Re: [psas-avionics] looking for a project?
Jamey == Jamey Sharp ja...@psas.pdx.edu writes: Jamey Anybody considering taking on the ODE solver task: we'll happily take Jamey a simple algorithm like Runge-Kutta first, if that's where you'd like Jamey to start, and you can improve from there. Anyone who does should take a look at ode(1). GNU has a version distributed as part of GNU Plotutils http://www.gnu.org/software/plotutils/. (I beleive the UI, like for most of plotutils, is based on Bell Labs' version.) -JimC -- James Cloos cl...@jhcloos.com OpenPGP: 1024D/ED7DAEA6 ___ psas-avionics mailing list psas-avionics@lists.psas.pdx.edu http://lists.psas.pdx.edu/mailman/listinfo/psas-avionics
[psas-avionics] looking for a project?
Is anyone lurking who's interested in working on sensor models or physics models? We've been needing your help for a while now. :-) Please reply to me privately if you just want to express interest; but if you have questions or suggestions please reply on-list. And if you don't like these projects there are probably more I'm forgetting. Sensor modelling: I want to see a good model of GPS. The Global Positioning System is an incredible piece of engineering work. If you're like me you'll learn a lot and have fun doing it. We have quite a bit of GPS expertise in this group, but it's currently locked up in people who need to be working on other things. I want a model that simulates carrier phase, carrier velocity, and time of week for each of several satellites. That'll involve some basic orbital mechanics and some understanding of how GPS works. Physics modelling: I don't care what piece of rocket physics you want to model as long as it involves torque. Our current simulator never causes the rocket to rotate at all. Two possibilities are to model crooked fins (which leads to spin) or to model wind (which applies more force at the fins than at the nose, causing the rocket to lean into the wind). If several people are interested you can work together or model different kinds of forces. And you can start with a very simple model and make incremental improvements. If you have an interest in more advanced physics, please jump in here. Better atmospheric drag models, for example, would be awesome. We also have a to-do item to implement a decent numerical differential equation solver, such as Runge-Kutta. That was a homework assignment in an undergraduate geology class I took, so it isn't that hard ;-) but we haven't had time. Ideally you'd implement these models in C as part of our simulator, but if you'd rather prototype in another language, such as Matlab, that would still be a good start. Just remember that we can't fly anything using, for example, Matlab toolboxes, so you need to be prepared to help reimplement any toolbox or library you use. I'm hoping that we don't have an opening for someone to model magnetometers, because Theo already made a lot of progress on that. Anyone know if Theo will be back once the summer is over? Jamey ___ psas-avionics mailing list psas-avionics@lists.psas.pdx.edu http://lists.psas.pdx.edu/mailman/listinfo/psas-avionics
Re: [psas-avionics] looking for a project?
Thanks for this feedback! If anybody takes on implementing better physics models, please consult with Dan for advice. :-) On Mon, Aug 24, 2009 at 1:15 PM, Ikirk...@pdx.edu wrote: Quoting Jamey Sharp ja...@psas.pdx.edu: Physics modelling: I don't care what piece of rocket physics you want to model as long as it involves torque. I'm curious why you want to involve torque? Our rocket actually needs to model fins as a force applied some distance from the center of mass and the center of pressure. Torque (or if you will, a moment) is generated when a force vector is crossed with a position vector, assuming the force and the vector sum of forces are NOT co-located. It's literally as simple as f x r (where x is the cross product and f and r are vectors). The reason this is done this way in industry is because lateral fin forces contribute to rotations *AND* translations, and this coupling is what makes the rocket a non-minimum phase system. It's crucial that this relationship is present in the simulation dynamics, because it creates the need for a different control strategy than would be needed if you neglect it. First disclaimer: I'm not that good at physics. That's why I'm asking for help. :-) I don't understand the distinction you're making. I understand that torque comes from applying force off-center (for some value of center that I understand only in abstract terms). I also understand that a force like wind will tend to have both a linear and a rotational effect on the rocket body. But right now Josh and I only understand how to implement linear components, so I'm looking for someone to add the rotational components to the model. Are you saying I don't want what I thought I wanted, or that I'm using the wrong terms? Our current simulator never causes the rocket to rotate at all. Two possibilities are to model crooked fins (which leads to spin) or to model wind (which applies more force at the fins than at the nose, causing the rocket to lean into the wind). If several people are interested you can work together or model different kinds of forces. And you can start with a very simple model and make incremental improvements. The role only model is given on the roll control page, though I owe an explanation to make it useful. It will be expanded to include a damping force caused by the AOA of the main fins during rotation. I've done a planar 3DOF model (vertical and horizontal translation, plus one rotation) that I plan to show on that page after I finish my fin testing. Finally, I will expand the models to the general 6DOF model with unlimited force/position inputs, but maintaining the rigid body assumption. I think higher order body mechanics are unnecessary here. Awesome! I take it you mean this page? http://psas.pdx.edu/rollcontrol/ We also have a to-do item to implement a decent numerical differential equation solver, such as Runge-Kutta. That may be a start, but there are better algos out there, especially if are dealing with numerically stiff equation sets. Investigating them will help with singularities that will certainly when you are considering cm scale fins with planetary scale orbits. Ooh, I look forward to having that class of problems. Anybody considering taking on the ODE solver task: we'll happily take a simple algorithm like Runge-Kutta first, if that's where you'd like to start, and you can improve from there. Jamey ___ psas-avionics mailing list psas-avionics@lists.psas.pdx.edu http://lists.psas.pdx.edu/mailman/listinfo/psas-avionics
Re: [psas-avionics] looking for a project?
On Mon, Aug 24, 2009 at 01:41:51PM -0700, Jamey Sharp wrote: Thanks for this feedback! If anybody takes on implementing better physics models, please consult with Dan for advice. :-) On Mon, Aug 24, 2009 at 1:15 PM, Ikirk...@pdx.edu wrote: Quoting Jamey Sharp ja...@psas.pdx.edu: Physics modelling: I don't care what piece of rocket physics you want to model as long as it involves torque. I'm curious why you want to involve torque? Our rocket actually needs to model fins as a force applied some distance from the center of mass and the center of pressure. Torque (or if you will, a moment) is generated when a force vector is crossed with a position vector, assuming the force and the vector sum of forces are NOT co-located. It's literally as simple as f x r (where x is the cross product and f and r are vectors). The reason this is done this way in industry is because lateral fin forces contribute to rotations *AND* translations, and this coupling is what makes the rocket a non-minimum phase system. It's crucial that this relationship is present in the simulation dynamics, because it creates the need for a different control strategy than would be needed if you neglect it. First disclaimer: I'm not that good at physics. That's why I'm asking for help. :-) I don't understand the distinction you're making. I understand that torque comes from applying force off-center (for some value of center that I understand only in abstract terms). I also understand that a force like wind will tend to have both a linear and a rotational effect on the rocket body. But right now Josh and I only understand how to implement linear components, so I'm looking for someone to add the rotational components to the model. Are you saying I don't want what I thought I wanted, or that I'm using the wrong terms? If I understood the distinction correctly, it sounds like the complaint related to the use of torque rather than force, where the latter creates the former as well as other things. If so: yes, we want to work in terms of forces, and the problems we have arise because we don't yet understand all the details of applying forces to the rotational components of our rocket model. Also, I'd like to know exactly what you meant by non-minimum phase system. I've read http://en.wikipedia.org/wiki/Minimum_phase and I still don't fully understand it. Our current simulator never causes the rocket to rotate at all. Two possibilities are to model crooked fins (which leads to spin) or to model wind (which applies more force at the fins than at the nose, causing the rocket to lean into the wind). If several people are interested you can work together or model different kinds of forces. And you can start with a very simple model and make incremental improvements. The role only model is given on the roll control page, though I owe an explanation to make it useful. It will be expanded to include a damping force caused by the AOA of the main fins during rotation. I've done a planar 3DOF model (vertical and horizontal translation, plus one rotation) that I plan to show on that page after I finish my fin testing. Finally, I will expand the models to the general 6DOF model with unlimited force/position inputs, but maintaining the rigid body assumption. I think higher order body mechanics are unnecessary here. Awesome! Definitely awesome! I look forward to seeing that. - Josh Triplett ___ psas-avionics mailing list psas-avionics@lists.psas.pdx.edu http://lists.psas.pdx.edu/mailman/listinfo/psas-avionics