I have one of the Guardians and have flown it lot on some R/C models.
In "3D" mode, it tries to hold wherever it is pointed and not let the plane rotate in pitch, yaw, or roll, when the transmitter sticks are at neutral. But it has NO IDEA where "up" is. It only resists rotation. if I put it at the highest settings for all three axes, in 3D, put the plane into a 45 degree dive, flip the switch to turn 3D mode on, then the plane will want to try to fly at a 45 degree dive, into the round if I let it
It does have some amount of "drift' over time.
But the above is not the magic answer for rocket boosts.
2D mode is the magic. I like to call it "wings level" mode. The system is turned on, with the plane wings and fuselage level. The Guardian senses the G load as it start up and remembers therefore what direction gravity is (WAS originally) coming from. If for example the plane is tilted nose-up 10 degrees at power-up, then the Guardian will have a 10 degree nose-up pitch error when activated in 2D mode. During flight, when in 2D mode, the Guardian holds the wings level, and the fuselage level. The amazing thing to me is that it remembers this even when flying around for 10-15 minutes, all sorts of turning and maneuvers, it still levels the wings and fuselage properly when letting the control sticks go to center.
Of course, for vertical rocket flight, we want "the fuselage" (rocket body) to be pointed straight up. But it is as simple as having the Guardian mounted horizontally the same orientation as in a plane, with the "top" side facing straight up just like it does for airplane flight. Simply that in that case, the airplane "roll" axis (aileron) becomes the rocket YAW axis, and the airplane Yaw axis (rudder) becomes the rocket ROLL axis. Pitch is the same (well, the rocket body is at 90 degrees compared to a plane fuselage, but the Guardian's sensing and control of the pitch axis is the same axis of rotation).
I think Eagle Tree is being "cagey" in their explanation of how it works. They make a reference to gravity...... but once in the air the plane creates its own G-forces that have nothing to do with Earth Gravity. My take on it is that it uses gravity mostly for the initialization process, then it uses the gyro chips and accelerometers (self-created G force vectors plus gravity vector) to keep track of exactly where it was oriented at start-up, and recalculates the numbers "on the fly". So, it seems to be a micro Inertial Measurement Unit. Which for a relatively inexpensive consumer non-military device I would expect to have some drift over minutes, but none that I could determine with airplane flights.
For a supersonic rocket, I do not think the velocity by itself would be critical. The greater issues would be the structural integrity/strength of the control surface pivot system, servos, pushrods, and control arms. That will be true whether using a $70 Eagle Tree or some military-grade guidance package. And making sure that the guidance system does not cause a way-too-high response that could create extreme g-loads that not only might make the rocket, or some critical part fail structurally, but possibly exceed the limits of the guidance system (But I do not think it is THAT hard to do. Being aware and concerned enogh to worry about it is half the battle already won) .
So, it's not so much whether the Eagle Tree Guardian is suited for it, but is the rocket designer suited to use any guidance system, for a given specific flight envelope? Anyone who wants to throw ANY kind guidance system into a supersonic HPR rocket, without doing any smaller scale test flights first, is either a super-genuis, or is fooling himself tremendously as disaster is very likely to occur without having had any practical experience.
Now having said that, for a supersonic rocket, you'd want to design the canard system and/or Guardian for less control effectiveness than for a rocket that would be flying 100-200 mph. On my Sunguidance rockets, I did experiments that made use of small canards, and large canards. For a really fast flying rocket I'd use smaller canards.
Also, the mechanical linkage ratio between servo output shaft and the shaft of the canard. For example, it could be set up for a 1:1 ratio, 30 degrees of servo shaft rotation produces 30 degrees of canard rotation. But for a very fast rocket, that could have higher stresses, it would be better to "gear it down", so to speak, such as a 1:2 ratio (50%), so that 30 degrees of servo rotation causes 15 degrees of canard rotation. That also has the benefit of the stress on the servo being reduced, so the canard system would require only 50% as much power from the servo as for a 1:1 set-up (this is why I made the gearing-down reference). And this issue of less power required by the servo is also true of using smaller canards. On the flip side, larger canards require more servo power. So no matter what, the servos need to be up to the job, in power and speed.
And, also, of course, the Eagle Tree (or whatever system) can be adjusted for less gain so that it is giving the servos enough control motion to steer the rocket gently on course, without being over-controlled. The down side to reducing the control effectiveness for supersonic speed, is that it will then not be able to control the rocket very well at lower speeds. The ideal situation for a hobby rocket with guidance would be for the guidance system to be able to self-adjust the gain during flight, high gain at launch, less and less gain as it goes faster. The higher-tech way to account for that would be to use airspeed sensors.... another way would be to program in a variable rate of gain (values determined by flight simulations) depending on elapsed time after liftoff (but then would need a foolproof launch detect). But the Guardian could not do that - for a programmable system you'd need something like an Arduino-based R/C model flight stabilization unit (Flyduino).
So, it is a mix of those above THREE things: The aerodynamic responsiveness of the system (Canard size in relation to the rest of the model, including CG location, size of rear fins, and
most importantly the Moment of Inertia), the mechanical rotation of the canards in relation to the servo motion, and the Guidance System responsiveness (depending on the gain).
In the animated GIF below, that is from a flight of my Sunguidance model (A bit larger gif than my avatar on all of my postings). It was set for a really "hard ride", very responsive. Now, that was not so great for an onboard camera. so I later made changes so that it would not over-control like that, but still had plenty of control authority.
georgesrockets.com/GRP/video/Airvid/Sun-Pics/SunTesTAnimation3.gif
Now, your concerns about the G-load rating of the Eagle Tree. There is one thing that I could think of as a likely problem. Hybrids. Some of the Hybrids, especially the Hypertek, have a LOT of vibration. Until people learned better, a number of hybrids (often Hypertek) rockets using accelerometer-based altimeters for electronic ejection, crashed because the vibrations confused the h*** out of the accelerometers so they could not accurately detect apogee, often never firing.
- George Gassaway
If we handed the Hobby King guy a sample, we'd have to shoot him and take it back. :kill: I think that might get us in trouble. :dark:
