Re: Jog wheel

["Dean L. Hedin" <dlh2001-at-comcast.net>]

Well, I can't seem to get the joystick to function on my EMC box.
"Joy2key" complains that it can't open /dev/js0.  I'll have to play
with it some more.

"Joy2key" looks to be somewhat configurable.  For example, you can optionally
have it send repeated keystokes if you hold the stick in one direction. So it 
looks like it might work ok with EMC.  Perhaps it may handle the keyboard up
/down message issue better.

In any case and after reading your post ...
I think the best way to approach the joystick jog would be with a seperate
program to run that handles the joystick input and generates NML messages
to EMC.   That way, one could have it so that the jog rate could be increased
as the joystick deflects more.  

I was thinking that a better control for the Z axis would be that you hold 
one of the buttons down and then the stick motion would only apply to z axis.

This frees up another button to be used for maybe a software estop or perhaps 
a logging feature.   I would love to be able to move the head to a position 
and record the coordinates (by pressing the button on the joystick), perhaps 
even directly to a g-code file.  I often do this manually to perform simple 
outline and routing tool paths with my machine.

Why am I thinking joystick and not wheels?  Well it seems to me that this is 
a peripheral that most low buget EMC'ers can swing and EMC ought to have 
some sort of limited support for this.  Jog wheels/encoders  need to be 
fabricated in some way.

However,  I would approach the jog wheel problem in similiar manner, 
I would somehow try to utilize an ordinary mouse.  It has optical quadrature
rotorary encoders already built in.  Maybe one could hack a mouse and attach
suitable handles to the optical encoders (or substitute one's own encoders, 
shafts, & handles)  

Then,  as in the case of the joystick  program mentioned above, one could 
then just translate mouse event data to jog motion messages to EMC.

The only implementation issue would be how to gracefully switch the mouse 
between the normal X desktop and the actual jogging function.  




On Saturday 01 June 2002 11:18 pm, Ray Henry wrote:
> Dean

> There are a bunch of cautions buried in this post.  And it attempts to deal
> with both the issue of a joystick and with a handwheel.  And it is not all
> my own work, Fred, Will, Matt, Steve, and I discussed some of this
> recently.
>
> Joystick
>
> It is easy enough to take a single keystroke and turn it into a motion
> command using Tcl/Tk and emcsh.  All that you would have to do is bind the
> keystroke to a specific jog command.  These bindings are already active
> with tkemcex.  I believe the also work with tkemc.

> Caution -- While the EMC motion modules themselves have no problem with
> jogging more than one axis at a time, part of the problem here is that you
> will get the key presses mixed up with the key releases if you try to run
> more than one set at at time from the keyboard or from these processes in
> tkemc.
>
> If we added a test variable to each of the jog processes that would prevent
> a second axis jog when one is already active we could prevent the keystroke
> interference problem.
>