Hi all,
While this is only my first post on this forum, I have been observing it for many years and have done my own fair share of tinkering with water rockets. I thought I might just add my own 2 cents to this discussion, since I have also been toying with the idea of using gyros in water rockets, with an arduino.
From what I understand with my own messing around, the combination of sensors used in this kind of deployment system really depends on what you are looking for. If you want to have a fully recorded model of a rocket’s orientation AND rate of change in orientation then you probably need to be looking at some of the 6DOF sensor systems, which are essentially a 3-axis accelerometer and a 3-axis gyro working together.
I am yet to test it in an actual rocket, but I have been thinking that a slightly cruder approach using just a 3-axis gyroscope (or potentially one with fewer axes) could be used to achieve similar results. A gyroscope measures the rate of change of the sensor’s orientation (in units like degrees per second). This does not then actually give the exact orientation of the sensor but it can show how quickly any orientation change may be occurring and in which axis. My thought was then to monitor this rate of angular change data in a rocket. As the rocket pitches over at apogee, this may not necessarily happen that quickly (depending on the rocket design and balance) but it should be a reasonably large non-zero value for a few fractions of a second – which should stand out from sensor noise even if you don’t have that much filtering. Essentially you would pick some value above the noise, and given the sensor data rate (equivalent to some time period), the pseudo-code would be:
- Code:
[Looped]
counter=0
N=chosen threshold value
If x>N
add 1 to counter
If counter > 100 (arbitrary number of consecutive counts above threshold desired to define a rocket pitchover)
Pitchover detected! - activate parachute deployment
end
end
This threshold value could be adjusted to suit a particular rocket, but fancier code could also be used for filtering and smarter detection of the pitchover could also be done using an integrator-type analysis where the area underneath a rate of angular change plot could be found – so that faster pitchover and slower pitchovers could both be detected by the same code as they would add to the same net change in rocket angle.
As I mentioned before I have been looking at a 3-axis gyro for flexibility but one axis would have be assigned as being parallel to the rocket axis, and thus ignored for the purposes of detecting the pitchover (as it would only record the axial spinning of the rocket).
I have tested this rough concept on the ground very roughly using an Arduino Uno (and so it was tethered to the desk) and it seems to work, but I am yet to adapt it across to an Arduino Pro Mini for more portable and realistic tests. The gyro I have been using here is a 3-axis gyro taken out of a Wii Motion Plus, an addon for the Nintendo Wiimote. They are pretty affordable ways of sourcing decent gyros, the onboard gyro measures at up to 1500 degrees per second which should be plenty quick enough for this application. Compared with the breakout boards from sparkfun you can probably get a Motion Plus for more like $15-25. It communicates with the arduino via I2C and code for Kalman filtering the sensor data is also around on the net. Plus, if you are wanting to upgrade to a 6DOF system then adding in the sensor from a Wii nunchuck adds the 3-axis accelerometers.
Some relevant links:
http://randomhacksofboredom.blogspot.com.au/2009/06/wii-motion-plus-arduino-love.htmlhttp://www.arduino.cc/cgi-bin/yabb2/YaBB.pl?num=1253202298That tilt board you found on sparkfun may work also, but it might depend on how violent the pitchover is - because everything should technically be in freefall should it? Only one way to find out though, and you might pick up the ‘wobble’ as the rocket tilts over…
Anyway just thought I would share some thoughts and I look forward to seeing how things go!
Cheers,
Kristian
Sun Jul 01, 2012 2:35 pm 
