Throw Mode for Rocket Powered Boost?

GeorgeG · February 16, 2019, 1:25am

Hi.

I’ve used Arducopter with some Mini-APM’s on some projects. My newest project will use “Throw Mode”. It will be a Rocket Boosted Quad, that after arming, and taking off 100% on rocket power, when it reaches apogee, I expect (!) for Throw Mode to stabilize the model into a hover.

So, can someone confirm to me that Throw Mode should work for this? It’s going to be accelerated by rocket thrust for about 1.5 seconds, then coast for a bit to apogee.

What does Throw Mode use for sensing apogee? Barometer only, or is it using inertial measurements, or a mix of both? If the Quad begins to veer to the horizontal under rocket thrust, I’m wondering if Throw Mode would detect apogee due to the barometer, or if not, whether the IMU might not detect apogee due to continued acceleration but by the horizontal axis. So this is a secondary issue, whether Throw Mode could likely kick in save a bad launch that is still thrusting but having reached apogee and losing altitude.

The rocket quad should be aerodynamically stable, but if one of the windmilling props spins differently there may be uneven drag to pull it to one side. Due to the design, it’s not practical to fold for boost and locking the blades to not spin on boost seems more of a risk of something to go wrong.

If there is a better place to ask, please let me know. I am asking in Copter 3.5 since I’m not sure how stable 3.6 is. I’ll be using a Pixracer for this, since the Mini-APM’s can’t run any version that has Throw Mode.

I know, the above may come off as an outlandish idea. But I have been flying model rockets since 1970, R/C planes since the late 70’s, and my last Quadcopter was this one (conventional in flight, not in looks or fabrication): https://www.rocketryforum.com/threads/flying-r-c-lunar-module-quadcopter-project.137174/

I will say that I am not planning to do a rocket boosted Lunar Module. But the design I do want to do this with will be a bit similar, but aerodynamically stable.

peterbarker · February 16, 2019, 2:29am

So, can someone confirm to me that Throw Mode should work for this? It¢s going to be accelerated by rocket thrust for about 1.5 seconds, then coast for a
bit to apogee.

throw mode should work for this.

What does Throw Mode use for sensing apogee? Barometer only, or is it using inertial measurements, or a mix of both? If the Quad begins to veer to the

The code is pretty straight forward:

github.com

ardupilot/ardupilot/blob/master/ArduCopter/mode_throw.cpp#L250


      
                      (double)velocity,
                      (double)velocity_z,
                      (double)accel,
                      (double)ef_accel_z,
                      throw_detect,
                      attitude_ok,
                      height_ok,
                      pos_ok);
              }
          #endif  // HAL_LOGGING_ENABLED
          }
          
          bool ModeThrow::throw_detected()
          {
              // Check that we have a valid navigation solution
              nav_filter_status filt_status = inertial_nav.get_filter_status();
              if (!filt_status.flags.attitude || !filt_status.flags.horiz_pos_abs || !filt_status.flags.vert_pos) {
                  return false;
              }
          
              // Check for high speed (>500 cm/s)

// Once a possible throw condition has been detected, we check for 2.5
m/s of downwards velocity change in less than 0.5 seconds to confirm

If there is a better place to ask, please let me know. I am asking in Copter 3.5 since I¢m not sure how stable 3.6 is. I¢ll be using a Pixracer for this,
since the Mini-APM¢s can¢t run any version that has Throw Mode.

What sort of g-forces are you going to be pulling here?

My guess is that the EKF is going to get unhappy.

You’ll probably clip the accelerometers, making the EKF unhappy…

I know, the above may come off as an outlandish idea. But I have been flying model rockets since 1970, R/C planes since the late 70¢s, and my last
Quadcopter was this one (conventional in flight, not in looks or fabrication):
FLYING R/C Lunar Module Quadcopter project | Rocketry Forum - Model Rocketry Forums

I will say that I am not planning to do a rocket boosted Lunar Module. But the design I do want to do this with will be a bit similar, but aerodynamically
stable.

You should probably check the legality of doing this in your jurisdiction,
BTW. It might not just be a matter of “don’t guide in boost phase”, it
might be “no guidance on anything that’s a rocket”…

GeorgeG · February 16, 2019, 2:57am

Thanks for the reply.

What sort of g-forces are you going to be pulling here?
My guess is that the EKF is going to get unhappy.
You’ll probably clip the accelerometers, making the EKF unhappy…

Estes D12 on a rocket of 10 to 14 ounces. Maybe as much as 10 G’s for the max thrust spike , dropping down to about 3 to 4 G’s after the first 1/2 second (actually less since this is going to be so draggy with those props windmilling once it gets moving).
But it would be good to know for some future rocket projects if there is a max G limit to keep under.

You should probably check the legality of doing this in your jurisdiction, BTW. It might not just be a matter of “don’t guide in boost phase”, it might be “no guidance on anything that’s a rocket”…

Thanks for the concern. But Model Rocketry has had onboard guidance since 1988, and I built one of the first successful ones (Sunguidance, flies towards the sun). Also mentored on a project that used an Eagle Tree Guardian for vertical guidance for rocket flights. Video of the Sunguidance project: https://www.youtube.com/watch?v=I6ZFSSBQNT0

There is also some incredible work being done by Joe Barnard of BPS.space , for vectored-thrust (gimbaled engine) vertical rocket guidance using similar sensors and microcontrollers as Multicopter FC’s use

peterbarker · February 16, 2019, 7:20am

Estes D12 on a rocket of 10 to 14 ounces. Maybe as much as 10 G¢s for the max thrust spike , dropping down to about 3 to 4 G¢s after the first 1/2 second
(actually less since this is going to be so draggy with those props windmilling once it gets moving).
But it would be good to know for some future rocket projects if there is a max G limit to keep under.

You can check the sensors present in the PixRacer
(Redirecting to latest version of document (master)) and look at the
data sheets for those.

Thanks for the concern. But Model Rocketry has had onboard guidance since 1988, and I built one of the first successful ones (Sunguidance, flies towards
the sun). Also mentored on a project that used an Eagle Tree Guardian for vertical guidance for rocket flights. Video of the Sunguidance project:
https://www.youtube.com/watch?v=I6ZFSSBQNT0

Interesting! Which jurisdiction?

There is also some incredible work being done by Joe Barnard of BPS.space , for vectored-thrust (gimbaled engine) vertical rocket guidance using similar
sensors and microcontrollers as Multicopter FC¢s use

Well, it sounds like ArduRocket would be legal somewhere in the world

Oh, I should also note that your GPS may go out to lunch, depending on
which it is and which “don’t use me on a ballistic missile” technology is
inside. Sorry, I don’t know what the thresholds for those are.

GeorgeG · February 17, 2019, 1:56am

You can check the sensors present in the PixRacer
(Redirecting to latest version of document (master)) and look at the data sheets for those.

I looked it up. Pixracer’s sensors can go as high as 16G’s, if they are configured to do so. If so, this model won’t exceed that, in any case as I figure max G at thrust spike to be 10G’s in the first half second.

As for model/ hobby rocketry and guidance, my “local” jurisdiction is the United States (any specific local laws regarding rockets would apply regardless of guidance or not). There are safety codes that do not allow models to be flown towards targets, whether ballistically or otherwise. So, so long as a rocket with guidance onboard is configured/set to fly UP and not target anything (on the ground or in the air), that’s OK. Having conducted the successful Sunguidance project in 1988, I’ve been VERY leery of anyone trying to work up a rocket guidance system that could be used to target something. The irony is that with modern GPS precision automated flight, anyone wanting to do harm with a dangerous payload could do so far easier with a propeller-driven model rather than a rocket (to a fixed ground target, anyway).

Oh, I should also note that your GPS may go out to lunch, depending on which it is and which “don’t use me on a ballistic missile” technology is inside. Sorry, I don’t know what the thresholds for those are.

IIRC, the threshold for GPS cut-out is supersonic velocity, or perhaps 700 mph. Not a hardware technical limitation but a software limitation to prevent the “Civilian” form of GPS from being abused for nefarious purposes (with a supersonic vehicle anyway) .

BTW - I want to keep this project under wraps from most of the model rocket community until I have it ready to test fly (didn’t announce my Lunar Module Quad till after the first flight). But it’s not likely that anyone on the rocket forums would be checking out this group. So, I’m converting a famous classic Estes model rocket to fly as a Quadcopter, the Mars Lander. There’s a photo of someone else’s model at this link: https://i.imgur.com/b6Ocw5q.jpg

And this is a drawing I’ve worked up for the Quadcopter version. https://i.imgur.com/3aYwUm5.gif
(legs will be at 45 degree orientation to the arms, was simpler for a basic planning drawing for my own use to not bother with a 45 degree view of the legs).

As a kid (1970), I got one of these kits but it was WAY too complicated for my skill level at the time to build it, so I cannibalized the parts. And this is the 50th anniversary year of that kit.

But the Chicken or the Egg part of this is that when I found out about Throw Mode 3 years ago, I decided to make a rocket boosted Quad. Then I started to think of what sort of design to do (nothing complex with folding/deployable arms). Maybe a totally custom design, or maybe adapt an old design. The instant that the Mars Lander came to mind, that was it, the perfect model to do this with. Well, it has some special challenges, but making such an iconic classic kit into a rocket boosted quad is worth it. And such a fitting match with the Lunar Lander.

peterbarker · February 17, 2019, 7:13am

Well, that sounds good. As a first, easy cut, you could presumably make a
body-less quadcopter (just include the PixRacer and its external sensors).
They do tend to get frustrated when their efforts to spin props don’t work
but you can arm the vehicle and ensure that it maintains its concept of
its attitude the entire flight.

Yours,