DX-4 Rocket Build Thread
- Thread starter Joe Walsh
- Start date
Help Support The Rocketry Forum:
DAllen
Well-Known Member
How did I miss this thread the first time?!
op:
op:
noffie79
Well-Known Member
- Joined
- Jul 25, 2015
- Messages
- 1,340
- Reaction score
- 13
How did I miss this thread the first time?!
I missed this too, but I'm watching now!
dixontj93060
Well-Known Member
- Joined
- Feb 19, 2009
- Messages
- 13,083
- Reaction score
- 45
How did I miss this thread the first time?!
I missed it as I thought it was a simple Madcow build (but that's DX3).
JERRYR708
Well-Known Member
At least this thread is long enough now to get you through that bucket of popcorn without to much intermission.How did I miss this thread the first time?!
To get the control parameters, I made a simulation of the vehicle's behavior and tuned the control coefficients until I found a pairing that gave satisfactory response. Here is a video of the simulation:
https://drive.google.com/open?id=0B8B78hvl6NpyY1R3MW5HSExVeW8
Everything is now ready for an engine test.
https://drive.google.com/open?id=0B8B78hvl6NpyY1R3MW5HSExVeW8
Everything is now ready for an engine test.
So much for theoretical derivations of control coefficients. It looks like the parameters I derived lead to divergent response: https://drive.google.com/open?id=0B8B78hvl6NpyeFJHOWRSbmg5bHM. You can see how the magnitude of the oscillations steadily increases.
At first I thought I had accidentally given the wrong sign to one of the coefficients, but that turned out not to be the case. I'm going try again with more damping. I had hoped that the theoretical derivation would get very close to "good enough," but that doesn't seem to be the case. I guess this test program just took a turn for the expensive.
At first I thought I had accidentally given the wrong sign to one of the coefficients, but that turned out not to be the case. I'm going try again with more damping. I had hoped that the theoretical derivation would get very close to "good enough," but that doesn't seem to be the case. I guess this test program just took a turn for the expensive.
After testing several times with different control parameters, I think I've found a workable solution. There's still some oscillation, but without a large stockpile of motors to use in tests this will have to be enough.
https://drive.google.com/file/d/0B8B78hvl6NpycjZSaVplbFFfTzQ/view?usp=sharing
https://drive.google.com/file/d/0B8B78hvl6NpycjZSaVplbFFfTzQ/view?usp=sharing
KevinM
Well-Known Member
- Joined
- Sep 4, 2009
- Messages
- 129
- Reaction score
- 45
Hi Joe,
Are you using a PID for this controller, or are you now in to state space? From the point of view of a casual observer (with some control systems knowledge) your last video looked as though your system had too much inertia for the P-gain setting you had (causing it to overshoot your set point), where as now it looks like the P-gain term is so low that you're not really getting back to your set point. Did you try adding any D-term (with the previous P-gain setting) to slow down how fast you approach the set-point? I completely understand that it gets expensive to experiment like that and burn through motors...
It's great to see this thread moving along - it gives me motivation/reminder that I need to move from just "collecting components" to "start of build" on my own such project.
Are you using a PID for this controller, or are you now in to state space? From the point of view of a casual observer (with some control systems knowledge) your last video looked as though your system had too much inertia for the P-gain setting you had (causing it to overshoot your set point), where as now it looks like the P-gain term is so low that you're not really getting back to your set point. Did you try adding any D-term (with the previous P-gain setting) to slow down how fast you approach the set-point? I completely understand that it gets expensive to experiment like that and burn through motors...
It's great to see this thread moving along - it gives me motivation/reminder that I need to move from just "collecting components" to "start of build" on my own such project.
Kevin,
I'm glad you get some motivation from this project, thanks for lending your expertise to examine the control. I'm doing a PD controller; wanted to do state space but haven't had time to read the book I got on it.
The initial run had P = 1.0 and D = 0.3. The divergent response indicated to me that I needed less P-term or more D-term. So I tried P = 0.5 and D = 0.3. Still oscillating, but appeared to be stable (not diverging). So then I tried P = 0.5 and D = 0.4: that was better, but still oscillatory. Then I tried P = 0.5 and D = 0.5. This is the current configuration. The thrust is off-axis, so a little bit of tilt is expected.
I think you're right: perhaps I was a little too aggressive in reducing P. I think I might try P = 0.7 and D = 0.5. If there's an improvement, I'll fly with that.
I'm glad you get some motivation from this project, thanks for lending your expertise to examine the control. I'm doing a PD controller; wanted to do state space but haven't had time to read the book I got on it.
The initial run had P = 1.0 and D = 0.3. The divergent response indicated to me that I needed less P-term or more D-term. So I tried P = 0.5 and D = 0.3. Still oscillating, but appeared to be stable (not diverging). So then I tried P = 0.5 and D = 0.4: that was better, but still oscillatory. Then I tried P = 0.5 and D = 0.5. This is the current configuration. The thrust is off-axis, so a little bit of tilt is expected.
I think you're right: perhaps I was a little too aggressive in reducing P. I think I might try P = 0.7 and D = 0.5. If there's an improvement, I'll fly with that.
I finally launched this rocket on a D3 a couple weeks back. The rocket was initially stable but quickly went unstable and crashed.
Here is video of the launch: https://vimeo.com/230377307
Watch closely and you can see the rocket quickly accumulating rotational speed around the roll axis. Roll speed forces the gimbal to slew back and forth with the same period as the rotation rate, and my theory is that the max slew rate was eventually exceeded.
The roll speed was caused by the off-center motor configuration. I had reasoned that the torque induced by an off-center motor would average to zero as the gimbal moved around. In hindsight it was easy to think of very good reasons why this isn't true.
Here is video of the launch: https://vimeo.com/230377307
Watch closely and you can see the rocket quickly accumulating rotational speed around the roll axis. Roll speed forces the gimbal to slew back and forth with the same period as the rotation rate, and my theory is that the max slew rate was eventually exceeded.
The roll speed was caused by the off-center motor configuration. I had reasoned that the torque induced by an off-center motor would average to zero as the gimbal moved around. In hindsight it was easy to think of very good reasons why this isn't true.
LithosphereRocketry
Pining for the Fjords
- Joined
- Feb 19, 2017
- Messages
- 882
- Reaction score
- 93
Check out this guy's TVC projects for some inspiration, if nothing else...
www.bps.space
https://www.youtube.com/channel/UCILl8ozWuxnFYXIe2svjHhg
He has a much harder job - he's trying to land chuteless, SpaceX style...
www.bps.space
https://www.youtube.com/channel/UCILl8ozWuxnFYXIe2svjHhg
He has a much harder job - he's trying to land chuteless, SpaceX style...
That is so cool, thanks for sharing that. Looks like he ran into the same issue with the off-center motor mounting. Unfortunately, his post-flight summaries are too general for me to have understood the off-center thrust to be the cause of the crashes, so seeing that ahead of time would not have helped. Very cool regardless.
CORZERO
Well-Known Member
- Joined
- Feb 5, 2015
- Messages
- 714
- Reaction score
- 7
I signed up for the product notification!
Looks like a fun system to play with. I watched a few of his videos. It appears as though many of his test flights were significantly underpowered, and used BP motors. Not sure why. I wonder if his success rate would increase if he applied an appropriate amount of thrust? Pretty cool project, at any rate.
Last edited:
KevinM
Well-Known Member
- Joined
- Sep 4, 2009
- Messages
- 129
- Reaction score
- 45
This is a complete guess on my part, but I'd guess that the answer is tied to his main end goal of sitting the rocket back down using retro-rocket fire. The higher he goes, the more energy he'll have to burn off on the way back down (to a point) and certainly the harder it will be for his video equipment to observe what's going on as he comes over the top - it seems to me to be to his advantage (again, to a point) to minimize how much impulse he's using for his test flights.
As a third-party observer to both, the BPS rockets strike me as having a different goal from Joe's target of a stable upward trajectory under power. (Joe, feel free to jump in here and correct my flightpath if I'm off here!
BTW, I'm glad to year you're still at it!)KMc
