Adding an optimized passive radiator to a micro speaker box raised its output by 5 dB at 200 Hz compared with the same box without one (Jiang et al., Archives of Acoustics 2017). That trick is behind the rubber-capped “bass radiators” at the ends of many sealed, waterproof Bluetooth speakers. The cost is excursion (how far a cone travels from its rest position): in a worked example the tightest distortion limit stopped a woofer at 2.28 mm, while the loosest would have allowed 15.9 mm. Lab reviews show what happens past that point: Rtings measured 5.98 dB of compression at max volume on the first-generation Marshall Emberton, and SoundGuys clocked 6 hours 16 minutes of real playback on a speaker rated for 14. We aggregated data from Klippel GmbH engineering notes, Archives of Acoustics, Rtings, SoundGuys, DXOMARK, an IEC 60529 test lab and manufacturer spec sheets (labelled as vendor claims). For the market side of the same product category, see our portable speaker market statistics.
TL;DR
- An optimized passive radiator raised micro speaker-box output by 5 dB at 200 Hz (Jiang et al., Archives of Acoustics 2017).
- A retail 2.5-inch passive radiator is rated at 73 Hz resonance with 3.2 g of moving mass (Dayton Audio ND65-PR page, vendor claim).
- Usable excursion on a phone micro-speaker rose from 106 µm under a fixed linear model to 344 µm with adaptive nonlinear control (KLIPPEL, Active Transducer Protection Part 1).
- A distortion compensator changed micro-speaker output level by only -0.1 to 0.4 dB (Hu, Lu and Qiu, UTS repository 2016).
- Most speakers Rtings tested reach 80 to 100 dB SPL at max volume (Rtings dynamics test 2021).
- Compression at max volume ranged from 0.39 dB (UE HYPERBOOM) to 5.98 dB (Marshall Emberton) (Rtings dynamics test 2021).
- The JBL Charge 5 scored 46 for volume, against 141 for the best speaker in its comparison set (DXOMARK 2021).
- A 1.8 kg Marshall Middleton claims 87 dB SPL, one decibel less than the 0.36 kg Willen II (Marshall spec pages, vendor claims).
- IPX7 requires surviving 1 m of water for 30 minutes (DLS Electronic Systems, IEC 60529 immersion testing).
- The JBL Flip 7 lasted 6 hours 16 minutes at 80 dB, versus a 14-hour rating (SoundGuys 2025).
- A single-cell smart amp delivers 6.1 W at 3.6 V into 4 ohms, using an 11.5 V boost (Texas Instruments TAS2563 page, vendor claim).
1. The Excursion Ceiling: How Xmax Is Defined and Measured
Xmax is usually read as a hard mechanical stop. It is not: it is a distortion budget. Both the older AES2-1984 wording and the newer IEC 62458 defaults set the line at roughly 10%, but they measure different symptoms, which is why two spec sheets quoting “Xmax” for similar drivers are often not comparable. The force factor (Bl, the motor’s push per ampere) and the compliance (Cms, how easily the suspension gives) both sag as the coil moves outward, and whichever sags first sets the ceiling.
The worked example in the 2026 revision of the displacement-limit note is the most useful number in this section. The inductance limit capped the woofer at 2.28 mm, while the Doppler limit alone would have allowed 15.9 mm, almost seven times more (15.9 ÷ 2.28 = 6.97, our arithmetic). In other words, a cone can usually travel much further than it can travel cleanly. For a waterproof speaker that cannot use a port to relieve the driver, that clean-travel number is the real bass budget.
| Metric | Value | Source |
|---|---|---|
| Xmax criterion in the older AES2-1984 standard: peak voice-coil displacement at which motor linearity deviates by | 10% | KLIPPEL Application Note AN 4, updated April 4, 2012, quoting AES2-1984 |
| IEC 62458 default displacement limits: force factor falls to (share of rest value) / compliance falls to (share of rest value) | 82% (about 10% modulation distortion) / 75% (about 10% harmonic distortion) | KLIPPEL Application Note AN 5, revision 1.2, May 22, 2026 |
| Example woofer with 57 Hz resonance: tightest limit (inductance) vs loosest limit (Doppler) | 2.28 mm vs 15.9 mm peak | KLIPPEL Application Note AN 5, May 22, 2026 |
| Worked Xmax test on a 70 Hz driver, two tones at resonance and 8.5 x resonance in a 4:1 amplitude ratio | 2.66 mm (1.9 mm rms x 1.4), reached at 4.65 Vrms when 3rd-order intermodulation hit 10% | KLIPPEL Application Note AN 4, April 4, 2012 |
| Retail 2.5-inch passive radiator: rated linear excursion / peak-to-peak travel | 5 mm / 10 mm | Dayton Audio ND65-PR product page (vendor’s own claim) |
Context note: the 2012 application note is the most recent available version of that worked example. The 10% threshold marks the end of the range the standard treats as usable, not a recommended everyday operating point.
2. Passive Radiators: Bass From a Sealed Box
A passive radiator is a cone with no magnet or voice coil. The air trapped inside the cabinet pushes it, and its mass and suspension tune the box the way a port would. The reason it suits waterproof designs is plain: a port is an open tube into the enclosure, while a passive radiator is a sealed membrane. The only peer-reviewed micro speaker-box measurement we found put the gain at 5 dB at 200 Hz, after the radiator’s mass, stiffness and size were optimized with finite-element modelling.
The vendor numbers show the scale involved. A 2.5-inch radiator with 15.6 cm² of cone area and 5 mm of linear travel moves 7.8 cm³ of air (15.6 cm² x 0.5 cm, our arithmetic), which matches the 0.008 L displacement volume on its spec page. The same vendor advises that radiators should displace at least twice as much air as the active driver they support, which helps explain why all three Marshall models cited in this roundup carry two of them. Size still wins: Marshall’s own spec pages claim a 75 Hz floor for the 0.36 kg Willen II and 50 Hz for the 1.8 kg Middleton, five times heavier.
| Metric | Value | Source |
|---|---|---|
| Output gain from adding an optimized passive radiator to a micro speaker box, vs the same box without one | +5 dB at 200 Hz | Jiang, Kwon, Kim and Hwang, Archives of Acoustics 42(4), 2017 |
| 2.5-inch passive radiator: resonance / moving mass / displacement volume | 73 Hz / 3.2 g / 0.008 L | Dayton Audio ND65-PR product page (vendor’s own claim) |
| Claimed low-frequency limit: 0.36 kg speaker with one 2-inch driver and 2 radiators vs 1.8 kg speaker with two 3-inch woofers and 2 radiators | 75 Hz vs 50 Hz | Marshall Willen II and Middleton spec pages (vendor’s own claims) |
| Lab verdict on JBL Charge 5 bass, published August 5, 2021 | Lacking extreme low-end extension | DXOMARK JBL Charge 5 speaker test |
| Lab verdict on JBL Flip 7 bass, test bench 1.0.1, updated July 28, 2026 | Bass lacks weight and punch, and compresses near max volume | Rtings JBL Flip 7 review |
| Method for measuring a passive radiator without removing a motor: plane-wave tube, two-load technique (5 Brigham Young University authors, November 7, 2025) | Diaphragm-assembly masses showed favorable agreement with destructive measurements | Anderson et al., Journal of the Acoustical Society of America 158(5), 2025 |
Outlier note: the 2025 JASA paper describes passive radiators as notoriously difficult to characterize, because the usual electrical-impedance tests need a motor. That helps explain why independent passive-radiator data is so thin, and why the Jiang study (most recent available data: Archives of Acoustics, 2017) remains the main published benchmark.
3. Large-Signal Distortion in Small Drivers
“Large signal” means drive levels high enough that the driver stops behaving like a linear spring and motor. The best open measurements come from phone-size micro-speakers, which are smaller than the 2-inch and 3-inch drivers in portable speakers but obey the same physics. The striking finding is that air itself becomes a distortion source: nonlinear damping from air flowing through the gaps accounted for 45% of peak distortion in air, but only 6% when the same driver was measured in a vacuum.
That matters for small sealed drivers: the same study traced more than half of the driver’s mechanical resistance to viscous air flow in the gap and leaks. It also saw the fundamental compress by 6 dB as drive rose from 0.1 V to 2 V, and the total quality factor (the sharpness of the resonance peak) halve from about 2 to 1. A separate Nanjing University and University of Technology Sydney team showed the fix does not have to cost loudness: their compensator changed output level by only a fraction of a decibel. For the silicon end of the same size class, see our MEMS micro-speaker bandwidth statistics.
| Metric | Value | Source |
|---|---|---|
| Phone micro-speaker in a 2012 large-signal study: outer size / radiating area / resonance | 11 x 15 x 4 mm / about 1 cm² / about 700 Hz | KLIPPEL, Modeling the Large Signal Behavior of Micro-Speakers, 2012 |
| Compression of the fundamental at 700 Hz when drive rose from 0.1 V to 2 V (total quality factor fell from about 2 to 1) | 6 dB | KLIPPEL, Modeling the Large Signal Behavior of Micro-Speakers, 2012 |
| Share of peak distortion caused by nonlinear damping, measured in air vs in a vacuum | 45% vs 6% | KLIPPEL, Modeling the Large Signal Behavior of Micro-Speakers, 2012 |
| Closed-box micro-speaker in a compensation study: size / resonance / units tested | 0.9 x 1.6 cm² / 840 Hz / 4 units | Hu, Lu and Qiu, Compensating the distortion of micro-speakers in a closed box, UTS repository manuscript, 2016 |
| Change in output level (power spectral density) after the distortion compensator was switched on | -0.1 dB to 0.4 dB | Hu, Lu and Qiu, UTS repository manuscript, 2016 |
| Perceptual distortion score from the HEAD acoustics MDAQS model: JBL Flip 7 vs JBL Flip 6 | 3.5 vs 2.5 | SoundGuys JBL Flip 7 review, updated June 13, 2025 |
Recency note: both micro-speaker studies are the most recent available open data (2012 and 2016). Below its resonance, the 2012 driver’s relative distortion exceeded 40%, which is the region where a small driver is asked to reproduce bass.
4. DSP Bass Limiting, Compression and Maximum Volume
Every compact speaker eventually hits the ceiling from section 1, and what it does next is a DSP decision. Many turn the bass down as volume rises (dynamic range compression), which is how lab reviews end up describing bass that thins out at the top of the dial. Rtings measured 5.98 dB of compression at max volume on the first-generation Marshall Emberton, against 0.39 dB on the UE HYPERBOOM.
How the protection models the driver matters as much as whether it exists. In a smartphone micro-speaker test, protection built on a fixed linear model allowed 106 µm of usable movement in the worst case, while adaptive nonlinear control allowed 344 µm, more than three times as much (344 ÷ 106 = 3.2, our arithmetic); the model-based schemes each add 3 ms of audio delay. Spec sheets hint at the same trade-off from the other side: Marshall’s heaviest model in this comparison claims the lowest maximum level, which suggests the published figure reflects tuning and limiting choices as much as hardware.
| Metric | Value | Source |
|---|---|---|
| Typical maximum volume across speakers tested by Rtings | 80 to 100 dB SPL | Rtings speaker dynamics test, published April 6, 2021 |
| Loudest example cited by Rtings: JBL PartyBox 100 at max volume | Over 104 dB SPL | Rtings speaker dynamics test, 2021 |
| Compression at max volume: UE HYPERBOOM / Marshall Emberton (first generation) / Sonos Five (mostly in the bass) | 0.39 dB / 5.98 dB / 7.3 dB | Rtings speaker dynamics test, 2021 |
| DXOMARK volume score: JBL Charge 5 vs best speaker in the comparison set (Bluesound Pulse Mini 2i) | 46 vs 141 | DXOMARK JBL Charge 5 speaker test, August 5, 2021 |
| Usable excursion on a smartphone micro-speaker (worst case): protection with a fixed linear model vs an adaptive linear model vs adaptive nonlinear control | 106 µm vs 200 µm vs 344 µm | KLIPPEL, Active Transducer Protection Part 1: Mechanical Overload |
| Claimed maximum SPL at 1 m: Willen II (0.36 kg) / Emberton III (0.67 kg) / Middleton (1.8 kg) | 88 dB / 90 dB / 87 dB | Marshall Emberton III, Willen II and Middleton spec pages (vendor’s own claims) |
| Speaker amplifier chip with built-in bass DSP: output per channel at 24 V into 8 ohms, 1% THD+N, plus 3-band dynamic range compression and excursion protection | 30 W | Texas Instruments TAS5825M product page, datasheet revision H, January 9, 2023 (vendor’s own claim) |
Context note: DXOMARK reports that the Charge 5 did not reach the minimum loudness its protocol requires for a full maximum-volume evaluation, and heard only a subtle hint of dynamic compression at max volume. The Rtings and DXOMARK figures are the most recent openly published values (2021); newer Rtings measurements sit behind a membership wall.
5. IP67 and IP68 Sealing Versus the Battery Power Budget
Waterproofing and battery power are the two constraints wrapped around the driver. IP ratings come from IEC 60529: the first digit covers dust (6 means dust tight) and the second covers water. IPX7 means surviving 1 m of water for 30 minutes, and IPX8 only means “deeper than that”, with depth and time set by the manufacturer. A sealed cabinet rules out an open port, which pushes designers back to the passive radiators in section 2.
The battery side is about voltage as much as capacity. Single-cell amplifier ratings are quoted at 3.6 V, and a bridged amplifier on that rail can swing at most about (3.6 V)² ÷ (2 x 4 ohms) = 1.6 W of sine power into 4 ohms (our arithmetic, ideal lossless case). TI’s TAS2563, rated at 6.1 W from 3.6 V, gets there by boosting its output-stage supply to 11.5 V, and every extra watt comes out of runtime. Runtime claims deserve the same scepticism as loudness claims: the measured Flip 7 runtime of 6.27 hours was about 45% of its 14-hour rating (6.27 ÷ 14, our arithmetic), although the two figures were taken at different volumes. For how those cells age, see our battery degradation statistics and headphone battery capacity fade data.
| Metric | Value | Source |
|---|---|---|
| IPX7 immersion test (IEC 60529) | 1 m of water for 30 minutes | DLS Electronic Systems, IPX7 and IPX8 Immersion Testing |
| IPX8 immersion test (IEC 60529) | Deeper than 1 m; depth and duration set by the manufacturer or a test plan | DLS Electronic Systems, IPX7 and IPX8 Immersion Testing |
| JBL Flip 7 IP68 rating: claimed immersion / claimed drop height onto concrete | 1.5 m for 30 minutes / 1 m (vendor claims reported by SoundGuys) | SoundGuys JBL Flip 7 review, June 13, 2025 |
| JBL Flip 7 battery: measured at a steady 80 dB from 1 m vs rated at moderate volume | 6 h 16 min vs 14 h (16 h with Playtime Boost) | SoundGuys JBL Flip 7 review, June 13, 2025 |
| Rtings verdict on JBL Flip 7 battery life | Falls short of the advertised time | Rtings JBL Flip 7 review, July 28, 2026 |
| Single-cell smart amplifier: output at 3.6 V vs 12 V supply (4 ohms, 1% THD+N) / boost voltage | 6.1 W vs 10 W / 11.5 V | Texas Instruments TAS2563 product page, datasheet revision D, January 17, 2024 (vendor’s own claim) |
| Marshall Emberton III: amplifier / claimed battery life / water rating | 2 x 38 W Class D / 32+ hours / IP67, 1 m for 30 minutes | Marshall Emberton III spec page (vendor’s own claim) |
Context note: Ultimate Ears rates the Wonderboom 4 at IP67, says it floats, drop-tests it from 1.5 m and claims 14 hours of playtime (Ultimate Ears Wonderboom 4 page, vendor’s own claims). We found no independent measurement of how much the waterproof membranes in these speakers reduce output, so no such figure appears here.
Summary: Portable Bluetooth Speaker Passive Radiator and Excursion by the Numbers
| Metric | Value | Source |
|---|---|---|
| Output gain from an optimized passive radiator in a micro speaker box | +5 dB at 200 Hz | Jiang et al., Archives of Acoustics 2017 |
| 2.5-inch passive radiator resonance / moving mass | 73 Hz / 3.2 g | Dayton Audio ND65-PR (vendor claim) |
| Claimed low-frequency limit, 0.36 kg vs 1.8 kg speaker | 75 Hz vs 50 Hz | Marshall spec pages (vendor claims) |
| Example woofer: inductance limit vs Doppler limit | 2.28 mm vs 15.9 mm | KLIPPEL Application Note AN 5, 2026 |
| Closed-box micro-speaker resonance | 840 Hz | Hu, Lu and Qiu, UTS repository 2016 |
| Output change from distortion compensation | -0.1 dB to 0.4 dB | Hu, Lu and Qiu, UTS repository 2016 |
| MDAQS distortion score, JBL Flip 7 vs Flip 6 | 3.5 vs 2.5 | SoundGuys 2025 |
| Typical max volume of tested speakers | 80 to 100 dB SPL | Rtings dynamics test 2021 |
| Volume score, JBL Charge 5 vs best in set | 46 vs 141 | DXOMARK 2021 |
| Claimed max SPL at 1 m, Willen II / Emberton III / Middleton | 88 / 90 / 87 dB | Marshall spec pages (vendor claims) |
| IPX7 immersion requirement | 1 m for 30 minutes | DLS Electronic Systems (IEC 60529) |
| IPX8 immersion requirement | Deeper than 1 m, set by manufacturer | DLS Electronic Systems (IEC 60529) |
| JBL Flip 7 claimed IP68 immersion | 1.5 m for 30 minutes | SoundGuys 2025 (vendor claim) |
| JBL Flip 7 measured vs rated battery life | 6 h 16 min vs 14 h | SoundGuys 2025 |
| Single-cell smart amp output at 3.6 V into 4 ohms | 6.1 W | Texas Instruments TAS2563 (vendor claim) |
| Marshall Emberton III claimed battery life | 32+ hours | Marshall spec page (vendor claim) |
Methodology and Sources
- KLIPPEL Application Note AN 4, Peak Displacement Xmax, updated April 4, 2012: klippel.de
- KLIPPEL Application Note AN 5, Displacement Limits due to Driver Nonlinearities, revision 1.2, May 22, 2026: klippel.de
- W. Klippel, Modeling the Large Signal Behavior of Micro-Speakers, 2012: klippel.de
- W. Klippel, Active Transducer Protection Part 1: Mechanical Overload (references run to 2015): klippel.de
- Jiang, Kwon, Kim and Hwang, Analysis and Optimization of Micro Speaker-Box Using a Passive Radiator in Portable Device, Archives of Acoustics 42(4), 753-760, 2017: doi.org/10.1515/aoa-2017-0078
- Anderson, Neu, Gregg, Young and Leishman, Experimental assessment of passive radiator parameters using normal-incidence sound transmission measurements, JASA 158(5), 3697-3710, November 2025: doi.org/10.1121/10.0039708
- Hu, Lu and Qiu, Compensating the distortion of micro-speakers in a closed box with consideration of nonlinear mechanical resistance, manuscript in the UTS repository, 2016: opus.lib.uts.edu.au
- Rtings, speaker dynamics test (SPL and dynamic range compression at max volume), April 6, 2021: rtings.com
- Rtings, JBL Flip 7 review, test bench 1.0.1, updated July 28, 2026: rtings.com
- SoundGuys, JBL Flip 7 review by Dave Carr, updated June 13, 2025: soundguys.com
- DXOMARK, JBL Charge 5 speaker test, August 5, 2021: dxomark.com
- DLS Electronic Systems, IPX7 and IPX8 Immersion Testing (IEC 60529): dlsemc.com; IEC overview of the IP code: iec.ch
- Dayton Audio ND65-PR passive radiator product page (vendor claims): daytonaudio.com
- Marshall spec pages for Willen II, Emberton III and Middleton (vendor claims): Willen II, Emberton III, Middleton
- Ultimate Ears Wonderboom 4 product page (vendor claims): ultimateears.com
- Texas Instruments TAS2563 and TAS5825M product pages (vendor claims): TAS2563, TAS5825M
- Data watch: independent, open measurement data on portable-speaker excursion is scarce, so several core figures come from older studies (2012, 2016, 2017) and are flagged as most recent available data. The micro-speaker studies used phone-size drivers, not the 2-inch to 3-inch drivers in portable speakers, so treat their numbers as physics illustrations rather than portable-speaker averages. The Rtings dynamics page states two different thresholds for a good result (over 92 dB and over 95 dB for loudness, under 1 dB and under 2 dB for compression) and two different reference levels (76 dB and 85 dB SPL); we only quote its measured examples. Rtings now keeps its measured Flip 7 loudness, compression and battery figures behind a membership, so newer numbers likely exist but were not openly readable. SoundGuys’ battery test ran at 80 dB while JBL’s rating assumes moderate volume, so the gap is not a like-for-like comparison. We found no independent measurement of acoustic loss through waterproof membranes; manufacturer datasheets we tried were unreachable, so that topic was dropped rather than estimated. Manufacturer figures (Dayton Audio, Marshall, Ultimate Ears, Texas Instruments, JBL) are vendor claims and were not independently verified.
Last updated: October 27, 2026. We update this roundup quarterly, and the next refresh is expected when the next revision of the IEC 62458 displacement-limit application note or the next Journal of the Acoustical Society of America study on passive-radiator measurement is published.