From Zero to Centimeter Precision: A Complete Guide to Building an RTK System with UM982

Practical guide to building a centimeter‑accurate RTK positioning system using the UM982 module – with hardware tips, software tuning, and real‑world solutions.

In precision agriculture, drone navigation, and robotics, centimeter‑level positioning accuracy has become an industry necessity. The UniCore UM982 module delivers a cost‑effective solution with its full‑constellation, full‑frequency support, dual‑antenna heading, and on‑chip RTK technology. This guide takes you through the technical core of the UM982—from hardware connections to algorithm optimization—to build a complete centimeter‑level positioning system.

1. UM982 Core Technology Overview

As a leading GNSS module from China, the UM982 performs at the forefront of global standards. Built on the NebulasIV chip with 22nm process, it integrates 1408 super channels and simultaneously tracks BDS, GPS, GLONASS, Galileo, and QZSS across all frequency bands.

Key performance metrics comparison:

Note: 1 ppm adds 1 mm error per kilometer of baseline, suitable for wide‑area mobile applications.

The UM982’s Dual‑RTK engine is a key advantage. It allows independent processing of primary and secondary antennas and cross‑validates results for reliability. Field tests show heading accuracy of 0.1° with a 1 m baseline, ideal for flight control and agricultural machinery guidance.

2. Hardware System Integration

A complete RTK system requires careful hardware design. Below is a proven configuration:

Core components:

• UM982 module (primary + secondary antenna)
• Industrial‑grade GNSS antennas (right‑hand circular polarization ceramic recommended)
• 4G DTU module (for CORS differential data)
• Power management circuit (5V/2A input, low‑noise LDO)
• Surge protection (TVS + gas discharge tube)
• Shielding can (to reduce EMI)

# Example power configuration with TPS7A4700 LDO
power_config = {
    "input_voltage": 5.0,      # V
    "output_voltage": 3.3,     # V
    "max_current": 1.5,        # A
    "noise_level": 4.2,        # μVrms
    "psrr": 78,                # dB @ 1kHz
    "enable_pin": "GPIO12"
}

Antenna placement golden rules:

• Primary‑secondary antenna spacing ≥1 m (can relax to 30 cm for drones)
• Keep ≥5 mm clearance between antenna plane and metal surfaces
• Avoid large metal reflectors near antennas
• Prefer mounting with clear sky view

Real‑world lesson: In one project, positioning drift occurred due to motor interference. Spectrum analysis revealed strong 2.4 GHz EMI. Solutions: add ferrite cores to antenna cables, shift PWM frequency to 1.8 GHz, and apply metal shielding.

3. Software Configuration & Optimization

Proper configuration unlocks the UM982’s full potential. These validated settings improve real‑world performance:

$CMD,CONFIG,NMEA,OUTPUT,RATE,GGA,1*2F
$CMD,CONFIG,NMEA,OUTPUT,RATE,RMC,1*21
$CMD,CONFIG,NMEA,OUTPUT,RATE,GSV,5*2B
$CMD,CONFIG,NMEA,OUTPUT,RATE,GST,1*2E
$CMD,CONFIG,NMEA,OUTPUT,RATE,HDT,10*2A

RTK parameter tuning:

For precision agriculture, we developed a dynamic tuning algorithm that adjusts parameters based on speed and environment, improving UTM coordinate stability by 40% in high‑speed seeding scenarios.

4. Troubleshooting Common Issues

Cold start timeout checklist:

• Check antenna impedance (should be ≈50 Ω)
• Verify 3.3V power ripple (<50 mVpp)
• Confirm ephemeris validity via $GPGGA
• Test RF front‑end gain (LNA ≥28 dB)
• Update firmware to ≥V2.1.5

Common error codes:

In a drone project, frequent E25 errors were traced to 4G network jitter. Mitigations included local differential data buffering (≥5 s), TCP retransmission, a backup LoRa link, and IMU‑based short‑term position prediction.

5. Advanced Application Scenarios

Agricultural autosteering integration:

• CAN bus architecture at 500 kbps
• J1939 protocol parsing
• PID steering control
• AB line navigation logic
• Millimeter‑wave radar obstacle detection

// Simplified steering control example
void SteeringControl(Position current, Position target) {
    float cross_track = CalculateCTE(current, target);
    float heading_err = NormalizeAngle(target.heading - current.heading);

    float steer_angle =
        0.8 * atan(2.0 * WHEELBASE * cross_track / pow(LOOKAHEAD_DIST, 2))
        + 0.2 * heading_err;

    CAN_Send(STEER_PID, Kp*steer_angle + Kd*(steer_angle - last_angle));
    last_angle = steer_angle;
}

Drone swarm cooperative positioning: Master node runs RTK, slaves use relative positioning, 5G time synchronization, distributed Kalman filtering, and vision‑aided compensation. In a Jiangsu smart farm, 12 UM982‑based drones achieved ±2.5 cm cooperative accuracy with 95% spray overlap, improving efficiency by 300% over single‑unit operation.

6. Performance Testing & Validation

Static baseline test: Base station (UM982 + choke ring antenna) and rover with baseline 10–50 m, sampling ≥4 hours, recording raw observables and position solutions.

Dynamic test benchmarks:

In a rigorous test with a 2 km baseline, we achieved 0.6 cm + 0.3 ppm horizontal accuracy, validating UM982’s long‑baseline performance for applications like bridge monitoring.

7. Emerging Technology Fusion

Multi‑sensor fusion unlocks new capabilities. Ongoing experiments include:

• Visual‑Inertial‑GNSS tight coupling: UM982 provides global reference, IMU offers high‑rate attitude, visual SLAM builds local maps, factor graph optimization fuses data with adaptive weighting.
• 5G + RTK hybrid positioning: Combines UM982 raw measurements, 5G base station ranging, and inertial data for seamless indoor‑outdoor continuity.

In a robotics project, this hybrid approach improved positioning continuity by 90% in transitional environments like tunnels and warehouses.

With the rollout of B2b‑PPP services and QZSS L6E support, UM982 will offer centimeter‑level service even without local base stations, expanding its reach to remote and maritime applications.

Resources & Development Board

Explore the open‑source hardware design and firmware examples:

• YanTechHub / NavCore-UM982-RTK-GNSS-Board on GitHub