Where do we go from here, was [CAD_CAM_EDM_DRO] Re: A crazy idea
Posted by
bsptrades
on 2002-06-07 00:46:23 UTC
--- In CAD_CAM_EDM_DRO@y..., "Carol & Jerry Jankura"
<jerry.jankura@s...> wrote:
I would agree, the wintell Mafia and basic PC are headed and always
have been away from the path that real-time and the machine world
need. The PC is a one user tool in the consumer world made to run
shrink wrapped software and such.
Once upon a time control systems were custom boxes built to do a
task and often very expensive unique designs , I know I have built
many. The PC came along and some clever folks figured out that you
could use this inexpensive platform for other tasks. Not that it is
really well suited or robust but the fact that it can be adapted to
do the job and is available for cheap on the consumer market. Now the
trends are once again migrating away from this application.
This should pose no real problem though in the shadow of the PC is
an ever growing DSP market , micro controllers, PDA's and such that
will easily adapt to the control tasks. Similar to EMC one needs to
design the current process in a standard modular fashion in a
portable platform such as C. I can port any standard C module onto a
$5 AVR processor or TI DSP the down side is the mass availability and
knowledge base is lacking compared to the PC platform. I don't worry
too much however with the processing power of new chips an
alternative will emerge. I have seen micros such as the SX chips in
the 100mhz range being packaged similar to the basic stamp but with
more robust language support.
What I do wonder about is the current trend I see so many trying for
the machine control. What was one adapting the common hardware to do
a specific job is being treated more as a software application. A
control system is not a PC any longer and should not be treated as
one in my opinion. The PC is hooked to the machine tool to be a
controller, to get great real-time performance it is unrealistic to
expect desktop response and user functions as well. I believe the two
should be completely separate as possible, the PC runs user software
and the machine controller is optimized to run the machine. You
should not need a zillion MHz processor to run a lathe if you don't
expect the dang mouse, sound card and floppy to all work at the
same time with it.
We use rt-linux all the time for simulation work, man in the loop
critical control processes. The control system computer does not have
a hard disk,keyboard, mouse or video screen. You build a clean small
kernel, boot via compressed image from CD to ram disk or use a flash
drive. You communicate via X-windows from the dev station. I believe
this is the best setup for a machine controller as well. In a sense a
black box dedicated to the machine, lean mean and bullet proof. E-
stop big deal next power up it reboots and starts over you can't
corrupt a nonexistent hard disk.
In short I believe folks should view the CNC machine as a black box
no matter what the current method. You are not using a PC for the PC
environment on the machine tool you use the PC because it is a
powerful black box that can be made to do it. This is not only
good real-time policy but keeps the concept modular so that when the
next black box comes along you can use that.
There is a reason that FANUC does not come standard with 8000 screen
savers a web browser and solitaire. Not that they can't do it but
more so they don't want to.
Again my opinion but the CNC controller should be just that, first
and foremost a machine controller. Optimize for machine control then
if you can do more great but don't count on it.
Brian
BSP
<jerry.jankura@s...> wrote:
> I'm wondering when the time will come that the machine control sideof the
> house will realize that the office automation side of the house isno longer
> going to produce machines that have the attributes required forautomation
> and that they'll have to start developing their own hardware.Windows
>
> Already, we have "black boxes" that take a serial output from a
> based machine to assure that the timing necessary for fast,synchronized,
> moves can be met. How long until someone expands that black box toinclude
> the controller software (G-Code interpreter, that is) and uses thePC as a
> "terminal" to send high level commands to the box, and then untilsomeone
> integrates the video into the box to eliminate the PC.the basic
>
> Whether it's I/O mapped or memory mapped, you still can do I/O at
> chip level.Jerry ,
>
> -- Jerry
>
> |Yes, old PC's with serial and parallel ports will be
> |around for a while, but eventually the supply of old
> |equipment dries up. When was the last time you ran
> |across a machine with an 8" floppy drive? a 5-1/4"
> |floppy drive?
I would agree, the wintell Mafia and basic PC are headed and always
have been away from the path that real-time and the machine world
need. The PC is a one user tool in the consumer world made to run
shrink wrapped software and such.
Once upon a time control systems were custom boxes built to do a
task and often very expensive unique designs , I know I have built
many. The PC came along and some clever folks figured out that you
could use this inexpensive platform for other tasks. Not that it is
really well suited or robust but the fact that it can be adapted to
do the job and is available for cheap on the consumer market. Now the
trends are once again migrating away from this application.
This should pose no real problem though in the shadow of the PC is
an ever growing DSP market , micro controllers, PDA's and such that
will easily adapt to the control tasks. Similar to EMC one needs to
design the current process in a standard modular fashion in a
portable platform such as C. I can port any standard C module onto a
$5 AVR processor or TI DSP the down side is the mass availability and
knowledge base is lacking compared to the PC platform. I don't worry
too much however with the processing power of new chips an
alternative will emerge. I have seen micros such as the SX chips in
the 100mhz range being packaged similar to the basic stamp but with
more robust language support.
What I do wonder about is the current trend I see so many trying for
the machine control. What was one adapting the common hardware to do
a specific job is being treated more as a software application. A
control system is not a PC any longer and should not be treated as
one in my opinion. The PC is hooked to the machine tool to be a
controller, to get great real-time performance it is unrealistic to
expect desktop response and user functions as well. I believe the two
should be completely separate as possible, the PC runs user software
and the machine controller is optimized to run the machine. You
should not need a zillion MHz processor to run a lathe if you don't
expect the dang mouse, sound card and floppy to all work at the
same time with it.
We use rt-linux all the time for simulation work, man in the loop
critical control processes. The control system computer does not have
a hard disk,keyboard, mouse or video screen. You build a clean small
kernel, boot via compressed image from CD to ram disk or use a flash
drive. You communicate via X-windows from the dev station. I believe
this is the best setup for a machine controller as well. In a sense a
black box dedicated to the machine, lean mean and bullet proof. E-
stop big deal next power up it reboots and starts over you can't
corrupt a nonexistent hard disk.
In short I believe folks should view the CNC machine as a black box
no matter what the current method. You are not using a PC for the PC
environment on the machine tool you use the PC because it is a
powerful black box that can be made to do it. This is not only
good real-time policy but keeps the concept modular so that when the
next black box comes along you can use that.
There is a reason that FANUC does not come standard with 8000 screen
savers a web browser and solitaire. Not that they can't do it but
more so they don't want to.
Again my opinion but the CNC controller should be just that, first
and foremost a machine controller. Optimize for machine control then
if you can do more great but don't count on it.
Brian
BSP
Discussion Thread
mariss92705
2002-06-06 07:21:50 UTC
A crazy idea
bjammin@i...
2002-06-06 07:31:46 UTC
Re: [CAD_CAM_EDM_DRO] A crazy idea
stephen_stallings
2002-06-06 07:39:51 UTC
Re: A crazy idea
Doug Harrison
2002-06-06 07:45:28 UTC
Re: [CAD_CAM_EDM_DRO] A crazy idea
mariss92705
2002-06-06 08:05:44 UTC
Re: A crazy idea
mariss92705
2002-06-06 08:33:15 UTC
Re: A crazy idea
stephen_stallings
2002-06-06 08:43:39 UTC
Re: A crazy idea
mariss92705
2002-06-06 09:10:54 UTC
Re: A crazy idea
waynegramlich
2002-06-06 11:00:45 UTC
Re: A crazy idea
Vajk Fekete
2002-06-06 11:52:29 UTC
Re: [CAD_CAM_EDM_DRO] Re: A crazy idea
Larry Edington
2002-06-06 12:04:02 UTC
Re: [CAD_CAM_EDM_DRO] Re: A crazy idea
Alan Marconett KM6VV
2002-06-06 13:18:06 UTC
Re: [CAD_CAM_EDM_DRO] A crazy idea
mariss92705
2002-06-06 14:08:19 UTC
Re: A crazy idea
Carl Mikkelsen, Oasis
2002-06-06 14:48:07 UTC
Re: [CAD_CAM_EDM_DRO] A crazy idea
JanRwl@A...
2002-06-06 16:50:44 UTC
Re: [CAD_CAM_EDM_DRO] A crazy idea
Carol & Jerry Jankura
2002-06-06 20:32:40 UTC
Where do we go from here, was [CAD_CAM_EDM_DRO] Re: A crazy idea
bsptrades
2002-06-07 00:46:23 UTC
Where do we go from here, was [CAD_CAM_EDM_DRO] Re: A crazy idea
bjammin@i...
2002-06-07 04:23:24 UTC
Re: [CAD_CAM_EDM_DRO] Re: A crazy idea
turbulatordude
2002-06-07 05:15:04 UTC
Where do we go from here, was [CAD_CAM_EDM_DRO] Re: A crazy idea
bjammin@i...
2002-06-07 05:27:26 UTC
A crazy idea
Carol & Jerry Jankura
2002-06-07 19:54:37 UTC
RE: Where do we go from here, was [CAD_CAM_EDM_DRO] Re: A crazy idea