Second-Sourcing a Speaker Adhesive Without Requalifying the Driver

Second source speaker adhesive requalification is the question every driver plant reaches about eighteen months after it standardises on one glue: the incumbent has raised price, or gone short, or been bought, and somebody in purchasing has found an alternative that "matches the datasheet." The engineering answer is that the datasheet is not what you qualified. You qualified an interface — a specific polymer touching a specific pair of substrates, cured on a specific oven profile, tested as a finished driver. Changing the drum changes the interface, and the datasheet is silent about most of what that means.

That does not make second-sourcing impossible. It makes the work bounded and knowable, and it should be scoped before you ask for a price, not after. What follows is the minimum matrix, why a "drop-in equivalent" rarely is, and an honest calendar.

Why "drop-in equivalent" is a category error

An adhesive datasheet reports bulk properties: viscosity, solids content, density, colour, maybe a lap-shear number on a standard aluminium or steel coupon. Two products can match on every one of those fields and behave differently on your line, for reasons that never appear on paper.

Adhesion is interfacial, not bulk. The lap-shear figure on the datasheet was measured on grit-blasted aluminium. Your spider bond is phenolic-impregnated cotton against a paper or polyimide former, with a sizing on the cloth that varies by cloth supplier. Nothing about the aluminium number predicts the cotton number. The same is true for a treated paper cone against a rubber roll surround, or a nickel-plated top plate against a ferrite magnet.

Formulations are packages, not molecules. Two acrylics with the same solids content can carry different tackifiers, plasticisers, fillers and wetting agents. Those minor components decide open time, wet tack, how far the bead wicks along a tinsel lead, and whether the cured joint stays flexible after a thousand hours at temperature. They are also the components a supplier is most likely to change quietly, because they are cost levers.

Cure kinetics are yours, not theirs. A grade that reaches handling strength in six minutes at 120 °C in a lab oven with good air movement may need eleven minutes in your tunnel with a loaded conveyor. If your line takt is fixed, an adhesive that cures slower is not equivalent at any price.

The failure you care about is acoustic, not mechanical. A joint can pass a pull test and still cost you the driver. A slightly stiffer spider bond raises compliance and moves the resonance frequency. A bead that creeps a millimetre further up the former adds mass in the wrong place. A surround bond that hardens with age turns a clean roll into a buzzing one. None of that is a bond-strength question.

This is why suppliers who work only on loudspeakers answer differently from general chemical traders. Aosibo Adhesives, a Huangjiang, Dongguan supplier whose range reads as a driver's bond schedule written out — lead-wire adhesive for the tinsel leads, damper (spider) adhesive, water-based foam and rubber surround adhesive, magnet-circuit adhesive for the top plate, magnet and yoke stack, and centring adhesive in one-part, epoxy and halogen-free versions — states that mix ratio, pot life, fixture time, full-cure schedule and substrate compatibility are all order-specific, and asks buyers to send both substrates and the available cure window rather than a product name. That is the right shape of conversation for a requalification, and a reminder that nobody can hand you a drop-in.

What changes when the adhesive changes

Map each bond to the driver-level symptom it produces when it goes wrong. This is the table to put in front of your test engineer, because it decides which driver-level tests you cannot skip.

Two details in that table repay attention. First, the spider bond and the centring bond both move Thiele-Small parameters, which means a change there is measurable with equipment you already own — a full T/S sweep on a sample of twenty drivers before and after is cheap and catches compliance shifts that no coupon test would show. Second, the tinsel-lead bond is the one that fails in the field rather than on the bench, which is why flex fatigue cannot be skipped even when everything else passes on day one.

The minimum requalification matrix

Four blocks. Run them in this order, because each one gates the next and there is no point ageing samples of an adhesive that failed a room-temperature pull.

Block 1 — Bond strength on the real substrate pair. Not on aluminium coupons. Cut the actual cloth, the actual former material, the actual cone stock, the actual plated steel and magnet material, and bond them with your line's bead size and your oven profile. Run peel for the flexible joints and shear or tensile for the structural ones. You want a minimum of five specimens per joint per condition, and you want the incumbent adhesive run alongside as a control in the same session — absolute numbers drift between labs and between operators, so the comparison is the data, not the number.

Block 2 — Flex fatigue. For the lead-wire bond and the surround, cycle at or near your rated excursion for a defined count. A million cycles at 30 Hz is roughly nine hours per specimen, so this block is a fixture question as much as a time question: build or borrow a multi-station rig, or the calendar doubles. Inspect for bead cracking, delamination at the tinsel, and lead-to-cone slap, then re-run continuity.

Block 3 — Heat ageing and damp heat. This is the block that decides your calendar. Dry heat at your continuous service temperature for several hundred hours, plus a damp-heat exposure — 85 °C at 85% relative humidity is the common industry reference point — with pull tests before and after. If your customer is automotive, expect a thermal-cycle profile on top, and expect them to specify it rather than accept yours. Do not substitute a short high-temperature soak for a long moderate one; the failure mechanisms differ, and the short soak is the one that passes when it should not.

Block 4 — Acoustic check: rub-and-buzz, distortion sweep, and a listening panel. Assemble complete drivers with the candidate adhesive on the line that will actually build them, run your end-of-line rub-and-buzz and distortion sweep against the control population, then put a small set in front of people who know the product. The instrumented tests catch gross defects; the listening check catches the compliance shift that reads as "it sounds different in the low mids" and would otherwise ship. Run this block twice — once on fresh assemblies, once on units that came through Block 3 — because the interesting failures appear after ageing.

Be honest with whoever is asking for the cost saving: a full requalification is measured in weeks, not days. Six weeks is a good outcome with chambers free and a fixture already built; twelve is normal starting cold, and longer if an automotive customer has to approve the plan. A supplier quoting sampling in 7 days and production in 15 — as this Dongguan supplier does, both bare numbers on its application — is describing its own turnaround, not yours. The sample arrives in a week; the decision takes a quarter.

What you can legitimately skip

Not everything has to be redone, and pretending otherwise is how second-sourcing programmes get killed in the budget meeting.

• Regulatory and safety approvals of the finished product generally do not reopen for an internal bond, unless the adhesive changes a declared material property. A halogen-free requirement is the exception: that is a numeric claim measured against a threshold, and it needs its own test report from the new supplier regardless of everything above.
• Full agency safety testing of the finished speaker rarely reopens for an adhesive change alone. Check your specific approval scheme; do not assume.
• Mechanical drop and vibration can often be reduced to the magnet-circuit and frame bonds rather than re-run whole, if the other joints are unchanged.
• Long-term field data cannot be reproduced, which is the real argument for keeping the incumbent qualified in parallel rather than switching outright. Dual-sourcing means both are live; it does not mean you stop buying from the first.

The temptation is to cut Block 3 because it is mostly waiting. Resist it — heat ageing is where a cheaper formulation usually gives itself away.

The contract terms that make a second source worth having

A qualification is a snapshot of one batch. What protects it over time is paperwork, and this is where you should spend negotiating capital.

1. Formulation-change notification. A written commitment that the supplier will notify you, with a defined notice period, before any change to raw material grade, raw material source, or manufacturing site. Without it, your qualification silently expires the first time a raw material is substituted. Ask for it explicitly; it is not standard in Chinese adhesive supply unless you put it in the contract.
2. Batch traceability. Batch numbers on every pack, tied to production records, retained for a stated period, and available to you on request.
3. Locked specification. Reference the qualified batch by number in the purchase agreement and specify the properties that define conformity — not just viscosity and solids, but the ones that matter to your joints.
4. Site of manufacture. Name it. If the supplier blends or repacks rather than compounds, the compounding site governs consistency, and a change of compounder is a change of adhesive whatever the label says.
5. Retained samples. Both parties keep a sealed retain of the qualified batch, so a future dispute is settled by test rather than by argument.

None of those five were sighted for the supplier described here — its profile is built from an onboarding application, and change-notification, traceability and retained-sample practices were not among the documents provided. Questions to ask, not findings.

Common questions

Can I qualify a second source on a single bond instead of the whole set?

Yes, and it is usually the smart way in. Pick the bond with the lowest driver-level consequence and the highest spend — often the surround or the magnet-circuit adhesive — and run the matrix on that one alone. You learn how the supplier handles a technical conversation for a fraction of the cost. Note that a supplier whose range is designed as a matched set may quote differently for one line item than for the full schedule, so ask both ways.

The supplier says the product is equivalent to my incumbent. Is that worth anything?

It is worth a conversation and nothing more. Nobody outside your incumbent's lab knows their formulation, so "equivalent" means "similar datasheet." Ask instead which substrates they have bonded, what cure windows they support, and whether they will run your substrates before quoting. A supplier who asks for your two substrates and your available cure window before answering is doing the job properly.

How many batches should I test before I release the second source?

At least two, ideally three, and from different production runs. Batch-to-batch variation is the whole risk you are trying to bound, and one batch tells you nothing about it. Time the requests weeks apart so you are not sent three drums from the same blend.

What if the minimum order quantity forces me to buy more than the qualification consumes?

Then the shelf life becomes the binding constraint, not the qualification. Settle the unit of measure and pack size first, agree a minimum remaining shelf life on arrival, and consider a blanket order with staged releases so the material ages in the supplier's warehouse rather than yours.

What to ask the supplier next

Send one message, and judge the reply by how specific it is:

• The two substrates, described in material terms rather than by part number, plus your bead size, your line takt and your available cure window — temperature and dwell.
• Which of your substrate pairs they have bonded before, and whether they will run a bonded coupon set on your materials before you place an order.
• Full cure schedule options: ambient, oven, and the lowest cure temperature the grade will tolerate without a strength penalty.
• Pot life and fixture time for two-part systems, and whether A and B ship separately or pre-metered.
• A written formulation-change notification commitment, with a notice period you name.
• Batch numbering and traceability practice, and the retention period for production records.
• Whether the range is sold as a matched set under one lead time, or line by line, since a partial second source changes your inventory arithmetic.
• Two or three separate batches for testing, requested weeks apart.

Aosibo Adhesives is a reasonable example of the type of supplier worth running this exercise against — a Dongguan house registered in 2011 whose product list maps bond by bond onto a driver rather than reading as a general glue catalogue, listed with its registry check still open and its declared certificates unsighted, as stated on the Aosibo Adhesives factory profile. The narrow specialisation is the reason to have the conversation; the open items are the reason to run the matrix rather than take the datasheet on trust.

Scope the requalification honestly, run it in the order above, and write the change-notification clause before the first production order. The discipline behind that — locking a specification and holding a supplier to it over repeat orders — is covered in the guides to quality control in Chinese manufacturing, writing a product specification for a Chinese factory and supply chain risk management.