Arduino Nano 33 BLE Sense Rev2
The Arduino Nano 33 BLE Sense Rev2 packs a Nordic nRF52840 ARM Cortex-M4 at 64MHz with BLE 5.0 and seven onboard sensors — a 9-axis IMU, microphone, gesture sensor, barometric pressure, and humidity — into the Nano form factor. It is purpose-built for TinyML and sensor fusion projects where the sensors are the product, not add-ons.
Best for TinyML and sensor fusion projects with onboard sensors, skip if you need WiFi or external sensor flexibility.
Where to Buy
Pros
- Seven onboard sensors — 9-axis IMU, MEMS microphone, gesture, light, proximity, pressure, and humidity
- Optimized for TinyML with TensorFlow Lite Micro support via Arduino library
- BLE 5.0 via nRF52840 — excellent for wearables and wireless sensor projects
- 1MB flash holds TinyML models without external storage
Cons
- No WiFi — BLE only for wireless communication
- 64MHz ARM Cortex-M4 is slower than ESP32-S3 for inference workloads
- Onboard sensors are fixed — you cannot swap them for different sensors
- Micro-USB instead of USB-C
The Onboard Sensor Array
The Rev2 includes seven sensors soldered directly to the board: a BMI270 accelerometer/gyroscope and BMM150 magnetometer form a 9-axis IMU for motion tracking. The APDS-9960 provides gesture detection, ambient light sensing, and proximity measurement. An MP34DT06JTR MEMS microphone captures audio. The LPS22HB measures barometric pressure, and the HS3003 measures temperature and humidity.
This sensor density is unique — no other board in this comparison includes any onboard sensors. For TinyML projects where you need training data from multiple sensor types simultaneously, having everything pre-wired and calibrated eliminates weeks of hardware integration work.
TinyML Capabilities
Arduino's TensorFlow Lite Micro library targets the Nano 33 BLE Sense specifically. The 1MB flash stores compressed TFLite models, and the 256KB SRAM holds inference buffers. The 64MHz Cortex-M4 runs simple classification models — gesture recognition, keyword spotting, anomaly detection — at real-time speeds.
For larger models or image-based ML, the ESP32-S3 with 8MB PSRAM and 240MHz dual-core is significantly more capable. The Nano 33 BLE Sense's advantage is the integrated sensor suite — you can go from unboxing to collecting training data in minutes.
Full Specifications
Processor
| Specification | Value |
|---|---|
| Architecture | ARM Cortex-M4 |
| CPU Cores | 1 |
| Clock Speed | 64 MHz |
Memory
| Specification | Value |
|---|---|
| Flash | 1 MB |
| SRAM | 256 KB |
Connectivity
| Specification | Value |
|---|---|
| Bluetooth | 5.0 |
I/O & Interfaces
| Specification | Value |
|---|---|
| imu | BMI270 + BMM150 (9-axis) |
| microphone | MP34DT06JTR MEMS |
| gesture_sensor | APDS-9960 (gesture, light, proximity) |
| pressure_sensor | LPS22HB |
| humidity_sensor | HS3003 |
| GPIO Pins | 14 |
| ADC Channels | 8 |
| SPI | 1 |
| I2C | 1 |
| UART | 1 |
| USB | Micro-USB (native) |
Power
| Specification | Value |
|---|---|
| Input Voltage | 5 V |
| operating_voltage | 3.3 V |
Physical
| Specification | Value |
|---|---|
| Dimensions | 45 x 18 mm |
| Form Factor | Arduino Nano |
Who Should Buy This
Onboard 9-axis IMU (BMI270 + BMM150) captures motion data. APDS-9960 detects hand gestures. TensorFlow Lite Micro runs classification models on the 64MHz M4. No external sensors to wire.
The onboard MP34DT06JTR MEMS microphone captures audio. 1MB flash stores TFLite keyword spotting models. Arduino's TensorFlow Lite library simplifies deployment.
No WiFi — BLE only. The ESP32-S3 with external BME280 provides WiFi + BLE + environmental sensing with more memory for web dashboards.
Better alternative: ESP32-S3-DevKitC-1
Frequently Asked Questions
What sensors are on the Arduino Nano 33 BLE Sense Rev2?
BMI270 + BMM150 (9-axis IMU), MP34DT06JTR (MEMS microphone), APDS-9960 (gesture, light, proximity), LPS22HB (barometric pressure), and HS3003 (temperature + humidity). Seven sensors total.
Can the Nano 33 BLE Sense run TensorFlow Lite?
Yes. Arduino provides an official TensorFlow Lite Micro library optimized for this board. The 1MB flash stores models, and the 64MHz M4 runs inference. Best for small models — gesture recognition, keyword spotting, anomaly detection.
Nano 33 BLE Sense vs ESP32-S3 for ML?
The ESP32-S3 is faster (240MHz vs 64MHz) with more memory (8MB PSRAM vs 256KB SRAM) and supports camera input. The Nano 33 BLE Sense has seven onboard sensors for immediate data collection. Choose based on whether you need integrated sensors or raw compute power.
Does it have WiFi?
No. The nRF52840 provides BLE 5.0 only. For WiFi, choose the Arduino Uno R4 WiFi, Arduino Nano ESP32, or any ESP32 board.
Can I use external sensors with the Nano 33 BLE Sense?
Yes. It has SPI, I2C, and 8 ADC channels for connecting external sensors alongside the onboard ones. The onboard sensors use I2C addresses that you cannot change, so check for conflicts with external I2C devices.