Micro Speaker Sample Development: What OEM Buyers Should Expect
1. What a Sample Round Is Actually For
Short answer: A sample round tests one question. If it tests more than one, the answer is not interpretable, and the round has been spent on ambiguity rather than information.
The most common expectation to correct is that a sample proves the design. It does not. A sample proves one thing: that a specific configuration, built under specific conditions, behaves in a specific way. Everything else — whether the product will sound right, whether the structure can accommodate it, whether the cost works — is a separate question requiring separate evidence.
This is why the round structure matters. A round that asks one question and changes one variable returns an answer. A round that asks three questions and changes three variables returns three answers that cannot be separated, and the only way forward is another round — which is how programmes that expected three rounds end up needing eight.
There is a related point about what a sample cannot tell you. A sample built in the supplier's standard fixture tells you about the driver under those conditions. It does not tell you what the product will do, because the product's cavity, front surface, grille and sealing are all different from the fixture. Closing that gap requires either measuring the sample in a mock-up of the product, or accepting that the first installed measurement is still to come.
2. How to Plan the Rounds
Short answer: Plan rounds around questions, not around calendar time. Each round should change one variable and measure the installed condition, not just the driver.
1. Round one: feasibility. Does the band definition produce the intended result in the available volume? Use a known driver in the cavity the structure will provide, and measure the assembled condition.
2. Round two: the parameter that matters most. Change the single variable with the largest effect — usually cavity volume, diaphragm thickness or the magnetic circuit — and measure the same conditions.
3. Round three: confirmation and comparison. Build the candidate and a reference unit, measure both under identical conditions, and confirm the difference is the one intended.
4. Production-intent round: the configuration as it will be built, measured on the acceptance basis that will be used at goods-in.
The number of rounds is less important than what each one changes. A programme that needs five rounds because each changes one variable is in better shape than one that needs three and changes three things in each, because the first converges and the second oscillates. It is also worth stating explicitly which rounds are optional, so that a round can be skipped when an earlier answer is decisive.
3. What to Measure, and Under What Conditions
Short answer: Measure where the product operates, and state the basis with every figure. A measurement at a point outside the operating band can look adequate while the product does not.
The measurement points should sit inside the band the product uses. For a voice product that means points across the speech band rather than a single conventional frequency; for a music product it means points covering the range the content occupies. Selecting the test points is part of the development work, not an administrative detail, because it determines what the published figures describe.
Every figure should carry its basis: frequency, distance, drive voltage, cavity volume and mounting surface. This is what makes a round's result comparable with the previous round's and with the eventual acceptance test. Where the sensitivity and resonance figures for the same model state different cavity volumes, the matched pair should be requested before either is used, because the two numbers then do not describe the same installed state.
Table 1 lists the conditions that should be recorded for each round, and what each one is for. The list is short, and omitting any of it is what makes a later round's result impossible to compare.
Table 1: Conditions to record for each sample round.
Condition | Why it has to be recorded | What goes wrong without it |
Cavity volume and sealing method | Determines the installed resonance through FC = Fs × √(1 + Vas / Vb) | A round that appears to have changed the driver when only the cavity changed |
Mounting surface or baffle used | A small front surface reflects differently from a large test baffle | Level differences attributed to the driver that come from the fixture |
Drive voltage and impedance | Sets the power the driver receives, and therefore the operating point | A result measured at a different power than the previous round |
Measurement frequency points | Determines which part of the response the figures describe | A figure at a point outside the operating band that looks adequate |
Reference unit measured the same day | Makes a comparison meaningful rather than indicative | A comparison between measurements taken weeks apart under different conditions |
What was changed since the previous round | The only way to attribute a difference to a cause | A result that cannot be acted on, requiring another round to interpret |
4. What to Do When a Sample Is Close but Not Right
Short answer: A near miss is informative, but only if the direction of the error is identified. Change one parameter in the direction the measurement indicates, not several at once.
The instinct when a sample is close is to adjust several parameters together in the direction that seems right. That is the fastest way to lose the information the round produced. The measurement already indicates which parameter is out: if the installed resonance is above target, the cavity or the driver's compliance is the lever, not the magnet. If the level in the band is short, the magnetic circuit or the diaphragm is the lever, not the cavity.
It is also worth distinguishing between a result that is wrong and a result that is right for a different question. A driver that meets its published figures in the test fixture but not in the product has not failed; it has answered a different question. The fix is to measure the installed condition, which usually requires the product's cavity and front surface rather than a better driver.
5. What a Sample Cannot Tell You
Short answer: Four things are outside what a sample can establish: production consistency, long-term behaviour, cost at volume, and the assembled product's result.
5. Production consistency. A sample is one build. Consistency comes from process control and from data across units, not from a single part.
6. Long-term behaviour. Environmental and ageing performance requires qualification testing over time, which a sample round cannot compress.
7. Cost at volume. Sample pricing reflects the build, not the production cost structure, and the two are not proportional.
8. The assembled product's result. A sample measured in the supplier's fixture does not include the product's cavity, grille, sealing or front surface.
Project Case Study (Hongsheng) — Three Cavity Volumes in One Round
The requirement came from an infant monitoring product that combined a camera with two-way talk and presence alerts. The customer's starting point was a competitor's reference unit, and their stated targets were a soft, non-harsh sound at a level that felt right to them. The structure allowed about 15 cc of cavity and a thickness limit of 13 mm, which constrained the driver to roughly 30 mm. They had evaluated existing drivers from several suppliers without resolving it.
The first sample round was designed around a single question, and it answered it decisively. Using the same 28 mm driver, three cavity conditions were built and measured on the same test day: a 30 cc cavity, a 20 cc cavity and a 10 cc cavity, with the competitor's reference unit measured alongside on identical conditions. Because the driver was held constant, the differences between the three curves were attributable to cavity volume alone. The 30 cc and 20 cc conditions tracked each other closely across the midrange and into the high band, while the 10 cc condition sat measurably lower and lost part of the upper-band extension. The competitor's reference unit was strongest in the upper band, which identified where the real gap was.
That answer set the direction for the second round. Because the harshness the competitor comparison had identified was an upper-band issue rather than a low-frequency one, the change made was in the dome material rather than in the driver size or the cavity: the dome was changed from polyester film to aluminium, extending the usable frequency range and reducing the perceived edge without lowering overall level. One variable, in the direction the first round had indicated.
The final configuration used the 15 cc cavity the structure allowed, with a 28 mm driver whose free-air resonance is 220 Hz, giving an installed resonance of about 380 Hz — inside the target of 400 Hz. The delivered part measures 100 ± 3 dB sensitivity at 2.83 Vrms, 10 cm and 2 kHz, rated 2.0 W with a 2.5 W maximum.
The programme reached design freeze with the customer confirming the result met the requirement. The driver was integrated as a single assembly with its cavity, so the structure only had to accommodate one part. The transferable conclusion about sample development is that the first round's value came from holding the driver constant and varying one thing, which turned a listening preference into a measurement that pointed at the next change.
Holding the driver constant and varying only the cavity volume turned a listening preference into a measurement. The first round's answer then pointed at the dome material rather than at the driver size.
The distinction matters because it determines whether a round converges. Hongsheng can build sample rounds against a mock-up of the product's own cavity and front surface, and states the cavity condition alongside each published configuration, so a round's result can be compared against the published basis.
6. What to Confirm Before Each Round
Short answer: Six items before each round: what changed, what is being measured, under what conditions, and what answer would make the round decisive.
9. What exactly changed since the previous round, stated as one variable wherever possible.
10. What question this round answers, and what result would make the next round unnecessary.
11. The conditions that will be recorded: cavity volume, mounting surface, drive voltage and impedance, and measurement points.
12. Whether a reference unit will be measured on the same day under the same conditions.
13. Whether the sample will be measured in a mock-up of the product's cavity and front surface, or only in the supplier's standard fixture.
14. What the round cannot tell you, so that its result is not asked to carry more weight than it should.
7. FAQ: Micro Speaker Sample Development
Q1: How many sample rounds should I plan for?
Plan around questions rather than a fixed number. Feasibility, the single most important parameter, confirmation against a reference, and a production-intent build is a common structure. The number matters less than what each round changes; a round that changes three variables costs more than a round that changes one.
Q2: Why does changing one variable per round matter so much?
Because a round that changes three things returns three answers that cannot be separated. The programme then spends its remaining rounds working out which change produced which effect, which is how a three-round plan becomes an eight-round one.
Q3: What should the first round test?
Feasibility: whether the band definition produces the intended result in the available volume. Use a known driver in the cavity the structure will actually provide, and measure the assembled condition rather than the driver alone.
Q4: The sample meets its datasheet but not my requirement. What now?
Measure the installed condition. A sample measured in the supplier's fixture has not included your cavity, front surface, grille or sealing. In one programme, three cavity volumes with the same driver produced curves that differed across the whole band, and the 10 cc condition lost upper-band extension as well as low-frequency output.
Q5: What can a sample not tell me?
Production consistency, long-term behaviour, cost at volume, and the assembled product's result. Consistency comes from process control and data across units; environmental and ageing performance needs qualification testing; sample pricing is not production pricing.
Q6: How do I choose the measurement points?
Put them inside the band the product uses. For voice that means across the speech band rather than a single conventional frequency; for music it means covering the range the content occupies. Selecting the test points determines what the published figures describe.
Q7: The sample is close but not right. Which parameter do I change?
The one the measurement points at. If the installed resonance is above target, the lever is the cavity or the driver's compliance, not the magnet. If the level in the band is short, the lever is the magnetic circuit or the diaphragm. Changing several at once loses the information the round produced.
Hongsheng can supply sample rounds measured on the acceptance basis that will be used at goods-in, so the last round of a programme and the first incoming inspection describe the same condition.
8. Summary: One Question Per Round
Sample development converges on the number of variables per round, not on the number of rounds. A round that changes one thing and measures the installed condition returns an interpretable answer; a round that changes three returns ambiguity, and the cost of that ambiguity is measured in rounds. The measurement conditions — cavity volume, mounting surface, drive voltage and impedance, and the points at which the response is read — are what make one round comparable with the next, and the test points should sit inside the band the product actually uses.
Equally important is knowing what a sample cannot establish. Production consistency, long-term behaviour, cost at volume and the assembled product's result all lie outside a single build, and a programme that expects a sample to prove them will keep requesting rounds that cannot answer. Hongsheng can build sample rounds against a mock-up of the product's own cavity and front surface, and states the cavity condition alongside each published configuration, so a round's result can be compared against the published basis rather than against a different fixture.
Next step If you are planning sample development for a micro speaker, the shortest route to a configuration worth testing is to state five things: the space available for the driver and its cavity, the impedance the amplifier will drive, the power the rail can supply, the level required at the intended listening position, and the single question the first round has to answer. With those specified, our engineering team can recommend a suitable configuration for evaluation, or state plainly which part of the acoustic design has to change first.
More in This Series
· Custom Micro Speaker Development: From Product Requirements to Mass Production →https://www.hsdz-spk.com/news/588.html
· What to Expect When Ordering Micro Speaker Samples from China → https://www.hsdz-spk.com/news/589.html