MPU6050 with ESP32: Wiring, Code and Calibration

What is the MPU6050 and what does it measure?

The MPU6050 is an inertial measurement unit, or IMU, with six degrees of freedom: it measures linear acceleration and the effect of gravity on the X, Y, and Z axes, and reports angular velocity on the same three axes. All six ADC channels have sixteen bits. This piece does not have a magnetometer; Therefore, the absolute direction to the north cannot be obtained from its raw data alone.

Robot balance control, tilt detection, motion recording, motion controller, vibration monitoring and roll and pitch estimation are suitable applications. For a stable angle, gyroscope and accelerometer data should be combined with an algorithm such as Complementary or Kalman. Simple integration of the gyroscope drifts over time due to bias and noise.

Appropriate applications and important limitations

How to read the MPU6050 datasheet correctly?

First, see the difference between MPU-6000 and MPU-6050: MPU-6000 has I2C and SPI interface, but MPU-6050 only supports I2C and has a separate VLOGIC pin. The numbers below are for the raw MPU6050 chip, not necessarily the power input of a ready-made module.

A 16-bit output does not imply angular accuracy of 1/65536 of a revolution. Noise, initial bias, temperature-dependent bias, scale-factor error, mechanical mounting and vibration all affect the result. The gyroscope datasheet specifies initial zero-rate output tolerance and its temperature variation, so measuring bias at the start of a project is essential.

Ranges, sensitivity and filter selection

Smaller range gives more numerical sensitivity but saturates in large movement. For the training sample, ±4g and ±500°/s are suitable balance; This is not a fixed version choice for all projects. If your bot has a hard hit or fast spin, record the actual Peak before increasing the range.

The Adafruit library offers bandwidth choices of 260, 184, 94, 44, 21, 10, and 5 Hz. The 21Hz filter in the sample code reduces noise on the table, but may be slow to capture fast motion. Design bandwidth, sampling rate and control delay together.

The difference between raw chip, GY-521 and Adafruit board

The VDD limit of 2.375 to 3.46 volts belongs to the chip itself. The Adafruit board shown in the picture has a regulator and level shifter according to the official manual, its Vin operates at 3-5V and it has a tenk ohm pull-up resistor on SDA and SCL. This specification should not be extended to any GY-521 blue board or unnamed module.

Some GY-521s have a feed regulator, but Level Shifting and Pull-up may be different. Before applying 5V, check the schematic of the same version and the actual SDA/SCL voltage. A safe and simple connection for the ESP32 is to power the document board with 3.3V and keep the I2C logic at 3.3V.

Pins, I2C Address and Key Registers

The official Register Map defines the register SMPLRT_DIV in 0x19, CONFIG in 0x1A, GYRO_CONFIG in 0x1B, and ACCEL_CONFIG in 0x1C. Acceleration, chip temperature, and gyro continuous data are from 0x3B to 0x48. PWR_MGMT_1 is in 0x6B and WHO_AM_I is in 0x75.

Wiring MPU6050 with ESP32 and Arduino Uno

GPIO21 and GPIO22 are just default Generic ESP32. On ESP32-S3, C3 and custom boards, replace the actual pins on your board. The official Arduino-ESP32 documentation recommends using Wire.setPins before begin or the proper form of Wire.begin(SDA,SCL) to change pins.

Step-by-step setup in Arduino IDE

1. From the Library Manager, install the Adafruit MPU6050, Adafruit Unified Sensor and Adafruit BusIO libraries.
2. Select the correct ESP32 or Arduino Uno board and port.
3. Do the wiring with power off and then check the module voltage and I2C line level.
4. Run the i2c-scanner.ino file; 0x68 should be seen with AD0 down.
5. Place the sensor on a fixed surface, upload the original code and open the Serial Monitor at 115200 baud.
6. Do not touch the sensor until the end of 500 calibration samples; Then slowly move each axis and check the units.

MPU6050 practical code for ESP32 and Arduino

#include <Wire.h>
#include <Adafruit_MPU6050.h>
#include <Adafruit_Sensor.h>

#if defined(ESP32)
constexpr int SDA_PIN = 21;
constexpr int SCL_PIN = 22;
#endif

constexpr uint8_t MPU6050_ADDRESS = 0x68;
constexpr uint16_t CALIBRATION_SAMPLES = 500;
Adafruit_MPU6050 mpu;
float biasX = 0, biasY = 0, biasZ = 0;

void calibrateGyro() {
  sensors_event_t a, g, t;
  for (uint16_t i = 0; i < CALIBRATION_SAMPLES; ++i) {
    mpu.getEvent(&a, &g, &t);
    biasX += g.gyro.x;
    biasY += g.gyro.y;
    biasZ += g.gyro.z;
    delay(5);
  }
  biasX /= CALIBRATION_SAMPLES;
  biasY /= CALIBRATION_SAMPLES;
  biasZ /= CALIBRATION_SAMPLES;
}

void setup() {
  Serial.begin(115200);
#if defined(ESP32)
  Wire.begin(SDA_PIN, SCL_PIN);
#else
  Wire.begin();
#endif
  Wire.setClock(100000);

  if (!mpu.begin(MPU6050_ADDRESS, &Wire)) {
    Serial.println(F("MPU6050 not found"));
    while (true) delay(10);
  }
  mpu.setAccelerometerRange(MPU6050_RANGE_4_G);
  mpu.setGyroRange(MPU6050_RANGE_500_DEG);
  mpu.setFilterBandwidth(MPU6050_BAND_21_HZ);
  calibrateGyro();
}

void loop() {
  sensors_event_t a, g, t;
  mpu.getEvent(&a, &g, &t);
  Serial.print(F("aX=")); Serial.print(a.acceleration.x, 3);
  Serial.print(F(" aY=")); Serial.print(a.acceleration.y, 3);
  Serial.print(F(" aZ=")); Serial.print(a.acceleration.z, 3);
  Serial.print(F(" | wX=")); Serial.print(g.gyro.x - biasX, 4);
  Serial.print(F(" wY=")); Serial.print(g.gyro.y - biasY, 4);
  Serial.print(F(" wZ=")); Serial.println(g.gyro.z - biasZ, 4);
  delay(100);
}

Code analysis and output unit

• The condition #if defined(ESP32) specifies the Generic ESP32 pins; Uno uses Wire hardware foundations.
• Wire.setClock(100000) is a conservative board to start with and stays below the 400kHz datasheet ceiling.
• mpu.begin(0x68, &Wire) checks the WHO_AM_I ID through the library and performs initialization.
• The acceleration output of the library is in m/s² and the gyroscope output is in rad/s; These are not LSB raw code.
• The average of 500 samples only estimates the gyroscope bias at rest and the instantaneous temperature. This method does not correct for scale factor, misalignment, or temperature variations.
• The constant bias of the accelerometer is not reduced, because even in the stationary state, the axis parallel to gravity sees approximately one g, and it is wrong to blindly remove it.

What is the use of the internal temperature sensor?

Temperature registers sit between the accelerometer and gyroscope data, and the datasheet gives the Temperature = TEMP_OUT / 340 + 36.53 relation to convert the raw code to Celsius. This temperature is related to the inside of the chip and does not take the place of the accurate ambient temperature sensor.

Its useful application is recording the temperature of the IMU itself along with the gyroscope bias. For accurate product, one can measure the bias at several controlled temperatures and build a compensation model; An offset is not guaranteed at room temperature for the entire -40 to +85 degree range.

Common errors and troubleshooting of MPU6050

PCB design tips, power supply and mechanical installation

• Place the decoupling capacitors and feed path according to the Typical Operating Circuit datasheet near the chip.
• Install the IMU near the rigid part of the PCB; Board bending and screw tension can change MEMS bias.
• Avoid placing the IMU next to a motor, power inductor, or local heat source without testing.
• document the axes printed on the module with the product coordinate device; Apply software 90 degree rotation explicitly.
• For long cable, check bus capacitance, pull-up and SDA/SCL waveform with oscilloscope; 400kHz is an interface ceiling, not a requirement.
• Use fixed timing or Data Ready interrupt for real-time control; delay(100) is for educational display only.
• Perform the final calibration after assembly in the actual frame and within the product's temperature range.

conclusion

A reliable setup of the MPU6050 is not just about connecting four wires. You should consider chip and module difference, AD0 address, measurement range, bandwidth, output unit, gyro bias and mechanical installation at the same time. The code in this article is a repeatable baseline for six-axis reading; The next step is to estimate the angle, design the Sensor Fusion and calibrate it to fit the actual product.

Frequently asked questions

Detailed sources and follow-up

• TDK Official Page for MPU-6050 and Obsolete Status
• MPU-6000/MPU-6050 Official Datasheet, Revision 3.4
• Official MPU-6000/MPU-6050 Register Map in Documents section of TDK page
• Adafruit MPU6050 Board Official Manual
• Official BSD-licensed Adafruit MPU6050 library
• Official documentation for I2C on Arduino-ESP32
• Photo Credit: Bryan Siepert, License CC BY-SA 3.0
• VL53L0X training with ESP32 and Arduino
• SHT31 training with ESP32 and Arduino
• Free course of sensors in 7 steps
• motion and position training category
• Sensors Learning Center at LCENJ

Raw chip numbers are taken from PS-MPU-6000A-00 Rev.3.4 and registers from RM-MPU-6000A-00 Rev.4.2. The 3-5V and 10kOhm pull-up specifications are extended to the Adafruit documentation board only. Obsolete status is monitored from the current TDK Product Center page.