For illustrative purposes, we will use the small 3'' multicopter depicted above, but the tuning sequence we developed at IAV GmbH will work on almost any other multicopter. This method uses the latest available ArduPilot WebTools, some of the new features of ArduCopter 4.3 and best practices from the Ardupilot Community.
We will detail the firmware parameters to change, the sequence of how to change them, help you to find the value for each parameter, explain which test flights to conduct and the order in which to conduct them and help you document all your steps for traceability. This applies industry-proven software configuration management (SCM) techniques to tuning an ArduCopter vehicle. You will be able to tune multiple vehicles (think production line) using this method and still have individual traceability of each parameter change on each vehicle.
This post is divided into the following sections:
- Getting the hardware right
- Proper initial configuration
- First flight: Motor Thrust Hover and Harmonic Notch data collection
- Second flight: MagFit
- Third flight: Evaluate the aircraft tune - part 1
- Fourth flight: Evaluate the aircraft tune - part 2
- Autotune flight(s)
- Performance evaluation flight
- Windspeed Estimation flight(s)
- Baro Compensation flight(s)
- System Identification flights
- Productive configuration
- Position controller
- Precision land
- Guided operation without RC transmitter
- Conclusion
Some rules of thumb for hardware requirements:
- Robust frame construction: A stable and rigid frame is crucial for stable and safe flight behavior. Carbon frames are recommended but not essential, and remember carbon is an electrical conductor.
- ESC telemetry: Use only ESCs that provide at least RPM telemetry. It simplifies Notch filter tuning and improves its response-time and accuracy.
- Vibration reduction: Vibrations reduce the efficiency, stability and lifespan of the drone. All propellers must be carefully balanced. They are the source of most of the vibrations. All components must be securely fastened to minimize vibrations and avoid damage caused by vibrations.
- Protection of sensors from external disturbances:
- Vibration-dampening mounting of the FC
- Separation of compass and high-current paths: To reduce electromagnetic interference, it is recommended to spatially separate the compass (likely integrated with GNSS receivers) from high-current paths and magnetic sources such as motors and power distribution systems/cables.
- Protection of barometer from airflow: The barometer must be protected from wind and airflow or turbulence generated by the propellers. A small piece of open-cell foam placed over the sensor acts as a low-pass filter, ensuring accurate altitude measurements.
- GNSS systems are likely to be affected by USB3 devices. Keep possible negative influences in mind while using USB3 components.
- Proper cable management: Cables and wires must be organized sensibly to prevent entanglement or damage during flight. It must be ensured that no cables hinder movable parts such as propellers or gimbal mechanisms, or are damaged by them. Flexible, silicone-coated cables for data transfer save weight and reduce vibration transmission. Weak connectors are prone to loosening under the influence of vibration.
- Weight distribution: An even weight distribution of the drone with the FC at the center of gravity improves stability and flight control. Components such as batteries, sensors, cameras, and other payloads must be positioned evenly to achieve uniform weight distribution and maximum fit between the geometric and physical center of gravity.
- Battery placement: The battery is often located in the center of the frame to ensure stability during flight. It must be ensured that the battery is rigidly mounted and secured to prevent slipping or unintentional disconnection during operation. Additionally, when properly attached, the battery acts as an inertial mass and helps dampen vibrations. Beware of landing directly on the battery since most of the batteries do have a resistant shell.
- Voltage monitoring: to dynamically scale the PIDs and maintain stable flight in low battery conditions.
- Current monitoring: to compensate for the dynamic magnetic field caused by the high motor currents.
- FC Power supply: Must provide enough current for the flight controller, GNSS receivers and other payloads operating on 5V.
- Roll/Pitch/Yaw-Imbalance: When mounting the motors on the arms, especially on round arms, make sure that all motors and propellers are perfectly level so that the thrust produced is directed straight down. Misaligned motors will cause the multicopter to drift. Depending on which motors are misaligned and their direction of misalignment, the multicopter drifts laterally forward, backward, left, right, or axially around the Z-axis, and efficiency is reduced accordingly.
- Helical GNSS antennas: These kinds of antennas are the preferred choice for drones and their benefits justify the extra cost.
- STM32 H7 processor family: Flight controllers that use these processors, have enough processing power and memory to run ArduCopter firmware without restrictions.
Use tools like ecalc for multirotor to find a suitable set of components to meet your needs.
.PNG)
With these requirements in mind, and for our specific application, we have selected the following components:
| Type | Part |
|---|---|
| Frame | Diatone Taycan MX-C |
| Flight Controller | Matek H743 SLIM V3 |
| ESC | T-Motor F45 4in1 ESC V2 |
| Motors | 4x Diatone Mamba 1404 5000kv |
| Propeller | 4x HQProp 3018, 3-Blade |
| BEC with voltage/current monitor | Holybro PM02 V3 |
| Battery | SLS X-Cube 4S 1800mAh 40C/80C |
| GNSS receiver | Holybro H-RTK F9P Helical |
| SDCard | Any fast Micro-SDCard > 8 GiB |
| RC Receiver | TBS Crossfire Nano RX se |
| RC Transmitter | Radiomaster TX16S with EdgeTx and Yaapu scripts |
| Remote ID transmitter | Holybro Remote ID transmitter |
We connected the components as depicted below. The figure excludes the LiPo battery and the PM02 BEC with a voltage/current monitor.
| Component | Flight controller connections |
|---|---|
| T-Motor F45 4in1 ESC V2 | G, G, Vbat, not Connected, S4, S3, S2, S1, Cur, Rx8 (SERIAL5) |
| Holybro PM02 V3 | not connected, G, Vbat2, Curr2, not Connected, not Connected |
| Holybro H-RTK F9P Helical | 5V, Tx2, Rx2, CL1, DA1, not connected, not connected, 3V3, Buzz, G |
| TBS Crossfire Nano RX se | G, 5V, Rx6, Tx6 |
For reproducibility and quality purposes, we configure the vehicle with a well-defined sequence of intermediate parameter files. Each file modifies just a small set of the over 1200 parameters on the flight controller. By splitting the process into small manageable steps, we reduce the probability of making a mistake or missing a step and allow interleaving parameter changes with test flights.
Each intermediate parameter file is a text file, editable in any common text editor (excluding MS Word) like Notepad++, nano or code. It contains the official Ardupilot parameter documentation in the form of comments in the lines preceding the parameter. By using this you save the time of looking up the online documentation for each parameter.
It contains the reason why we changed the parameter in a comment on the same line as the parameter and is used to trace each parameter change to the reason for that parameter change. Use it! You can thank me later.
Comments start with the '#' character. A small example with a single parameter is shown below:
# Arming with Rudder enable/disable
# Allow arm/disarm by rudder input. When enabled arming can be done with right rudder, disarming with left rudder.
# Rudder arming only works with throttle at zero +- deadzone (RCx_DZ). Depending on vehicle type, arming in certain
# modes is prevented. See the wiki for each vehicle. Caution is recommended when arming if it is allowed in an auto-throttle mode!
# 0: Disabled
# 1: ArmingOnly
# 2: ArmOrDisarm
ARMING_RUDDER,0 # We find it safer to use only a switch to arm instead of through rudder inputs
We provide you here with around 44 of these intermediate parameter files (*.parm). If you are working with multiple vehicles, create a separate directory for each vehicle with a descriptive identifiable name. Copy the 44 intermediate parameter files into them. Edit the files to match the specific requirements of each vehicle. Now you have traceable documentation records for every parameter change on each of your vehicles.
If you are in the business of manufacturing multicopters and maintain high-quality standards that result in the production of multiple, nearly identical vehicles, you can reuse most intermediate parameter files across these vehicles.
Only three intermediate parameter files: 03_imu_temperature_calibration_results.param, 12_mp_setup_mandatory_hardware.param and 25_inflight_magnetometer_fit_results.param are specific to each vehicle instance.
All other intermediate parameter files can be used without modifications across all instances (or serial numbers) of the same product model.
This is a list of the software used in this process
| Software | Version | Description |
|---|---|---|
| Mission Planner | latest beta | Ground control station (PC software) used for configuring and operating the vehicle |
| ArduCopter | 4.4.3 or 4.5.0-DEV | Flight controller firmware |
| BLHeliSuite32 | 32.9.06 | PC software to flash and configure ESCs with BLHeli_32 ARM firmware |
| BLHeli_32 ARM | 32.8 | ESC firmware with Bidir Dshot support |
| EdgeTx companion | 2.9.2 | PC software for configuring and updating EdgeTX based RC transmitters |
| EdgeTx | 2.9.2 | Radiomaster TX16S firmware with touch screen support |
| Yaapu scripts | 2023-11-08 | Display vehicle telemetry on the Radiomaster TX16S |
| Simple text editor: Notepad++, nano or code | any | Allows editing plain text files without undesired text changes. Do not use microsoft word! |
| python | >= 3.8 | Python interpreter used to run automation scripts |
| offline IMU temperature calibration tool | github master | Offline IMU temperature calibration script will graph the uncorrected and corrected calibration for the accelerometers and gyroscopes |
| MAVExplorer | latest | Analyze dataflash (.bin) log files |
| Scripted MagFit flightpath generation | latest | Lua script to generate a MagFit flight path |
| ecalc for multirotor | online | Finds a suitable set of components that meet your needs |
| ArduPilot Hardware Report | online | Provides an overview of connected hardware from a .bin log and visualization of sensor position offsets. |
| ArduPilot Filter Review Tool | online | Aid in configuring and validating the Harmonic Notch filters |
| ArduPilot Filter Analysis | online | Bode plot tool to give insight into gyro low-pass and notch filter attenuation and phase lag |
| ArduPilot PID Review Tool | online | Review PID tune in the frequency domain. Step response estimate is generated. |
| ArduPilot MAGFit in flight compass calibration | online | Do compass calibration using a flight log |
| Ardupilot Log Viewer | online | Log viewer, analyzer and plotter. Can also do MagFit |
| Add grid to image | online | Add a grid overlay to any image |
| SketchAndCalc | online | Calculate areas |
Always connect the vehicle battery before connecting the USB cable (if you are using one) between the PC and the flight controller. Always disconnect the USB cable (if you are using one) between the PC and the flight controller before disconnecting the vehicle battery.
Start by downloading the latest Mission Planner ground control station software for your Microsoft Windows PC and flash the latest stable ArduCopter firmware for your flight controller. Start Mission Planner on the PC and connect to the vehicle using a USB cable.
2.1 Configure IMU Temperature calibration using the offline calibration tool
For this procedure follow these steps:
- Download the 02_imu_temperature_calibration_setup.param parameter file to your PC, edit it with a text editor to meet your needs and in Mission Planner while connected to the flight controller select Compare Params, review the changes and press Continue followed by Upload Params.
- Power off the flight controller and remove the battery.
- Place the flight controller without battery in a freezer capable of reaching your vehicle's minimum expected operation temperature (-18°C in our case).
- Once the flight controller is completely cooled down to its minimum expected operation temperature, take it out and power it. Do not move the flight controller for one or two hours.
- If you have a buzzer connected, you will hear a short periodic beep while the calibration is in progress. When the calibration is complete, a completion tune will play. If you have notification LEDs on your flight controller they will cyclically flash red, blue and green while calibrating. If you do not have a buzzer connected nor notification LEDs, monitor the
INS_TCALn_ENABLEparameters. On completion, theINS_TCALn_ENABLEparameters will change to 1 (enable) for each calibrated IMU. - Power it off, and remove the micro SDCard
- Copy the latest
.binlog file in the micro SDcard from/APM/LOGSto your PC - Insert the micro SDcard back into the flight controller
- Then use the Offline Calibration Tool to calculate the value of the temperature calibration parameters.
- Download the tempcal_IMU.py python script to your PC.
- Execute
python tempcal_IMU.py <your_bin_log_file.bin> - the generated
tcal.parmfile should be similar to the 03_imu_temperature_calibration_results.param parameter file. - Upload the tool-generated
tcal.parmfile (not ours) to your flight controller.
The graphic below depicts the accelerometer drift versus time and the board temperature versus time. The temperature curve, depicted in black, is logarithmic as expected. The other curves are smooth, proving that the flight controller was not moved in the process and the calibration is valid. As can be seen, before the calibration temperature changes caused a big change in accelerometer/gyro drift. After the calibration, temperature changes will cause no significant accelerometer/gyro drift changes.
The vehicle must be completely assembled, with all cables connected, but WITHOUT propellers attached.
Follow mounting the autopilot documentation to determine the correct value of the AHRS_ORIENTATION parameter. For its configuration, we created a 04_board_orientation.param parameter file. Download it to your PC, edit it with a text editor and in Mission Planner while connected to the vehicle select Compare Params, review the changes and press Continue followed by Upload Params.
In our setup, we use EdgeTX firmware on the RC transmitter. Download and install EdgeTX Companion to your PC. Start it and configure it as depicted below.
After that, use a micro SDcard to update the firmware on the Radiomaster TX16S and copy the yaapu scripts to the /WIDGETS/yaapu directory on the micro SDcard.
Once the RC transmitter is running EdgeTx you can load the Taycan MX-C EdgeTX configuration file into EdgeTX companion and upload it to the radio.
Or simply copy only the settings that you require, EdgeTX companion is very flexible.
In our setup, we used an advanced RC receiver that cannot be fully configured using Mission Planner's SETUP >> Mandatory Hardware >> Radio Calibration menu.
For its configuration, we created a 05_remote_controller.param parameter file. Download it to your PC, edit it with a text editor and in Mission Planner while connected to the vehicle select Compare Params, review the changes and press Continue followed by Upload Params.
The RC transmitter we used has a big color display where telemetry data is displayed, nevertheless, we use telemetry data for real-time flight monitoring with Mission Planner or QGroundControl. For its configuration, we created a 06_telemetry.param parameter file. Download it to your PC, edit it with a text editor and in Mission Planner while connected to the vehicle select Compare Params, review the changes and press Continue followed by Upload Params. Once this is operating we no longer need the USB connection to the vehicle.
In our setup, we used a Bi-directional Dshot ESC that cannot be fully configured using Mission Planner's SETUP >> Mandatory Hardware >> Servo Output menu.
For its configuration, we created a 07_esc.param parameter file. Download it to your PC, edit it with a text editor and in Mission Planner while connected to the vehicle select Compare Params, review the changes and press Continue followed by Upload Params.
The step above configured ESC communication pass-thru. In our vehicle, we use BLHeli_32 ARM ESC firmware. So we use BLHeliSuite32 Version 32.9.0.6 to configure the ESCs. Flash the Firmware version described in the table above. Configure the parameters to match the figures below.
In our setup, the battery voltage is measured directly at the flight controller Vbat pin and the current is measured at the 4-in1 ESC Curr pin.
For its configuration, we created a 08_batt1.param parameter file. Download it to your PC, edit it with a text editor and in Mission Planner while connected to the vehicle select Compare Params, review the changes and press Continue followed by Upload Params.
To be on the safe side we used a Holybro PM02 as a redundant secondary voltage and current monitor. One other option would be the CBU 2-8s DroneCAN Battery Monitor and Current Sensor. For its configuration, we created a 09_batt2.param parameter file. Download it to your PC, edit it with a text editor and in Mission Planner while connected to the vehicle select Compare Params, review the changes and press Continue followed by Upload Params.
GNSS receivers very often contain a magnetometer (compass) sensor. So they need to be configured before proceeding to the next step. For its configuration, we created a 10_gnss.param parameter file. Download it to your PC, edit it with a text editor and in Mission Planner while connected to the vehicle select Compare Params, review the changes and press Continue followed by Upload Params.
On Mission Planner select SETUP >> Mandatory Hardware and work yourself through all the submenus as described below. DO NOT SKIP ANY STEP.
This relates to the FRAME_CLASS and FRAME_TYPE parameters.
Answer the questions that Mission Planner asks, select Add suggested settings for 4.0 and up (Battery failsafe and Fence) and upload the calculated parameters to the flight controller by pressing Upload to FC.
Follow the ArduPilot wiki instructions and calibrate the accelerometers. For small vehicles use:
- Calibrate Accel
- Calibrate level
For very large vehicles:
- Simple Accel Cal
Follow the ArduPilot wiki instructions and calibrate the compass(es).
Disable internal compasses if the battery or power wires are close to the flight controller.
Do not select Automatically learn offsets it makes little sense on a multicopter. And we will do in-flight MagFit later
If you have a large vehicle you might want to use large vehicle MagCal instead.
Follow the ArduPilot wiki instructions and calibrate the Remote Control.
Turn on your RC Transmitter and move the sticks around. Make sure all transmitter channels move across their entire range.
Change the parameters according to your requirements.
Do not make changes here, these parameters will be set later on the Motor/Propeller order and direction test section
Define the flight modes you plan to use.
Do not use POSHOLD, use LOITER instead.
Both only work outdoors because they require a good GNSS signal quality with low variance.
If that is not possible, GPS glitches will occur and ALTHOLD flight mode is recommended instead.
These are very important and can save your vehicle in case of failure.
Configure at least Radio Failsafe, Battery Failsafe and Geofence
This is just informational. No need to change anything.
Change the parameters according to your requirements.
The changes you did in the steps above have been stored in your vehicle. Most of the changed parameters are vehicle-instance specific and can not be reused between two vehicles, no matter how similar they are. We provide an 12_mp_setup_mandatory_hardware.param parameter file only for illustrative purposes. You must not upload our vehicle-specific file to your vehicle.
Most people wrongly assume that they are now done with tuning. But that is wrong, there are still steps to do before the first flight.
The next configuration steps are exclusively done by editing intermediate parameter configuration files and in Mission Planner while connected to the vehicle selecting Compare Params, reviewing the changes and pressing Continue followed by Upload Params.
In our case, a 3'' small drone, the props are expected to rotate at speeds higher than 400Hz and we have an STM32 H7 family processor. So the scheduler loop rate should be increased to 800Hz. For the inflight MagFit step, we must activate Lua scripting.
Download the 13_general_configuration.param parameter file to your PC, edit it with a text editor to meet your needs and in Mission Planner while connected to the vehicle select Compare Params, review the changes and press Continue followed by Upload Params.
For this test, you need to:
- Visit the ArduPilot Hardware report on your PC and upload the
.binfile you got in the previous section. - Take a look at the results
It should look like the following picture. If it doesn't, go back and perform the missing calibration(s).
Download the 14_logging.param parameter file to your PC, edit it with a text editor to meet your needs and in Mission Planner while connected to the vehicle select Compare Params, review the changes and press Continue followed by Upload Params.
The table below explains which bit is responsible for which .bin dataflash log message(s):
| Message | description | Log rate | LOG_BITMASK | ||
|---|---|---|---|---|---|
| field name | bit | value | |||
| SIDD | System ID data | SCHED_LOOP_RATE / 1 | ATTITUDE_FAST and ATTITUDE_MED | 1 and 0 | 3 |
| SCHED_LOOP_RATE / 2 | ATTITUDE_FAST | 0 | 2 | ||
| SCHED_LOOP_RATE / 4 | ATTITUDE_MED | 0 | 1 | ||
| SCHED_LOOP_RATE / 8 | - | - | 0 | ||
| SIDS | System ID settings | SCHED_LOOP_RATE / 1 | ATTITUDE_FAST and ATTITUDE_MED | 1 and 0 | 3 |
| SCHED_LOOP_RATE / 2 | ATTITUDE_FAST | 0 | 2 | ||
| SCHED_LOOP_RATE / 4 | ATTITUDE_MED | 0 | 1 | ||
| SCHED_LOOP_RATE / 8 | - | - | 0 | ||
| RATE | Desired and achieved vehicle attitude rates | SCHED_LOOP_RATE / 1 | ATTITUDE_FAST and ATTITUDE_MED | 1 and 0 | 3 |
| SCHED_LOOP_RATE / 2 | ATTITUDE_FAST | 0 | 2 | ||
| SCHED_LOOP_RATE / 4 | ATTITUDE_MED | 0 | 1 | ||
| SCHED_LOOP_RATE / 8 | - | - | 0 | ||
| ATSC | Scale factors for attitude controller | SCHED_LOOP_RATE / 1 | ATTITUDE_FAST and ATTITUDE_MED | 1 and 0 | 3 |
| SCHED_LOOP_RATE / 2 | ATTITUDE_FAST | 0 | 2 | ||
| SCHED_LOOP_RATE / 4 | ATTITUDE_MED | 0 | 1 | ||
| SCHED_LOOP_RATE / 8 | - | - | 0 | ||
| ATT | Canonical vehicle attitude | SCHED_LOOP_RATE | ATTITUDE_FAST | 0 | 1 |
| 10 Hz | ATTITUDE_MED | 1 | 2 | ||
| IMU | Inertial Measurement Unit data | SCHED_LOOP_RATE | IMU_FAST and ATTITUDE_FAST | 18 and 0 | 262145 |
| 25 Hz | IMU | 7 | 128 | ||
| GPS | Information received from GNSS systems attached to the autopilot | ~ 5 Hz | GPS | 2 | 4 |
| GPA | GPS accuracy information | ||||
| UBX1 | uBlox-specific GPS information (part 1) | ||||
| UBX2 | uBlox-specific GPS information (part 2) | ||||
| GRAW | Raw uBlox datas | ||||
| GRXH | Raw uBlox data - header | ||||
| GRXS | Raw uBlox data - space-vehicle data | ||||
| TERR | Terrain database information | ||||
| PM | autopilot system performance and general data dumping ground | 1 Hz | PM | 3 | 8 |
| XKF0 | EKF3 beacon sensor diagnostics | 25Hz if ATTITUDE_FAST is set, else 10Hz | - | - | - |
| XKF1 | EKF3 estimator outputs | ||||
| XKF2 | EKF3 estimator secondary outputs | ||||
| XKF3 | EKF3 innovations | ||||
| XKF4 | EKF3 variances | ||||
| XKF5 | EKF3 Sensor innovations (primary core) and general dumping ground | ||||
| XKFS | EKF3 sensor selection | ||||
| XKQ | EKF3 quaternion defining the rotation from NED to XYZ (autopilot) axes | ||||
| XKV1 | EKF3 State variances (primary core) | ||||
| XKV2 | more EKF3 State Variances (primary core) | ||||
| XKT | EKF3 timing information | ||||
| AHR2 | Backup AHRS data | ||||
| POS | Canonical vehicle position | ||||
| CTRL | Attitude Control oscillation monitor diagnostics | 10Hz | CTUN | 4 | 16 |
| RFND | Rangefinder sensor information | ||||
| PRX | Proximity Filtered sensor data | ||||
| PRXR | Proximity Raw sensor data | ||||
| BCN | Beacon information | ||||
| CTUN | Control Tuning information | 100Hz | |||
| PSCN | Position Control North | 10Hz | NTUN | 5 | 32 |
| PSCD | Position Control Down | ||||
| PSCE | Position Control East | ||||
| RCIN | RC input channels to vehicle | 10Hz | RCIN | 6 | 64 |
| RCI2 | (More) RC input channels to vehicle | ||||
| RSSI | Received Signal Strength Indicator for RC receiver | ||||
| VIBE | Processed (acceleration) vibration information | 10Hz | IMU or IMU_FAST or IMU_RAW | 19 or 18 or 7 | - |
| CMD | Executed mission command information | on event | CMD | 8 | 256 |
| MAVC | MAVLink command we have just executed | ||||
| BAT | Gathered battery data | ?? | CURRENT | 9 | 512 |
| BCL | Battery cell voltage information | ||||
| MCU | MCU voltage and temperature monitoring | ||||
| POWR | System power information | ||||
| RCOU | Servo channel output values 1 to 14 | 10Hz | RCOUT | 10 | 1024 |
| RCO2 | Servo channel output values 15 to 18 | ||||
| RCO3 | Servo channel output values 19 to 32 | ||||
| OF | Optical flow sensor data | ?? | OPTFLOW | 11 | 2048 |
| PIDN | Proportional/Integral/Derivative gain values for North | SCHED_LOOP_RATE if ATTITUDE_FAST is set else 10Hz | NTUN and PID | 12 and 5 | 4128 |
| PIDE | Proportional/Integral/Derivative gain values for East | ||||
| PIDR | Proportional/Integral/Derivative gain values for Roll | SCHED_LOOP_RATE if ATTITUDE_FAST is set else 10Hz | PID | 12 | 4096 |
| PIDP | Proportional/Integral/Derivative gain values for Pitch | ||||
| PIDY | Proportional/Integral/Derivative gain values for Yaw | ||||
| PIDA | Proportional/Integral/Derivative gain values for Altitude | ||||
| MAG | Information received from compasses | ?? | COMPASS | 13 | 8192 |
| CAM | Camera shutter information | on event | CAMERA | 15 | 32768 |
| TRIG | Camera shutter information | ||||
| MOTB | Motor mixer information | 10Hz | MOTBAT | 17 | 131072 |
| ACC | IMU accelerometer data | SCHED_LOOP_RATE | IMU_RAW | 19 | 524288 |
| GYR | IMU gyroscope data | ||||
| VSTB | Motor mixer information | SCHED_LOOP_RATE | VIDEO_STABILISATION | 20 | 1048576 |
| FTN | Filter Tuning Message - per motor | SCHED_LOOP_RATE | FTN_FAST | 21 | 2097152 |
| FTNS | Filter Tuning Message | ||||
| FTN1 | FFT Filter Tuning | 25 Hz | |||
| WINC | Winch | 10Hz | Any | any | any |
Start by checking the motor numbering with the Mission Planner Motor test without propellers. Remember the correct order is A, B, C, D and not 1, 2, 3, 4.
Make sure the MOT_SPIN_ARM is high enough so that all motors spin reliably.
Make sure the MOT_SPIN_MIN is at least 0.03 higher than MOT_SPIN_ARM.
!!! If you get this test wrong or skip it, you might destroy your vehicle and endanger yourself !!!
Do not try to test the attitude controller without propellers, it will not work, such tests are inconclusive and meaningless. Do not try to test the attitude controller with the vehicle tied down on the ground, it will not work, such tests are inconclusive and meaningless. At this point, the vehicle does not react to roll, pitch, yaw or thrust inputs.
Now mount the propellers on the vehicle.
After that, follow the setting Motor Range to adjust the motor demand by the flight controller to the limitations of the ESC.
These three parameters linearize (correct) the non-linear thrust vs. PWM response of the propulsion system (battery, ESC, motors, propellers). They are crucial for the stability of the vehicle. If this is set too high we see an increase in gain at the lower end of the thrust range and a decrease in gain at the upper end. Therefore when set to high you can see instability at low throttle and if set too low you can see instability at high throttle.
The automation on SETUP >> Mandatory Hardware >> Initial Tune Parameters gave you a good starting point on their values.
But we recommend using a Trust Stand or Olliw method or ArduPilot DIY Thrust Stand to determine their value.
At the time of writing Automatic MOT_THST_EXPO estimation lua script is not yet ready for production use.
If you have a very small, or a very large vehicle that requires non-default PID values for a safe flight, change them in the 16_pid_adjustment.param parameter file Usually, smaller vehicles require lower than default PID rate values. Larger vehicles usually require higher than default PID rate values.
Read and follow ArduPilot's Remote ID setup instructions.
You might have to build OpenDroneID firmware for production.
Edit the 17_remote_id.param file with a text editor to meet your needs and in Mission Planner while connected to the vehicle select Compare Params, review the changes and press Continue followed by Upload Params.
Configure the gyro noise reduction notch filters with a estimation of the operation parameters as described on the 18_notch_filter_setup.param parameter file. Edit it with a text editor to meet your needs and in Mission Planner while connected to the vehicle select Compare Params, review the changes and press Continue followed by Upload Params. The estimation will be improved after the first flight.
For more detailed information visit the First flight
Test the initial setup on the ground in stabilize flight mode by using as little RC transmitter throttle as possible without taking off. At this sweet spot, inspect all axes (roll, pitch and yaw) by providing small RC transmitter stick inputs. If the multicopter behaves correspondingly, the setup is good to go.
After some careful test maneuvers switch to ALTHOLD and hover for 30 to 40 seconds one to two meters above the ground. Land and disarm.
Immediately check for hot motors.
If the motors are too hot, check the .bin dataflash log, high or oscillating RATE.*out values indicate which PID gain you should reduce to remove the output oscillations causing the motors to heat up.
Load the .bin log from the first flight onto the online Ardupilot Filter Review tool
Follow the instructions from Peter Hall on his Blog Post to configure the Harmonic Notch filter(s).
The graph below is a bode diagram of the gyro signals before and after the low-pass and Harmonic Notch filters. As you can see, the filters remove most of the vibration noise from the gyro sensors.
Below is the configuration we used.
In our case, we got the 19_notch_filter_results.param parameter file. Edit it with a text editor to meet your needs and in Mission Planner while connected to the vehicle select Compare Params, review the changes and press Continue followed by Upload Params.
Load the .bin log from the first flight onto the online Ardupilot Log Viewer or into Mission Planner.
Take a look at the VIBE.VibeX, VIBE.VibeY, VIBE.VibeZ graphs they all should be below 15
According to common ArduPilot forum knowledge, and quoting @xfacta:
- Vibrations over 30 are very bad
- Vibrations over 20 are causing issues even if you don't know it yet
- Vibrations over 15 are in a grey area - it could go either way - check clipping, it must be zero
- Vibrations below 10 are good
Now upload the .bin log to the Hardware-Report Tool once again to check CPU load and loop times, which in our case look like this:
Use the .bin log from the first flight to set the parameters described on the 20_throttle_controller.param file. Edit it with a text editor to meet your needs and in Mission Planner while connected to the vehicle select Compare Params, review the changes and press Continue followed by Upload Params.
Now that the Harmonic Notch filter and the throttle controller are configured, the second flight will be safer. This flight will be used to calibrate the compass during a realistic operation scenario in the air.
Follow these steps:
- Download the
copter-magfit-helper.luaandadvance-wp.luascripts from Scripted MagFit flightpath generation and put them on the micro SDCard'sAPM/scriptsfolder - Activate MagFit by downloading the 24_inflight_magnetometer_fit_setup.param parameter file to your PC, edit it with a text editor to meet your needs and in Mission Planner while connected to the vehicle select Compare Params, review the changes and press Continue followed by Uploadd Params
- Perform the MagFit figure-eight flight and land
- Download the latest
.bindataflash log file from the micro SDcard's/APM/LOGSfolder - Load it into MAVExplorer using the command line:
MAVExplorer.py filename.binor into the ArduPilot MAGFit in flight compass calibration using an internet browser. - Select the area where the multicopter performed the Figure eight (exclude the takeoff and landing flight sections)
- Perform the MagFit calculations. In our case, we got the 25_inflight_magnetometer_fit_results.param parameter file
- Upload the tool-generated
.paramfile, not ours, to the vehicle.
After that repeat the steps described in 2.12 ArduPilot Hardware Report. The report should now look like this:
Follow the first part of evaluating the aircraft tune.
After landing take a look at the RATE.*out values in the .bin log file, they all should be below 0.1.
Download the 28_evaluate_the_aircraft_tune_ff_disable.param parameter file, and in Mission Planner while connected to the vehicle select Compare Params, review the changes and press Continue followed by Upload Params.
Follow the second part of evaluating the aircraft tune.
After landing take a look at the RATE.*out values in the .bin log file, they all should be below 0.1.
Download the 29_evaluate_the_aircraft_tune_ff_enable.param parameter file, and in Mission Planner while connected to the vehicle select Compare Params, review the changes and press Continue followed by Upload Params.
The Autotune is an automated iterative process:
- It changes the parameter values of the attitude PID controllers
- Tests the overshoot and settling-time of the control loop using the new PID values
- If they are within the desired requirements, the process is over. If not, it gets repeated from the beginning
If the battery gets depleted before you can complete the Autotune flight(s), download the latest .bin log file from the micro SDCard directory /APM/LOGS.
Take a look at the ATUN messages.
They show how the PID values change over time.
You should use the latest PID values from the ATUN messages to set the starting point of your next PID Autotune with a fresh battery.
But be careful, these new PID values might be unstable and cause your vehicle to crash.
To be on the safe side perform the third and fourth flights again according to the instructions above.
This way, the Autotune will restart from the partially optimal values it found before the battery got depleted, instead of starting from scratch.
An example of the relevant ATUN message data of an interrupted yaw Autotune .bin dataflash log is depicted below:
The correspondence between the PIDs' initial values and the ATUN message fields is shown in the respective tables for each of the four Autotune axes in the sections below.
The tune of the vehicle must be done in the vehicle's most agile configuration. That is, the vehicle will be at its minimum take-off weight with fully charged batteries.
This is why we will do Autotune with multiple flights, one axis per flight.
Typically the quality of the Autotune results for each axis is proportional to the value of the ATC_ANG_RLL_P, ATC_ANG_PIT_P, and ATC_ANG_YAW_P parameters for their respective axis.
Also the higher the values, the tighter the tune.
If you get low values, improve the hardware and revisit the previous sections to further reduce the vibrations.
Autotuning in low-wind conditions is desirable, but if that is not possible you can increase the AUTOTUNE_AGGR parameter value to 0.110 or even 0.120.
That is a workaround and will not produce as good results as low-wind conditions autotune.
We set up the autotune as a flight mode, and as such it will use the underlying ALTHOLD flight mode.
If you want to use the LOITER flight mode as the underlying mode during autotune you need to set an RC channel function switch to autotune.
- Download the 30_autotune_roll_setup.param parameter file to your PC and upload it to the flight controller. It will activate the roll axis Autotune.
- Outdoors on a non-windy day (or indoors in a big warehouse like we at IAV do) take off and fly in either
AltHoldorLoiterflight mode. - At about 2 meters high, select
Autotuneflight mode in the RC transmitter to engage Autotune. - Use the RC transmitter sticks to correct the vehicle position if it gets too high, too low or too close to obstacles.
- Once the Autotune is completed, land and disarm the vehicle without changing the flight mode.
You should get something like the 31_autotune_roll_results.param parameter file
The autotune might have found a poor solution, here are some indicators of a poor tune:
- The resulting
ATC_ANG_RLL_Pparameter value is smaller than 4.5 - The resulting
ATC_RAT_RLL_Dparameter value is equal to theAUTOTUNE_MIN_Dparameter value
If the battery got depleted before Autotune completion, change the initial PID parameters as shown in the table below:
| PID parameter name | PID parameter value based on the ATUN field |
|---|---|
| - | ATUN.Axis=0 |
| ATC_RAT_RLL_P | ATUN.RP |
| ATC_RAT_RLL_I | ATUN.RP |
| ATC_RAT_RLL_D | ATUN.RD |
| ATC_ANG_RLL_P | ATUN.SP |
| ATC_ACCEL_R_MAX | ATUN.ddt |
- Download the 32_autotune_pitch_setup.param parameter file to your PC and upload it to the flight controller. It will activate the pitch axis Autotune.
- Outdoors on a non-windy day (or indoors in a big warehouse like we at IAV do) take off and fly in either
AltHoldorLoiterflight mode. - At about 2 meters high, select
Autotuneflight mode in the RC Transmitter to engage Autotune. - Use the RC transmitter sticks to correct the vehicle position if it gets too high, too low or too close to obstacles.
- Once the autotune is completed, land and disarm the vehicle without changing the flight mode.
You should get something like the 33_autotune_pitch_results.param parameter file
The autotune might have found a poor solution, here are some indicators of a poor tune:
- The resulting
ATC_ANG_PIT_Pparameter value is smaller than 4.5 - The resulting
ATC_RAT_PIT_Dparameter value is equal to theAUTOTUNE_MIN_Dparameter value
If the battery got depleted before Autotune completion, change the initial PID parameters as shown in the table below:
| PID parameter name | PID parameter value based on the ATUN field |
|---|---|
| - | ATUN.Axis=1 |
| ATC_RAT_PIT_P | ATUN.RP |
| ATC_RAT_PIT_I | ATUN.RP |
| ATC_RAT_PIT_D | ATUN.RD |
| ATC_ANG_PIT_P | ATUN.SP |
| ATC_ACCEL_P_MAX | ATUN.ddt |
- Download the 34_autotune_yaw_setup.param parameter file to your PC and upload it to the flight controller. It will activate the yaw axis Autotune.
- Outdoors on a non-windy day (or indoors in a big warehouse like we at IAV do) take off and fly in either
AltHoldorLoiterflight mode. - At about 2 meters high, select
Autotuneflight mode in the RC transmitter to engage Autotune. - Use the RC transmitter sticks to correct the vehicle position if it gets too high, too low, or too close to obstacles.
- Once the Autotune is completed, land and disarm the vehicle without changing the flight mode.
You should get something like the 35_autotune_yaw_results.param parameter file
The autotune might have found a poor solution, here are some indicators of a poor tune:
- The resulting
ATC_ANG_YAW_Pparameter value is smaller than 4.5
If the battery got depleted before Autotune completion, change the initial PID parameters as shown in the table below:
| PID parameter name | PID parameter value based on the ATUN field |
|---|---|
| - | ATUN.Axis=2 |
| ATC_RAT_YAW_P | ATUN.RP |
| ATC_RAT_YAW_I | ATUN.RP * 0.1 |
| ATC_RAT_YAW_FLTE | ATUN.RD |
| ATC_ANG_YAW_P | ATUN.SP |
| ATC_ACCEL_Y_MAX | ATUN.ddt |
This particular Autotune axis is only relevant for small, agile vehicles.
- Download the 36_autotune_yawd_setup.param parameter file to your PC and upload it to the flight controller. It will activate the yaw D axis Autotune.
- Outdoors on a non-windy day (or indoors in a big warehouse like we at IAV do) take-off and fly in either
AltHoldorLoiterflight mode. - At about 2 meters high, select
Autotuneflight mode in the RC transmitter to engage Autotune. - Use the RC transmitter sticks to correct the vehicle position if it gets too high, too low or too close to obstacles.
- Once the Autotune is completed, land and disarm the vehicle without changing the flight mode.
You should get something like the 37_autotune_yawd_results.param parameter file
Make sure that your resulting ATC_RAT_YAW_D parameter value is different from AUTOTUNE_MIN_D value.
If that is not the case then the autotune failed to find a proper ATC_RAT_YAW_D.
The cause is probably too high noise values at the input of the Yaw D controller.
If the battery got depleted before Autotune completion, change the initial PID parameters as shown in the table below:
| PID parameter name | PID parameter value based on the ATUN field |
|---|---|
| - | ATUN.Axis=3 |
| ATC_RAT_YAW_P | ATUN.RP |
| ATC_RAT_YAW_I | ATUN.RP * 0.1 |
| ATC_RAT_YAW_D | ATUN.RD |
| ATC_ANG_YAW_P | ATUN.SP |
| ATC_ACCEL_Y_MAX | ATUN.ddt |
Now that the yaw axis is tuned, the autotune should be able to improve the roll and pitch axis tune.
- Download the 38_autotune_roll_pitch_retune_setup.param parameter file to your PC and upload it to the flight controller. It will activate the roll and pitch axis Autotune.
- Outdoors on a non-windy day (or indoors in a big warehouse like we at IAV do) take off and fly in either
AltHoldorLoiterflight mode. - At about 2 meters high, select
Autotuneflight mode in the RC transmitter to engage Autotune. - Use the RC transmitter sticks to correct the vehicle position if it gets too high, too low or too close to obstacles.
- Once the Autotune is completed, land and disarm the vehicle without changing the flight mode.
You should get something like the 39_autotune_roll_pitch_retune_results.param parameter file
As you can see in the picture below, the Stabilize Roll, Pitch and Yaw P gains achieved with this method are high. The maximum stabilize P gain that autotune strives for is 36, and that was achieved in the roll axis! This is a clear indication that the vibration noise filters and the PID control loops are working well together.
After using Autotune to find proper PID parameters, it is time to evaluate the performance of the vehicle's control loops in real flight.
Follow these steps:
- Power on the vehicle
- Download the 28_evaluate_the_aircraft_tune_ff_disable.param parameter file, and in Mission Planner while connected to the vehicle select Compare Params, review the changes and press Continue followed by Upload Params.
- Switch to
ALTHOLDflight mode and wait for home location acquisition. - Take-off at around 10m above the ground.
- Perform smooth maneuvers using the RC transmitter roll stick.
- Perform smooth maneuvers using the RC transmitter pitch stick.
- Perform smooth maneuvers using the RC transmitter yaw stick.
- Repeat the maneuvers with increasing aggressivity making sure you stay inside the stable envelope of the vehicle.
- Land and download the latest
.binlog file from/APM/LOGSto your PC - Use ArduPilot's PID Review Tool to review the PID step response of each PID.
- Download the 29_evaluate_the_aircraft_tune_ff_enable.param parameter file, and in Mission Planner while connected to the vehicle select Compare Params, review the changes and press Continue followed by Upload Params.
In our vehicle, we got a transient response of around 60ms in roll and pitch and around 110ms in yaw.
Step response Roll:
Step response Pitch:
Step response Yaw:
If you are satisfied with the performance, increase ATC_THR_MIX_MAX to 0.9 (default is 0.5) to increase prioritization of attitude control over throttle.
This can reduce the pitch overshoot sometimes seen (especially on copters with large propellers) in AltHold if the vehicle suddenly slows after performing a fast-forward flight.
Take a look at the RATE.*out values in the .bin log file, they all should be below 0.1.
Follow ArduCopter's airspeed estimation Wiki and/or use the Lua script provided by Yuri in the forum.
In our case, the frontal area looks like this:
and the side area looks like this:
Divided by 1,000,000 to convert from mm² to m², the frontal area is 0.01097 m² and the side area is 0.01455 m². The weight of our drone is 560g, therefore the ballistic coefficients are
- EK3_DRAG_BCOEF_X = 0.56 kg / 0.01097 m² = 51.0399
- EK3_DRAG_BCOEF_Y = 0.56 kg / 0.01455 m² = 38.4798
Download the 40_windspeed_estimation.param parameter file to your PC, edit it with a text editor to meet your needs and in Mission Planner while connected to the vehicle select Compare Params, review the changes and press Continue followed by Upload Params.
Now do the flight to collect the data to Calculate the Propeller Drag Coefficient.
After that, open the logs with MAVExplorer to get the needed values.
Display the absolute values for acceleration in X and Y, as well as GPS speed by inputting graph abs(IMU.AccX) abs(IMU.AccY) GPS.Spd.
Then crop it so you only see the tests in AltHold flight mode.
It should look like this:
Note that our test flight was quite noisy but there is enough data to extract. Next crop it so you see one acceleration into the wind and the consecutive deceleration. It should look like this:
Get the current wind speed, that is the GPS speed when AccY reaches zero and the GPS speed has stabilized. In this case, it is:
- windspeed = 2.35 [m/s]
Next, get the groundspeed at the start of the test. That is the GPS speed when the vehicle starts to decelerate after the little bit of jitter is over. In this case, it is:
- groundspeed = 3.9 [m/s]
With this information, you can calculate the vehicle's airspeed, which is:
- airspeed = windspeed + groundspeed = 6.25 [m/s]
Next get the maximum acceleration during the test, which is the acceleration at the time of the groundspeed measurement. In this case, it is:
- max_accel = 4.2 [m/s²]
With the air density at the time of testing and the previously calculated ballistic drag coefficient (EK3_DRAG_BCOEF_X for front and back, EK3_DRAG_BCOEF_Y for left and right side) you can now calculate the bluff body drag, which is 1/2 * air density * airspeed^2 / BCOEF.
In this case, it is:
- Bluff body drag = 0.5 * 1.260 [kg/m³] * (6.25 [m/s])² / 38.4798 [kg/m²] = 0.6395 [m/s²]
With that, you can now calculate the momentum drag, which is max_accel - bluff body drag. In this case, it is:
- Momentum drag = 4.2 [m/s²] - 0.6395 [m/s²] = 3.5605 [m/s²]
Now you can calculate the momentum drag coefficient EK3_DRAG_MCOEF, which is momentum drag / airspeed.
In this case, it is:
- EK3_DRAG_MCOEF = 3.5605 [m/s²] / 6.25 [m/s] = 0.5697 [1/s]
For better accuracy, you should do that for all directions and take the average. In our case, we got:
- front: 0.4628 [1/s]
- back: 0.4757 [1/s]
- left: 0.5426 [1/s]
- right: 0.5697 [1/s]
- average: 0.5127 [1/s]
Note that these are quite high values due to the ducts around the props. For a normal copter with open propellers, it should be in the range of 0.1 to 0.2.
After it is set, do another flight and check that the windspeed and direction are correctly estimated.
Follow ArduCopter's baro compensation Wiki and/or use the Lua script provided by Yuri in the forum.
Download the 41_barometer_compensation.param parameter file to your PC, edit it with a text editor to meet your needs and in Mission Planner while connected to the vehicle select Compare Params, review the changes and press Continue followed by Uploadd Params.
Now do the flight to collect the data and analyze the logs to see if the barometer is correctly compensated and insensitive to wind.
These steps are optional. Their goal is to build a mathematical model of the vehicle that can later be used to further optimize the control loops of the vehicle according to a set of constraints (requirements).
Documentation is available on Fabian Bredemeier's Identification of a multicopter section at ArduCopter's_wiki.
Download the 42_system_id_roll.param parameter file to your PC, edit it with a text editor to meet your needs and in Mission Planner while connected to the vehicle select Compare Params, review the changes and press Continue followed by Upload Params.
Now do the flight to collect the data for the roll rate system identification.
Download the 43_system_id_pitch.param parameter file to your PC, edit it with a text editor to meet your needs and in Mission Planner while connected to the vehicle select Compare Params, review the changes and press Continue followed by Upload Params.
Now do the flight to collect the data for the pitch rate system identification.
Download the 44_system_id_yaw.param parameter file to your PC, edit it with a text editor to meet your needs and in Mission Planner while connected to the vehicle select Compare Params, review the changes and press Continue followed by Upload Params.
Now do the flight to collect the data for the yaw rate system identification.
Download the 45_system_id_thrust.param parameter file to your PC, edit it with a text editor to meet your needs and in Mission Planner while connected to the vehicle select Compare Params, review the changes and press Continue followed by Upload Params.
Now do the flight to collect the data for the thrust system identification.
This describes how to use IAV's multi-objective optimization to achieve even better (according to a predefined set of constraints) PID tuning.
One other approach is described by Bill Geyer in his Blog post: Predicting Closed Loop Response For Faster Autotune.
Download the 46_analytical_pid_optimization.param parameter file to your PC, edit it with a text editor to meet your needs and in Mission Planner while connected to the vehicle select Compare Params, review the changes and press Continue followed by Uploadd Params.
Some changes should be made for everyday productive operation.
Download the 47_everyday_use.param parameter file to your PC, edit it with a text editor to meet your needs and in Mission Planner while connected to the vehicle select Compare Params, review the changes and press Continue followed by Upload Params.
The most inner angle rate and angle control loops have been tuned. Now let's tune the position controller.
Edit the 48_position_controller.param file with a text editor to meet your needs and in Mission Planner while connected to the vehicle select Compare Params, review the changes and press Continue followed by Upload Params.
These are optional, and only make sense if you have extra hardware on your vehicle to support it. Download the 49_precision_land.param.param parameter file to your PC, edit it with a text editor to meet your needs and in Mission Planner while connected to the vehicle select Compare Params, review the changes and press Continue followed by Upload Params.
These are optional, and only make sense if you do beyond visual line-of-sight (BVLOS) autonomous flights using a companion computer. Download the 50_guided_operation.param.param parameter file to your PC, edit it with a text editor to meet your needs and in Mission Planner while connected to the vehicle select Compare Params, review the changes and press Continue followed by Upload Params.
We presented a sequence of small, methodic steps that result in a fully operational and safe drone. Beginning with informed hardware decisions, appropriate hardware configuration and concluding with a finely tuned vehicle equipped with robust, fast-acting control loops. Each step is documented in its own intermediate parameter file, ensuring reproducibility and traceability. Each file is numbered, ensuring that the sequence of steps is clear. The number of test flights was reduced to a minimum, and their order was optimized. This process was developed for our specific multicopter, but it can be tailored to any other.
| PID controller | Intermediate parameter file(s) used to configure and tune it |
|---|---|
| Position Z acceleration | 20_throttle_controller.param |
| Roll rate | 31_autotune_roll_results.param, 39_autotune_roll_pitch_retune_results.param |
| Pitch rate | 33_autotune_pitch_results.param, 39_autotune_roll_pitch_retune_results.param |
| Yaw rate | 35_autotune_yaw_results.param, 37_autotune_yawd_results.param |
| Roll | 31_autotune_roll_results.param, 39_autotune_roll_pitch_retune_results.param |
| Pitch | 33_autotune_pitch_results.param, 39_autotune_roll_pitch_retune_results.param |
| Yaw | 35_autotune_yaw_results.param, 37_autotune_yawd_results.param |
| Position XY velocity | 48_position_controller.param |
Many thanks to the ArduPilot's developers and community.
This work has been sponsored by the company I work for IAV GmbH. We provide engineering and consulting for robotic systems including multicopters. Feel free to contact us for help or development support.
Your vehicle is now properly tuned according to AduPilot's standard procedures and some of IAV GmbH's know-how.
Enjoy,
Jan Ole Noack
Amilcar do Carmo Lucas