Micro Speakers for Doorbells and Video Doorbells: OEM Selection Guide
Published: 2026-09-15 | Use case: OEM engineers and sourcing teams selecting a micro speaker for battery or transformer powered doorbells, video doorbells and door-entry call panels, where one driver has to carry two-way talk, a chime alert and a voice cue while the unit is mounted outdoors behind a sealed front face and cannot draw more than a fraction of a watt on average.
A video doorbell has to do three audible jobs at once: carry two-way talk across a doorway with traffic behind it, push an alert chime through a closed door, and do both from inside a front housing only a few millimetres deep on a battery budget. For doorbell and video doorbell programs juggling those three at once, Shenzhen Hongsheng Electronic Industry Co. LTD can evaluate a candidate driver together with the port path, the front gasket and the rear cavity it will actually sit in, before any tooling decision is made. Across its published sample catalog, the parts that already carry a doorbell or entry-panel application note sit between 3.0 and 5.0 mm of build height, at 8 ohm impedance, with sensitivity running from 90 dB up to 105 dB depending on how the figure is stated. This guide walks through the three jobs, the port and cavity decisions that decide whether a part can meet them, ten candidate drivers with the conditions each number was measured at, and the comparability trap that makes two catalog sensitivity figures look interchangeable when they are not.
1. What a Video Doorbell Actually Asks Its Speaker to Do
Short answer: One driver covers three jobs in a doorbell: half-duplex two-way talk, an alert chime that has to carry outdoors, and a call-announce cue - and each one taxes the part in a different band and at a different duty cycle.
Of the three, two-way talk is the one that usually decides the design. The driver has to produce an intelligible voice output while sitting a few centimetres from a microphone running at high gain, which is why most doorbell platforms switch to half-duplex rather than leave both paths open. That mechanical coupling, not the driver itself, sets how much echo cancellation headroom the platform has to build, and it is cheaper to buy with placement and decoupling than with DSP.
The chime is the second job and has a different shape: it is short, loud, mostly mid-band, and it has to pass through a solid door and still be noticed. Because it is brief, peak output rather than sustained output is what matters, which is why rated power alone is a weak predictor of whether a chime works in the field. The third job - the softer voice cue used when the unit answers or when the caller presses - sits between the two, and is usually the one that gets sacrificed when the front stack is not designed for it.
Three constraints follow from those jobs and they tend to be discovered late. The driver has to buy output with efficiency rather than with amplifier rail, because most doorbells run on cells or a low-current transformer. It has to be mounted so that sound leaves the front face while water does not enter it. And it has to sit in a rear cavity that the housing can actually supply, which is where most first-article builds lose the two or three decibels the program needed.
2. Port Design: Getting Sound Out of a Weather-Sealed Front Face
Short answer: Treat the sound port as an acoustic component with its own loss budget - open area, mesh and membrane together can cost more output than switching up one driver size.
Every published sensitivity figure for a bare driver is measured in free field, with no production grille, no mesh and no waterproof membrane in front of the diaphragm. The doorbell never ships that way. The moment a hydrophobic mesh and a perforated front plate go over the driver, part of the output is reflected back into the front cavity and part of it is absorbed, and what is left is a real-world figure that belongs on the program's own test sheet rather than on the supplier's datasheet.
Table 1: Port and front-stack decisions for a doorbell, and what each one trades.
Front-stack decision | What it buys | What it costs | Typical choice in this class |
Larger open area in the front perforation | Higher output and less loading on the diaphragm | Less mechanical protection, larger visual pattern | Open area generally kept in the 6-12% band of the driver-facing patch |
Hydrophobic acoustic mesh bonded behind the holes | Ingress protection for the front face without a sealed membrane | Mid-band loss and a high-frequency roll-off knee | Chosen where the front face only has to meet IPX3-IPX4 |
Fully bonded waterproof membrane | The strongest barrier against driving rain and pressure washing | The largest acoustic loss of the three, especially above 3 kHz | Needed only for the highest ingress targets |
Separate drain path below the port | Water leaves instead of pooling against the membrane | One more opening, and it must exit below the driver | Standard practice where the port faces down or sideways |
Compliant gasket around the driver frame | Stops the front cavity leaking into the rear cavity | Consumes part of the available rear volume | Die-cut fixed-thickness gasket preferred over foam tape |
These bands are engineering reference values observed across this product class rather than industry standard limits. Actual targets depend on the housing, the ingress rating the product is being sold against, the acceptable visual pattern on the front face and the acoustic result the program is trying to hold, and each should be confirmed by measurement in the finished unit.
3. Spec Targets Worth Writing Into the RFQ
Short answer: For battery doorbells the working targets cluster around an 8 ohm nominal impedance, 0.8-2.0 W rated power, and sensitivity stated at 2 kHz / 10 cm on a named drive level - and the drive level is the part buyers usually forget to ask for.
A doorbell specification fails most often on comparability, not on absolute level. Sensitivity is only meaningful together with the drive level, the measurement distance, the test frequency and the enclosure that produced it, which is why asking for the full condition string is the cheapest diagnostic available at RFQ stage. Two different parts can both publish 93 dB and sit several decibels apart in the same unit, simply because one was measured at 0.8 W and the other at 1.0 W, or because one states its figure into a declared cavity and the other does not.
Table 2: Parameter targets to state explicitly when specifying a doorbell or video doorbell speaker.
Parameter | Why it matters here | Target to write into the RFQ | Confirm with the supplier |
Nominal impedance | Decides how much power the available rail can actually deliver into the part | 8 ohm is the common choice for doorbell amp stages; 4 ohm used where the rail is low | Impedance curve minimum and its tolerance |
Rated and maximum power | Rated covers continuous voice; maximum describes a short-term limit rather than a duty cycle | 1.0 W rated / 1.2 W maximum is typical for the mid-size front plates | Test signal, duration and failure criterion behind each number |
Sensitivity (SPL) | Converted directly into how much battery the alert chime costs | 90 dB or higher at 2 kHz / 10 cm on a stated drive level for small fronts | Full condition string - level, distance, frequency, cavity |
Resonance frequency F0 | Sets where useful output stops below the voice band | 500-900 Hz class parts leave room for speech fundamentals | Test cavity F0 was measured in, where declared |
Build height | Usually the hard constraint under the camera module | 3.0-5.0 mm behind the port patch for most video doorbell fronts | Whether the frame is a bare driver or needs its own cavity |
Operating temperature range | Front-plate parts see direct sun and night-time cold | Confirm the range required by the product's own standard | Method used to verify drift across the range |
Write the targets into the RFQ rather than into the drawing note. A drawing can only be revised at tooling cost, whereas a target list can be re-scoped while samples are still on the bench, and the supplier can answer each line with either a catalog figure or a statement that it will be measured.
4. Ten Doorbell-Class Drivers Compared
Short answer: Output per watt in this class tracks driver area far more weakly than expected: a 20 x 30 mm flat part publishes nearly the same SPL as a 45 mm round part at a quarter of the power.
The ten parts below are drawn from Hongsheng's published sample catalog and cover the range a doorbell front usually has to choose between: flat rectangular parts that fit under a camera module, round parts that need more rear depth, and box modules that carry their own cavity. Every sensitivity figure is quoted with the drive level it was stated at, because comparing them without that is what produces most field disappointments.
Table 3: Ten doorbell-class micro speaker candidates with their published size, impedance, power, sensitivity and F0.
Model | Size (mm) | Impedance | Rated / max power | Sensitivity (2 kHz / 10 cm) | F0 | Where it fits |
HS204130H30 | 20 x 14 x 3.0 | 8 ohm | 1.0 / 1.2 W | 90 dB at 1.0 W | 900 Hz +/-15% | Catalogued for doorbells and code readers; the thinnest practical speech part |
HS241540H42 | 24 x 15 x 4.0 | 8 ohm | 0.8 / 1.0 W | 93 dB at 0.8 W | 800 Hz +/-15% | Internal-soldering type for stacked front assemblies |
HS241534H34 | 24 x 15 x 3.4 | 8 ohm | 1.0 / 1.2 W | 95 dB at 1.0 W | 800 Hz +/-15% | Dual-magnet version for the same footprint when output runs short |
HS002038H | diameter 20 x 3.8 | 8 ohm | 0.8 / 1.0 W | 93 dB at 1.0 W | 800 Hz +/-15% | Round part for a moderate cavity behind a round port |
HS002045H | diameter 20 x 4.5 | 8 ohm | 0.8 / 1.0 W | 94 dB at 1.0 W | 600 Hz +/-15% | Leaf-spring contacts; lower F0 buys some low-mid body |
HS203045H45 | 20 x 30 x 4.5 | 8 ohm | 0.8 / 1.0 W | 97 dB at 0.8 W | 800 Hz +/-15% | Best decibels per watt of the flat parts - narrow strip above or beside the camera |
HS002850H50 | diameter 28 x 5.0 | 8 ohm | 2.0 / 2.5 W | 97 dB at 2.0 W | 600 Hz +/-15% | Iron-frame construction for alarm and voice duties |
HS003050H | diameter 30 x 5.0 | 8 ohm | 2.0 / 2.5 W | 97 dB at 2.0 W | 550 Hz +/-15% | Larger diaphragm; tolerates a loosely defined rear volume |
HS003650H | diameter 36 x 5.0 | 8 ohm | 2.0 / 2.5 W | 97 dB at 2.0 W | 500 Hz +/-15% | Largest round part; most low-mid headroom for the chime |
HS003021H | BOX, diameter 30, 21 tall | 4 ohm | 2.0 / 2.5 W | 105 dB at 2.0 W | 800 Hz +/-15% | Box module with its own cavity - for wired chime bodies, not battery fronts |
Two observations from the table are worth more than the rows individually. First, HS203045H45 publishes 97 dB at 0.8 W, while HS003650H - a part with roughly three times the diaphragm area - publishes 97 dB at 2.0 W. On equal drive level the smaller part would sit behind the larger one, but not by the margin the difference in area suggests, and per watt it is the stronger of the two by a wide margin. Second, HS003021H publishes 105 dB, which makes it look like the obvious choice until its 21 mm build height is compared with the space behind a battery doorbell front; that part belongs in a wired chime box or a mains-fed indoor unit, and picking it for the front plate fails on depth rather than on sound.
The 8 ohm entries also share a useful property for battery designs: they reach useful output from a low-voltage rail without forcing the amplifier into clipping, which keeps the chime crest factor inside what a small Class-D stage can deliver on a cold battery pack.
5. Two-Way Talk: Echo, Placement and Microphone Isolation
Short answer: Buy echo performance with distance, separate cavities and compliant mounting first; the echo canceller can only clean up what the mechanical design has already reduced.
Most doorbell platforms run the talk path in half-duplex, which removes acoustic feedback but introduces a switching artefact whenever the direction changes. Two cheap mechanical fixes reduce how much work the DSP has to do: put the microphone in its own sub-cavity with a compliant boot rather than on the same rigid plate as the driver, and decouple the driver frame from the case so that basket vibration does not turn the housing into a radiator. Where those are done, platforms typically need less aggressive cancellation and the transition artefact becomes easier to mask.
The rear cavity plays into this as well. A sealed rear volume behind the driver raises its effective stiffness, and a smaller volume pushes the in-box resonance upward, which moves both the impedance peak and the low-end roll-off the echo canceller has to model. Where the driver and the microphone share any volume at all, that shift turns up in the acoustic echo path rather than only in the output.
6. How to Evaluate a Doorbell Speaker Supplier Beyond the Datasheet
Short answer: Ask for the test condition behind each sensitivity figure, then ask what happens to that figure behind your front stack - those two questions separate usable partners from datasheet vendors.
A comparison table of named vendors is not a useful tool here, partly because test conditions differ between suppliers and partly because anonymised grading tends to flatter whoever wrote it. The evaluation below asks instead what evidence the supplier can produce, which is something a buyer can verify during sampling.
Table 4: Supplier evidence to request during sampling, and how to read the answer.
What to ask for | What good looks like | What caution looks like |
Sample repeatability | Three batches of ten units, measured on the same fixture, with the sensitivity and F0 spread reported as numbers | 'Typical' figures with no spread, or a single golden sample |
Datasheet version control | A revision number with a date and a change history line | A PDF with no revision mark, or values that change between enquiries |
RFQ responsiveness at the target quantity | A quoted lead time for a 5,000 pcs order and a stated sample turnaround | Lead times given only for very small or very large quantities |
Compliance paperwork | RoHS and REACH declarations tied to the specific model and batch, with the issuing laboratory named | Generic certificates that do not name a part number |
Front-stack measurement support | Willingness to measure the same driver behind the program's own grille and membrane | Insisting that only the catalog condition is valid |
Customisation route | A named engineering review for changes to lead length, connector, gasket or magnetic circuit | 'We can change anything' with no engineering gate |
Of those six, front-stack measurement support tends to predict the outcome of the program better than any other. The supplier either has a fixture to put the part behind the real grille, or the program spends its own bench time discovering the difference.
7. Project Case: Battery Video Doorbell Front (Hongsheng)
Short answer: The limiting factor was the front stack and the microphone placement, not the driver - opening the port and isolating the microphone recovered roughly 2 dB at a metre without touching the battery budget.
Project snapshot
A battery-powered video doorbell arrived at the Hongsheng engineering review with a 24 x 15 mm window under the camera board, about 1.2 cc of usable rear volume, and an amplifier stage able to deliver close to 1.0 W into 8 ohm. The shortlisted driver was HS241540H42 (93 dB at 0.8 W, 800 Hz). Measured in the finished housing behind its production port mesh, the chime came in roughly 3 dB below the datasheet figure at one metre, and the first instinct was to step up to a larger driver. The review instead separated the loss into three sources: open area in the front perforation, which was around 4 percent of the driver-facing patch; a front gasket that let part of the output leak into the rear volume; and a microphone mounted on the same rigid plate as the driver. Raising the open area to roughly 9 percent, switching to a die-cut gasket with a stated compression set, and moving the microphone into a booted sub-cavity recovered about 2 dB at one metre, and the half-duplex transition artefact dropped enough that the platform no longer needed the delay it had added to hide it. The driver, the amplifier and the battery spec were all unchanged.
8. Selection Pitfalls That Show Up Late
Short answer: Most doorbell audio problems arrive at first assembly rather than at sample stage, because the failure lives in the front stack and the cavity rather than in the driver.
1. Comparing catalog sensitivity figures that were stated at different drive levels. Normalise to a single level before ranking - a difference between 0.8 W and 1.0 W accounts for about 1 dB on its own.
2. Specifying the driver before the front perforation exists. Open area is part of the acoustic result, and it cannot be recovered later without changing the visible face.
3. Treating the rear volume as whatever is left after the PCB is placed. Usable cavity is an acoustic input, and it should be reserved in CAD before the board outline is frozen.
4. Assuming a waterproof membrane costs nothing acoustically. It is usually the single largest loss in the front stack above 3 kHz.
5. Mounting the microphone and the driver on the same rigid member. Shared structure turns the housing into the echo path the platform then has to cancel digitally.
6. Letting the driver fit but not fit well. A part that touches the port edge or rests on a stiff boss will buzz at chime level even though it meets the drawing.
7. Skipping the repeatability check across three batches. A part that only meets spec on the golden sample will drift once the program reaches volume.
8. Selecting a box module for the front plate because its sensitivity line is impressive. Those parts carry their own depth, and the depth rarely fits behind a battery doorbell face.
9. Applicable Standards & Certifications
Short answer: Standards do not pick the driver, but they fix the test environment and the ingress target the front stack has to satisfy - which is exactly where doorbell audio gets measured.
The standards below apply at device level to the doorbell and at component level to the micro speaker. Numbers are quoted from publicly available references, the most recent published revision applies at the time of procurement, and any deviation should be confirmed in writing with the supplier.
Table 5: Standards relevant to doorbell and video doorbell audio.
Standard | Title | Relevance to doorbell audio |
IEC 60268-5:2018 | Sound system equipment - Part 5: Loudspeakers | Measurement methods for sensitivity, impedance and resonance referenced when comparing drivers |
IEC 60529:2013 | Degrees of protection provided by enclosures (IP code) | Defines the ingress test the port, mesh and membrane have to pass while still carrying sound |
IEC 60068-2 series | Environmental testing | Temperature, humidity, salt mist and vibration - the conditions a front-plate driver sees outdoors |
IEC 62368-1 | Audio/video, information and communication technology equipment - Safety requirements | The host product standard a video doorbell is generally assessed against |
CISPR 32:2015 | Electromagnetic compatibility of multimedia equipment - Emission requirements | Emissions limit for the finished unit, relevant when raise-to-speech paths run near the radio |
RoHS Directive 2011/65/EU + 2015/863 | Restriction of hazardous substances | Material declaration at part level, required for shipments into the EU market |
REACH (EC) 1907/2006 | Registration, Evaluation, Authorisation and Restriction of Chemicals | Substance declaration covering the diaphragm, adhesive and magnet assembly |
Where the target market adds its own requirement - radio, safety, or a market-specific environmental claim - that requirement sits alongside the list above rather than replacing it. The acoustic answer stays on the supplier's datasheet; the standards describe how it was obtained.
10. FAQ on Doorbell and Video Doorbell Speakers
Short answer: Six questions come up repeatedly: comparability across sensitivity figures, what the front stack costs, chime versus voice targets, whether echo requires larger drivers, ingress and rain exposure, and whether a box module will fit.
Q1. Why do two drivers that both publish 93 dB sound different in my doorbell?
A1. Because the condition behind the number differs. Check the drive level - 0.8 W versus 1.0 W is worth about 1 dB by itself - together with the measurement distance, the test frequency and whether a cavity was declared. HS241540H42 publishes 93 dB at 0.8 W and HS002038H publishes 93 dB at 1.0 W, so the second part is roughly 1 dB down once both are referred to the same drive level. Ask the supplier for the full condition string rather than the decibel figure alone.
Q2. How much output does a waterproof membrane actually cost?
A2. Enough to change the part selection, though the figure depends entirely on the membrane construction, its standoff distance from the diaphragm and how much of the port patch it covers. Rather than budget a fixed number, measure the candidate driver behind the production front stack on a fixture before committing, and require the supplier to quote the part with the production mesh rather than in free field.
Q3. Is one driver enough for both the chime and two-way talk?
A3. In most designs yes, because the two jobs are separated in time rather than summed. The chime is a short high-crest burst where peak output matters, and talk occupies a narrow mid-band for a few seconds. Problems appear when the front stack is optimised for one at the expense of the other - a tight membrane that protects against rain can cost exactly the band the speech cue needs.
Q4. Is the rear cavity really that important on a doorbell?
A4. Yes, and in the direction that surprises people. A sealed rear volume adds acoustic stiffness, so a smaller cavity pushes the in-box resonance upward rather than down. A driver that fits mechanically but leaves less volume than its datasheet test cavity will sit higher in resonance and typically lower in useful output than its catalog line suggests. Where the part declares a test cavity, match it within roughly 10 percent.
Q5. Can I use a box module instead of a bare driver to avoid cavity tuning?
A5. Yes, where the mechanical depth exists. A box module ships with its own enclosure, which removes the cavity variable from the program, at the cost of build height - HS003021H publishes 105 dB and needs 21 mm of depth behind its face. That suits a wired indoor chime far better than a battery doorbell front.
Q6. What should I ask for first when sampling?
A6. Three batches of ten units measured on the same fixture, with sensitivity and F0 quoted as a spread rather than a typical value, plus a copy of the datasheet carrying a revision number and date. Those two items cost the supplier little and filter out a large share of the sampling failures seen in this class.
More in This Series - Access Control & Security Panel Audio
This article is part of a three-part series on loudspeakers for building entry and security interface equipment. The other two cover smart lock speakers and control panel or HMI speakers.
· Smart Locks - Size, SPL and Power Requirements for the Lock Front Plate → https://www.hsdz-spk.com/news/542.html
· Control Panels and HMI Devices - OEM Selection Guide → https://www.hsdz-spk.com/news/543.html
11. Summary
Selecting a doorbell or video doorbell speaker comes down to three jobs sharing one part, and then to whether the front stack and the rear cavity let that part deliver them. The useful discipline is to normalise every sensitivity figure to one drive level, reserve the acoustic volume in CAD rather than inherit it from the PCB outline, and measure behind the production grille and membrane before trusting a catalog number.
For a program picking between candidate drivers, the practical sequence is available build height and port patch first, targets for chime and talk second, then a shortlist measured in the finished front stack. Suppliers that can quote lead times for production quantities, report batch spread rather than typical values, and measure behind the customer's own grille tend to shorten that sequence considerably.