CAD CAM EDM DRO - Yahoo Group Archive

Re: [CAD_CAM_EDM_DRO] Re: Re: PIC-Servo

Posted by Jon Elson
on 2002-07-10 09:45:04 UTC
Christopher Morse wrote:

> So you are saying you are using EMC to run a servo
> without a step converter?

Of course! That's the best way to move a machine, for several
reasons. The EMC computer always know the true position, and can
declare a following error within 1 mS of the machine exceeding the
error bound set for that axis. (EMC now has a sliding-scale error
bound, too, so larger errors are permitted at higher speeds.)
Given the current position, desired position and desired velocity,
a new velocity command is computed every servo cycle, and sent
to DACs that output a +/- 10 V signal that commands the servo
amps to deliver that velocity. Most velocity servo amps also sense
motor current, and use that in an inner control loop to improve
response time and stability.

> PIC-Servos close the loop
> on position and are controlled by position commands
> via a serial line. The problem is to get EMC to
> communicate via a serial line.

Why do you want to? Position loops are the wrong way to approach
the problem. Velocity loops were determined to be the best method
of motion control back in the 1950's, and physics hasn't changed.

Is this a 2-way communication, such that the CNC computer has a
position readout, or does it just go to wherever the computer says
to go?

> With the new CMC chipset that buffers position
> commands it seems that all that is required is for EMC
> to output packets with the desired positions for all
> axes.

This could be done with the code that is in steppermod.o, as that
is what it does at a certain level. How many position updates a second
can the PIC-servo handle? I do 1000 position readings and compute new
velocity commands every second. EMC could go faster, if a fast, light
axis system needed it, 10 KHz would be no problem. A 333 MHz Pentium II
CPU can do a 3-axis servo calculation, INCLUDING the I/O, in 50 uS!

> Real-time output is not necessary since the
> PIC-Servo buffers up a bunch of commands. Has this
> type of communication been implemented in EMC?

EMC is controlled by a real-time clock in the motherboard. I think
you'd have a problem if the PIC-servo had its own idea of real-time
that drifted over time with the commands coming from EMC.

Jon

Discussion Thread

babinda01 2002-07-08 05:10:47 UTC PIC-Servo Christopher Morse 2002-07-08 17:30:48 UTC Re: [CAD_CAM_EDM_DRO] PIC-Servo andrew abken 2002-07-08 20:43:53 UTC Re: [CAD_CAM_EDM_DRO] PIC-Servo Brian Pitt 2002-07-08 21:23:00 UTC Re: [CAD_CAM_EDM_DRO] PIC-Servo Ray Henry 2002-07-09 07:03:41 UTC Re: Re: PIC-Servo Jon Elson 2002-07-09 10:25:53 UTC Re: [CAD_CAM_EDM_DRO] Re: Re: PIC-Servo dumbdrive 2002-07-09 13:59:04 UTC Re: PIC-Servo Joll503@A... 2002-07-09 16:00:50 UTC Re: [CAD_CAM_EDM_DRO] Re: PIC-Servo Christopher Morse 2002-07-09 17:31:33 UTC Re: [CAD_CAM_EDM_DRO] Re: Re: PIC-Servo dumbdrive 2002-07-09 17:45:35 UTC Re: PIC-Servo andrew abken 2002-07-09 18:17:03 UTC Re: [CAD_CAM_EDM_DRO] PIC-Servo andrew abken 2002-07-09 18:18:32 UTC Re: [CAD_CAM_EDM_DRO] Re: Re: PIC-Servo andrew abken 2002-07-09 18:21:51 UTC Re: [CAD_CAM_EDM_DRO] Re: Re: PIC-Servo andrew abken 2002-07-09 18:49:51 UTC Re: [CAD_CAM_EDM_DRO] Re: Re: PIC-Servo Jon Elson 2002-07-10 09:45:04 UTC Re: [CAD_CAM_EDM_DRO] Re: Re: PIC-Servo Jon Elson 2002-07-10 09:49:09 UTC Re: [CAD_CAM_EDM_DRO] Re: Re: PIC-Servo Jon Elson 2002-07-10 09:56:35 UTC Re: [CAD_CAM_EDM_DRO] Re: Re: PIC-Servo