Thinking about it, there are three friction components. Dry friction is proportional to load (i.e if you put a heavy load on the table it will be harder to turn). Viscous friction is proportional to speed. Stiction is the initial resistance to movement and is mainly dependent on how long the slide has been stationary and the load on it. Most machines will have a mix of the three. A well oiled dovetail slide for instance will have quite a lot of viscous friction and stiction. Your rotary table sounds more like it has a lot of dry friction. My previous comment only covers viscous friction.
I get the feeling that adding an inverse deadband would cause problems, especially for small movements. Sudden steps in your response curve are a recipe for oscillation. The biggest problem is that the friction is likely to vary a lot. If you put something heavy on the table, friction will go up. Worse, once it beds in friction will probably go down. With an inverse deadband you are likely to end up with violent oscillation if the friction drops. A decent amount of integral should be able to handle friction. Integral compensates for steady state loads (e.g dry friction and cutting loads). Say you command a small movement. P, D and the feed forwards won't produce enough torque to move the table. However Integral will build up rapidly until the table does move. Even if you are only one count out of the dead band, integral will eventually build up enough to move that one count. With decent tuning the existing PID component should be able to handle most loads. Les > Les, I thought that friction is not at all proportional to speed? > > It is a value that only depends on the direction (sign) of speed, not on the > value of speed. > > Am I mistaken? > ------------------------------------------------------------------------------ The modern datacenter depends on network connectivity to access resources and provide services. The best practices for maximizing a physical server's connectivity to a physical network are well understood - see how these rules translate into the virtual world? http://p.sf.net/sfu/oracle-sfdevnlfb _______________________________________________ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
