What to Check on a Silicone Case Sample Before You Cut Steel

The courier bag lands on a Thursday. Inside is a black case for your handheld terminal. It slides onto the device, the ports look right, the surface feels about right, and the instinct is to write "approved" and ask for a proforma invoice. Knowing how to evaluate a silicone case sample before approving mass production is a narrower exercise than it looks, because most of what you want to know is not present in the object in your hand.

If that sample was 3D printed, vacuum cast, or cut from a cured slab on a CNC, it is a shape study. It is a useful shape study. But it is not a moulded silicone part, and it cannot answer a moulded part's questions. Treating it as though it can is how buyers end up approving a case that fits beautifully in resin and comes back from the first production shot with a hard flash ridge running across the grip.

The commercial stakes are asymmetric. A wrong call on fit costs you a second sample round and a week or two. A wrong call after steel is cut costs you the tool modification, the re-sample, and the slot in the production schedule — and if the change is dimensional rather than cosmetic, sometimes the tool itself. That asymmetry is why the sample review deserves an hour with calipers rather than five minutes on a desk.

First work out what kind of sample you are holding

Ask one question before anything else: what process made this part, and which mould number did it come from? A supplier who has ready-made tooling for your device model can hand you a genuine moulded part almost immediately — WJM Silicone, for instance, keeps a library of 80-plus ready-made public molds for POS terminal models and quotes sampling from those in as little as 72 hours — what a public-mould library does and does not cover is worth checking against your exact device before you assume it applies. A part off an existing steel mould is a production part in every respect that matters. A part made in three days for a device that has no tool yet is something else entirely.

Write the answer into your file. Six weeks later, when someone asks whether the approved sample was a moulded part, nobody will remember.

Two follow-up questions separate a serious quoting engineer from a salesperson. First, was the sample post-cured? Moulded silicone normally goes through a post-bake to drive off cure by-products; a part that skipped it can smell faintly acidic and can shift a shade of colour once it has been through the oven. Second, what did the part weigh? Part weight is the cheapest proxy you have for wall thickness, and it is the number that later explains a unit price. Ask for it in grams to one decimal, and ask again on the production first article.

The three things a pre-tooling sample can honestly answer

If you are holding a non-moulded sample, there are exactly three questions it is qualified to settle. Run those three properly and defer the rest.

Dimensional fit, measured against the device rather than the drawing

Seat the device. Then stop looking and start measuring. Take calipers to the internal opening at four points — both across the width and along the length, top and bottom — and to the wall thickness at the thinnest visible section. Compare against the 3D data the case was built from, not against the case drawing, because the case drawing already contains the offset the designer chose.

The detail buyers miss: an additive sample is not dimensionally neutral. Printed parts typically grow along the build axis and shrink slightly in-plane, and the direction of that error depends on how the shell was oriented on the plate. So a printed case that grips the device perfectly can predict a moulded part that is loose, and one that is a fraction tight can predict a moulded part that is right. Ask which axis the sample was built on before you read anything into a snug fit.

Cutout registration, checked with the cable you actually ship

Registration is about position and aperture, and it is where most respins originate. Do not test a port with a fingertip. Test it with the exact connector that ships in the box, including its overmoulded boot — a bare plug clears an aperture that a moulded strain relief will not. Do the same with the charging cradle or dock, the belt clip, the hand strap and the scanner sled if the device has one. Cradle clearance is the single most common reason a case that fits the device fails in the field: the case is correct, the dock was never modelled.

For scanner and camera hardware, check the exit path rather than the hole. A window aperture that looks generous can still clip the edge of a scan cone or a wide-angle field of view, and on a translucent case wall a flash can pipe light sideways into the lens and put a halo on every image. Both are geometry problems that a printed shell will show you honestly, which makes them worth spending sample time on.

Device retention, with the fit mechanism named

Retention is where the two worlds diverge most sharply. A rigid printed shell holds a device by interference: it is either tight or it falls off. A moulded silicone case holds by stretching over the device and recovering. Those are different physics, and a printed sample cannot simulate the second one at all.

What you can still do is check the mechanism: how far up the display face the lip runs, whether the corners wrap far enough to survive an edge drop, whether the case must stretch over a protruding camera bump or a battery door, and whether removing it requires two hands. If the sample has to be split or cut to get the device in, that is not a failure of the design; it is evidence that you are reading a shape study, and it should push the retention judgement to the first moulded part.

What you cannot learn until the steel exists

Some properties do not exist until a compound is cured under heat and pressure in a closed cavity. List them explicitly and mark them as deferred, so nobody later claims they were approved.

Shrink belongs to the compound, not to the shape

Silicone shrinks on cure, and the amount depends on the grade, the hardness, the filler loading and the cure and post-bake schedule — commonly low single-digit percentages, but it is a number the moulder calculates for your specific compound, not a constant you can look up and apply. The tool is cut oversized by that factor. A CNC-cut or printed sample carries none of it, which means the dimensional agreement you just measured is provisional until the first moulded shot. Ask the moulder what shrink allowance they applied and to which grade, and keep the answer with the tooling file, because a later change from a 50 Shore A grade to a 70 Shore A grade can change it, along with much else you specified on the hardness line.

Flash lines land where the tool splits, and that is an ergonomic decision

Every moulded part carries a parting line. On a compression-moulded case it is a real, feelable edge before deflashing and a faint witness line after. Where it sits is decided when the tool is designed, and the default choice is whatever makes the tool simplest to cut and easiest to demould — which is frequently straight across the side wall your user's thumb rests on.

This is the question to raise before steel, not after: ask for the mould split line and gate locations to be shown on the 3D data at design review, and say which surfaces have to stay clean. Ask also how flash is removed — hand trimming produces a variable rim, tumble or cryogenic deflashing produces a consistent one — because the answer sets the edge quality you will receive for the life of the tool.

Grip, tear, colour and odour are all cure-dependent

Surface grip on a moulded case comes from the texture etched into the steel, not from a printed finish, and its tack changes with the compound and with any post-mould treatment. Tear resistance at thin sections — the bridge over a port, the lip over a display, the web at a corner — depends on the grade's tear strength and on how the cavity fills, and it is the property most likely to bite you at 3,000 units rather than at unit one. Colour has to be judged on a moulded, post-cured chip under controlled lighting; pigmented silicone is opaque and diffuse, and it will not match a paper chip the way an ink does. Push all of it to the first-article stage and say so in writing.

The pre-tooling sample checklist

Run this on the sample in front of you, in this order, and record the numbers rather than the verdict.

1. Log the provenance. Process, mould number if any, cavity, compound and hardness, post-cure yes or no, part weight in grams.
2. Photograph the sample as received, all six faces, with a scale in frame. This is your evidence if the production part differs.
3. Measure the internal opening at four points and the wall at the thinnest section. Record actuals, not pass or fail.
4. Seat the device. Check the display lip height, corner wrap and any camera or sensor bump clearance.
5. Test every aperture with the real accessory — shipped cable with its boot, dock or cradle, belt clip, hand strap, stylus, sled.
6. Check the optical paths: scanner window, camera field of view, flash isolation, any status LED that has to remain readable.
7. Check the acoustic and thermal openings: speaker, microphone, and any vent that has to stay unobstructed on a device that runs warm.
8. Note what the sample cannot answer and copy that list into the tooling agreement as deferred items.
9. Ask for the mould split line, gate positions and ejector layout on the 3D data before the tool is released for cutting.
10. Define the first-article gate: quantity, what gets measured, who signs, and where the retained counter-sample is stored.

Point 10 is the one buyers skip. Sample approval and first-article approval are two separate decisions, and only the second one is about the tool. A useful sequence is the one most OEM and ODM moulders already run internally — concept, 3D modelling, rapid prototyping, mould fabrication, material testing, pilot run, then volume — with your signature required at the prototype stage and again after the pilot run, not once at the beginning. If you want the full picture of how those stages compress or stretch, the sample-to-production timeline for China orders is the wider frame this checklist sits inside.

Common questions

Should I pay for the sample, and does it come off the order?

Sample charges are ordinary and a supplier who waives them is not doing you a favour worth much. What matters is the policy in writing: the amount, what it covers, and whether it is credited against the first production order. Ask before the sample ships, not after, and put the answer on the proforma. The related question — whether a design or scanning fee is refundable against tooling — belongs in the same email.

The sample fits perfectly. Do I still need a pilot run?

Yes, and for a reason that has nothing to do with distrust. A single sample tells you about one cavity on one shot. A pilot run tells you about cavity-to-cavity variation, about how the part behaves after a full post-cure cycle in a loaded oven, and about whether deflashing holds up when an operator is doing it at rate rather than at leisure. Agree the pilot quantity and the measurement plan in advance.

Who keeps the approved sample?

Both parties, sealed and signed, with the date and the mould number on the label. The retained counter-sample is the only object that settles a dispute about production quality later, and it is worthless if only one side holds it. Ask how long the factory stores retains and whether they will re-issue one after a tool refurbishment.

What to ask the supplier next

Before you release tooling money, send these questions in one email and keep the reply:

• Which process made the sample, from which mould, in which compound and hardness — and was it post-cured?
• What shrink allowance are you applying, for which grade, and what changes if the hardness changes?
• Show the mould split line, gates and ejectors on the 3D data before cutting steel.
• How is flash removed, and what rim finish should I expect at volume?
• What is the sample charge, and is it credited against the first order?
• What is the pilot-run quantity, what gets measured, and who signs the first article?
• Who owns the tool, where is it stored, and what is the shot-count maintenance schedule? (The who owns the mould question is worth settling in the same document; so is the true cost of tooling you are about to commit to.)

A shop with existing tooling for your device changes this conversation, because the sample stops being a study and becomes the part. That is the practical appeal of a public-mould library like the one at WJM Silicone in Longgang, Shenzhen, where sampling from a ready-made POS mould is quoted at 72 hours — you are evaluating a moulded case, not a prediction of one. For a custom device with no existing tool, the discipline above still applies, and how a first sample order normally runs is the process the questions attach to.