More on the NEW controller
Posted by
Arne Chr. Jorgensen
on 1999-09-21 00:49:10 UTC
Hi,
I am just adding a few words about the subject, before I hear any
comment on my previous input.
Some may have thought I just made a joke, and for those who don't
know what adaptive control is, here is a very simple analogy.
You stand 3 feet above ground and make a jump. You have adaptive
control. You know what forces to expect, so some muscles loose their
tension, other will do the opposite. If you did not have this
adaptive control, you would not be able to walk.
In the case of a controller, there has to be some a command path
added to the common one. This one will change the gain in the
servo. The model in EMC will know what forces the tool and table
will met in advance, and change the gains in the servo.
I believe that this has been planned for in the start of the EMC.
You have the inverse kinematics and such. This is exactly some of
the things that Dean Frank would like. What he is doing is to try
to measure some of these forces, but then it is already too late.
Sure he could gain something, but by working out these forces and
tune the controller to best met them, then you will gain something
better.
If this is possible, - would be up to Fred Proctor to tell. He
would have to give some details on a form of API for such a servo,
and we would have to make it. I do believe the STG board could be
used as a preliminary test bed. If he would be willing to undertake
some of the workload, is another question, or if we could get some
help from some University. At the moment, I don't have any job or
income, - so there is limits to what I could help with, - but I
guess there would be something I could help with. Time will show.
If not this is possible, then I don't see that there is much to
gain in an all digital controller. The best we have is a classical
one.
//ARNE
I am just adding a few words about the subject, before I hear any
comment on my previous input.
Some may have thought I just made a joke, and for those who don't
know what adaptive control is, here is a very simple analogy.
You stand 3 feet above ground and make a jump. You have adaptive
control. You know what forces to expect, so some muscles loose their
tension, other will do the opposite. If you did not have this
adaptive control, you would not be able to walk.
In the case of a controller, there has to be some a command path
added to the common one. This one will change the gain in the
servo. The model in EMC will know what forces the tool and table
will met in advance, and change the gains in the servo.
I believe that this has been planned for in the start of the EMC.
You have the inverse kinematics and such. This is exactly some of
the things that Dean Frank would like. What he is doing is to try
to measure some of these forces, but then it is already too late.
Sure he could gain something, but by working out these forces and
tune the controller to best met them, then you will gain something
better.
If this is possible, - would be up to Fred Proctor to tell. He
would have to give some details on a form of API for such a servo,
and we would have to make it. I do believe the STG board could be
used as a preliminary test bed. If he would be willing to undertake
some of the workload, is another question, or if we could get some
help from some University. At the moment, I don't have any job or
income, - so there is limits to what I could help with, - but I
guess there would be something I could help with. Time will show.
If not this is possible, then I don't see that there is much to
gain in an all digital controller. The best we have is a classical
one.
//ARNE