help with kinematics - standard XY movement but ..

For laser & plasma purposes, I'm designing a lightweight high speed XXY platform (ie two long X rails, one orthogonal Y gantry), but the whole thing is unusually single-belt driven by two fixed servos, ie no motors on the moving Y gantry in order to save weight. The single belt not only drives the X axis, but also the tool-carrying carriage on the Y axis.


Although the design is perhaps elegant, the kinematics are unusual, as both servos are intertwined, even for straight linear X or linear Y motions, and perhaps not even EMC can be configured to control such a device.


However, I hope I'm wrong, and I can collaborate with an EMC kinematics expert to see if it is feasible to control this system under EMC.


To interested parties, I can either fax, or scan & email a drawing to make this layout clear.


cheers
Doug

sounds interesting ....
I would like to see more details.

I guess it is possible with EMC. There is only one thing you need:
A mathematical description of the relation between motor-positions (joints) and X-Y positions.
You need it in both directions:
given the motor/joint positions, you must be able to calculate the X-Y positions, and given the X-Y psotions you must be able to calculate the corresponding joint positions.
The first one is called forward tranformation, the latter one is inverse transformation.

More details you find here www.linuxcnc.org/docview/html//motion_kinematics.html .

rgds, Gerd

The problem is not necesarily all that complex.

I have been looking at his kind of drive combination in order to implement a flying probe tester.

If you go with servo drive and actual position measurement ( not inferred measurement from a rotry shaft encoder) then the coupling between the two axes ( as one moves the other one will too) is not a problem but a challenge for the PID controller to cope with what can only be described as external noise being injected into the loop.

Here is a scan, sorry for the quality. The belt is double-sided timing belt of a long length, the ends joined at the carriage. There are 8 timing pulleys, of which 6 are idlers, and the motors could be placed anywhere.... perhaps diagonally to distribute the tension, or perhaps the same (left or right) side, in order to minimize the distance from the drivers to the motors.

Only the timing belt, pulleys and carriage are shown, the actual rail mechanism is not relevant to the feasibility of this design. It is only necessary that the gantry remains orthogonal to the X axis, which is greatly aided by the non-slippable characteristic of the (dual sided) timing belt & pulleys. It would be most convenient to put the encoders on the motor shafts, but steppers could also be used (with no feedback).

For example, in a laser system, only mirrors need be mounted on the moving gantry, providing for VERY high speeds and accelerations. This might also do for sensing/camera scanning applications.

The motors are totally independent, but their relationship to each other determines the direction of the carriage. For example, if they turn opposite of each other at the same speed, a constant motion along the X axis results. However, I have no idea how EMC might be programmed to accept such a scheme.

Anyones help would be appreciated!

cheers
Doug

Try again with the scan (too many pixels the previous attempt) - Doug

hi Doug,

There should be no problem to implement a kinematic transformer in EMC2, if it does not even already exist. Not sure though what is name of this type of mechanic, although it is well known from drawing tabletts and other things.

Bigger problems could be in the mechanic/accuracy and resolution.

With respect to your drawing I consider positive x move from bottom to top, and positive y from right to left.
Provided positive motor movement is clockwise, the equations are:

without warranty, quickly developed.

Take the bipod example from EMC, replace the equations inside with above ones, compile a kinematic file and that's it.

rgds, Gerd

Gerd wrote privately:
>Not sure though what is name of this type of mechanic, although it is well known from drawing tablets and other things.
>Bigger problems could be in the mechanic/accuracy and resolution.


If there is a consensus that EMC might be able to drive this, then I'll go ahead and build a prototype - for a plasma cutter, where small amounts of inaccuracy/resolution/jerkiness is hidden by the crudity of the cut.

However, if it works well, I might also retrofit it with a homebrew 80W or 120W laser. I already have a commercial 80W laser, but theres lots I don't like about it - hence my desire to implement EMC on my own (larger) machine.

Doug

andypugh wrote:

I am positive that EMC can drive this, it is very similar to a number of bipod designs.
There is a rather elegant one here used as a foam-cutter www.cnc-hotwire.de/

yes, and here is the EMC2 configuration for it: code.google.com/p/emc2hotwinch/

rgds, Gerd