ESP32-S3 DevKit & Dual LX7

Digital GPIO Matrix (45 Pins)

45 programmable GPIO pins with ultra-flexible IO MUX matrix operating at 3.3V logic.

• All output pins support hardware LEDC PWM up to 40 MHz with independent timers
• Direct connection to hardware SPI, I2S, I2C, and UART peripherals across any pins
• Strapping pins (GPIO 0, 45, 46) configure boot mode and internal LDO voltage

Analog & Capacitive Touch (ADC1 / ADC2)

Dual 12-bit SAR ADCs with up to 20 analog channels and 14 capacitive touch inputs.

• ADC1 (GPIO 1–10) functions reliably during active Wi-Fi data transfers
• ADC2 (GPIO 11–20) shared with wireless baseband; use ADC1 for precision telemetry
• 14 capacitive touch pins capable of detecting human touch through 10mm acrylic

Native USB, Camera & LCD Peripherals

High-speed dedicated silicon controllers for multimedia and direct bus communication.

• GPIO 19 (USB D-) and GPIO 20 (USB D+) connect directly to the native USB PHY
• Dedicated RGB/QSPI display controllers for fluid 60 FPS LVGL graphic user interfaces
• 3 independent hardware UART controllers and 2 I2C master/slave controllers

Native USB 2.0 FS

GPIO 19 (D-) & GPIO 20 (D+)

12 Mbps Full-Speed USB supporting CDC virtual serial, HID keyboard/mouse, MSC flash disk, and hardware JTAG debugging.

High-Speed SPI (FSPI/MSPI)

Octal SPI / Quad SPI

Up to 120 MHz bus speed for high-speed Octal Flash and Octal PSRAM streaming with DMA channels.

DVP Camera Interface

8-bit / 16-bit Parallel DVP

Connects to OV2640, OV3660, and OV5640 camera sensors with automatic DMA line buffering for real-time video.

I2S Audio & Voice

BCLK, WS, DOUT, DIN, MCLK

Dual I2S audio channels with PDM microphone decoding for wake-word speech detection and smart speakers.

Hardware UART (x3)

Dynamic IO MUX Mapping

3 fully independent UART controllers supporting baud rates up to 5 Mbps with automatic hardware flow control.

Two-Wire I2C (x2)

Default GPIO 8 (SDA) / 9 (SCL)

Dual I2C controllers supporting Standard (100 kHz), Fast (400 kHz), and Fast Plus (1 MHz) transmission modes.

Use the ESP-NN acceleration library with TensorFlow Lite Micro to execute INT8 matrix multiplications using dedicated Xtensa PIE vector instructions.

Pro tip: Enables 15 FPS person detection models and sub-100ms offline voice keyword spotting directly on the chip.

Debug firmware line-by-line using OpenOCD and GDB over the native USB Type-C port without purchasing an external ST-LINK or J-Link probe.

Pro tip: Set breakpoints, inspect register values, and trace FreeRTOS stack usage directly in VS Code.

Allocate massive memory buffers using heap_caps_malloc() to hold uncompressed video frames and deep learning neural network weights.

Pro tip: Always allocate large display framebuffers in SPIRAM to keep internal SRAM free for DMA and Wi-Fi stacks.

The integrated RISC-V ULP coprocessor monitors sensors and touch inputs while the dual 240 MHz main cores sleep in deep sleep at just 7 µA.

Pro tip: Wake up the main cores only when sensor readings exceed a defined analog threshold.

Drive parallel 16-bit RGB displays or SPI TFT displays at 60 FPS using the hardware RGB LCD peripheral and dual-framebuffer tearing prevention.

Pro tip: Combine LVGL 9 with ESP32-S3 DMA for smartphone-smooth scrolling menus and animations.

Cryptographically sign all application firmware using an RSA-3072 private key; the hardware bootloader verifies signatures before execution.

Pro tip: Combine Secure Boot with Flash Encryption to prevent firmware reverse engineering and physical cloning.

Edge Computer Vision & Object Detection

Pair with OV2640/OV5640 camera sensors to run TinyML image classification and person detection models directly on-device.

Offline Voice Recognition & Wake-Word Detection

Use hardware vector instructions and dual I2S microphone inputs to execute neural network voice trigger recognition without cloud latency.

Smart Touchscreen GUI Terminals

Drive 800x480 RGB/SPI displays using LVGL with 8MB Octal PSRAM acting as a continuous double-buffered frame buffer.

Secure Industrial IoT Gateways

Hardware cryptographic accelerators (AES-XTS, RSA-3072, ECC) safeguard telemetry data transmitted over Wi-Fi and Bluetooth Mesh.

// ESP32-S3 - High-speed Wi-Fi Server with JSON telemetry
#include <WiFi.h>
#include <WebServer.h>

const char* ssid = "Your_WiFi_SSID";
const char* password = "Your_WiFi_Password";
WebServer server(80);

void handleTelemetry() {
  String json = "{";
  json += "\"chip\":\"ESP32-S3\",";
  json += "\"cores\":2,";
  json += "\"cpu_freq_mhz\":" + String(getCpuFrequencyMhz()) + ",";
  json += "\"free_heap\":" + String(ESP.getFreeHeap()) + ",";
  json += "\"free_psram\":" + String(ESP.getFreePsram());
  json += "}";
  server.send(200, "application/json", json);
}

void setup() {
  Serial.begin(115200);
  WiFi.begin(ssid, password);
  while (WiFi.status() != WL_CONNECTED) {
    delay(500);
    Serial.print(".");
  }
  Serial.println("
Connected! IP: " + WiFi.localIP().toString());

  server.on("/api/stats", handleTelemetry);
  server.begin();
}

void loop() {
  server.handleClient();
}

// ESP32-S3 - Pin heavy AI/Sensor tasks to separate CPU cores
#include <freertos/FreeRTOS.h>
#include <freertos/task.h>

void aiInferenceTask(void *param) {
  while (1) {
    // Core 1 handles intensive AI neural network inferences
    vTaskDelay(pdMS_TO_TICKS(50));
  }
}

void wifiCommTask(void *param) {
  while (1) {
    // Core 0 handles wireless network packets and telemetry
    vTaskDelay(pdMS_TO_TICKS(10));
  }
}

extern "C" void app_main() {
  // Pin AI task strictly to Core 1
  xTaskCreatePinnedToCore(aiInferenceTask, "AI_Core1", 8192, NULL, 5, NULL, 1);

  // Pin Network task strictly to Core 0
  xTaskCreatePinnedToCore(wifiCommTask, "WiFi_Core0", 4096, NULL, 4, NULL, 0);
}