On Sunday, January 29, 2012 12:00:01 PM andy pugh did opine: > On 29 January 2012 13:23, John Thornton <bjt...@gmail.com> wrote: > > So, looking at one of the links for servo tuning is our > > > > FF0 = Velocity Feed Forward Gain. > > FF1 = Position Feed Forward Gain. > > FF2 = Acceleration Feed Forward Gain or Friction Feed Forward Gain. > > > > I would like to update the docs with some descriptive text in addition > > to "0th order feed forward gain" as that and the other descriptions > > only seem to make sense to a mathematics professor. Bingo John.
> There is a reason that mathematicians talk the way they do, it is to > be unambiguous in the general case. > I think the best solution would be for all LinuxCNC users to take a > maths degree, but that might be impractical. Well, not too practical, and certainly at my age. My math extends into the trig realm because it has had to in the electronics field, but the actual operation (sin,tan,hyp) are things that to me are not at all obvious, but I have an idea what the answer should look like so I grab my TI-35 or whatever, and see which gives me the good answer I need, usually without paying attention to which of those function families it was that gave me the correct answer. > The problem with your explanation is that it assumes a > position-command / position-feedback system and there are other > possibilities that are not that unusual. (closed-loop spindle control > is one) > > FF0: This term adds a value to the PID output directly proportional to > the input. It is useful for any system where a steady-state output > requires a non-zero input. The most likely case is a closed-loop > spindle speed control, where 10V might give 1000rpm, and a FF0 gain of > 0.01 would mean that the PID terms were only required as correction, > not the basic output value. This term would not normally be used with > a position-feedback system, except possibly to compensate for the > effects of a spring counterbalance. What if the spring is either too much or too little, this seems to invite the need for a non-symmetrical FF0 function. > FF1: This term adds a value to the PID output proportional to the rate > of change of the input. So, if the input is position, FF1 is > proportional to velocity demand. If the input is rpm then FF1 is > proportional to angular acceleration/torque demand. This term can > often be very useful to increase the responsiveness of a > position-feedback system with velocity-control servos. I don't know enough about this to comment since all my teeny stuff is stepper driven. > FF2: This adds a value to the PID output which is proportional to the > rate of change of the rate of change of the input (mathematically, the > second order differential with respect to time). In a typical > position-control loop this would correspond to the acceleration. (in a > velocity-control system it would correspond to the "jerk"). In the > position-control case this would be useful to compensate for the mass > of a heavy table, for example. Now, here is something that could even be useful to me. What is needed as I see it, is some way to convert the tables weight (mass actually) into something reasonably approximating the correct FF2 setting. Having example maths included in the description would ISTM, be a huge amount of help, lots more than the general recommendation of starting at .01 and working up. IOW, it the table weighs 30 pounds, or the table weighs 3000 pounds, it should be able to make an 'optimum' starting value a calculate-able item. I think it would have to be peak controlled by the individual stepgens MAX_ACCEL limits of course. > FF3: This value does not exist and has been inserted to see who is > still paying attention. A negative FF3 term could be used to add a > degree of jerk control to a position-feedback system. See, I'm still reading... And on the face of it, reads like it could be a useful to have function. But I didn't see the word 'stiction' in the above, and this is largely independent of mass in my mind. Its also quite real in my Z axis in particular. Probably somewhat reducible by more precise counterweight spring tensioning, but with the limited length of the bearing surface available, not likely a totally eliminate-able item. It IS going to rock and tilt with only a 3.5" bearing length on the posts ways. Not much, but I can measure it with a dial indicator. A users $0.02. Cheers, Gene -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author) My web page: <http://coyoteden.dyndns-free.com:85/gene> Do not handicap your children by making their lives easy. -- Robert Heinlein ------------------------------------------------------------------------------ Try before you buy = See our experts in action! The most comprehensive online learning library for Microsoft developers is just $99.99! Visual Studio, SharePoint, SQL - plus HTML5, CSS3, MVC3, Metro Style Apps, more. Free future releases when you subscribe now! http://p.sf.net/sfu/learndevnow-dev2 _______________________________________________ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
