ADI ADXL and ADXRS Chips for Drone IMU Design

A drone sensing design begins with chips: an accelerometer measures specific force, a gyroscope measures angular rate, and the processor combines those signals into an attitude estimate. For an Analog Devices design, that makes the ADXL accelerometer and ADXRS gyroscope families the primary component-selection problem. A pre-calibrated ADIS IMU module is the integrated alternative, not the main subject of this guide.

This distinction matters when buying parts. The same “three-axis accelerometer” label can describe a ±3 g analog-output device, a 20-bit low-noise digital sensor or a ±400 g impact recorder. Likewise, ADXRS products differ in axis count, range, digital versus analog output, package orientation and vibration performance. Those differences cannot be replaced by a generic “drone IMU chip” keyword.

The core comparison covers ADXL355, ADXL373, ADXL357, ADXL347, ADXL343 and ADXL335, plus the ADXRS290, ADXRS453 and ADXRS646 gyroscopes. The ADIS16470, ADIS16501, ADIS16505, ADIS16495 and ADIS16550 modules remain in a shorter section for buyers deciding whether to build a custom sensor chain or purchase a factory-calibrated six-axis assembly.

Core ADI drone sensor chip comparison #

ADXL accelerometer chips: the main selection layer #

ADXL355: low-g precision and low drift #

The ADXL355 is a low-noise, low-drift digital triaxial accelerometer with selectable ±2 g, ±4 g and ±8 g ranges, SPI/I²C interfaces, a 20-bit ADC and integrated temperature sensing. ADI specifies a 14-terminal hermetic LCC package and a −40°C to +125°C operating range. It fits low-frequency tilt, stabilization support and vibration characterization where drift and noise matter more than high-impact range.

The procurement decision must include the exact orderable suffix and packing code. It must also account for the LCC assembly process and board stress: a premium sensor can still produce poor results when mounted close to a flexing fastener or heat source.

ADXL357 and ADXL357B: wider precision ranges #

The ADXL357 and ADXL357B extend the low-noise digital architecture to ±10 g, ±20 g and ±40 g ranges. ADI publishes a 20-bit ADC, programmable filters, SPI and limited I²C support. ADI also positions ADXL357B as more suitable for tilt sensing in high-vibration environments.

This is not a direct “upgrade” from ADXL355. The correct part depends on acceleration headroom, target bandwidth and noise. Confirm whether the approved BOM permits ADXL357B; do not merge ADXL357 and ADXL357B into one undifferentiated stock line.

ADXL373: high-g event capture #

The ADXL373 is a micropower ±400 g digital accelerometer intended for impact and shock detection. ADI publishes selectable 160 Hz to 2560 Hz bandwidth, SPI/I²C access, interrupt-driven wake modes and 12-bit output at 200 mg/LSB. On a drone it can record a hard landing, collision, launch event or payload shock.

Its high range does not make it a better attitude sensor. It occupies a separate high-g event channel and should not be compared with ADXL355 solely on maximum acceleration.

ADXL343: general-purpose digital motion sensing #

The ADXL343 supports ±2 g, ±4 g, ±8 g and ±16 g ranges, SPI/I²C, a 32-level FIFO and activity, inactivity, tap and free-fall functions. It suits cost-sensitive motion detection or a secondary acceleration channel where embedded event logic reduces host workload.

It is not a specification-compatible substitute for ADXL355. Resolution format, noise, drift, package and registers differ, even though both devices expose three digital acceleration axes.

ADXL335: analog output for custom acquisition #

The ADXL335 is a ±3 g triaxial accelerometer with signal-conditioned analog outputs. One capacitor per axis sets bandwidth. The device can fit legacy controllers, teaching platforms and custom analog designs, but the ADC, reference, anti-alias filtering and sampling synchronization become system responsibilities.

The orderable forms include ADXL335BCPZ, ADXL335BCPZ-RL and ADXL335BCPZ-RL7. These suffixes describe supply format and must be carried through the RFQ when reel quantity or assembly process matters.

ADXL347: stop and verify the identity #

The requested ADXL347 code appears in historical third-party material, but this review found no current official ADI product page or datasheet that supports a reliable specification table. LimChip therefore treats it as an identity exception. The buyer should request an original label image, package marking, manufacturer datasheet and traceability, then ask engineering whether the intended code was ADXL343, ADXL345, ADXL346, ADXL357 or another device. Similar digits are not permission to substitute.

ADXRS gyroscope chips: angular-rate sensing #

ADXRS290: dual-axis stabilization #

The ADXRS290 is a dual-axis pitch-and-roll MEMS gyroscope with a ±100°/s range, SPI output, programmable filters and a published noise density of 0.004°/s/√Hz. ADI specifies less than 0.5 ms delay for a 30 Hz input at the widest bandwidth setting and positions the device for stabilization.

Its low noise is attractive for pointing and gimbal work, but the device does not measure yaw and ±100°/s can saturate in an aggressive maneuver or crash. A three-axis design needs another yaw-rate channel and careful time alignment.

ADXRS453: single-axis digital gyro for vibration #

The ADXRS453 measures a single angular-rate axis up to ±300°/s and sends a 16-bit rate word within a 32-bit SPI message. ADI publishes 0.01°/s/g vibration rejection, 16°/hour null-bias stability and internal temperature compensation.

It is offered in a conventional package for yaw response and a vertical-mount option that can orient the sensing axis for pitch or roll. Procurement must therefore verify the complete package code; “ADXRS453” without orientation and package context is not enough for production release.

ADXRS646: analog yaw sensing in severe vibration #

The ADXRS646 is an analog-output yaw-rate gyroscope with a minimum ±250°/s range that can be extended to ±450°/s. ADI emphasizes vibration immunity, a typical 12°/hour bias stability figure and severe-mechanical-environment applications.

Unlike ADXRS290 and ADXRS453, it places more analog responsibility on the designer. Supply, external capacitors, output scaling, ADC performance and reference behavior must all be reviewed. It is valuable for explaining why two “gyro chips” with similar rate ranges can require very different BOMs and firmware.

Supporting context: how the chips become an IMU #

A conventional six-degree-of-freedom IMU combines:

• a triaxial gyroscope that measures angular rate around roll, pitch and yaw axes; and
• a triaxial accelerometer that measures specific force along three axes.

Many products marketed as “nine-axis IMUs” add a triaxial magnetometer. That term must be checked carefully: the magnetometer may be inside the same package, on the same module or completely separate. It is also technically inaccurate to assume that every drone magnetometer uses a fluxgate structure. Compact UAV systems commonly use semiconductor magnetometer technologies, and the exact sensing principle depends on the selected device.

A barometer, GNSS receiver, optical-flow camera, radar altimeter or visual-inertial camera may be colocated with the flight controller, but these devices are aiding sensors rather than additional IMU axes.

Supporting context: from raw motion to stable flight #

1. Synchronous sampling and timing #

The controller reads gyro and accelerometer data at a rate appropriate to the airframe dynamics and control-loop bandwidth. A high output-data rate can help, but rate alone is not quality. Sample timing jitter, sensor latency, SPI transaction time, data-ready synchronization and filtering delay can all reduce phase margin.

ADI's integrated ADIS products expose SPI-compatible communications, data-ready behavior and, depending on model, external synchronization and delta-angle or delta-velocity outputs. Those functions can simplify deterministic acquisition, but the flight-control team must still measure end-to-end latency on the intended processor and firmware.

2. Calibration and preprocessing #

Raw readings contain bias, scale-factor error, cross-axis sensitivity, noise and temperature-dependent drift. The preprocessing stage typically applies:

• factory calibration supplied by the sensor or IMU vendor;
• board- or aircraft-level bias estimation at startup;
• axis transformation from sensor coordinates to the vehicle frame;
• temperature compensation;
• notch or low-pass filtering for motor and propeller vibration; and
• plausibility checks, saturation detection and redundant-sensor voting where required.

An IMU should not be calibrated while the aircraft is moving or while the temperature is changing rapidly unless the algorithm explicitly supports that condition. A perfectly still bench calibration can also fail in flight if the mounting structure transmits a strong narrow-band vibration into the sensor.

3. Attitude estimation #

Gyroscope integration gives fast short-term attitude response, but gyro bias accumulates into angle error. Accelerometer data provides a gravity-related reference when non-gravitational acceleration is limited; aggressive turns and translational acceleration make that reference temporarily unreliable. Magnetometer data can constrain heading, but motors, high-current wiring, steel fasteners and payloads can distort the local field.

Complementary filters, extended Kalman filters and other observers combine these measurements according to their uncertainty. The estimator should reduce the weight of an aiding source when its assumptions are violated—for example, during high linear acceleration or magnetic interference—rather than treating every sensor as equally trustworthy.

4. Closed-loop correction #

The attitude estimate is compared with the commanded attitude. The controller then adjusts motor torque through the electronic speed controllers. This control action is why low noise and low drift matter, but latency and vibration rejection are equally important: clean data that arrives too late can still destabilize a fast loop.

System boundary: what the chips cannot do alone #

Three popular claims require boundaries.

First, an IMU does not guarantee autonomous return-to-home after GNSS loss. It can propagate position and velocity for a limited interval, but inertial errors grow with time. A safe return function normally needs additional constraints such as visual odometry, optical flow, terrain sensing, radio ranging, a map, airspeed or renewed GNSS reception.

Second, centimetre-level hovering is not an IMU-only capability. It generally depends on an external position reference such as RTK GNSS, visual positioning or another local ranging system, plus an appropriately designed controller.

Third, a magnetometer does not “remove” gyro drift under every condition. It provides heading information only when the local magnetic field is sufficiently known and undisturbed. Magnetic validation and calibration remain system responsibilities.

ADIS integrated IMUs: the secondary build-versus-buy option #

The following comparison uses specifications published on ADI product pages and datasheets. It is a positioning guide, not a statement that one device is universally better. Confirm the current datasheet, exact orderable suffix and lifecycle before design release.

ADIS16470: compact, wide-range motion sensing #

The ADIS16470 combines a triaxial ±2000°/s digital gyroscope with a triaxial ±40 g accelerometer. ADI specifies 8°/hour gyroscope in-run bias stability, delta-angle and delta-velocity outputs, SPI communications, programmable synchronization modes and a −25°C to +85°C operating range. ADI lists unmanned and autonomous vehicles among its applications.

Its high gyro and accelerometer ranges can be useful on vehicles that experience fast rotation, impact or aggressive maneuvering. The trade-off is that range should not be selected in isolation: a platform focused on slow, precise pointing may prioritize a lower-range model with different noise and stability characteristics. Buyers should also distinguish the ADIS16470BMLZ device from evaluation and breakout-board ordering codes.

ADIS16501: a calibrated precision module for modern integration #

The ADIS16501-2BMLZ provides a ±500°/s triaxial gyro, ±14 g triaxial accelerometer, delta-angle and delta-velocity outputs, SPI, data ready, external synchronization modes and on-demand self-test. ADI specifies 2.7°/hour gyro in-run bias stability and a −40°C to +105°C operating range.

It is a strong comparison point for industrial or professional UAV programs that need better calibrated inertial performance than a commodity flight-controller IMU but cannot accommodate a large connectorized navigation module. Integration still requires a suitable land pattern, assembly control, thermal characterization and verification of the host SPI implementation.

ADIS16505: choose the suffix from the motion envelope #

The ADIS16505 family offers three gyro-range models: ADIS16505-1BMLZ at ±125°/s, ADIS16505-2BMLZ at ±500°/s and ADIS16505-3BMLZ at ±2000°/s. The family uses a compact 100-ball BGA module and provides factory calibration, SPI, external synchronization and inertial self-test functions.

The range suffix is not a cosmetic procurement detail. A lower range may suit slow stabilization or pointing, while an agile airframe can saturate it. Conversely, specifying the widest range without examining noise and control requirements can waste performance. Put the full suffix on the approved BOM and RFQ; “ADIS16505” alone is incomplete.

ADIS16495: tactical-grade performance with a larger module #

ADI positions the ADIS16495 as a tactical-grade 6-DoF inertial sensor for guidance, navigation, control, avionics and unmanned vehicles. The family offers ±125°/s, ±450°/s and ±2000°/s variants. ADI publishes 0.8°/hour gyro in-run bias stability and 3.2 µg accelerometer in-run bias stability for the ADIS16495-1, together with factory calibration over −40°C to +85°C.

This is not an automatic upgrade for every drone. The approximately 47 × 44 mm aluminum module, connector mechanics, power integration, cost and export/end-use considerations can dominate the design. It belongs in a different sourcing conversation from a small BGA IMU used on a compact flight-control PCB.

ADIS16550: autonomous-grade positioning needs system-level justification #

The ADIS16550BMLZ is positioned by ADI as an autonomous-grade 6-DoF inertial sensor. Published features include a ±300°/s triaxial gyro, ±14 g accelerometer, 1°/hour gyro in-run bias stability, factory calibration and a connectorized aluminum package. ADI includes dead reckoning in GPS-denied environments among its applications.

That statement does not mean indefinite unaided navigation. It indicates a performance class intended to reduce inertial error growth; mission duration, vibration, temperature, initialization and aiding architecture still determine the achieved result. Procurement should also review end-use restrictions, traceability and documentation before treating this class of IMU as an ordinary commercial sensor.

Vibration, placement and thermal design #

Propeller imbalance, motor commutation, frame resonance and aerodynamic loading can inject energy well above the aircraft's useful motion bandwidth. If that energy aliases into the sampled band or interacts with MEMS nonlinearities, software filtering may not recover the true motion.

During prototype validation:

1. measure the vibration spectrum at the proposed IMU location across motor speeds; 2. confirm that mechanical isolation does not introduce an uncontrolled low-frequency resonance; 3. keep the sensor away from rapidly cycling heat sources where practical; 4. use data-ready synchronization rather than uncontrolled polling when supported; 5. confirm axis orientation and sign conventions with deliberate single-axis motion; and 6. repeat bias and attitude tests across the expected temperature range.

For magnetometer-equipped systems, route high-current motor and battery conductors away from the magnetic sensor and repeat calibration with the final payload, wiring and fasteners installed.

Procurement and incoming-inspection checklist #

Do not approve a substitute only because it has the same six axes and SPI interface. A valid alternate must be reviewed for gyro range, accelerometer range, bias stability, angular random walk, noise density, bandwidth, filter delay, vibration sensitivity, axis alignment, package, power, startup behavior, calibration and software registers.

How to prepare a sensor-chip RFQ #

An actionable RFQ should include the full manufacturer part number, quantity, target delivery date, acceptable date-code window, required packaging, destination and whether the demand is for prototype, pilot build or production. Also identify whether the line is an ADXL accelerometer, ADXRS gyroscope or ADIS module, plus the required range, interface and package. Those details prevent a purchasing team from comparing devices that sit in completely different performance and integration classes.

Availability, stock, date code, lot condition and shipment timing must be confirmed at quotation. Distributor inventory should never be treated as proof that a specific IMU has been engineering-approved for the aircraft.

Conclusion #

A drone sensing BOM starts with the actual chips. ADXL accelerometers measure specific force across very different low-g, precision and shock ranges; ADXRS gyroscopes measure angular rate with different axis counts, interfaces and vibration behavior. Combining them into a useful IMU requires timing, alignment, calibration and filtering.

The main discrete choices in this guide are low-noise ADXL355/ADXL357, high-shock ADXL373, general-purpose digital ADXL343, analog-output ADXL335, dual-axis ADXRS290, digital single-axis ADXRS453 and analog yaw-rate ADXRS646. ADXL347 should not enter an approved BOM until its identity is resolved. ADIS16470, ADIS16501, ADIS16505, ADIS16495 and ADIS16550 are integrated alternatives when factory calibration and six-axis packaging justify the higher module-level commitment.

Official references #

• Analog Devices ADIS16470 product page
• Analog Devices ADIS16501 product page
• Analog Devices ADIS16505 product page
• Analog Devices ADIS16495 product page
• Analog Devices ADIS16550 product page
• Analog Devices ADXL355 product page
• Analog Devices ADXL357 product page
• Analog Devices ADXL373 product page
• Analog Devices ADXL343 product page
• Analog Devices ADXL335 product page
• Analog Devices ADXRS290 product page
• Analog Devices ADXRS453 product page
• Analog Devices ADXRS646 product page
• Analog Devices: Game of Drones—Part 1
• Analog Devices inertial measurement unit product category