Kenneth Lerman wrote: > > Our problem is generally different, though. We are changing the > *setpoint* and trying to get the system response to match the setpoint. > > Yes, this is a GREAT complication that so many articles on PID or servo systems in general totally ignore. And, it is "great" because it allows a very linear system, and therefore easily tractable mathematically, to be excited with non-linear stimuli, and therefore become not so tractable. Jerk is the ultimate non-linearity, as it can be unbounded even though the actual G-code is bounded. > Regarding comments Jon has made about 'I', some systems I've worked on > had something called "wind up limiting". The idea was to limit the > amount of integral that could be added to the system. > I think we also have limiting of integral wind-up. But, the problem comes right BACK to the changing setpoint. As soon as the commanded velocity changes, the integral history is no longer valid. If the direction changes, then the I is actually working AGAINST reducing error. But, of course, if you have a strong I term, you can't just sharply cut it to zero when an inflection occurs, that might be even worse. So, that's why the I becomes a curse, it is hard to make it work ONLY when it helps most, but not when it gets in the way.
Jon ------------------------------------------------------------------------------ LogMeIn Rescue: Anywhere, Anytime Remote support for IT. Free Trial Remotely access PCs and mobile devices and provide instant support Improve your efficiency, and focus on delivering more value-add services Discover what IT Professionals Know. Rescue delivers http://p.sf.net/sfu/logmein_12329d2d _______________________________________________ Emc-developers mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/emc-developers
