Quanser 3DOF Helicopter

This is a repository with supporting material (texts, data, codes, ...) for Quanser 3DOF Helicopter.

One way to interpret this laboratory/toy gadget is this that it represents an anchored (and simplified) two-rotor helicopter such as the famous CH-47 Chinook

Source: https://en.wikipedia.org/wiki/Boeing_CH-47_Chinook#/media/File:CH-47_Chinook_helicopter_flyby.jpg

Control systems challenge

From a control systems perspective, this sort-of-kind-of Chinook helicopter has two independent actuators – propellers driven by DC motors. Thrust induced by the two propellers together with the pitch angle (the angle of the Chinook around the transverse axis), determine both the vertical and the forward motion. Since the helicopter is suspended by a long arm anchored at a 2-DOF joint, the former motion exhibits itself as an elevation angle of the arm, while the latter exhibits itself as a travel angle (angle around a vertical axis) of the same arm. Designing a controller that tracks reference values of both the elevation and the travel angles by setting the voltages applied to the two DC motors while measuring the three angles (pitch, elevation, travel) using incremental encoders constitutes the control challenge with this laboratory gadget.

Instructions for running the experiment

1. Double-check that the room around the experiment is free of any obstacles.

2. Also observe how unstable the experiment is when seated on the box. We have to fix this (attach the base of the model to the box with screews and fill the box with some heavy load). But in the meantime, just be aware of possible tipping over.

3. The only possibility to change the configuration of the experiment is to change the location of the counterweight block (the brass-ish cube). You are certainly not expected to experiment with this setting, but just check that the fastening screw is in the fifth hole (on the top side this is labelled as corresponding to 71 g of equivalent mass on the helicopter side).

4. Switch on the VoltPAQ-X2 Amplifier – the switch is on the back side just above the power cable. Check that the gains are correctly switched to 3x on the front panel (just in case somebody changed this setting before you).

5. Connect the Q8-USB Data Acquisition Device to the power (it has no switch, just plug in the connector on the back side).

6. If you have not done so, download the code from our repository here, namely from /sotware/quanser-fixed2021 subdirectory. Note that you should certainly pick this directory and not the directory with the original code (that you can also download from the product webpage as the Simulink Courseware) because the latter contains a few problems that have already been fixed in our version.

7. Launch Matlab R2017a. Unfortunately the currently installed license of the QUARC library for interaction with the used data acquisition system is too old (some 2.x) for more recent versions of Matlab. New license (some 4.x) have just been ordered (April 2021) and as soon as we receive it, we will install the latest Matlab R2021a and the latest QUARC and the problem will be solved.

8. In Matlab switch to your local /software/quanser-fixed2021 (sub)directory and open the setup_lab_heli_3d.m file.

9. Double check that in this setup_lab_heli_3d.m on line 27 you have correctly set WITH_ADS = 'NO'; The possibility to set the option to YES is only included for possible extension of the whole laboratory model with some extra hardware (active disturbance system, ADS). Since we do not have it, you can also freely ignore the files with the ADS label in their names (such as the files setup_ads_configuration.m and q_heli3d_w_ads.mdl).

10. Note that if you want to record the measured data for later analysis (which you will certainly do), the routes that you are familiar with from purely simulation projects may not work here. The reason is that while using the Quanser QUARC system, you are running the Simulink model in External Mode, which prevents from using some Simulink functionality. In particular, if you use Simulink scope block(s) for logging the data, which is a common route, increasing the parameter called Limit data points to last will have no impact on how long the sequence of measurements will be stored in Matlab workspace (but do not forget to tick off the Log data workspace option and insert the variable name). If you ignore this issue, you may easily leave the lab assured that you simulated a minute or two only to discover at home that only the last few seconds have been saved. In order to prevent this, the Duration parameter needs to be changed somewhere else. Namely, go to manu Code, submenu External Mode Control Panel. Click on Signal & Triggering and change the Duration parameter there. If you need more on this, have a look at the secion on Data Collection in QUARC manual but if you are happy with logging to the variables in Matlab workspace through the Simulink scope blocks, you should be fine now.

11. Run the file. It will essentially build a mathematical model (by calling some other function from the directory) and design an LQR controller with an integral part.

12. Once you are done, open the q_heli3d.mdl Simulink model that contains the blocks for interfacing the real experimental model with the just designed controller (note that the similarly named file s_heli3d.mdl is just a simulation version of the model).

13. Build the model (by clicking the icon top right). This will take a few seconds.

14. If the build process finishes successfuly, click on the Connect to the target model. This may also take a few seconds.

15. Click the Run icon (or Ctrl+T) and the demo experiment starts.

16. Click the Stop to finish the experiment.

17. Now you are ready to replace the internals of the controller with your own stuff. Good luck.

18. Honestly, the mathematical model created upon calling the setup_heli_3d_configuration.m file deservers some polishing too. Maybe a significant one... After all, it is just a linear model. If you want to read about one particular modeling endeavour which resulted in a nonlinear model, have a look at the paper Brentari, Mirko, Paolo Bosetti, Isabelle Queinnec, a Luca Zaccarian. „Benchmark model of Quanser’s 3 DOF Helicopter". LAAS CNRS, February 2018. https://hal.laas.fr/hal-01711135.

Some comments on interpreting the measurements

Incremental encoders

Note that the angular sensors used in this setup are incremental encoders. This means that based solely on the measurements, the controller has no way to tell the initial angle. This simply means that whatever commanded (reference) value will be always regarded with respect to the initial angle. While this does not constitute a problem for the travel angle and the elevation problem, it may be a challenge for the pitch angle because the experiment can equally well start with the helicopter's nose up or down.

Incorrectly stated positive direction of travel angle in the documentation

Note that the documentation states the positive direction of the travel angle inconsistently with the measurements. The documentation says that the travel angle λ increases for counterclockwise motion, but this is not only in disagreement with the measurements but also does not comply with the conventions in aerospace. The travel angle in this experiment corresponds with the yaw angle ψ as used in the aerospace (the angle around the vertical body axis), for which the common conventions is that it is positive when turning to the right, which corresponds to the clockwise travel in the experiment. Until resolved, we keep it as an issue.

Source: https://commons.wikimedia.org/wiki/File:Yaw_Axis_Corrected.svg

Names and labels for the other two angles

While the other two angles appear to be defined consistently with the measurements, it may still be worth emphasizing that the pitch angle labelled ρ here is actually much more commonly labelled θ in aerospace texts. The angle increases when the aircraft's nose is going up.

Similarly, the angle called tilt here and labelled ε is actually equivalent to the more conventional roll angle φ from aerospace texts. The angle increases when the aircraft's (possibly just imagined) right wing is going down.

While the experiments can be viewed and operated independently of notational conventions in aerospace, it may be convenient to stick to them.