Active roll control system. Success

The Rocketry Forum

Help Support The Rocketry Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

Neway

Well-Known Member
Joined
Jul 31, 2010
Messages
53
Reaction score
0
Yesterday I took my master's degree in Aerospace engineering with thesis of the video below.
I'm Italian and my english in not the best. Could you review all text in the video and the description below? Thanks!

This video is a summary of my master's degree thesis in aerospace engineering took inside Università degli Studi di Palermo (Italy).
Rockets spin about their longitudinal axis due to an imperfect realization. This behavior could be a problem for telemetric systems or when guidance control on the other two axes are used.
In my thesis I designed, made and tested an automatic control system to wipe out the roll angular speed of a model rocket.
The system uses two control fins (aerodynamic control) moved by a flight computer equipped with a 3DOF gyroscope, an accelerometer and an atmospheric pressure sensor. I entirely made this flight computer: electronics, printed circuit and programming in C language. The code contain also two digital Kalman filters which compute an optimal estimation of the state-space from sensor data.
To design the digital controller and to predict the dynamic behavior of the entire rocket I implemented in Matlab / Simulink a 6 degree of freedom mathematical model. All aerodynamic parameters where estimated with real test flight or with McDonnell Douglas Digital DATCOM Fortran code (program which uses semi-empirical methods).
I also designed and realized the rocket motor from a commercial combustion chamber.


[YOUTUBE]qnUfsLCa8h8[/YOUTUBE]
 
Very nice!

Can you turn it into a commercial system and get Boostervision to carry them?
 
This is the most impressive thing I have every seen here.

Is the thesis available online, or is it in italian?
 
Thanks to all!

Can you turn it into a commercial system and get Boostervision to carry them?
Unfortunatly it isn't so easy, because the electronic is made for that fins, and the fins are designed on that specific rocket

Is the thesis available online, or is it in italian?
no it is in italian!
 
That is absolutely impressive. Well done sir!

Might I ask - does this stabilizer improve the efficiency of the rocket in regards to altitude? I would suspect that the "rolling" effect wastes energy and creates turbulence that would negatively affect altitude. Your device/stabilizer should improve the altitude of any given motor correct?
 
does this stabilizer improve the efficiency of the rocket in regards to altitude? I would suspect that the "rolling" effect wastes energy and creates turbulence that would negatively affect altitude. Your device/stabilizer should improve the altitude of any given motor correct?

Data from my flight aren't reliable because I used slightly different propellant reload. In theory the rocket should go higher because when spinning, the fins produce a lift and then an induced drag. I don't know the amount of drag produced, but it should be calculated
 
Hi,

Very cool! That is a very stable flight from the very start! Our team has been working on a similar system on our larger hybrid rocket, Iso-Haisu.

The first iteration failed to work because of unexpected aerodynamics in the canard control fins (not so unlike yours). Canard fins in front of the main fins can produce so much vortices in the air which hit the main fins that the interference actually reverses the effect of the control fins. The system worked for a few seconds after launch, but after that the control fins when to their end position trying to stop the roll, but it was just pushing it to roll more. (https://www.youtube.com/watch?v=Yvc_xCLJimk)

For the next launch we changed the two canards to a single control flap on one of the main fins. This worked beautifully. We are using a magnetometer for roll detection and are experimenting with algorithms, so the response is not as immediate as yours, but it proved the functionality. (https://www.youtube.com/watch?v=3Lhv978UPMI)

This spring we launched the rocket for the third time with exactly the same control system, and it went haywire. After a lot of investigation we realized that the algorithm we were using to detect roll was highly sensitive to the orientation of the rocket. During this flight the rocket tilted rather strongly north, causing a significantly larger roll to be measured. This combined with a small delay in the control loop most likely caused the system to fail. (https://www.youtube.com/watch?v=Iq1ycwDI2cE)

For our next launch we're planning on replacing the software so that the roll is detected directly from the movement of magnetic north. This poses some challenges on what to do if rolling over 180 degrees, but it would make the algorithm immune to magnetic amplitude effects. (Yes, gyros could be used and might even be more reliable, but we like to experiment with things. :) )

And naturally I've done some of the simulation using OpenRocket - which was my master's thesis. :)


Regarding the flight altitude - I don't think the effect will be measurable. If the rocket is rolling, it's inducing slightly more drag, but the roll actually reduces the amount of lift the fins are producing. When the canards are correcting the roll both the canards and main fins are producing lift (in opposing directions), and thus drag. My guesstimate is that it pretty much cancels out.


Cheers,
Sampo N.
 
This spring we launched the rocket for the third time with exactly the same control system, and it went haywire. After a lot of investigation we realized that the algorithm we were using to detect roll was highly sensitive to the orientation of the rocket. During this flight the rocket tilted rather strongly north, causing a significantly larger roll to be measured. This combined with a small delay in the control loop most likely caused the system to fail. (https://www.youtube.com/watch?v=Iq1ycwDI2cE)

For our next launch we're planning on replacing the software so that the roll is detected directly from the movement of magnetic north. This poses some challenges on what to do if rolling over 180 degrees, but it would make the algorithm immune to magnetic amplitude effects. (Yes, gyros could be used and might even be more reliable, but we like to experiment with things. :) )

I was surprised to learn a few years ago that the magnetic field lines are only about 22 degrees from vertical where I live. So if the rocket is angled 22 degrees from vertical toward the magnetic South, no roll orientation could be detected from the magnetic field at all, and any more pitch than that would result in a control reversal. Good-quality gyros are cheap and easily available these days. Better yet, you don't have to differentiate the gyro signal to calculate your control inputs.

Regarding the flight altitude - I don't think the effect will be measurable. If the rocket is rolling, it's inducing slightly more drag, but the roll actually reduces the amount of lift the fins are producing. When the canards are correcting the roll both the canards and main fins are producing lift (in opposing directions), and thus drag. My guesstimate is that it pretty much cancels out.

I think you're right that in an unstabilized rocket, the roll rate prevents the tail fins from producing net lift in the roll direction. If the canards are counteracting that motion, then the canards and the tail fins will both be producing lift in opposite directions, and more induced drag than if the rocket were spinning freely.
 
Another time thanks to all.

Sampo, I know you for OpenRocket. Great program! I read your degree thesis a couple of time.
I saw thase video on this forum some times ago. How much large were canard fins and what was their distance from aft fins? What was the maximum fin deflection?
 
WOW! Just wow! To see your project go from concept to reality with such complexity is fantastic. I can't even begin to comprehend the complexities of this project, then again, my field of study is soil science. Great work sir!
 
Great project, your work is excellent; using the 6DoF sensor was a brilliant move. Welcome to the roll stabilization club!

attachment.php
 
Hi,

I saw thase video on this forum some times ago. How much large were canard fins and what was their distance from aft fins? What was the maximum fin deflection?

The fins were rectangular with the root chord about 6cm and semispan about 10cm. Distance from the main fins was around 1.5-2m. You can see them in this photo: https://eetu.tunk.org/haisunaata/isohaisu.jpg

It's quite a large rocket so we made the control fins rather large as well. Also the further away the fin are is of the rocket centerline, the more effective the momentum is, so we made the fins that shape. Max. cant was about 10 degrees. Don't know whether the triangular canards like the ones you've used has less trouble with the main-fin interference.

Cheers,
Sampo N.
 
using the 6DoF sensor was a brilliant move.
No, for roll control I used only the roll axis of the gyro: roll speed is filtered by a Kalman filter to remove noise and sent to a PID controller. The 6DOF study I made was to tune the PID controller, so in Matlab / Simulink I was able to understand what happen to the whole rocket dynamic when fins are deflected.
Other two axes of gyro are only for data recording purpose (to look for rocket pitch and yaw oscillations). Accelerometer and barometer, with the help of another Kalman filter, needs only to have altitude, speed and acceleration (for data analysis).

Don't know whether the triangular canards like the ones you've used has less trouble with the main-fin interference.
Yes! I saw that photo some month ago!
Maximum deflection is good.
I looked in literature for canard - main fins interference, and I concluded that it is negligible whe rear fins are more than 4 MAC (mean aerodynamic chor) aft canard fins. I also made a simulation with Digital DATCOM, and the answer was near zero downwash angle. I think your problems came from structural twist
https://en.wikipedia.org/wiki/Control_reversal
 
Here are some photos:
PCB making
dscn0086bi.jpg p1030497m.jpg elettronica.jpg

3D
esplosopartimobili.jpg esplososistemadicontrol.png

Contol system (right) plus altimeters: gwiz LCX and Kalmalt (Homemade barometric recording altimeter with Kalman filter)


Control system vain


 
Question, Adriano: Does the system attempt to continue stabilizing after recovery deployment, or is it shut off and placed neutral at apogee?

Later!

--Coop
 
You can see it in the complete video: fins stop moving at about 40s.
To make things simple I programmed the control system to control the rocket for 35 second. After this time control fins are positioned at 0°.
Even if fins are moving during the first part of descend phase, they can't control the rocket because air flow on them is too slow (aerodynamic force negligible).

[YOUTUBE]g4eliZDPqzk[/YOUTUBE]
 
I noticed that early in the flight of the stabilized rocket, the booster suddenly twisted relative to the av-bay. Any explanation for that? I guess you didn't use shear pins...
 
Yes! You noticed well! I didn't use shear pins and it was an error. Next time I will use them.
The twisting movement was not a negative occurrence. This additional trouble proved that control system is able to recover the rocket from a difficult situation.
 
Hi,

I looked in literature for canard - main fins interference, and I concluded that it is negligible whe rear fins are more than 4 MAC (mean aerodynamic chor) aft canard fins. I also made a simulation with Digital DATCOM, and the answer was near zero downwash angle. I think your problems came from structural twist
https://en.wikipedia.org/wiki/Control_reversal

The original fins had a approx. 8mm aluminum rods on which we had a ~1mm alu plates screwed at about 1/3 of the plate's chord (at the center of the rods). We glued blue foam padding on the plates to make more of an airfoil shape and finished it with duct tape. The fin chord was 5cm, so I wouldn't have expected that to twist.

The rocket body itself consisted of three parts, the lower two which were screwed tightly together during flight (so it could be split up for transportation), and there was a jut preventing the payload bay from rotating against the lower section. It shouldn't be able to rotate more than a mm or so.

The MAC length of the main fins is about 35cm (root chord 50cm, tip chord 10cm), and the control fins' chord is 5cm. The distance from the control fins to the leading edge of the main fins is 130cm, so it's 3.7 times the main fin MAC or 26 times the control fin MAC. Do you think this should be sufficient that the interference should not affect the control?

In the video it seems that the control works for a few seconds, but as speed picks up the control reverses and (based on logged data) even though the control fins are completely opposing the roll, the roll just increases. Then again any possible structural twist will also be larger in higher speeds. What was your source on the interference effect?

Cheers,
Sampo N.
 
I don't remember the exact source of the 4 MAC rule. My flight dynamics professor said me the rule showing some data (I don't know where she took them), but I have another useful paper about canard interference on main wing:
https://ntrs.nasa.gov/archive/nasa/casi.ntrs.nasa.gov/20010020654_2001028128.pdf
I also ran a simulation with Digital DATCOM wich estimates the downwash angle generated from canard on the main wings, and the result was nearly zero.

I think that an aerodynamic interference cannot bring to a similar control inversion. The effect is too big!
You can also estimate the torque on the fin with this paper:
https://ntrs.nasa.gov/archive/nasa/casi.ntrs.nasa.gov/20010020654_2001028128.pdf

Structural effects are quite big. My fins are made with 6 ply of carbon fiber (total thickness 1mm) and they show some torque at the max load (flight speed 100 m/s, max deflection 12°)
 
Good Job Neway. Fantastic!:wink:

I think i'm going to send you an email/pm:grin:
 
Just awesome! I'm starting to study control and dynamics....Is your thesis available for reading??
Congrats!
 
Wow. I miss that one, flight stabilisation not only roll is what I would like to achieve.
 
Back
Top