Battery Analytics
Mathematical models proving 2.1 years of operational longevity on a single 18650 lithium-ion cell.
Power Consumption by Tier
A visual breakdown of the cascade's active current draw. Note how the device rapidly jumps from microamps to 200mA, but only for a fraction of a second.
| Tier | State / Action | Current Draw | Duration per Trigger |
|---|---|---|---|
| 01 | Deep Sleep (Base State) | ~0.01 mA (10 µA) | 99.9% of the time |
| 02 | Hardware Interrupt | ~80 mA | ~0.2 seconds |
| 03 | Microphone + Edge AI (TinyML) | ~82 mA | ~2.0 seconds |
| 04 | OV2640 Capture + Flash LED | ~150 mA | ~1.0 seconds |
| 05 | WiFi TX (Telegram POST) | ~200 mA | ~2.0 seconds |
Total Active Time per Alert: ~5.2 seconds.
Average Active Current: ~135 mA for 5.2 seconds.
Because the device sleeps immediately after Tier 5, the absolute active energy used per event is incredibly small.
Battery Life Scenarios
Assuming a standard 18650 Li-ion battery (3000 mAh, derated to 2400 mAh) and 2 false alarms per day.
~3 Days
If built out-of-the-box, the inefficient AMS1117 voltage regulator on the ESP32-CAM and the LM393 chips + LEDs on the sensor modules drain power continuously.
~39 Days
By bypassing the sensor circuit boards and wiring the bare sensor tips directly to the ESP32 pins with high-value pull-down resistors, the LM393 chips and LEDs are eliminated.
~2.1 Years
To reach commercial-grade battery life, the inefficient AMS1117 voltage regulator is desoldered and replaced with an ultra-low-dropout MCP1700 or HT7333 regulator.