.
Scope of this page, and how it differs from the two beside it
We publish three separate pages on this subject, and they are deliberately not the same page. Thin-wall injection molding explains the process itself — flow length, freezing, pressure, cooling, gate and vent design, defects and process windows. Electronics enclosure injection molding covers PCB-centric enclosure design, including bosses, snap-fits, EMI shielding, ingress protection and thermal paths.
This page covers the part in between: how a thin-wall housing is specified and accepted as a purchased part. What goes on the drawing, how the wall is called out zone by zone, how a cosmetic surface is defined in words a supplier can be held to, how the seam between two halves is made to close, and what validation you should require before a production run is released.
If you are still deciding whether your geometry can be moulded thin at all, the process page answers that. If your housing is an electronics enclosure, the enclosure page covers the board-driven constraints. If you already know the geometry works and you are writing the purchase specification, this is the page.
At a Glance
A Housing Is a Stack of Requirements, Not a Thin Part
Short answer: a thin-wall part only has to fill and eject. A housing has to fill, eject, hold its shape, close against its other half, survive handling and use, and look acceptable on the faces someone will judge. Those requirements pull against each other, and the specification is where you decide which one wins.
Most housing programmes that go wrong do not go wrong in the mould. They go wrong because two requirements were never written down and therefore never traded off: how flat the flange has to be for the seam to close, and how much cosmetic freedom the moulder has on the faces nobody sees. Both are cheap to decide at drawing stage and expensive to discover at first article.
| Requirement | What it pushes the design toward | What it costs if ignored |
|---|---|---|
| Fill the wall | Thinner, more uniform wall; higher flow material; gate placed to shorten the longest path | Short shots, or a process window so narrow that yield falls in production |
| Structural stiffness | Ribs and a deeper section rather than a thicker wall | A housing that passes drop but flexes enough to crack seals or pop clips |
| Assemble and stay closed | Snap-fits or screw bosses, with flange flatness controlled | Seams that gap, clips that release, or a rework loop on the assembly line |
| Cosmetic acceptance | Graded surfaces, gate and ejector placement away from grade A | Batches rejected on appearance after the tool is already cut |
| Sealing or ingress resistance | A gasket groove or sealing land held flat, with a controlled mating surface | A leak test that fails intermittently and cannot be traced to one dimension |
| Secondary operations | Features designed for painting, printing, plating or shielding | Masking labour, adhesion failures, or a coating that will not pass adhesion test |
| Volume and life | Steel grade, cavitation, cooling and ejection robustness | A tool that makes good parts at T1 and wears into flash by the third production run |
The order of that table is not arbitrary: it is roughly the order in which the decisions become expensive to change. Fill and stiffness are settled in DFM; cosmetic grading and stack-up are settled on the drawing; everything else is settled in the tool.
The Zone Wall Map: Specifying Thickness Where It Actually Varies
Short answer: a housing does not have a wall thickness. It has five or six different ones — the nominal wall, the rib wall, the boss wall, the snap hook, the flange land and any local pad — and a drawing that says “wall 1.X mm” specifies none of them.
This is the single most useful document you can put in a housing RFQ. It is also the one buyers most often omit, because the CAD model already contains the thicknesses and it seems redundant to restate them. It is not redundant. The model shows what the part is; the wall map states which thicknesses are functional requirements and which are consequences of the model — and only the first category constrains the moulder.
| On the drawing | Why it earns its place |
|---|---|
| Nominal wall, as a tolerance range rather than a single value | Lets the toolmaker and moulder optimise within a band instead of chasing one number |
| Rib wall and root radius, marked as requirement | Controls sink on the opposite visible face; without it, rib thickness drifts to whatever is easy to cut |
| Boss wall, plus the datum it is measured from | Bosses carry screws and inserts; a diameter tolerance without a datum is not inspectable |
| Minimum wall at the worst-flow location, with the flow length stated | Converts “thin” into a fill problem the moulder can actually evaluate |
| Flange flatness and land width | The seam is decided here, not at the assembly station |
| Reference dimensions for everything that is a consequence of the model | Keeps inspection on the requirements and stops argument over dimensions nobody intended to control |
One practical note: the wall map is also the cheapest early-warning system you have. If the map shows a nominal wall that thins to a fraction of itself at the far end of a long flow path, the part is a fill problem before a tool exists — and that is a conversation to have at quotation, not at T1.
Cosmetic Surface Grades: Turning “No Visible Defects” Into Something Enforceable
Short answer: “no visible defects” is not a specification, because visible is a matter of lighting, distance and who is looking. A graded surface schedule is what converts appearance from an opinion into an acceptance criterion.
This is where housing programmes generate the most bad feeling, and almost never because the moulder is careless. The buyer has a product in mind; the moulder has a part on a bench under workshop lighting. Neither has written down which faces are judged, from where, and against what. The result is a batch that passes dimensional inspection, passes the assembly trial, and is rejected on appearance by someone in a different room.
Two things make the schedule work. First, it has to be drawn on the part, not described in a note — a shaded view per grade is unambiguous in a way that prose is not. Second, the viewing condition has to be stated: distance, light source and angle. “Grade A under a 500 lux diffused source at 500 mm, viewed perpendicular” is a test. “Grade A, cosmetically acceptable” is an invitation to argue.
| Appearance issue | Where it comes from | What the specification can do about it |
|---|---|---|
| Sink opposite ribs and bosses | Local mass cooling and shrinking behind a thin visible wall | Set rib and boss wall relative to nominal, and require the sink check on the grade-A face |
| Weld line | Two flow fronts meeting, usually behind a hole or core | State that weld lines are not permitted on grade A, which forces gate and core layout to be resolved in DFM |
| Gate vestige | The gate remnant at the feed point | Nominate permitted gate locations and the finish allowance, rather than leaving gate position to the toolmaker |
| Ejector witness | Ejector pin marks on the face they push against | Assign ejector contact to grade B or C faces and state it on the drawing |
| Flow and blush marks | Shear at the gate, or hesitation in a thin section | Permitted within a written limit on B, not on A; the fix is process and gate design, not polishing |
| Colour and lot variation | Masterbatch, regrind ratio or resin lot | Fix the colour standard and the permitted regrind percentage in the purchase specification |
| Texture and gloss drift | Tool texture wear over the production run | Reference a texture standard on the drawing and require the texture to be re-verified at agreed intervals |
Note that only two of those seven are solved by the moulder alone. The rest are decided by what the drawing permits, which is why the grading exercise pays for itself the first time a batch is questioned.
Mating Halves: Why the Seam Is a Stack-Up, Not a Dimension
Short answer: gap and flush are two different requirements with two different causes. Gap is axial — driven by flange thickness and flatness. Flush is lateral — driven by the locating features. Specifying “the halves must fit” controls neither.
A two-part housing is one assembly made from two parts that are toleranced independently, often in two different tools, sometimes at two different suppliers. Each half can be perfectly in tolerance and the assembly still open at the seam, because the tolerances accumulated in the same direction. This is arithmetic, not workmanship, and it is the reason a stack-up review belongs at quotation stage.
Gap — the axial seam opening
- Flange thickness of half A and of half B, both with their tolerances
- Flatness of each flange, including warp after ejection
- Any gasket, seal or label sitting in the joint
- Snap-fit engagement depth, or screw clamping if the joint is fastened
- Whether the two halves come from one tool family or two
Flush — the lateral step at the seam
- Positional tolerance of the locating features — pins, ribs, steps or lips
- Draft on the mating walls, which moves the surfaces as they part
- Mould mismatch between cavity and core on each half
- Shrinkage anisotropy, worst in glass-filled materials
- Assembly method: a hand-closed seam and a screwed seam do not behave the same
The practical consequence: a housing drawing that tolerances the flanges but leaves the locating features as reference dimensions has specified the gap and left the flush to chance. In glass-filled materials especially, the same nominal geometry can produce a visible step on one side and none on the other, because flow direction sets the shrinkage direction.
| Decision | What it fixes | What it costs |
|---|---|---|
| Tolerance the assembly, not only the parts | Makes the seam an explicit requirement with an owner | One extra drawing view, prepared before the tool is quoted |
| Nominate the mating interface as the datum | Stops the stack-up from being measured from a cosmetic surface that itself moves | Rework of the datum scheme on both drawings, done early |
| Cut both halves in one tool where volume allows | Removes mould-to-mould variation from the stack altogether | Larger tool and tonnage; not economic at low volume |
| Control flange flatness, not just thickness | Addresses the dominant gap contributor, which thickness alone does not | A flattening or fixturing step after ejection on warprone geometries |
| Add a locating step or lip rather than relying on screws | Controls flush independently of clamping force | Tool complexity at the parting line |
| Choose a low-anisotropy material if flush is tight | Reduces the shrinkage direction effect that causes the step | Possible stiffness or cost penalty against a glass-filled grade |
What a Thin Wall Does to the Tool
Short answer: filling a thin wall needs higher pressure and higher injection speed, and the tool has to survive both. The consequences land on rigidity, venting, shut-offs, ejection and steel selection — not on the cavity dimensions, which are the part everyone checks.
The process mechanics behind this are covered in detail on the thin-wall injection molding page, including why flow length governs fill and why the achievable wall is set by geometry rather than by a universal limit. This section covers only what those mechanics do to the purchase decision.
| Tool aspect | What thin-wall moulding demands | Why the buyer should care |
|---|---|---|
| Cavity and core rigidity | More support and a stiffer stack, because cavity pressure is higher and deflection shows up as wall variation | Wall thickness variation across the part is a tool stiffness problem before it is a process problem |
| Shut-offs and parting line | Tighter shut-off integrity, since higher pressure finds any weakness as flash | Flash on a grade-A face is a reject; flash at a shut-off is a tool repair |
| Venting | Vents placed at the actual end-of-fill locations and cut to the material's depth | Inadequate venting produces burn marks and short shots that look like process faults |
| Ejection | More ejector area, balanced, because a thin part is easier to distort than to push out | Ejector witness on a visible face, or a part that bows on ejection and never recovers |
| Cooling layout | Balanced cooling with attention to the thick zones — bosses, ribs, pads | Cycle time and warp are decided here; both appear in the unit price |
| Steel selection | Grade chosen for the production volume and the abrasive or corrosive nature of the material | A tool that wears into flash mid-programme costs more than the steel it saved |
| Cavitation and family layout | Balance between cavitation, tonnage and the risk of one cavity drifting | Family tools are cheap per part until one cavity makes the whole shot suspect |
Tool construction itself is covered on injection mold making. The purchasing point is simpler: when a quote for a thin-wall housing tool comes in well below the others, ask which of the rows above was reduced. It is usually rigidity, venting or steel — and all three show up later as defects rather than as a line item.
Material by Housing Requirement, Not by Preference
Short answer: start from the requirement that would fail first — impact, stiffness, heat, chemical exposure, flame, UV or appearance — and let that choose the family. Starting from a familiar grade is how housings end up over-specified on one axis and marginal on another.
| Family | Chosen when the housing must | Watch for |
|---|---|---|
| ABS | Look good and mould easily at moderate cost; cosmetic parts with painting or printing | Modest heat and chemical resistance; not the grade for outdoor or solvent exposure |
| PC/ABS | Balance impact resistance with appearance and processability | Notch sensitivity and stress cracking under some solvents and moulding stresses |
| PC | Take impact and hold stiffness, including translucent or clear covers | Higher melt viscosity, so thin walls are harder to fill; drying discipline is critical |
| PA6 / PA66, unfilled | Toughness and chemical resistance at temperature | Moisture absorption changes dimensions in service; drying and conditioning matter |
| PA with glass fibre | Stiffness in a thin wall without thickening it | Anisotropic shrinkage — warp and seam flush both get worse; glass wear on the tool |
| PPS, often glass-filled | High heat and chemical exposure, electronics and under-hood environments | Cost, and a narrow process window; tool steel and heater layout need to suit it |
| Flame-retardant grades | Meet a stated flammability requirement for the end product | Specify the standard and the grade together; FR additives affect toughness, colour and processing |
| UV-stabilised grades | Survive outdoor exposure without chalking or colour shift | Stabilisation is a grade property, not a coating; confirm it is in the resin, not assumed |
| TPE / TPU | Provide grip, sealing or damping, usually overmoulded onto the housing | Bonding to the substrate is a material pair decision, covered on the overmoulding route |
Two specification habits are worth adopting. State the flammability requirement as a standard and a grade together, not as “flame retardant”, because that phrase does not tell a moulder which resin to buy. And state the regrind policy explicitly — permitted percentage, and whether regrind is allowed on grade-A parts — because colour and impact both move with it. Material data and grade comparisons are on the materials page.
Validation and Acceptance: What Proves a Housing Is Right
Short answer: a dimensional report proves the parts match the drawing. It does not prove the housing works. The acceptance schedule has to include the assembly, the impact requirement and, where claimed, sealing — because those are the failures a drawing cannot catch.
| Stage | What is checked | What it can still miss |
|---|---|---|
| First article inspection | Every requirement dimension on both halves, against the drawing and the datum scheme | A part that measures correctly but does not close, because the stack-up was never analysed |
| Assembly trial | The two halves together with inserts, gaskets, screws or clips, in the intended sequence | Long-run drift, and assembly at the customer's line rate rather than at a bench |
| Impact or drop test | The housing's actual requirement: drop height, orientation, surface and pass criterion you specify | Cracks that start at a boss root and only appear after repeated drops |
| Sealing or ingress test | The stated protection level, on assembled units, with the production gasket and closing force | A seal that passes on a hand-assembled sample and fails on a torqued production unit |
| Capability study | Whether the functional dimensions can hold their tolerance in series, not just on the first shot | Nothing about appearance; capability on a cosmetic face is a grading question, not a Cpk one |
| Material and colour verification | Grade certificate against the specified resin, and colour against the agreed standard under agreed light | Lot-to-lot variation that only appears across several production runs |
| Secondary-operation checks | Coating adhesion, print durability, texture match, shield continuity where applicable | Adhesion failures that develop in transit or after ageing |
Two of those are frequently skipped and both are cheap: the assembly trial and a written pass criterion for the drop test. A drop requirement without a stated height, orientation and failure definition cannot be passed or failed, only debated. Our quality system and certification status are documented on the quality certifications page.
Cost Drivers in a Housing Programme
Short answer: the thinnest wall that can be filled is not the cheapest wall. Below a certain point the material saving is smaller than the yield and cycle-time penalty it buys, and the difference shows up as scrap rather than as a line on the quotation.
| Driver | How it moves | Where the buyer has leverage |
|---|---|---|
| Part weight | Falls with wall thickness, roughly with the skin area | Wall map and rib strategy; use ribs for stiffness rather than a thicker skin |
| Cycle time | Rises as walls thin, because injection speed and packing have to increase | Cooling layout and gate position; accept a slightly thicker wall where it buys stability |
| Machine tonnage | Set by projected area and cavity pressure, not by part weight | Cavitation choice; a smaller cavitation tool on a smaller machine can beat a large tool |
| Tool cost | Rises with rigidity, venting, steel grade and cavitation | Volume forecast honesty — a tool built for the real volume, not an aspirational one |
| Yield and scrap | Worsens as the process window narrows | The wall map again: designing margin into the thinnest zone usually pays for itself |
| Secondary operations | Painting, printing, shielding, insert fitting, assembly | Cosmetic grading — every face moved from A to B removes finishing and inspection |
| Inspection | Follows the number of controlled dimensions and the grade-A area | Marking consequences of the model as reference dimensions instead of requirements |
This is also where a housing differs from a simple moulded part commercially: the tool is often a minority of the total programme cost, and finishing plus inspection can exceed it. A quote that is competitive on tooling and expensive on finishing is not a competitive quote.
Housing RFQ Checklist
Short answer: send the assembly as well as the parts. Almost every avoidable problem on this page is invisible when the two halves are quoted separately.
| Input | What it lets us do |
|---|---|
| Assembled 3D model, plus each half | Run the gap and flush stack-up instead of quoting each part blind |
| 2D drawings with datum scheme and tolerances | Separate requirement dimensions from reference geometry |
| Zone wall map | Evaluate fill at the thinnest zone and at the longest flow length |
| Cosmetic grade per surface, with viewing condition | Place gates and ejectors, and build inspection around the graded area |
| Material and grade, flame and UV requirements | Confirm the resin route and any tool steel implication |
| Mating parts, inserts, gaskets, screws | Include them in the stack-up and in the assembly trial |
| Annual volume and programme life | Choose cavitation, steel and cooling to suit the real life of the tool |
| Validation list and pass criteria | Quote the testing that will actually be required, not a nominal allowance |
| Secondary operations and who owns them | Define the boundary between moulding, finishing and your assembly |
| Target date and current failure, if any | Address the real problem rather than restating the drawing |
Commercial terms, packaging and logistics are handled on the quotation page. The list above is the technical input that determines whether the quote is meaningful.
Frequently Asked Questions
What counts as a thin-wall plastic housing?
There is no single millimetre figure, because whether a wall is thin depends on how far the plastic has to flow to fill it rather than on thickness alone. A housing is thin-walled in practice when the wall is thin relative to its flow length, so the melt risks freezing before the cavity is full. That is why the same wall thickness can be routine on a small housing and difficult on a large one.
How thin can a plastic housing wall be?
It depends on the material, the flow length from gate to the furthest point, the wall uniformity and the tool. Rather than quoting one number, the achievable wall is established during DFM against your actual geometry, and confirmed on T1 parts. A supplier who gives a single figure for all housings is guessing.
Why do two housing halves not line up?
Because the gap at the seam is a stack-up, not a single dimension. It accumulates the two halves' wall and flange tolerances, the positional tolerance of the mating features, mould-to-mould variation if two tools are used, and the effect of warp on the flange. Tolerance stack-up analysis has to be done on the assembly, not on each drawing in isolation.
What is a cosmetic surface grade and why specify one?
It is a written classification of which surfaces are judged visually and to what standard. Surfaces handled as grade A are judged in the assembled product under defined lighting; grade B permits minor marks outside the primary view; grade C is functional only. Specifying the grade by surface prevents both disputes over acceptable marks and unnecessary cost from over-finishing hidden faces.
What causes sink marks on a housing's visible face?
Local mass on the far side. Ribs, bosses and screw pillars pull material as they cool and shrink, and the visible wall opposite them sinks. It is controlled by rib and boss wall thickness relative to the nominal wall, by packing, and by cooling on the thick side, not by polishing the defect out afterwards.
Should ribs be thinner than the housing wall?
Yes, as a rule, because a rib root at full wall thickness acts as a local mass and produces a sink mark on the opposite face. The usual direction is a rib wall thinner than the nominal wall, with a generous root radius, and the numeric value is a DFM decision made against your material and wall rather than a fixed site-wide figure.
Does a thin-wall housing need a different mould?
It usually needs a more demanding one. Filling a thin wall takes higher injection pressure and speed, which raises the load on the tool, so rigidity, support, shut-off integrity, venting and ejection all become more critical, and steel selection has to suit the production volume rather than the cheapest grade that will cut the cavity.
What validation should a housing programme include?
Typically dimensional first-article against the drawing, an assembly trial with the mating halves and any inserts, a drop or impact test to your requirement, sealing or ingress testing where the housing is specified to resist it, and a capability study on the functional dimensions once the process is stable. Which of these apply is defined by your product requirement, not assumed by the moulder.
How do housing wall thickness and cost relate?
Thinner walls reduce material per part but raise the demands on machine, tool and process, and can reduce yield. The cost optimum is not the thinnest wall that can be filled; it is the wall that meets the structural and cosmetic requirement with a stable process. Below that point, scrap and cycle variability usually cost more than the material saved.
What should I send for a housing quote?
The assembled 3D model plus each half's drawing, a zone-by-zone wall thickness map, the cosmetic surface grade for each face, which dimensions are functional and which are cosmetic, the material and any flame or UV requirement, the mating parts and inserts, annual volume, and the validation you will require. Sending the assembly as well as the parts is what makes a stack-up review possible.
Related Pages
| Page | What it covers |
|---|---|
| Thin-wall injection molding process guide | Flow length, freezing, pressure, cooling, gates, vents, defects and process windows |
| Electronics enclosure injection molding | PCB-centric enclosure design, EMI shielding, ingress protection and thermal paths |
| Custom plastic shell components | Shell parts, surface processing and colour options |
| Home appliance plastic casings | Appliance casing parts and their specific cosmetic demands |
| Power supply plastic cases | Housings for power supplies, with heat and flame considerations |
| Communication equipment housings | Housings for communication equipment |
| Moulded threaded inserts | Installing metal threads into moulded bosses |
| Two-shot injection molding | Multi-material housings, sealing lips and soft-touch features |
| Injection mold making | Tool construction, steel selection and tool life |
| Engineering materials | Resin families, grades and property data |
| Quality certifications | Quality system and certification status |
| Injection molding services | The full moulding service and how the routes fit together |
Send the Assembly, Not Just the Two Halves
The problems on this page — a seam that will not close, a sink on the visible face, a finish dispute at first article — are all decided before the tool is cut. Send the assembly model, the two drawings, the wall map and the cosmetic grading, and the review will come back with the fill risk at the thinnest zone, the stack-up on the seam, and the places where the specification leaves something undecided.
What this page does and does not claim
This page deliberately publishes no universal minimum wall thickness, no standard seam gap and no blanket tolerance for thin-wall housings. Those values are set by the material, the flow length, the joint design and the tool — quoting a single figure for all housings would be a guess dressed as a specification, and it is the most common way a moulding quotation becomes unachievable later.
It also does not repeat the process physics from our thin-wall process guide or the board-driven enclosure rules from our electronics enclosure guide. Those pages are linked rather than summarised, so that each page keeps one job and none of them contradicts the others.
Where a number is required, it is established against your part in a DFM review and confirmed on trial parts. Technical content is reviewed before publication and the page carries a revision date.
