From Speaker Sample to Mass Production: A Real OEM Micro Speaker Project
1. What a Sample Round Can and Cannot Settle
Short answer: A sample round settles whether a configuration is in the right region. It cannot settle acceptance, because acceptance depends on a basis that has to be fixed before the numbers mean anything.
A sample round answers a narrow question: is this configuration close enough to be worth developing? It cannot answer the acceptance question, because acceptance depends on the measurement basis — the drive voltage, the distance, the frequency point, and whether the driver is measured free-air or installed. Two rounds measured on different bases produce two numbers that cannot be compared, and a programme that keeps changing its basis will keep reopening decisions it has already made.
There is a second thing a sample round cannot settle, and it is the one that decides consumer programmes: whether the sound is subjectively right. Sensitivity and resonance can both sit inside their targets while the product still sounds wrong, because the two figures say nothing about how the output is distributed across the band the listener actually hears. In this programme the customer had already tried several existing drivers and none of them resolved the complaint, even though the figures on paper were unremarkable.
So the first task in a sample programme is not to build a better part. It is to fix the basis — the reference sound, the cavity volume, the measurement conditions — so that later rounds produce comparable answers.
2. The Listening Session Came Before the Specification
Short answer: The customer's requirement was a subjective match to a reference product. The first useful round was therefore a listening session with a reference cavity and drive electronics, not a datasheet comparison.
The product is a camera-based unit with infant-care and two-way talk functions: it lets a parent talk to a child and notifies them when something needs attention. The customer came from a surveillance-camera background, where the convention is a low-cost driver that is simply audible and loud. For a product aimed at infants the requirement inverts: the sound has to be soft and non-fatiguing, because the listener is a child. The customer had benchmarked a specific reference product and had tried several existing drivers without matching it.
The structural envelope was tight. Only about 30 mm of driver could be fitted, the total thickness budget was 13 mm, and the usable cavity volume was 15 cc. Three construction routes had already been tried — a 20×30 mm bare driver, a 15×11 mm BOX module and a multimedia driver — and none of them resolved the complaint.
The root cause had three parts, and the third is the one most often missed. A cavity had to be designed at all, or the sound would be diffuse. The installed resonance had to come down below the target band. And the bandwidth had to be wide, because a narrow response concentrates energy and is what makes a small driver sound sharp rather than soft. The first two are conventional. The third is the one that changed the construction.
3. Scanning the Cavity Volume in One Round
Short answer: Four curves were captured in a single round — the reference product and three cavity volumes — which established where the returns on volume flatten out.
Rather than send parts back and forth, the trial was run in one round with the volume as the variable. The customer's reference product was measured alongside three cavity volumes using the same driver, so the four curves differ only in the volume behind the driver.
Table 1: The trial round, with the cavity volume as the only variable.
Configuration | What it established | Outcome |
The reference product the customer was benchmarking against | The subjective target every later round is judged against | Set the listening reference for the programme |
30 cc cavity, same driver | The largest volume in the trial set | Close to the reference in the band that matters |
20 cc cavity, same driver | Mid volume | Already close to the 30 cc result — the returns on volume were flattening |
10 cc cavity, same driver | Smallest volume in the trial set | Clearly below the larger volumes; the low end was audibly reduced |
15 cc cavity with the revised driver | A volume below the trial maximum, paired with a construction change | Selected for design freeze |
That last row is the economically important one. Because 20 cc and 30 cc were already close, the largest available volume was not the right answer. A smaller volume — the 15 cc the structure could actually give up — combined with a revised driver construction reached the same region, which is what made the mechanical design acceptable to the customer.
4. Project Case Study — From a Listening Session to Design Freeze
Project Case Study (Hongsheng) — From a Listening Session to Design Freeze
The customer built a camera-based infant-care product with two-way talk, benchmarked against a specific reference unit, and had tried a 20×30 mm bare driver, a 15×11 mm BOX module and a multimedia driver without matching it. The structure allowed about 30 mm of driver, a 13 mm total thickness and 15 cc of usable cavity. Hongsheng brought a standard cavity and an amplifier-driven version to the customer's site for a listening session, and the improvement was immediately audible — which converted the discussion from part selection to specification. A 28 mm driver with a resonance of 220 Hz free-air was then fitted into the 15 cc cavity, giving an installed resonance of 380 Hz against a recorded specification of 400 Hz ±15%. Hongsheng proposed changing the dome from PET to aluminium, which raises its rigidity and widened the response from 8 kHz to 18 kHz as recorded in the project file; that change is what addresses the sharpness rather than the level. The driver was also made as an integrated unit, which reduced both the structural space and the cost, and the customer designed the housing around the driver rather than the other way round, so no customer tooling was required. The delivered configuration measured 100 ± 3 dB at 2.83 Vrms, 10 cm and 2 kHz, with a 2.0 W rating and a 2.5 W maximum, at a tax-inclusive unit price of 4.0 RMB. The project reached design freeze with the customer confirming the result met the requirement.
One-line conclusion: a sample programme converges when the basis is fixed first — the reference sound, the volume, the measurement conditions — and the part is chosen against that basis rather than before it.
5. What the Physics Predicted, and What the Scan Confirmed
Short answer: The two recorded resonances — 220 Hz free-air and 380 Hz installed in 15 cc — are consistent with an enclosed volume of roughly 30 cc equivalent compliance, and the same relationship explains why the 10 cc curve fell away.
Installing a driver in a sealed cavity raises the system resonance, because the trapped air acts as a spring in series with the suspension. The relationship is that the installed resonance equals the driver's own resonance multiplied by the square root of one plus the ratio of the driver's compliance-equivalent volume to the cavity volume. Using the two recorded figures — 220 Hz free-air and 380 Hz installed in 15 cc — the implied compliance-equivalent volume is roughly 30 cc, which is a normal figure for a 28 mm driver.
The same relationship predicts the volume scan. Holding the driver fixed and shrinking the volume raises the installed resonance, and the effect accelerates as the volume gets smaller relative to the compliance-equivalent volume. That is exactly the shape of the trial: 30 cc and 20 cc sit close together, while 10 cc is clearly lower, because at that point the cavity is no longer large compared with the driver's own compliance. It is also why a volume below the trial maximum could be chosen — the curve had already flattened before the largest volume was reached.
For programmes where acceptance is subjective, Hongsheng can bring a reference cavity and drive electronics to a listening session and record the volume scan alongside it, so that the volume decision is made against evidence rather than against the largest space available.
6. What Gets Frozen at Design Freeze
Short answer: Six items, and only one of them is the part number. The other five are the basis that makes the part number mean something.
Table 2 lists what has to be frozen before a programme can move to production. The pattern is that the part is the last item, not the first.
Table 2: What a design freeze has to fix, and what happens if it is left open.
What is frozen | Why it has to be fixed | What happens if it is left open |
The acoustic target — band, level and resonance | It is the acceptance basis for every later round | Later rounds optimise against a moving target |
The measurement basis — voltage, distance, frequency point, free-air or installed | Without it two results cannot be compared | Sample-to-sample differences become unreadable |
The cavity volume and how it is sealed | The installed resonance depends on it | A silent volume change moves the resonance out of band |
The driver construction — diaphragm, dome and materials | It sets the bandwidth and the low end | A material substitution changes the sound without changing any figure |
The structural match — integrated unit or separate driver and cavity | It determines the space and cost the product carries | A late change reopens the mechanical design |
The cost basis | It is what the customer approves commercially | A cost change reopens the whole decision |
Two of these deserve emphasis in this programme. The dome material is a bandwidth decision disguised as a detail: changing it from PET to aluminium is what widened the response and addressed the sharpness, and it changes no figure that a specification usually lists. The structural match is the other: making the driver an integrated unit reduced both the space the product had to give up and its cost, which is what made the mechanical design acceptable.
7. What to Confirm Before the First Production Order
Short answer: Six confirmations turn a design freeze into a production order — starting with the measurement basis and the volume that was actually approved.
1. The measurement basis recorded alongside every accepted figure: drive voltage or power, distance, frequency point, and free-air or installed.
2. The cavity volume that was approved, and how it is sealed — the installed resonance depends on both.
3. Which construction details are acoustic decisions, such as dome and diaphragm materials, so that they are covered by change control.
4. Whether the driver is an integrated unit or a separate driver and cavity, and what that implies for the space and cost the product carries.
5. The acceptance route: whether the customer judges subjectively, instrumentally, or both, and what the subjective reference is.
6. The unit cost at the approved quantity, and which elements of it would move if the volume or the construction changed.
8. FAQ: From Sample to Mass Production
Q1: Why run a listening session instead of sending samples?
Because the requirement was a subjective match to a reference product, and no instrument reading substitutes for that judgement. Running the session with a reference cavity and drive electronics settled in one afternoon what several sample exchanges had not.
Q2: Why scan the cavity volume instead of using the largest volume available?
Because the scan showed the returns flattening: 30 cc and 20 cc were already close while 10 cc was clearly lower. That made a smaller volume defensible, which is what allowed the mechanical design to work.
Q3: How does cavity volume affect the installed resonance?
The trapped air acts as a spring in series with the suspension, so a smaller volume raises the installed resonance. Using the two recorded figures — 220 Hz free-air and 380 Hz installed in 15 cc — the implied compliance-equivalent volume is roughly 30 cc.
Q4: Why did the dome material matter more than the level?
Because the complaint was sharpness rather than quietness. Widening the response by changing the dome from PET to aluminium distributes the energy that was previously concentrated, which is what makes a small driver sound soft. It changes no figure a specification usually lists.
Q5: What does an integrated driver unit change?
It reduces the structural space the product has to give up and lowers the cost, because the cavity is part of the supplied assembly rather than something the housing has to form. It also removes a sealing variable from the customer's design.
Q6: Did the customer have to change their tooling?
No. The customer designed the housing around the driver rather than the other way round, so the programme reached design freeze without a customer tooling change.
Q7: What result did the programme deliver?
100 ± 3 dB at 2.83 Vrms, 10 cm and 2 kHz, an installed resonance of 380 Hz against a recorded specification of 400 Hz ±15%, a 2.0 W rating with a 2.5 W maximum, at a tax-inclusive unit price of 4.0 RMB. The project reached design freeze with the customer confirming the requirement was met.
9. Summary: Freeze the Basis, Not Just the Part
The programme is worth recording because the decisive work happened before the part was chosen. A listening session established what the sound had to be; a single-round volume scan established how much cavity the product actually needed; and a construction change — the dome material — addressed the complaint that no figure in the specification described. Only then was the part number settled.
The transferable rule is that a design freeze has to fix the basis as well as the part. The acoustic target, the measurement conditions, the cavity volume and its sealing, the construction details that carry acoustic meaning, the structural match and the cost basis — those six items are what make a part number mean something in production. A programme that freezes the part and leaves the basis open will keep re-opening the same discussion, and the sample count will grow without the answer changing.
Next step If you are evaluating a micro speaker for a consumer monitoring product, 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 target level at the intended listening position, and the reference sound the product is being judged against. 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/594.html
· Micro Speaker Sample Development: What OEM Buyers Should Expect → https://www.hsdz-spk.com/news/595.html