On Jan 5, 2009, at 7:12 PM, Stephen Wille Padnos wrote: > Greg Bentzinger wrote: > >> [snip] >> I am not proposing using an encoder for coordinated movement feedback >> like a servo would require, that would still be handled by >> counting the >> stepgen output. What I propose would be having an aux input for the >> encoders and a separate DRO display on the GUI like XA/YA/ZA. With >> the >> proper ini file setting an "in position check" could be done (with a >> acceptable tolerance factor) after each G00 or fixed cycle position >> move. If an out of tolerance position is detected an alarm could >> occur >> (the easy way) or a correction move could be applied (much more >> difficult). Normally if a G00 is issued the tool is free of the >> work so >> applying a correction should be as safe, if the tool failed to clear >> the work, then the damage was all ready done. >> >> > EMC2 already supports using an ecoder for feedback, and will stop if a > following error is detected. This happens during moves and while > stopped. It is trivial to connect an encoder to appropriate hardware > (parallel port and software counting for low pulse rates, more > advanced > devices for high step rates). The motion controller already compares > actual position to expected position, so if you connect an encoder, > this > will work with steppers. The only difference between stepper systems > and servo systems is that with stepper systems, the feedback is > usually > synthesized by the step generator - the motion controller basically > sees > how many steps were output instead of where the motor actually is. To > use real feedback, just use the encoder input to the motion controller > feedback input instead of the faked feedback from the step generator. > >> I'm guessing this would be very resource intensive. So a alternative >> might be to use a non-modal user defined M or G code to compare >> and/or >> reset position. This option would be handy for people using Step/dir >> servo drives where they are unable to adjust the in-drive following >> error such as Gecko 320/340 drives. Geckos allow 128 count error >> before >> tripping a servo fault. >> >> > It's not so resource intensive, it's already done ;) > I think all you need to do to reset the commanded/expected > positions is > to hit F2/F2 - machine off, machine on. When the machine is in the > off > state, command position is set to feedback position continuously, > so you > can do things like jog and not get a following error as soon as you go > back to machine on. To match a gecko, you'd want to set the EMC2 > following error tolerance a little less than the gecko I think (so > EMC2 > will detect a problem before the drive faults and has to be reset). > >> This won't help a machine fighting mid-band resonance, or one >> overloaded in a cut. That is a hardware/software issue where the >> builder needs to know the weakness and limits of the machine. However >> these comparisons if logged could help the builder tweak accel & max >> vel settings to an acceptable range that will run error free 95+% of >> the time. >> >> > Right, you can't fix mechanical issues this way. You can plot the > difference between commanded and actual position using halscope. > > - Steve
If my memory serves me correctly then there might be an intermediate position. I think code exist by JMK to use an rotary encoder plus a linear scale. Could this be extended to separate the I from PD in steppers and use an encoder on I to drive final position. Maybe I've been smoking the wrong stuff .... who knows. More caffeine certainly won't fix things. :-) Dvae > > > ---------------------------------------------------------------------- > -------- > _______________________________________________ > Emc-users mailing list > Emc-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/emc-users ------------------------------------------------------------------------------ _______________________________________________ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
