Jon, That was a good explanation of the possible problems that can occur when using linear scales for position feedback. What I still don't understand though, is how the knee milling machines in the shop that have linear scales, continue to run every day with very good accuracy, and no problems with servo hunting issues. I would think that a rotary encoder reading the rotary travel of a ball screw would cause a cumulative inaccuracy that increases with increasing travel of the table. A linear scale is measuring the actual movement,so in theory should be more accurate especially at greater distance. Can you tune out errors in ball screws using EMC, and get repeatable accuracy for most of the length of travel? Terry On Sat, Jul 3, 2010 at 1:19 PM, Jon Elson <[email protected]> wrote:
> Terry Neilson wrote: > > Jon > > Good to see your upgrade on a Bridgeport, you must have learned a lot > while > > doing it. I have used these Proto Trak gage wheel type encoders in the > shop. > > They have wipers and they work surprisingly well. I do not want to > include > > them on an upgrade though. > > > I'd never trust them to stay aligned. I sometimes home my machine and > then touch off with an edge finder, and work all > day without checking the alignment again. > > I would be curious if I could replace them with linear scales and > integrate > > them into an upgrade instead of rotary encoders. I would think that it > would > > be more accurate due to direct feedback of actual table position. > Yes, this is possible. There are a couple caveats. One is that you can > easily get a lot finer position resolution > with a shaft encoder. For instance, I have 1000 cycle/rev encoders > directly on 5 TPI ballscrews. So, the encoders > provide 4000 quadrature counts/rev * 5 TPI = 20000 counts/inch, or > 0.00005" (50 micro-inch) per count. > A typical old-style glass scale is 0.0005", or ten times coarser > resolution. Actually, anything made after 1975 or > so is really metric, so approximately .0004" = 10 um. You can now get > finer resolution scales, but they will > cost more. > > The other caveat is if there is any backlash in the ballscrew, you will > get servo hunting behavior. The servo motor > will endlessly hunt back and forth across the backlash distance so it > can nudge the table and encoder over one encoder > count. You can suppress this so some extent with the deadband setting. > So, if your backlash is under .001", it may work out OK. If your > backlash is several thousandths of an inch, I think you will be unhappy > with the way the machine works. > > You can suppress the effect of the backlash when settling, but you can > never suppress it while the machine is moving. > It effectively causes the servo loop to be momentarily in an "open loop" > condition every time the motor torque is reversed, > and that will make it very difficult to tune the servo response. So, > unless your ballscrews are very tight or adjustable, > you may find this to be unworkable. What I'm trying to convey here is > that anything that introduces a sharp discontinuity > in the control loop at any point can make the entire system unstable. > But, if your ballscrews are in prime condition, then go for it! > EMC certainly can handle linear encoder feedback, and your velocity > servo amps will handle the velocity control part of the > job. > > Jon > > > ------------------------------------------------------------------------------ > This SF.net email is sponsored by Sprint > What will you do first with EVO, the first 4G phone? > Visit sprint.com/first -- http://p.sf.net/sfu/sprint-com-first > _______________________________________________ > Emc-users mailing list > [email protected] > https://lists.sourceforge.net/lists/listinfo/emc-users > ------------------------------------------------------------------------------ This SF.net email is sponsored by Sprint What will you do first with EVO, the first 4G phone? Visit sprint.com/first -- http://p.sf.net/sfu/sprint-com-first _______________________________________________ Emc-users mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/emc-users
