On Fri, 07 Oct 2011 12:21:19 -0500 Jon Elson <[email protected]> wrote:
> andy pugh wrote: > > On 7 October 2011 03:10, Bruce Klawiter <[email protected]> wrote: > > > >> I am converting my Anilam M to EMC2 with Pico Systems, PPMC. > >> I am having a hell of a time doing the PID tuning with the linear > >> scales 10um. > > > > Not an answer to your question, but I do wonder how Anilam did it. > > After all, the machine used to work. > > > > It might be instructive to consider the differences between the > > controllers, to see what is currently not happening, but used to. > > > > Ignoring position control, does the machine achieve good velocity > > control for any given input voltage to the amps? (this might be > > rather "exciting" to test). > > > It is mostly working now. He has a following error of about 3 > encoder counts, max. > That sounds as good as it is going to get. Bruce stated that the > original machine could > hold .0005" tolerances on parts. I find this hard to believe with an > encoder resolution > of .0004". Maybe a good machinist could get it to make consistent > parts with some > dimension held to +/- .0005", but it is not believable that arbitrary > parts could > be made with accuracy nearly equal to the resolution. > > There may also be some encoder noise problem, or maybe RT latency > issues in the system. > > Jon > Just to cite some experience along these lines. Machine is a Mazak V5 but well used. Backlash is 0.0035 or so and following error is x and y in on the order of 0.0005 at 200 ipm, resolution is 1/25400. I had and occasion to make some oblong blocks approx .75 x 1.5 inches. Made 20 of them so this was a good chance to at least calc SD. SD came out at 0.001. This was with normal feeds and speeds. Rougher at +.003 followed by finisher. BTW - dimensions were biased positive by a thou or two; which could presumably be dialed in to parts +/- the SD. I think all machinists should try this kind of thing on their own machines to get a sense of what is actually achievable. Note: this was on 1020 steel. I would hope that Al would be better. :-) This kind of testing can be pretty sobering to those hoping for aerospace tolerances off well used machines. Reality test are seldom fun. :-) Damn! Hope this helps. Dave > ------------------------------------------------------------------------------ > All of the data generated in your IT infrastructure is seriously > valuable. Why? It contains a definitive record of application > performance, security threats, fraudulent activity, and more. Splunk > takes this data and makes sense of it. IT sense. And common sense. > http://p.sf.net/sfu/splunk-d2dcopy2 > _______________________________________________ > Emc-users mailing list > [email protected] > https://lists.sourceforge.net/lists/listinfo/emc-users ------------------------------------------------------------------------------ All of the data generated in your IT infrastructure is seriously valuable. Why? It contains a definitive record of application performance, security threats, fraudulent activity, and more. Splunk takes this data and makes sense of it. IT sense. And common sense. http://p.sf.net/sfu/splunk-d2dcopy2 _______________________________________________ Emc-users mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/emc-users
