Re: Low cost Servo Controller
Posted by
Ray Henry
on 1999-09-16 07:38:28 UTC
Bertho. "Velocity levels" - Right. We're talking the same thing. I'm
just using the wrong words.
-----snipped Bertho's om -----
-----snipped Jon's om-----
way, you have to consider both the resolution of the position loop and the
resolution of the velocity loop. I stalled on building Jon's drives but
winters coming and I'll get back to them. His board essentially allows
continuous variation in motor velocity.
IMO a stepper system with most common drivers has only two velocities -
step and stop. So 128 velocities could allow for much better motion
control. If you made the steps between those velocities non-linear
(exponential or brake point or ?) you could build a servo that would go
very slow for cutting and very fast for rapid. The next couple of
questions to answer would be how often would you send that velocity byte to
the amp and how would you measure actual velocity and where would you
compare it to desired velocity.
Off topic - I heard a shop owner comment about one of his machines the
other day. "Well it's slow, only 750 ipm rapid but it's flexibility and
quick setup makes up for some of the loss." And I'm wondering if 50 ipm is
enough!
Ray
just using the wrong words.
-----snipped Bertho's om -----
>It is more like what resolution do we need on feed rate. 1 IPM to 128 IPMJon. Oops, copy error. My dislexia is showing.
>does not sound enough so I guess I talked myself into using two bytes
>for velocity.
-----snipped Jon's om-----
>Actually, 16384 discrete levels, + direction.head.
>I find very good results with EMC and the STG card, and my servo amps.
>I am using 1000 cycle/rev encoders, so I get 4000 counts/rev, x 5 revs/inch,
>to get .00005" / encoder count. This is part of the secret.
>
>On my mill, I often bore out circular holes using circular interpolation.
>If it is a critical part, I then bore it out to final size with a boring
>But, just doing the circular interpolation with an end mill gives a VERYOnce you get mechanical things like backlash and axis rigidity out of the
>good circle, maybe with a variation in diameter of .002" or so. Most
>of that is due to a little bit of backlash in my ballscrews. The results
>are VASTLY better than my AB7320 did, but then it was never properly
>tuned for my machine. I also now have the following error holding
>within .005" all the way up to 105 IPM, with a following error of
>about .0005" or so at low feed rates.
way, you have to consider both the resolution of the position loop and the
resolution of the velocity loop. I stalled on building Jon's drives but
winters coming and I'll get back to them. His board essentially allows
continuous variation in motor velocity.
IMO a stepper system with most common drivers has only two velocities -
step and stop. So 128 velocities could allow for much better motion
control. If you made the steps between those velocities non-linear
(exponential or brake point or ?) you could build a servo that would go
very slow for cutting and very fast for rapid. The next couple of
questions to answer would be how often would you send that velocity byte to
the amp and how would you measure actual velocity and where would you
compare it to desired velocity.
Off topic - I heard a shop owner comment about one of his machines the
other day. "Well it's slow, only 750 ipm rapid but it's flexibility and
quick setup makes up for some of the loss." And I'm wondering if 50 ipm is
enough!
Ray