Loading…
QFN-24 · MPU-6500
Active6-Axis IMU (Gyro + Accel)
MPU-6500
TDK InvenSense · QFN-24 · 6-Axis IMU (Gyro + Accel)
Price on Request
Get a competitive quote within 24 hours. Volume discounts available. Shenzhen stock ships in 1-2 business days.
Overview
The MPU-6500 is TDK InvenSense's second-generation six-axis MotionTracking device, and it is the part that moved the MPU-6050 architecture forward for designs that need speed. Three MEMS gyroscopes, three MEMS accelerometers and a Digital Motion Processor share a single 3 × 3 × 0.9 mm 24-pin QFN built on InvenSense's CMOS-MEMS wafer-bonded platform, with on-chip 16-bit ADCs and signal conditioning for every axis, a 512-byte FIFO that lets the host burst-read data and then sleep, an auxiliary I2C master port for hanging an external magnetometer or pressure sensor off the same device, and a die rated to survive 10,000 g of shock.
Gyroscope specifications: programmable full-scale range of ±250, ±500, ±1000 and ±2000 °/s with sensitivity of 131, 65.5, 32.8 and 16.4 LSB per °/s respectively, ±3 percent initial sensitivity tolerance, ±0.1 percent nonlinearity, ±2 percent cross-axis sensitivity, ±5 °/s initial zero-rate offset and ±0.24 °/s per °C ZRO variation over -40 °C to +85 °C. Rate noise spectral density is 0.01 °/s/√Hz with total RMS noise of 0.1 °/s at a 92 Hz digital low-pass setting, the programmable filter spans 5 Hz to 250 Hz, and output data rate is programmable from 4 Hz to 8000 Hz. Accelerometer specifications: ±2, ±4, ±8 and ±16 g full-scale with 16,384, 8,192, 4,096 and 2,048 LSB per g, ±3 percent initial tolerance, ±60 mg zero-g offset, ±0.5 percent nonlinearity, 300 µg/√Hz noise power spectral density in low-noise mode, a 5 Hz to 260 Hz programmable filter, and output data rates of 4 Hz to 4000 Hz in low-noise mode or 0.24 Hz to 500 Hz duty-cycled in low-power mode. An on-chip temperature sensor runs at 333.87 LSB/°C.
Power and interface: gyro operation draws 3.2 mA, the accelerometer 450 µA at 4 kHz, all six axes together 3.4 mA, low-power accelerometer motion detection as little as 7.27 µA at 0.98 Hz update rate, and full-chip sleep 6 µA. VDD and VDDIO both run from 1.71 V to 3.6 V, so the part interfaces directly to 1.8 V, 2.5 V or 3.3 V hosts. Register access uses I2C at up to 400 kHz with slave address b110100X — 0x68 when AD0 is low and 0x69 when AD0 is high, which allows two devices on one bus — or SPI at 1 MHz; sensor and interrupt registers can additionally be read over SPI at up to 20 MHz, which is the mode high-rate flight controllers use. The DMP supports gesture recognition, programmable interrupts, angular position output and low-power pedometer functions while the host sleeps.
Two sourcing facts matter. First, identification: the WHO_AM_I register (0x75) returns 0x70 on a genuine MPU-6500, versus 0x68 on an MPU-6050, 0x71 on an MPU-9250 and 0x12 on an ICM-20602. This single read is the most reliable field check and it catches the most common fraud in this category — MPU-6050 dies remarked as MPU-6500, and breakout boards labelled MPU-6500 that carry MPU-6050 or clone silicon. Second, lifecycle: TDK's product page marks the MPU-6500 as not recommended for new designs and steers new sockets toward the ICM-20602, ICM-20608 and ICM-42688-P, which are available in the same 3 × 3 mm QFN-24 footprint and offer better noise and lower power. We list the part as Active and hold verified stock, but for a new long-life product we recommend designing in an ICM-206xx with the MPU-6500 as an alternate. Note also that the 3 × 3 mm package is not footprint-compatible with the 4 × 4 mm MPU-6050, so this is never a drop-in board swap. MOQ is 500 pcs on tape-and-reel or 50 pcs loose, shipping in 1-3 days from Shenzhen.
Key Features
- 6-axis IMU: 3-axis gyro ±250 to ±2000 °/s and 3-axis accel ±2 g to ±16 g
- 16-bit ADCs on every axis with programmable digital filters and ODR to 8 kHz
- SPI sensor reads at up to 20 MHz, or I2C at 400 kHz with address 0x68 / 0x69
- 512-byte FIFO, auxiliary I2C master port and on-chip DMP with pedometer support
- 3 × 3 × 0.9 mm QFN-24, 1.71 V to 3.6 V, 3.4 mA with all six axes active
Typical Applications
- Multirotor flight controllers and stabilised camera gimbals
- Self-balancing robots, Segway-style platforms and rover attitude control
- Wearable motion tracking, step counting and gesture recognition
- 3D mice, VR controllers and motion-based game peripherals
- Sports telemetry: racket swings, cycling power meters, ski and surf sensors
Specifications
- Gyroscope Full-Scale Range
- ±250 to ±2000 °/s
- Accelerometer Full-Scale Range
- ±2 to ±16 g
- ADC Resolution
- 16 bits
- Gyro Rate Noise Density
- 0.01 °/s/√Hz
- Digital Interface
- I2C (400 kHz) / SPI (1 MHz, 20 MHz sensor read)
- Supply Voltage (VDD / VDDIO)
- 1.71 to 3.6 V
- Supply Current (6-Axis Active / Sleep)
- 3.4 mA / 6 µA
- Operating Temperature
- -40 to 85 °C
- Pins
- 24
- RoHS
- Compliant
Procurement Tips
Require a WHO_AM_I (0x70) test record with every MPU-6500 lot, and add an ICM-20602 alternate to the BOM while the footprint is still the same 3 × 3 mm QFN-24 — it protects the build once TDK's NRND status turns into an allocation problem.
Frequently Asked Questions
How does the MPU-6500 differ from the MPU-6050?
Same 24-pin QFN family and the same six-axis concept, but the MPU-6500 adds a 20 MHz SPI sensor-read path (the MPU-6050 is I2C-only at 400 kHz), a wider 1.71 V to 3.6 V supply versus 2.375 V to 3.46 V, accelerometer output data rates to 4000 Hz, lower motion-detection current and a smaller 3 × 3 mm package versus 4 × 4 mm. Register maps differ, WHO_AM_I returns 0x70 instead of 0x68, and the packages are not footprint compatible, so it is a redesign rather than a swap. Choose the MPU-6500 or an ICM-206xx when you need SPI throughput for high-rate control loops.
Does the MPU-6500 have a DMP, and is the firmware available?
The silicon includes the Digital Motion Processor, and the product specification describes support for gesture recognition, programmable interrupts, angular-position output and low-power pedometer functions while the host sleeps. In practice, InvenSense's public MotionApps support was far better documented for the MPU-6050, so most MPU-6500 designs run sensor fusion on the host instead — a Madgwick or Mahony filter, or a complementary filter, at 200 Hz to 500 Hz. Budget engineering time for host-side fusion unless you have access to the DMP firmware image for this part.
What I2C address and SPI settings should I use?
I2C: the 7-bit slave address is b110100X, so 0x68 with AD0 low and 0x69 with AD0 high — two devices can share a bus, but AD0 must never float. The maximum bus speed is 400 kHz. SPI: chip select on nCS, AD0 becomes SDO, and register access runs at up to 1 MHz while sensor and interrupt register reads can be clocked at up to 20 MHz. When using SPI, disable the I2C slave by setting I2C_IF_DIS in USER_CTRL so the device cannot switch modes unexpectedly. Always verify WHO_AM_I (0x75) returns 0x70 before trusting any data.
How do I calibrate the gyro and accelerometer in production?
For the gyroscope, hold the board perfectly still on a level surface, collect 1,000 to 4,000 samples and average them; the result is the zero-rate offset, which you store and subtract — the ±5 °/s initial tolerance makes this mandatory for anything beyond a hobby demo. For the accelerometer, capture readings in six orientations (each axis pointing up and down), then solve for a three-axis offset and scale matrix. Temperature compensation helps further, since ZRO drifts ±0.24 °/s per °C. Store the coefficients in flash or EEPROM and re-verify them after any thermal process such as reflow or conformal coating.
MPU-6500 versus ICM-20602, MPU-9250 or BMI160 — what should a new design use?
The ICM-20602 and ICM-20608 are TDK's direct successors: same 3 × 3 mm QFN-24 footprint, Active lifecycle, lower noise and better bias stability than the MPU-6500. The MPU-9250 adds a co-packaged three-axis magnetometer for a nine-axis solution but is itself approaching end of life, so prefer ICM-20948 if you need magnetics. The Bosch BMI160 is a strong low-cost alternative with lower power and an integrated step counter. For a new long-life design choose an ICM-206xx or BMI160 and keep the MPU-6500 as an alternate source; for an existing MPU-6500 board, verified stock remains available.
Is the MPU-6500 obsolete or still in production?
TDK's product page carries a not-recommended-for-new-designs status marker and the company directs new sockets to the ICM-20602, ICM-20608 and ICM-42688-P, but the part is still manufactured and remains widely available in volume, which is why we list it as Active. Practically: it is safe to buy for existing designs, spare parts and moderate-lifetime products, but you should qualify a successor now rather than face a forced redesign later. We can supply both and help you map the register differences.
How do I detect counterfeit or remarked MPU-6500 parts?
Read WHO_AM_I at register 0x75: 0x70 means genuine MPU-6500, 0x68 means an MPU-6050 die, 0x71 means MPU-9250, 0x12 means ICM-20602 and anything else points to a clone. Then run a functional check: verify that all four gyro full-scale settings change the sensitivity by the expected factors of 2, that the accelerometer reports close to 1 g on the vertical axis at 16,384 LSB/g in the ±2 g range, and that the temperature sensor tracks ambient. Physically inspect the 3 × 3 mm QFN lid marking and the laser-etched lot code. Breakout boards labelled MPU-6500 very often carry MPU-6050 silicon, so always test the chip ID.
What is the MOQ and lead time?
MOQ is 500 pcs on tape-and-reel for the QFN-24, or 50 pcs loose for prototyping and small runs. Shenzhen stock of chip-ID-verified parts ships in 1-3 business days; the TDK authorized channel is typically 12-20 weeks. For volume builds we recommend adding an ICM-20602 or ICM-20608 alternate on the same BOM line so your assembler has a compliant second source.
Alternatives & Equivalents
View full cross-reference table →Need MPU-6500 pricing and availability?
Submit an RFQ — we reply within 24 hours with stock, pricing, lead time and packaging options.
Request a Quote →