XYZ Table Construction 3

[xyzjpk]

Need help building a large xyz table to study pendulum motion. The setup would be similar to a large gantry system. Each axis needs to be 3 ft long. The pendulum would be 10 lbs or so and would hang off the vertical z axis. All three axes need to be able to move at the same time between 4 and 8 inches/sec. I am looking to do this on the cheap, but have only found linear slide actuators that cost over $1000 each.

 

[unclesyd]

Come back with a better description of the movement needed.
What is the precision of the movement needed?

Are all at the same rate and time? You mentioned linear actuators.

Are do you need to federate the axis?

Are you needing the full 30"movement?

 

[GregLocock]

3 windscreen wiper motors: $30
9 feet of studding: $5
LED and receiver for each leadscrew: $6
3 cardboard discs with two or more holes in: $0

(Knowing how to put that lot together right first time: $100000)






Cheers

Greg Locock

 

[xyzjpk]

Thanks unclesyd-

Precision is not too important, maybe to within an eighth of an inch or less.

The x and y axes need a maximum speed of 4 in/sec while the z axis needs a maximum speed of 8 in/sec.

The axes will move at different rates defined by the controller. The goal is to define some control system.

The axes can all be connected to each other. The z is on the y and those are on the x axis. Alternative apparatus setups are welcomed.

I will need 3 ft of travel in all directions.


Thanks GregLocock-

How do the LED and receiver work with the cardboard disks?

 

[GregLocock]

You use the LED to count the number of revolutions of the disk, this tells you where the platform is. A windscreen wiper motor may struggle to get to your required speed in Z, as it has a fair weight to lift. A starter motor will do in that case.



Cheers

Greg Locock

 

[MikeHalloran]

Er, study pendulum motion?

As in, move the fulcrum in 3D to keep the ball fixed relative to the ground, and watch the ball rotate with a camera at a fixed location?

Greg may have underbid the job.

You may be missing an axis; see 'Foucault' and 'precession'.

To do something like that, i.e., construct a Cartesian gantry that provides a dynamically translating coordinate system, >>>without introducing errors of its own<<< in the motion of the pendulum, requires axis translators with extraordinarily low trajectory error, extraordinarily fine position resolution, and an extraordinarily large bit rate on the position feedback, or a completely analog control system. I.e., it's the sort of system where you'll be able to detect, and will have to correct for, torque ripple associated with motor commutation, and the passage of individual balls within ball bearings.

It might be easier to build the pendulum in the regular way, no mean feat by the way, and make the camera dynamically track its motion, with appropriately placed targets on the pendulum reflecting laser beams into four quadrant optodetectors feeding fast servos.

Either way, it's a _huge_ project.

I hope I've misinterpreted what you propose, and it really is something you can knock together with building materials and old car parts.



Mike Halloran
NOT speaking for
DeAngelo Marine Exhaust Inc.
Ft. Lauderdale, FL, USA

 

[GregLocock]

I bid to the spec!

Evil grin.

Actually you are right, I did underbid, getting the tables moving squarely will take more threaded rods and gears or chains.

None the less, guessing that this is not a commercial project, using linear slide actuators seems OTT.



Cheers

Greg Locock

 

[zekeman]

What is the purpose of this project? Is the pivot axis vector in the same direction? Are you collecting dynamic behavior while the sistem is in motion?
I think you should state the goals so that the forum can better serve your request.

 

[xyzjpk]

Thanks MikeHalloran and zekeman. I will try to clarify the project more. The pendulum will start out at a predetermined position relatively high above the ground. After it is released, the goal is to have a control system to actuate the actuators on each axis and get the pendulum to land safely on the ground. I will do control analyses to test out the control system. If the control system doesn't work, I will need to spend time looking more at the general pendulum dynamics with the xyz table.

Thanks Greg. As I am a young engineer with little mechanical experience, it looks like I need some help locating the hardware for building the xyz table.

 

[MikeHalloran]

Er, land safely on the ground?

As in, wait for the pendulum motion to damp itself out, then bring the 'ground' slowly up to it, or slowly drop the fulcrum?

OR,

Lift and translate the fulcrum rapidly after release, so it rotates around the ball, bringing the ball to a gentle stop just after 1/4 of an oscillation?

OR,

Lift and translate the fulcrum around the ball exactly as the ball would otherwise rotate around the fulcrum, and then damp out the fulcrum's inverted pendulum motion, while the ball remains stationary in space?

OR,

... what?



Mike Halloran
NOT speaking for
DeAngelo Marine Exhaust Inc.
Ft. Lauderdale, FL, USA

 

[xyzjpk]

Thanks Mike.
I will need to bring the pendulum to the ground and not the ground to the pendulum.
There really won't be much swinging action. The pendulum will be falling, so damping will need to come mainly from the translating actuators. The goal will be to damp out dynamics in the ball that hangs directly down off the vertical z axis.

 

[btrueblood]

Is the pendulum in free fall, or constant downwards velocity? Are you modulating the z-axis (vertical axis)?

 

[xyzjpk]

Thanks btrueblood. Ideally, I want constant downward velocity. I will probably throw in some accelerations to test robustness of the control system.

 

[nbucska]

XYZ:
Why don't you just use a computer model? it should be
easier, faster and perhaps even cheaper...


<nbucska@pcperipherals DOT com> subj: eng-tips
read FAQ240-1032

 

[GregLocock]

Just a sidepoint. You are actually intending to absorb energy from the pendulum. Your control strategy may need to cope with negative damping and other such lovelies.

I agree with nbuscka, I'd strongly suggest modelling it in Nastran 4D or equivalent before building it. Matlab/Simulink or Scilab would probably have an example of the 2d case. I'm pretty sure one of the above has an example showing how to balance a jointed pendulum above the platform.





Cheers

Greg Locock

 

[btrueblood]

xyz,

Are you trying to simulate a parachute deployment/landing for cargo drops or similar? Or just develop a general model and verify using a test rig? Reason being, as Greg pointed out, the mechanical system may not have the same energy dissipation as an aerodynamic system.

 

[unclesyd]

Agreeing that electric actuators are the best you might want to take a look at air cylinders. We use a good number of these in several of our manufacturing processes.

Depending on your desired precision you could use, in order good to better, precision air cylinders, guided air cylinders, or rodless cylinders.

One trick that I used to extend a table was to use Al faced honeycomb, virtually no increase in mass but very high strength and rigidity. You might check in optical bench catalogues extrusions for the gantry or use the ubiquitous 4" Al cable tray material.

I would check some of the surplus site as at time there are a lot of tables, actuators, and positioners available.

Here is one company we used.
They have some very good technical literature

http://www.branhampneumatics.com/bp2003/products.cfm?catID=42

Here is a supplier of honeycomb materials we used.

http://www.hexcelcomposites.com/Markets/Products/Honeycomb/Hexweb_attrib/Default.htm