Your speaker adhesive sample lands in a week, and without a test plan written before it ships it is a bottle of glue with an opinion attached. Sampling stated at seven days. Production stated at fifteen. Those are the numbers on the Aosibo Adhesives listing — a Huangjiang, Dongguan supplier registered in 2011 whose range covers the loudspeaker bond schedule: lead-wire, damper, surround, magnet-circuit and centring adhesives, with two-part epoxy and acrylic systems and cyanoacrylates alongside. Read as days, which is how they are read here, that means a small bottle can be in your hands roughly two weeks before you would need to place a production order to hold a launch date.
The trap is that seven days of supplier time is easy and five days of your own time is not. Most adhesive samples get a cursory bond, a tug, a nod, and a purchase order — and then the problems appear in month four, when you have a qualified bond, a running line and no room to move. A structured week of testing costs you a technician's afternoon and answers the three questions that matter: does this bond my actual substrates, does it cure inside the window my line actually has, and is what arrives in the drum the same thing that arrived in the bottle.
Here is a sample test plan you can run in seven working days, plus the specification you have to send before the sample ships to make the results mean anything.
Before it ships: five things that decide whether the sample is testable
A sample is a physical answer to a question. If the question was vague, the answer is decoration.
Send both substrates, not a product name. Adhesive suppliers state this directly and they mean it: mix ratio, pot life, fixture time, full-cure schedule and substrate compatibility are order-specific, and the useful inquiry sends the two materials you are bonding and the cure window you have available. Cone paper from two mills behaves differently. A phenolic-impregnated cotton spider and a Nomex-type spider are not the same bonding surface. Send swatches or parts if you can; send the material spec if you cannot. The same discipline that makes a good written product specification for a Chinese factory applies here, with substrates in place of drawings.
State your cure window as a constraint, not a preference. "Ambient" and "60 °C for 20 minutes" produce different recommendations, and a supplier who does not know your oven capacity will quote the grade that performs best rather than the grade that fits.
Ask for the sample to come from a production batch, with the lot number on the pack. A lab-mixed sample tells you whether the chemistry can work. A production-batch sample tells you what you will actually receive. Both are legitimate; only one of them is a baseline you can compare a future delivery against. If the supplier can only send a lab mix, ask them to say so — it is not a failing, but it changes what your sample proves, and it is one of the signals that helps you place a supplier as a repacker, a blender or a compounder.
Ask for enough material to test twice. A 20 g bottle vanishes in one afternoon of bracketing. Ask for a quantity that lets you run the plan and still seal half of it as your reference retain.
Ask for the technical data sheet and the safety data sheet to travel with the sample, plus the date of manufacture and the storage temperature. On cyanoacrylates and two-part systems the remaining shelf life when the container is opened at your end matters more than transit time — a point worth settling at sample stage, because a sample that is already nine months old is testing a different material from the one you will buy.
Day 0 and the seven-day plan
Twenty minutes of receiving checks
Do these before anything is bonded. They catch more problems than the bond tests do.
- Pack condition and closure. Weeping, a swollen bottle or a crusted cap on a cyanoacrylate ends the evaluation there.
- Date of manufacture and lot number present on the pack. If there is no lot number, ask why now rather than at first delivery.
- Net weight. Put the pack on a scale and compare against declared content. Underfill is common in hand-filled samples and is a cheap early read on filling discipline.
- Appearance and phase state. Water-based dispersions settle; a stir should redisperse them. Sediment that will not redisperse, a skin, or a separated layer in a two-part component is a formulation or storage problem.
- Viscosity at your line temperature, not at 25 °C. If your workshop runs at 18 °C in January, measure at 18 °C. Many application problems are a viscosity curve, not an adhesion failure.
- Odour and solvent character, noted rather than judged — it previews your ventilation and dangerous-goods conversations.
Record all of it: this becomes the incoming-inspection sheet for every future delivery, and building it now costs nothing.
The seven-day schedule
| Day | What you run | What it answers | Fail signal |
|---|---|---|---|
| 1 | Receiving checks; viscosity at line temperature; open time and wet tack with a stopwatch | Is the material as described, and can the operator work with it at your pace? | Skin or tack loss before your takt time elapses |
| 1 | Nominal-condition bonds on your real substrates, minimum 5 replicates per bond | Does it bond what you actually have? | Any wetting failure on any substrate |
| 2 | Cure-window bracket: nominal, nominal −10 °C, nominal −20% dwell | Does it cure in the window your line actually has on a bad day? | Strength collapse at the low corner |
| 2 | Fixture-time measurement at each bracket condition | Can the part be handled at your takt time? | Fixture time exceeding your conveyor interval |
| 3 | Full-cure pull and peel tests on nominal set; record failure mode | Real strength, and whether more strength would buy anything | Adhesive failure at the interface at low load |
| 3 | Cured-line inspection: squeeze-out, voids, wicking, staining on visible parts | Cosmetic and contamination risk | Wicking into a moving part or a visible cone face |
| 4 | Accelerated conditioning: heat soak at your service temperature; damp-heat if applicable | Does the bond hold when hot? | Strength drop that puts the bond below margin |
| 5 | Post-conditioning retest of the same replicates | How much margin the bond loses in service | Loss beyond your qualification threshold |
| 5 | Function test: a complete driver or sub-assembly, powered | Buzz, rub, rocking, outgassing near the diaphragm | Any audible artefact absent from the control |
| 6 | Bad-day simulation: over-application, under-application, contaminated surface, mis-mixed AB | How forgiving the material is with a real operator | Cliff-edge behaviour rather than gradual degradation |
| 7 | Documentation; seal and label the reference retain; write the second-sample request | Turns a week of work into a baseline | No retain kept — the most common omission |
Two notes on running it. First, keep a control: your incumbent adhesive, bonded the same day by the same operator on the same substrate lot. Absolute numbers from a workshop pull tester mean little; the delta against a control you trust means a lot. Second, run five replicates minimum per condition and record every one. Three replicates cannot distinguish a weak bond from a bad operator.
Reading the results
Read the failure mode, not just the number
This is the detail that separates buyers who evaluate adhesives from buyers who measure them, and it takes no equipment.
When a bonded test piece breaks, look at the broken faces:
- Adhesive failure — the glue line has released cleanly from one substrate, leaving that surface bare. The adhesive is not wetting or keying to that material. More cure time will not fix it; a different chemistry or a surface treatment might.
- Cohesive failure — the glue line has torn through itself, leaving adhesive on both faces. The bond to both substrates is sound and you are measuring the adhesive's own strength. This is the result you want at a load above your requirement.
- Substrate failure — the cone paper tore, the spider fabric delaminated, the former buckled. The adhesive is stronger than the part. Buying a stronger adhesive buys you nothing at all, and this is the point where "we upgraded the glue" becomes a waste of money.
- Mixed mode — the usual real-world result. Record the approximate percentage of each; a drift in that percentage between batches is an earlier warning than a drift in the pull number.
A test report that says "4.2 MPa" tells you less than one that says "cohesive, 90%, at 4.2 MPa, control 3.8 MPa cohesive."
Bond-by-bond: what each joint needs proving
The speaker range is bought as a matched set — lead-wire, damper, surround, magnet-circuit, centring — because a driver fails at whichever bond you tested least. Each needs a different question asked of it.
Lead-wire. The tinsel lead flexes for the life of the driver. Static pull strength is nearly irrelevant; flex fatigue and the stiffness of the cured fillet are what matter. Bond a set, then cycle them by hand or on a jig and look for cracking at the edge of the fillet. A bond that is too hard concentrates stress and fails faster than a softer one with half the pull strength.
Damper (spider). Carries the whole moving assembly. Test in peel as well as shear — the joint sees both — and test after heat soak, because the spider bond sits close to the voice coil.
Surround. Foam and rubber rolls are low-surface-energy and different from each other. Test both if you run both. For water-based grades, add a freeze-thaw cycle on the liquid adhesive itself: a pallet that freezes in transit or in an unheated warehouse can be finished regardless of the printed shelf life, and finding that out with a sample bottle in a domestic freezer costs nothing.
Magnet circuit. Shear on steel and on the magnet material, at service temperature. Also check whether cured squeeze-out interferes with the gap.
Centring adhesive. Where a grade is offered in more than one chemistry — one-part, epoxy and halogen-free versions are commonly listed as alternatives — sample the chemistry that fits your cure window rather than the one the supplier recommends by default, and if you sample the halogen-free grade, ask which threshold and which test method the claim was measured against and request the report. That is a numeric claim, and it needs a number and a method behind it.
What a seven-day sample cannot tell you
Be honest with your sourcing review about the limits — overclaiming a sample result is how the wrong bond gets qualified.
- Long-term ageing. A week of accelerated conditioning is an indicator, not a service-life prediction.
- Batch-to-batch variation. One sample is one batch. The cheapest probe is to order a second sample four to six weeks later, with a different lot number, and run the Day 1 and Day 3 tests again. Two data points is not a distribution, but it will catch a supplier whose "same grade" is not the same twice.
- Shelf life. You cannot test twelve months in a week. Ask for shelf life from date of manufacture, ask for the storage temperature, and put a minimum remaining shelf life on arrival into the purchase terms.
- Whether production material matches the sample. This is the classic risk, and it is why the reference retain matters more than any single test. Seal half the sample, label it with the lot number and the date, and store it as your baseline. When the first production delivery arrives, run the Day 0 checks and one bond test against that retain.
- How the supplier behaves under pressure. A stated 15-day production lead is a claim about a normal month. What it looks like in a bad one, you learn later — which is why the gap between a good sample and stable mass production is worth planning for explicitly rather than assuming away.
Common questions
Seven days for a sample — is that realistic for adhesives?
For a stock grade in a small pack, yes; a supplier holding finished material can pick, fill, label and ship inside a week. For a grade adjusted to your substrates or your cure window, seven days would be quick, and you should expect a first sample of the nearest standard grade followed by an adjusted one. Ask which of the two you are being sent, because it changes what the sample proves. General sample-order sequencing with Chinese factories applies, with the addition that adhesive samples have a clock on them from the date of manufacture.
How much sample material should I ask for?
Enough to run the plan twice over plus a sealed retain — in practice a few hundred grams for a line-fed adhesive, one or two standard small packs for an instant adhesive. Suppliers rarely refuse; the cost is filling labour, not material. And test on your own substrates from your current production lot, noting the lot on the report: supplier-supplied coupons show the adhesive at its best on materials it was developed against.
The sample passed. What is the first thing to check on the first production delivery?
The Day 0 receiving sheet, the lot number and date of manufacture, and one bond test against your sealed retain. That is under an hour of work and it catches the substitution risk that no amount of sample testing can address on its own.
What to ask the supplier next
Send this with the sample request rather than after it:
- Will the sample come from a production batch or a lab mix, and what is the lot number and date of manufacture?
- Sample pack size, and can it be sized to allow duplicate testing plus a retained reference?
- Technical data sheet and safety data sheet for each grade sampled, with the sample.
- Viscosity at 25 °C and, if available, at 20 °C — plus density, so consumption can be converted to volume.
- Recommended cure schedule, and the lowest temperature and shortest dwell at which the grade reaches full properties.
- Pot life and fixture time for two-part systems, and whether A and B ship separately or as a cartridge.
- Shelf life from date of manufacture, storage temperature range, and whether the grade is freeze-sensitive.
- Whether a second sample from a different lot can be supplied four to six weeks later, for a variation check.
- Confirmation in writing of both lead times, so the test week sits inside the schedule rather than on the critical path.
Aosibo Adhesives is a workable example to run this against: stated sampling of seven days and production of fifteen, a bond-schedule range built around the loudspeaker, and cure, viscosity and substrate data described as order-specific and shared at inquiry rather than published. Its factory profile on ChinaMakersHub lists what is stated and what has not been confirmed, which is the right starting position for a sample request. Pair the test plan above with a clear statement of the cure window your line actually has and the data-sheet fields worth arguing about, and the seven days the supplier spends will be matched by seven days that tell you something.