DECT NR Plus Intercom

Thought for 1 min 10 sec

flux

I’ll finish the battery-management validation report and update the project spec first, with no schematic changes.

Worked for 18 mins 31 sec

Created a project doc

Battery Management Validation Report

View Document

Status: Phase 2 validation gate complete for schematic planning. No schematic capture has started. Scope: single protected removable 18650 L...

flux

Finished. I created the Battery Management Validation Report and updated the Project Specification before any schematic work.

Key locked decisions: MAX17055ETB+T + BQ25895RTWT + TUSB320LAIRWBR, 1.0 A default charge, 1.5 A only after validation, 2.0 A engineering-only, BQ25895 charge disabled until firmware configures safe settings, and mandatory cell-temperature/USB-C current validation.

Suggested next steps:

Approve these battery-management requirements as the schematic baseline.
Select the exact protected 18650 cell and holder.
Continue Phase 2 with radio/MCU/audio architecture comparison before schematic capture.

Approve and compare

Approve baseline

Pick cell holder

Compare core architecture

Check report issues

No schematic yet

Talk to Flux to get started.

Phase 1 Feasibility Report - DECT NR+ Intercom Beltpack

Status: Draft for user approval before Phase 2 component selection.

This report covers Phase 1 only. No schematic or PCB layout has been created yet.

Executive Recommendation

Proceed to Phase 2 only under this architecture assumption: Revision A should be a four-node, low-latency, one-hop DECT NR+ intercom prototype with an automatically elected temporary coordinator, not a guaranteed true mesh product.

The Nordic nRF9151 is a credible DECT NR+ radio candidate for the European 1880-1900 MHz allocation, but the current public Nordic DECT NR+ implementation supports FT/PT star networking rather than direct mesh/multi-hop routing. Multi-hop should remain an explicit experiment or future partner-stack option, not a committed Revision A feature.

A dedicated application/audio MCU is recommended for Revision A unless benchmarking proves the nRF9151 application core can simultaneously handle radio control, codec work, mixing, UI, security, and recovery logic with comfortable CPU/RAM margin.

Key Sources Used

Nordic nRF9151 product page and product brief: 64 MHz Arm Cortex-M33, 1 MB flash, 256 KB RAM, DECT NR+ support, power-class claims: https://www.nordicsemi.com/Products/nRF9151
Nordic nRF9151 product specification DECT NR+ section: TX power and RX sensitivity: https://docs.nordicsemi.com/r/bundle/ps_nrf9151/page/dect.html
Nordic nRF Connect SDK Hello DECT sample: FT/PT IPv6 data over DECT NR+: https://docs.nordicsemi.com/bundle/ncs-latest/page/nrf/samples/dect/hello_dect/README.html
Nordic DECT NR+ PHY hello sample: low-level PHY experiment: https://developer.nordicsemi.com/nRF_Connect_SDK/doc/latest/nrf/samples/dect/dect_phy/hello_dect/README.html
Nordic DevZone answer on mesh support: mesh networking is currently not supported in Nordic implementation: https://devzone.nordicsemi.com/f/nordic-q-a/127492/dect-nr-multi-hop-mesh-networking-support-in-dect-nr-nrf9151-mfw-nr-_nrf91x1_2-0-0
ETSI DECT overview and EN 301 406-2 DECT-2020 NR harmonized standard: https://www.etsi.org/technologies/dect and https://www.etsi.org/deliver/etsi_en/301400_301499/30140602/03.01.01_60/en_30140602v030101p.pdf
CEPT/ECO DECT licensing references for 1880-1900 MHz and nominal 250 mW / 24 dBm class operation: https://docdb.cept.org/document/774 and https://docdb.cept.org/download/100
Opus codec RFC 6716: frame durations 2.5-60 ms: https://datatracker.ietf.org/doc/html/rfc6716
ETSI TS 103 634 LC3plus: 2.5 ms, 5 ms, 10 ms frame durations and low algorithmic delay: https://www.etsi.org/deliver/etsi_ts/103600_103699/103634/01.05.01_60/ts_103634v010501p.pdf
Nordic nRF5340 product page/brief: 128 MHz Cortex-M33 application core, 1 MB flash, 512 KB RAM, I2S/PDM/USB/security, Zephyr/NCS support: https://www.nordicsemi.com/products/nrf5340
USB Type-C sink CC behavior: 5.1 kOhm Rd and source current advertisement: https://www.usb.org/sites/default/files/USB%20Type-C%20Spec%20R2.0%20-%20August%202019.pdf
Kenwood K1 practical pinout references: https://www.wildtalk.com/knowledge-base/kenwood-2-pin-wiring-data/ and https://pinoutguide.com/HeadsetsHeadphones/kenwood_protalk_pinout.shtml

Requirements Table

Table

Area	Requirement	Phase 1 feasibility	Notes / assumptions
Users	Four identical beltpacks	Feasible	Four nodes is a reasonable first test size.
Audio receive	Receive all active talkers continuously	Feasible with custom audio transport	Requires packet scheduling, decoding, jitter buffering, mixing, and output limiting.
PTT transmit	Transmit mic audio only while external PTT is pressed	Feasible	PTT detection through Kenwood K1 interface; no PTT button on PCB.
Simultaneous talkers	More than one operator can transmit at once	Feasible but not turnkey	Needs radio airtime budget and collision-free scheduling or coordinator-managed streams.
Local mix	Mix received voice streams into mono earpiece	Feasible	MCU/audio DSP load must be measured for 1 encode + up to 3 decode/mix paths.
Base station	No permanently required base station	Feasible as elected coordinator	A temporary FT/coordinator can be one of the beltpacks. This is not the same as true mesh.
Coordinator election	Automatic coordinator election if needed	Feasible with custom firmware	Nordic stack supports FT/PT roles; automatic election/reassociation behavior must be implemented and measured.
Mesh / multi-hop	Investigate packet forwarding or mesh routing	High risk	Nordic says mesh is not currently supported in its implementation. Treat as experiment or partner-stack option.
Recovery	Recover if one beltpack leaves	Feasible, but coordinator loss is risky	Normal peer loss should be manageable; coordinator loss requires measured re-election time.
Radio	DECT NR+ in EU 1880-1900 MHz	Feasible	Verify exact band, firmware, channel plan, and RED/ETSI compliance.
Antenna	Replaceable external antenna, 50 ohm path, pi match, test provision	Feasible	Requires RF layout discipline, enclosure/body testing, and certification constraints.
Range	400 m outdoor LOS target, indoor church coverage	Outdoor target plausible, indoor not guaranteed	Link budget supports outdoor target with limited margin under body/fade assumptions; indoor walls destroy margin quickly.
UI	1.3 inch monochrome OLED, encoder/mute, channel switch, menu button	Feasible	Use low-power display and debounced controls.
Headset	Kenwood K1 two-jack interface, PTT/mic/earpiece/protection	Feasible with validation	K1 variants differ; provide 0 ohm options and require measuring final accessory.
Power	Removable protected 18650, 6 h minimum, 10-12 h target	Plausible	Coordinator/relay mode may challenge 10-12 h; must measure with target radio/codec load.
Charging	USB-C charge, power-path, safe single-cell Li-ion charging	Feasible	1.5-2 A charge likely needs switch-mode charger and thermal validation.
PCB	Revision A 6-layer impedance-controlled, testable prototype	Feasible and recommended	Avoid blind/buried vias; separate RF, audio, power, digital zones.
Open source	Open-source product intent	Partially constrained	Hardware and application firmware can be open; radio modem firmware and codec licenses may not be fully open.

Architecture Block Diagram

Diagram

DECT NR+ Feasibility Analysis

What is favorable

The European 1880-1900 MHz DECT allocation is a strong fit for a church intercom because it avoids crowded 2.4 GHz WiFi/BLE audio assumptions and is purpose-built for local-area low-latency communications.
Nordic nRF9151 publicly supports DECT NR+ and has published DECT NR+ sensitivity around -105 dBm for beacon/MCS0 and -103 dBm for unicast/MCS1, with up to about 19-21 dBm conducted TX power depending on band/configuration.
Nordic's Hello DECT sample demonstrates IPv6 data over DECT NR+ using FT/PT roles, and the shell/sample tooling supports scanning, association, and network management experimentation.

What is not proven

The current public Nordic stack does not provide a finished four-user intercom, group voice, synchronized low-latency audio, or mesh routing implementation.
Nordic's own DevZone response states that mesh networking is not currently supported in Nordic's implementation. This directly affects the user's preferred decentralized multi-hop experience.
The nRF9151 application processor is 64 MHz Cortex-M33 with 1 MB flash and 256 KB RAM. That may be adequate for radio control and simple firmware, but it is not safe to assume enough headroom for 1 encode + 3 decodes + audio mixing + OLED UI + security + routing.

Recommended Phase 1 position

Use nRF9151 as the DECT NR+ radio candidate.
Treat FT/PT star operation with elected coordinator as the baseline.
Treat true mesh/multi-hop as an experiment, not a committed Revision A capability.
Add a dedicated application/audio MCU in the architecture unless benchmarks prove it unnecessary.

Mesh and Multi-Hop Feasibility Matrix

Table

Requirement	Supported directly by chipset	Supported by current official SDK	Requires custom firmware	Requires custom MAC/networking work	Technically uncertain	Recommended validation experiment
EU 1880-1900 MHz DECT NR+ operation	Yes, subject to band/config	Yes, with correct firmware/regional config	Low	No	Regulatory/country details	Bring up nRF9151 DKs in EU overlay/config and verify legal channel/TX settings.
One FT with multiple PTs	Radio-capable	Yes, Hello DECT / MAC+DLC direction	Medium	Low	Four-node voice load	Run one FT and three PTs, measure latency/jitter/packet loss.
Peer-to-peer with no coordinator	PHY may allow experiments	Not as a direct turnkey user experience	High	Medium to high	High	Test whether two nodes can exchange low-latency payloads without persistent FT role.
Dynamic coordinator election	No automatic product feature	Role support exists, election not turnkey	High	Medium	Medium/high	Implement highest-ID or priority election, kill FT, measure recovery time.
Multi-hop routing	Standard has mesh concepts, chipset not enough by itself	No, Nordic says mesh not currently supported	Very high	Very high or partner stack	High	Try three-node relay at application layer; measure duplicate suppression, loop prevention, and added latency.
Store-and-forward/scheduled relaying	Not direct	No direct support found	Very high	Very high	High	Emulate relay packets with sequence numbers and TTL; measure jitter under two-hop load.
Simultaneous voice streams	RF link likely capable at low bitrates	Data transport exists, voice stack not turnkey	High	Medium	Medium	Send synthetic audio payloads for 1, 2, 3, and 4 talkers at 5/10/20 ms packet periods.
Low-latency group audio	Not direct	No complete sample found	High	Medium/high	High	End-to-end mouth-to-ear test with codec and jitter buffer.
Automatic frequency/interference management	DECT technology supports coexistence mechanisms	SDK has scan/association tools; full behavior must be verified	Medium	Medium	Medium	Interference/channel occupancy test in a real church environment.
Authenticated pairing and encrypted voice/control	Hardware/security primitives likely available	MAC security exists, product pairing is custom	High	Low/medium	Medium	Build four-node trusted-peer provisioning demo and lost-device removal flow.

Fallback Network Architecture

If true mesh audio is not practical with the current Nordic stack, use this fallback:

Each beltpack stores a network ID, device ID, and trusted-peer credentials.
On power-up, each node scans for an existing trusted coordinator/FT.
If no coordinator is found, nodes wait a randomized interval and elect a coordinator deterministically, for example by priority plus device ID.
PT nodes associate with the elected FT and exchange voice/control packets.
If the coordinator disappears, PT nodes stop transmitting voice, continue local UI indication, rescan, and run election again.
If an optional stationary powered relay is later added, it uses the same network identity and a higher coordinator priority, but no unique beltpack hardware redesign is required.
Any future relay packet format must include source ID, stream ID, sequence number, hop count/TTL, and recent-packet cache to prevent duplicate audio and routing loops.

This gives the desired user experience of no permanently required base station while staying honest about current stack limitations.

Audio Bandwidth, Codec, and Latency Budget

Audio target

Mono speech.
16 kHz or 24 kHz sampling.
Speech bandwidth suitable for intercom intelligibility, not music fidelity.
One-hop target below 40 ms where realistically possible.
Multi-hop target below 80 ms where realistically possible.
PTT-gated microphone; acoustic echo cancellation not mandatory.
Firmware-adjustable sidetone is recommended so operators hear a low-level copy of their own microphone while transmitting.

Codec comparison

Table

Codec	Strengths	Risks	Phase 1 recommendation
LC3plus	DECT-aligned, low latency, 2.5/5/10 ms frames, good speech quality at lower bitrates	Licensing/patent/commercial-use review required; embedded CPU/RAM benchmark required	Best final candidate if licensing and implementation are acceptable.
Opus low delay	Mature, open-source, excellent quality, supports 2.5-60 ms frames	Low-delay CELT mode can be CPU-heavy; SILK speech modes add delay and are not used below 10 ms frames	Benchmark on target MCU; strong alternative if licensing simplicity matters.
ADPCM	Very low CPU, low algorithmic delay, simple to debug, tolerant for prototypes	Higher bitrate, lower quality/noise robustness than LC3plus/Opus	Best Revision A fallback for early radio/audio transport testing.
Uncompressed PCM	Simplest lab payload, no codec CPU, deterministic	16 kHz 16-bit mono is 256 kbps per talker before overhead; poor airtime efficiency	Use only for early lab latency/transport tests.

Preliminary bitrate assumptions

Table

Mode	Payload rate per talker before packet overhead
16 kHz, 16-bit PCM mono	256 kbps
24 kHz, 16-bit PCM mono	384 kbps
IMA ADPCM at 16 kHz, 4 bits/sample	64 kbps
LC3plus / Opus speech target	Approximately 16-48 kbps depending quality/frame settings

One-hop latency budget target

Table

Element	Conservative target
ADC/audio codec buffering	1-3 ms
Audio frame collection	5-10 ms
Codec algorithmic delay and CPU time	5-12 ms
Radio scheduling and packet transfer	5-10 ms, must measure
Jitter buffer	5-10 ms
Mixing/limiting/DAC	1-3 ms
Preliminary one-hop total	22-48 ms

A sub-40 ms one-hop target is plausible only with short frames, tight radio scheduling, low jitter, and measured CPU headroom. A two-hop path likely adds another scheduling/forwarding/jitter interval, so below 80 ms is possible but not assured.

nRF9151 Alone vs Dedicated Application/Audio MCU

nRF9151 application-core constraints

64 MHz Arm Cortex-M33.
1 MB flash and 256 KB RAM.
Radio modem and DECT NR+ firmware on network/modem side.

This is attractive for a compact design, but Phase 1 should not assume it can handle all audio/network/UI/security tasks simultaneously.

Recommended Revision A direction

Use nRF9151 for DECT NR+ radio plus a dedicated application/audio MCU candidate such as nRF5340 or STM32U5. The nRF5340 is especially attractive for early work because it has Nordic/Zephyr ecosystem alignment, 128 MHz Cortex-M33 application core, 512 KB RAM class resources, I2S/PDM/USB, hardware security, and existing audio-development ecosystem experience. STM32U5 remains a valid alternative if more RAM, peripheral flexibility, or non-Nordic independence becomes more important.

Phase 2 should compare nRF5340, STM32U5, STM32H5, and any lower-power audio MCU options using measured codec benchmarks.

Preliminary RF Link Budget

Assumptions

Table

Parameter	Assumption	Notes
Frequency	1.89 GHz	Middle of European 1880-1900 MHz DECT band.
Legal class reference	Up to about 250 mW / 24 dBm nominal or EIRP class depending national wording	Must be checked against ETSI/CEPT/country rules and final antenna gain.
nRF9151 conducted TX power	19-21 dBm	Published Nordic values vary by band/config; use 20 dBm planning value.
Antenna gain	0 dBi planning value	Avoid relying on antenna gain for range; keep EIRP legal.
Feed/connect losses	0.7 dB each end	SMA/U.FL/cable/launch estimate.
Receiver sensitivity	-103 dBm	Conservative unicast/MCS1 value; -105 dBm published for beacon/MCS0.
Body loss allowance	8 dB total	Belt-worn placement can be worse; must test.
Fade margin	20 dB	Outdoor/body-worn reliability planning margin.
Indoor masonry wall loss	8-15 dB per wall	Highly variable; church construction can be worse.

Free-space path loss formula

FSPL(dB) = 32.44 + 20 log10(f_MHz) + 20 log10(d_km)

At 1.89 GHz:

Table

Distance	FSPL
100 m	78.0 dB
400 m	90.0 dB
1 km	98.0 dB

Link calculation at 400 m outdoor LOS

Using TX conducted 20 dBm, TX feed loss 0.7 dB, 0 dBi antennas, RX feed loss 0.7 dB, 8 dB body allowance, and 20 dB fade margin:

Received level before fade/body margin at 400 m = 20 - 0.7 + 0 - 90.0 + 0 - 0.7 = -71.4 dBm.

After 8 dB body allowance and 20 dB fade margin = -99.4 dBm equivalent planning level.

Compared with -103 dBm sensitivity, remaining planning margin is about 3.6 dB.

Interpretation

The 400 m outdoor line-of-sight target is plausible but not guaranteed; it has limited margin under conservative body/fade assumptions.
Indoor 400 m coverage cannot be guaranteed. Two masonry walls at 10-15 dB each can consume 20-30 dB, turning a viable outdoor link into a marginal or failed indoor link.
Multi-hop can improve coverage by replacing one long obstructed path with shorter paths, but only if routing, duplicate suppression, timing, and latency are implemented and measured.
Do not add an RF power amplifier in Revision A. It is not justified unless legal, supported by DECT NR+, compatible with certification, and required after measured link results.

Antenna Arrangement Evaluation

Table

Option	Pros	Cons	Recommendation
Board-mounted SMA connector	Simple, good for lab, direct 50 ohm path	User torque and belt impacts load PCB solder joints; hard to isolate from enclosure stress	Acceptable for bench prototypes only with mechanical bracing.
Reverse-polarity SMA	Can discourage ordinary antennas in some markets	Not legally or technically justified here by default; may confuse sourcing	Do not use unless certification/legal strategy specifically requires it.
Internal U.FL to enclosure-mounted SMA	Best mechanical isolation; replaceable external antenna; enclosure absorbs torque	U.FL has limited mating cycles; pigtail loss and assembly complexity; needs strain relief	Recommended for belt-worn product and likely Revision A if enclosure/mechanics are included.
Direct board-edge SMA	Good RF path, robust if edge mounted with mounting legs and enclosure support	Still transfers force into PCB; board outline/enclosure constraints matter	Good Revision A compromise if mechanically reinforced and kept away from body/battery.

Recommended direction: enclosure-mounted bulkhead SMA connected by short internal coax/U.FL or equivalent board RF connector, with strain relief, pi matching network at the RF launch, conducted-test option, and defined antenna keepout. For a bare development PCB, a reinforced edge-launch SMA plus RF test connector is acceptable.

Do not claim the antenna is tuned or certified until VNA tuning, radiated testing, body-worn tests, and regulatory review are complete.

Preliminary Power Budget

Battery assumption

Protected 18650 Li-ion cell: 3000 mAh nominal, 3.6 V nominal, about 10.8 Wh nameplate.
Usable system energy after discharge limits and regulator losses: roughly 8.5-9.5 Wh for planning.

Load assumptions to validate

Table

Subsystem	Mostly listening	Normal mixed use	Coordinator/relay	Worst continuous TX/RX	Notes
nRF9151 radio/modem	120-200 mW	200-450 mW	450-800 mW	800-1100 mW	Must measure with chosen MCS, TX power, and duty cycle.
Application/audio MCU	40-100 mW	80-180 mW	150-300 mW	200-400 mW	Codec count dominates.
Audio codec/amp/mic bias	25-80 mW	40-120 mW	40-150 mW	80-200 mW	Earpiece impedance and output level matter.
OLED and controls	20-80 mW	20-80 mW	20-80 mW	20-80 mW	OLED content/brightness must be managed.
Regulators/monitoring overhead	30-80 mW	50-120 mW	80-160 mW	100-220 mW	Depends on topology and load.
Total planning power	235-540 mW	390-950 mW	740-1490 mW	1200-2000 mW	Wide range until measured.

Runtime estimate

Using 9.0 Wh usable energy:

Table

Use case	Planning power	Estimated runtime	Comment
Mostly listening	0.35-0.55 W	16-26 h	Likely exceeds target if radio receive duty and OLED power are controlled.
Normal mixed use	0.65-0.95 W	9.5-14 h	Meets preferred 10-12 h only if codec/radio duty are efficient.
Continuous relay/coordinator	1.0-1.5 W	6-9 h	Meets minimum but may miss 10-12 h. Coordinator rotation may help.
Worst simultaneous TX/RX	1.5-2.0 W	4.5-6 h	Borderline; avoid defining this as normal operating condition.

The 6 h requirement is plausible. The 10-12 h target is plausible for normal use but not guaranteed for a heavily loaded coordinator/relay unless power measurements are favorable.

USB-C and Charging Feasibility

A basic USB-C sink can use Rd pull-downs on CC1/CC2, but a device drawing more than default current must detect source current advertisement or use a controller.
5 V USB-C without PD is enough for Revision A if the charger observes advertised current limits and total system load.
9 V USB-PD does not automatically help a single-cell design. With a linear charger it worsens heat; with a switch-mode charger it may reduce cable current but adds complexity.
For 1.5-2 A single-cell charging, a switch-mode charger/power-path controller is strongly preferred.
Thermal example: a linear charger at 5 V input, 3.6 V battery, 1.5 A dissipates about (5.0 - 3.6) * 1.5 = 2.1 W, before system-load effects. That is too hot for a small beltpack without severe thermal foldback.
The selected charger must include NTC/JEITA behavior or equivalent thermal safety, input current limit, power-path/load sharing, charger state reporting, and no unsafe USB backfeed.

Headset and Kenwood K1 Feasibility

Common practical K1-style wiring varies, but a conservative starting point from public references is:

Table

Connector	Contact	Typical function
2.5 mm TRS	Tip	Speaker / earpiece audio
2.5 mm TRS	Ring	Accessory/data/variant-dependent
2.5 mm TRS	Sleeve	Ground/common
3.5 mm TRS	Tip	Bias/supply variant-dependent
3.5 mm TRS	Ring	Microphone audio
3.5 mm TRS	Sleeve	PTT, often short to ground

Revision A must not hard-code one accessory assumption. Include ESD protection, RF filtering, microphone input protection, speaker output protection, PTT detect, mic-bias filtering, and 0 ohm/solder-jumper options for variants. Add clearly labeled MIC, PTT, EARPIECE_OUT, GND, and MIC_BIAS test points.

Production warning: the exact final headset/accessory must be measured before production. K1-compatible accessories are not electrically identical.

Major Technical Risks

Table

Risk	Severity	Why it matters	Mitigation
Nordic stack mesh support absent	High	Multi-hop is a major requested feature but not directly supported	Baseline elected-coordinator star architecture; run explicit relay experiments; consider partner stacks.
Coordinator failure recovery time	High	Intercom must recover automatically	Implement election experiment early and measure audio outage.
Group voice latency/jitter	High	Church operators need natural timing	Synthetic payload tests before hardware audio; short frames; small jitter buffer; measure worst case.
Codec CPU/RAM load	High	nRF9151 may be too small for all processing	Benchmark 1 encode + 3 decodes; use dedicated MCU if margin is poor.
Codec licensing/open-source compatibility	Medium/high	Open-source product may conflict with codec/IP terms	Compare LC3plus, Opus, ADPCM licensing in Phase 2.
RF body loss and antenna detuning	High	Belt-worn RF range can collapse	External antenna, keepout, VNA tuning, body-worn range tests.
Indoor range expectations	High	400 m indoor is unrealistic	State as target only; test in real venue; allow optional relay later.
USB-C/charger thermal safety	High	1.5-2 A charge can overheat	Switch-mode charger, NTC, current limit, thermal testing.
K1 headset pinout variations	Medium/high	Wrong wiring can damage audio/produce no PTT	Jumpers, protection, test points, accessory validation.
Audio noise/RF ingress	Medium/high	Earpiece intelligibility and RF immunity are critical	Physical zoning, filtering, analog LDO, shielding/grounding review.
Regulatory certification	High	DECT radio and replaceable antenna require compliance work	RED/ETSI plan, conducted/radiated pre-scan, authorized antenna list.
Mechanical connector stress	Medium/high	Belt-worn device sees repeated force	Bulkhead antenna, reinforced jacks/USB, mounting holes, enclosure support.

Recommended Development-Kit Experiments

Use at least five nRF9151 DKs if possible: four active nodes plus one sniffer/test/control node.

Basic DECT NR+ bring-up
- Run PHY hello and Hello DECT samples.
- Confirm firmware version, band, TX power, channel selection, RSSI, packet loss, throughput, and latency.
Star network with one FT and three PTs
- Run one coordinator/FT and three PTs.
- Measure packet latency distribution, jitter, packet loss, reconnect time, and power.
Coordinator failure and election
- Implement deterministic election, for example highest trusted device ID or configured priority.
- Kill the coordinator and measure time to recovered audio-capable network.
Synthetic audio payload emulation
- Send 20/40/80 byte payloads every 2.5/5/10/20 ms.
- Test one, two, three, and four simultaneous talkers.
- Measure airtime saturation and worst-case jitter, not only average latency.
Codec benchmark on target MCU candidates
- Benchmark ADPCM, Opus low-delay, and LC3plus if available/licensable.
- Measure 1 encode + 3 decodes + mixing + PLC + UI load.
- Record CPU, RAM, stack, and worst-case frame time.
Real audio loop
- Add external I2S codec/eval board.
- Test mic to encode to radio to decode to earpiece.
- Measure mouth-to-ear latency, dropout behavior, sidetone, limiter behavior, and earpiece output level.
Body-worn RF test
- Belt-worn, under clothing, antenna vertical/horizontal, body blocking path, indoor NLOS, masonry walls, and realistic church movement.
- Measure link margin, dropout bursts, and antenna placement sensitivity.
Charger/power thermal test
- Evaluate 1 A, 1.5 A, and 2 A charge cases with real cell and enclosure approximation.
- Measure charger IC, cell, USB connector, and enclosure temperature rise.

Phase 1 Go / No-Go Criteria for Phase 2

Proceed to component selection if the user accepts these Phase 1 conclusions:

Baseline network is elected-coordinator FT/PT star, not guaranteed true mesh.
Multi-hop is a research task and may require major custom MAC/routing work or a partner stack.
Revision A likely includes a separate application/audio MCU.
The external antenna will be replaceable but not claimed tuned/certified until measured.
400 m outdoor LOS is a target, not a guarantee; 400 m indoor coverage is not a valid claim.
10-12 h runtime is a normal-use target; continuous coordinator/relay operation may be closer to 6-9 h until optimized.
USB-C charging above 1 A requires careful current-limit and thermal design.
Kenwood K1 accessory compatibility requires measurement and configurable pinout options.

Phase 2 Starting Direction If Approved

Candidate categories to research and select in Phase 2:

DECT NR+ radio: Nordic nRF9151 variants/reference circuit and required crystals, SIM/eSIM exclusion if unused, RF match, power rails.
Application/audio MCU: nRF5340 vs STM32U5 vs STM32H5 or other suitable MCU.
Audio codec/amp: low-power I2S codec with electret mic bias, programmable gain, mono headphone/earpiece driver, hardware mute, click/pop suppression.
Charger/power path: single-cell Li-ion switch-mode charger/power path with USB-C current-limit support and NTC.
Fuel gauge: coulomb counter or model-gauge suitable for removable 18650.
Regulators: efficient buck for digital/radio and low-noise LDO for audio/RF as needed.
USB-C: connector, ESD, CC/current-detection or sink/PD controller.
RF connector/antenna: reinforced board-edge SMA or enclosure SMA with internal coax; pi match and test connector.
Kenwood K1 jacks: mechanically compatible 3.5 mm and 2.5 mm jacks at approximately 11 mm spacing with reinforcement.
OLED and controls: 1.3 inch monochrome OLED, encoder, channel rotary switch, menu button, power switch.

Approval Checkpoint

Phase 1 is complete. Approval is required before Phase 2 component selection.

Executive Recommendation
Key Sources Used
Requirements Table
Architecture Block Diagram
DECT NR+ Feasibility Analysis
What is favorable
What is not proven
Recommended Phase 1 position
Mesh and Multi-Hop Feasibility Matrix
Fallback Network Architecture
Audio Bandwidth, Codec, and Latency Budget
Audio target
Codec comparison
Preliminary bitrate assumptions
One-hop latency budget target
nRF9151 Alone vs Dedicated Application/Audio MCU
nRF9151 application-core constraints
Recommended Revision A direction
Preliminary RF Link Budget
Assumptions
Free-space path loss formula
Link calculation at 400 m outdoor LOS
Interpretation
Antenna Arrangement Evaluation
Preliminary Power Budget
Battery assumption
Load assumptions to validate
Runtime estimate
USB-C and Charging Feasibility
Headset and Kenwood K1 Feasibility
Major Technical Risks
Recommended Development-Kit Experiments
Phase 1 Go / No-Go Criteria for Phase 2
Phase 2 Starting Direction If Approved
Approval Checkpoint