[wilcorp70]

For my senior design in aerospace engineering at NC State, we are building a reduced-gravity simulator to test rovers on. I have kept the idea of building a CNC machine in the back of my mind for a while for when I get out of college and can get a shop set up and have often lurked on these and other CNC forums. So when I saw that this thing would pretty much be a 2 axis CNC machine without the milling, I knew I should come here for solid information.

The design consists of one axis suspended above a “sandbox” test area, a carriage containing a servo slides along the top of the rails and is attached to the rover below by cable. The servo motor maintains a constant torque on the chord attached to the rover in order to simulate reduced gravitational pull using the equation: τ=r(mfg)sin⁡θ.
Where r is the radius of the servo reel, m is the mass of the test body, f is the fraction of earth’s gravity being simulated, and g is earth’s gravitational acceleration.

The carriage has to be powered and move along with the rover below. The carriage must travel 5ft and be able to travel at least 6 in/sec, I would like it to be more like 10 or 12 in. The carriage will carry a servo with a maximum weight of 50 lbs, as well as suspend a 40-50 lb weight below.

We have some spare parts from previous projects including two 5.5 ft long 80/20 Series 10 pieces, and six 1.5 ft pieces. There are also two 4in x 3.5in x 5ft aluminum I-beams.

We currently have about $500 budgeted for this part of the build. That can go up if necessary, but I hope not for just one axis. I was thinking we could use the 80/20 as rails, mounted on top of the I-beams for rigid support. I'm not sure whether to try to go with a ball screw or rack and pinion (leaning toward the latter). I also don't know where to get wheels that would work with the 80/20 as they seem to be obsessed with their teflon bearings.

Ignoring the vertical supports, can anyone help us figure out a configuration including motors and controllers that would work within our budget? We can have parts machined, but prefer something that can bolt together. Any help would be appreciated and links to parts would be amazing.

[doorknob]

While I'm not sure that I can address your general questions, it seems to me that your horizontal positioning requirements do not call for the kind of high precision or repeatability that you might need with a CNC router, and so if that assumption is correct, then you might find that a less-expensive drive method than using a ball screw or rack-and-pinion might be a chain and sprocket arrangement.

[marx911]

just curious why would you try and use a motor to "reduce" your gravity when a simple counter weight would allow you to focus on keeping the gantry and "rover" in sync?

if you are running in a true sandbox then wheel slip is going to make keeping the gantry and rover sync a true issue as I see it, the issue at that point would a pulling or dragging effect

just my thoughts but if you had 2 pulleys and ran it up over across then down you could hang a counter weight and that would clear your cable to the rover and would give you the reduced gravity effect ( 50 lb rover mars gravity 38% hang 31 pound counter weight net rover weight 19 lbs)



I'm sure much smarter guys than me will ring in with cool solutions but that just my 2 cents

Good luck

[Andre' B]

I agree on the torque motor, it seems an over complication over just using a long spring or a counter weight. The counter weight would introduce undesired inertia effects, but a spring or bungee cord could be very light and have minimal impact and would be a lot smoother and way more reliable then a torque motor setup.
I guess it comes down to is the problem to design the simulator or get the test data for the rovers.

Did you research how NASA did this for the Apollo program?
They did some very interesting stuff with basic old school engineering. Seems like nowadays everyone (mostly management so then can look cool at the next meeting) wants to use the latest high tech to solve problems our ancestors would have solved with a stick and a rock.
Anyway turning off the comments on the current state of engineering in the US.


Positioning the carriage over the load is not so much a CNC problem as it is just a servo loop problem.

You need a sensor that can measure the angle of the supporting cable in the X,Z plane and in the Y,Z plane, or 2 separate sensors.

The servo loops would take those angles and move the X and Y axis motors to keep the angle as close to zero as possible.

Now I am not the person to tell you how to use them but I can tell you that EMC has all the software components to make this fairly simple.

LinuxCNC.org
And the group here.
EMC/Linux (Enhanced Machine Control) - CNCzone.com-The Largest Machinist Community on the net!

[wilcorp70]

just curious why would you try and use a motor to "reduce" your gravity when a simple counter weight would allow you to focus on keeping the gantry and "rover" in sync?

just my thoughts but if you had 2 pulleys and ran it up over across then down you could hang a counter weight and that would clear your cable to the rover and would give you the reduced gravity effect ( 50 lb rover mars gravity 38% hang 31 pound counter weight net rover weight 19 lbs)

We thought about that early on, but that solution had a few problems. We wanted to be able to test rovers which might have a changing mass (taking samples, deploying probes, etc.). Using the counterbalance also makes it slow and tedious to test multiple designs as you have to manually change the counterweight. Hanging the counterweight also introduces another swinging (could be rigid but might still oscillate) mass into the system; while right now our servo weighs approximately the same as the rover test body, it can easily be upgraded to a servo that lifts twice the weight without being much if any heavier. None of these problems are insurmountable, and effectively all we are doing here is removing the counterweight and using the motor to apply the exact same torque on the pulley.

As for dragging, that is the point of motivating the carriage independently, the controller will be taking input from a potentiometer on the cable which will be attached to change with the angle. Using PID control we then tell the stepper to move to reset the potentiometer to zero. This should mean that no matter if the rover is slipping the carriage should stay directly over it. I hope I said that clearly enough. Our design is somewhat similar in concept to the one that NASA commissioned in 2011 from astrobotic,

it seems to me that your horizontal positioning requirements do not call for the kind of high precision or repeatability that you might need with a CNC router, and so if that assumption is correct, then you might find that a less-expensive drive method than using a ball screw or rack-and-pinion might be a chain and sprocket arrangement.

Thanks so much for replying, a chain and sprocket is a good idea, and it is one that we are thinking about. Our proof of concept version simply used the I-beams as rails, skateboard bearings on the carriage all pulled by a cable. So a chain setup would already be a vast improvement. Our ultimate goal (read: really hard) though is to be able to simulate larger asteroid gravity ~.02g. At that high of a gravity reduction, small things start having a larger effect proportionally. An error of .02g for simulating mars gravity is trivial, whereas its double the gravity of an asteroid. We may never get that kind of accuracy and there are lots of potential sources of error like in a CNC project, errors we will probably spend the entire semester ironing out. Since they gave us the money to do it, we want to try to nip this potential error source in the bud early so we can focus on smaller details, rather than worrying we may not have gone with an accurate enough positioning system.

Thanks for all of the comments, I told my team to scour this forum for ideas as it is full of wonderful and intelligent people who know what they are talking about. We have to present a basic proposal by Tuesday, so I'm hoping we can have a rough parts list with lead times by then.

[wilcorp70]

I'm not sure what the best way to attach the 80/20 to the I-beam would be. Any ideas? Im also looking for a source for some wheels that will work in the T-Slot.

[RomanLini]

I'd like to throw a totally different idea in the mix.

If you used 3 high ceiling mounted points in a triangle, with thin cables to the rover you don't need to move them. Instead you put a tension sensor on each point and a servo winch, to adjust the tension on each of the 3 cables.

With the right servo motor the cable tension is pretty much proportional to motor current, so you may even be able to dispense with the tension sensing and just use winch torque control.

If the rover is sufficiently below the triangle and doesn't move far, you can probably just run equal winch tension.

As the rover moves the winches will automatically supply or retrieve cable so that tension is maintained.

If the rover moves in a larger XY range, you just sense the rover XY position (overhead camera?) and use a bit of math, and adjust the 3 winch tensions accordingly so the force applied to the rover is always vertical.

[GSSTUART]

What about using a lighter that air balloon? No tracking problems, simply attach it to the rover and it would follow anywhere. If this is an exercise in axis programing and control that would miss the point, but if all you are trying to do is reduce the effect of gravity with a constant upward pull this might do it. A surplus weather balloon of the appropriate size with a lighter than air gas should be within your budget. You would need to use the gym or some large indoor area for the testing.

[harryn]

It might just me, but I don't really see what the 80 20 is buying you other than cost. That material is great for proto frame building, but if you want smooth, straight rails, other conventional materials will likely be cheaper. For example, one way to do this is to entirely skip the I beams and 80 20, and just pour a self leveling epoxy pad - such as precision epoxy does. You can mount your rails right on to this material, and it will be very flat and easy to install - likely flatter than you will achieve with I beams and 80-20.

Skate bearings are fine - but consider going a step forward and using the entire skate wheel, not just the bearings. The action will be smoother and exhibit much less wear on your rails. Swiss style skate bearings will be more conducive to your goals than other bearings, since they are designed for this kind of application.

The main reason to use the skate bearings without the polyurethane wheels on a cnc router, is that this would introduce a small amount of flex under cutting loads, and the finished cut would have vibration marks. This isn't really the case for your project.

It is an interesting project - good luck.