Battery-less Audio Node Design

flux

I renamed the project and captured the concept in Project Specification. I also completed a datasheet-grounded feasibility check in Feasibility and Power Analysis.
Conclusion: the exact 2 mm cube off-the-shelf design should not proceed to schematic as stated—the microphone and EFR32-class SoC packages alone exceed the volume, and the 01005 buffer capacitor is orders of magnitude too small for RF/audio burst energy. Recommended next step is to choose a revised target: larger proof-of-concept PCB, true custom bare-die/SiP research package, or reduced-function NFC-style node.
mmkmk

mmkmk

Design a larger proof-of-concept PCB around these components.

flux

I’ll design a practical proof-of-concept schematic around the same RF SoC, PDM mic, Schottky rectifier, and GRM02 buffer concept. I’ll size the power path first, add the components and support circuitry, wire the schematic, then run review and document the firmware/power notes.

flux

Designed the larger proof-of-concept schematic around the EFR32MG22, SPK0641HT4H-1 mic, SMS7621 rectifier bridge, GRM02 benchmark capacitor, larger reservoir cap, AP2112K 3.3 V rail, SMA RF test port, coil input, SWD header, reset, decoupling, and RF match.
I documented the implementation in Proof-of-Concept Design Notes and added firmware bring-up guidance in Firmware Starter. Schematic connectivity is complete; remaining review issues are expected layout airwires and generic-part MPNs to resolve before fabrication. Next step is PCB layout setup and placement—ready when you want me to proceed.
Set board stackup
Auto-place components
Run ERC/DRC
Resolve BOM MPNs
Add test points
Sign up to send

Refine this doc
Ask about this doc
Feasibility and Power Analysis — RF-Harvested 2 mm Audio Node
Executive Conclusion
The proposed design is not feasible as stated using the listed off-the-shelf components in a 2 mm x 2 mm x 2 mm envelope. The concept can be pursued as either a larger proof-of-concept PCB or a custom chip-scale SiP/bare-die research package, but schematic capture for the listed exact architecture would produce an electrically and mechanically invalid design.
Datasheet / Library Findings
Knowles SPK0641HT4H-1 PDM Microphone
  • Supply voltage: 1.6 V min, 1.8 V typ, 3.6 V max.
  • Current consumption:
    • Performance mode: 630 uA typ, 710 uA max at 1.8 V.
    • Low-power mode: 230 uA typ, 275 uA max at 1.8 V.
    • Sleep current: 26 uA typ.
  • Package: 4 mm x 3 mm x 1.1 mm.
  • Pinout: DATA, CLOCK, SELECT, Vdd, and four GND pins.
  • Impact: the microphone alone exceeds the 2 mm cube footprint and consumes about 0.414 mW in low-power mode at 1.8 V.
Silicon Labs EFR32MG22-Class Wireless SoC
  • Library-available packages are QFN/TQFN, typically 4 mm x 4 mm for 32-pin variants and 5 mm x 5 mm for 40-pin variants.
  • Supply absolute max: 3.8 V on supply pins.
  • MCU active current examples:
    • 3.0 V, no DC-DC: ~39 uA/MHz typical for 76.8 MHz HFRCO while loop from flash.
    • 3.0 V with DC-DC: ~27 uA/MHz typical for 76.8 MHz HFRCO while loop from flash.
  • TX current:
    • 0 dBm output: ~4.3 mA system current typical.
    • 6 dBm output: ~8.4 mA system current typical.
  • Flash:
    • Endurance: 10,000 erase cycles.
    • Program one 32-bit word: 44 us typ, up to 1.45 mA program current.
    • Page erase: 12.9 ms typ, up to 1.34 mA erase current.
  • Impact: QFN packaged EFR32MG parts exceed the target footprint; frequent flash writes/erases are not suitable for continuous audio persistence.
Skyworks SMS7621-060 Schottky Diode
  • Package: 0201 footprint, approximately 0.600 mm x 0.300 mm x 0.226 mm.
  • Application: sensitive detector, sampling, and mixer circuits.
  • Electrical:
    • Reverse voltage max: 2 V.
    • Forward voltage at 1 mA: 260 mV to 320 mV.
    • Capacitance: 0.18 pF max at 0 V.
    • Series resistance: 12 ohm max.
  • Impact: useful for RF detection/rectification, but four diodes in a bridge add significant voltage loss and occupy nontrivial area. A bridge is also not ideal for ultra-low harvested voltages.
Murata GRM022D80G104ME15L 01005 MLCC
  • Package: 0.4 mm x 0.2 mm x 0.2 mm.
  • Capacitance: 0.1 uF nominal, +/-20%.
  • Rated voltage: 4 VDC.
  • Temperature characteristic: X6T, +22% / -33% over -55 C to +105 C.
  • Impact: suitable as local decoupling, not as the main burst energy reservoir.
Energy Buffer Calculation
Usable capacitor energy is:
E=12C(VHI2−VLO2)E = \frac{1}{2} C \left(V_{HI}^{2} - V_{LO}^{2}\right)
For a 0.1 uF GRM02 capacitor charged from 3.0 V down to 1.8 V:
E=0.5⋅0.1μF⋅(3.02−1.82)=0.288μJE = 0.5 \cdot 0.1\mu F \cdot (3.0^2 - 1.8^2) = 0.288\mu J
A modest radio burst at 15 mW for 2 ms requires:
E=15mW⋅2ms=30μJE = 15mW \cdot 2ms = 30\mu J
Required ideal capacitance from 3.0 V to 1.8 V:
C=2EVHI2−VLO2=10.4μFC = \frac{2E}{V_{HI}^{2} - V_{LO}^{2}} = 10.4\mu F
That is about 104 nominal 0.1 uF GRM02 capacitors before DC-bias derating, temperature tolerance, leakage, ESR, rectifier loss, and MCU/mic overhead. This is incompatible with the target volume.
Mechanical Volume Check
Using only the two largest active devices from datasheet/package data:
  • Microphone: 4 mm x 3 mm x 1.1 mm = 13.2 mm^3.
  • EFR32MG22 4 mm x 4 mm x 0.85 mm representative QFN height estimate = 13.6 mm^3.
  • Combined: approximately 26.8 mm^3 before coil, rectifier, storage capacitors, antenna/matching, interposer, wire bonds, encapsulation, acoustic cavity, and assembly tolerances.
The full system target is 8 mm^3 for a 2 mm cube, so the packaged active devices alone exceed the volume by more than 3x.
Flash Data Integrity Issue
Writing internal flash every few milliseconds is not viable for audio capture:
  • Page erase is about 12.9 ms typical, longer than a few-ms write interval.
  • Endurance is 10,000 erase cycles. At repeated millisecond-scale updates, hot pages would wear out quickly.
  • Audio data volume is high; even 16 kHz x 16-bit mono raw audio is 256 kbit/s = 32 kB/s.
  • Better options: RAM buffering plus occasional commit, event-only audio snippets, external FRAM/MRAM, or no local audio storage.
Harvesting Realism
  • Ambient RF harvesting at a 2 mm antenna/coil scale is generally nW to low-uW unless a dedicated power transmitter is very close.
  • 2.4 GHz wavelength is about 125 mm; a 2 mm antenna is about lambda/62 and very inefficient.
  • Inductive coupling is more plausible than far-field RF, but only with close alignment, resonant coil design, and a stronger external field.
  • A diode bridge is likely too lossy for weak harvested sources; a dedicated energy-harvesting PMIC or custom rectifier is required.
Option A — Practical Proof-of-Concept PCB
  • Increase board/package size substantially, likely at least 10 mm to 20 mm class.
  • Use the EFR32MG22 or similar wireless MCU, SPK0641HT4H-1 microphone, proper decoupling, crystal/load network, antenna matching, and a real energy-harvesting front end.
  • Use a larger reservoir capacitor or micro-supercapacitor.
  • Useful for firmware, RF power cycling, audio buffering, and protocol validation.
Option B — True 2 mm Research Prototype
  • Treat as custom SiP/bare-die project, not standard PCB.
  • Replace packaged MEMS microphone and EFR32MG QFN with bare die or custom ASIC.
  • Use custom integrated rectifier/PMIC, thin-film or trench capacitor, and chip-scale coil/antenna.
  • Use event-triggered or very short audio snippets, not continuous high-fidelity audio.
Option C — NFC/RFID-Like Audio Token
  • Use near-field inductive power only.
  • No 2.4 GHz active RF burst unless energy budget permits.
  • Offload most processing/storage to the reader; node captures short snippets or acoustic events.
Decision
Do not proceed with schematic capture for the original exact 2 mm off-the-shelf design. It would violate the mechanical envelope and energy budget. Proceed only after choosing a revised target: larger proof-of-concept PCB, custom chip-scale assembly, or reduced-function NFC/RFID-like node.

    Executive Conclusion

    Datasheet / Library Findings

    Knowles SPK0641HT4H-1 PDM Microphone

    Silicon Labs EFR32MG22-Class Wireless SoC

    Skyworks SMS7621-060 Schottky Diode

    Murata GRM022D80G104ME15L 01005 MLCC

    Energy Buffer Calculation

    Mechanical Volume Check

    Flash Data Integrity Issue

    Harvesting Realism

    Recommended Revised Architecture Options

    Option A — Practical Proof-of-Concept PCB

    Option B — True 2 mm Research Prototype

    Option C — NFC/RFID-Like Audio Token

    Decision

Documents

    RF-Harvested Audio Node — Project Specification

    Feasibility and Power Analysis — RF-Harvested 2 mm Audio Node

    Proof-of-Concept Design Notes

    Firmware Starter — EFR32MG22 POC Audio Harvester

Assets

No assets added.

RF-Harvested Audio Node

RF-Harvested Audio Node thumbnail
Battery-less 2 mm cube audio node concept powered by inductive/RF harvesting, with MEMS microphone capture, RF SoC burst transmission, rectification, and ceramic buffer storage.

Properties

Properties describe core aspects of the project.

Pricing & Availability

Distributor

Qty 1

Arrow

$4.65–$5.85

Digi-Key

$5.94–$6.17

HQonline

$0.95

LCSC

$12.35–$12.41

Mouser

$8.20

TME

$0.63

Verical

$3.71–$81.04

Controls