On Feb 3 2013 8:20 PM, TJoseph Powderly wrote: > On 02/03/2013 07:47 PM, EBo wrote: >> On Feb 3 2013 3:42 PM, Peter C. Wallace wrote: >>> On Sun, 3 Feb 2013, John Kasunich wrote: >>> >>>> Date: Sun, 03 Feb 2013 17:08:37 -0500 >>>> From: John Kasunich <jmkasun...@fastmail.fm> >>>> Reply-To: EMC developers <emc-developers@lists.sourceforge.net> >>>> To: EMC developers <emc-developers@lists.sourceforge.net> >>>> Subject: Re: [Emc-developers] custom EDM configuration questions >>>> >>>> >>>> >>>> On Sun, Feb 3, 2013, at 03:59 PM, EBo wrote: >>>> >>>>> possibly, but a more principled hack would be to interface it at >>>>> the >>>>> point where it calls the low lever Catmull-Rom spline, and just >>>>> evaluate >>>>> it in reverse. I do not know how difficult that would be, but it >>>>> would >>>>> require the ability to move in the negitive direction. I have >>>>> written >>>>> things like this for moving along splines with custom code, but I >>>>> would >>>>> have to look into how to do this in EMC-s low level motion >>>>> planning. >>>>> Probably the easiest thing to do would be to set up a >>>>> "reverse-motion" >>>>> pin in HAL, and then evaluate it internally, or modify the >>>>> adaptive >>>>> move >>>>> command to allow positive/negitive feedback voltage which would >>>>> imply >>>>> forward/reverse of motion. Anyway that is my thought off the >>>>> cuff. >>>>> >>>> Where things get ugly is when you try to do this in a general way. >>>> For >>>> example, what happens if you ask for negative feed while in a >>>> spindle >>>> synchronized move (single-point threading or rigid tapping)? >>>> Actually, >>>> that issue must have already been dealt with - adaptive feed would >>>> cause the problems in that case even for positive scale factors. >>>> But >>>> that was just an example. What about probing moves? What about >>>> the blend between successive non-co-linear straight lines? What >>>> about >>>> a profile that consists of many 0.002" long lines? Can you back >>>> up >>>> more than 0.002"? >>>> >>>> Sure, you can say "an EDM will never make a spindle synchronized >>>> move", but changes to the motion module need to do sane things >>>> regardless of what kind of a machine is being controlled. >>>> >>>> >>> What about different motion modules optimized different classes of >>> applications? >> Sure if you can describe realistic situations... >> >> >> ------------------------------------------------------------------------------ >> Everyone hates slow websites. So do we. >> Make your web apps faster with AppDynamics >> Download AppDynamics Lite for free today: >> http://p.sf.net/sfu/appdyn_d2d_jan >> _______________________________________________ >> Emc-developers mailing list >> Emc-developers@lists.sourceforge.net >> https://lists.sourceforge.net/lists/listinfo/emc-developers > i've written and investigated Linuxcnc for EDM for years now > maybe this will be of interest > > there are 3 basic forms of edm and each has unique needs > > all are EDM though and cannot tolerate the 'in position on time' > nature > of linuxcnc ( needs a different rule set for where the tool tip must > be > now ) > the tool tip must be in a position dictated by the process, > independant > of time and independant of the position where we were last check. > > the 3 types of edm are ( besides edg and a few even rarer modes) > 1) hole drilling... only moves along a single axis. usually a > single > joint > 2) sink edm ... retracts to a given point > ( maybe 3-8 axis, maybe 1, but the retract path is not > neccesarily > the approach path ( nor often the same )) > because it retacts to a safe point, this is 'sky blue', you > need'nt > have begun or been there before, its just safe > 3) wedm.. retracts along the programmed path thru one or more path > pieces > ( multiple gcodes/ multiple sequenctial tangential geometric > entities) > this requires a memory ( at least waypoints, like the segments > spit > out by the traj planner ) > > 1 & 2 are HAL solvable > 3 requires changes to linuxcnc position planning ( at least a > history > that can be un-done, like gps waypoints ) > > my own tests are only for sink edm ( some for hole poppers but my > HiPresure pump is down now ) > > i suggest something much simpler than the > VtoF to HAL to thresholds to motion proposal > i simply use a window comparator and output ADVANCE or RETRACT > signals > to hal > > hal knows the beginning and ending positions (limits) of the motion > and an OFFSET component's input is inc'd/dec'd according to the > process > ( according to the window comparator ) > > the machine cannot exceed the depth programmed ( could be x y z c as > you > like ) > so it has 'spark out' with a little hal-foo and timer ( or a higher > threshold in mine ) > It is not locked at the bottom, simply watch for it to be at the > bottom > for a len of time w/o backing off... thats a sign the work is really > completed > > the cut could go bad and retract to the RETRACT limit, and I abort at > that point > ( no where to run ) > > the cut could get done and spark out at the bottom > > right now i'm playing with feeding JMK's sine/cosine stepper example > to > implement orbiting from this lower 'roughing point' > > i specialize in sink edm. I just rebuilt 2 OKAMOTO's that were > terribly > dead and abused. > they cut fine now. I wrote all the macros for ELox system 8/9/10 > fanuc's > i wrote all the code for AGies during the 1980's > i wrote all the PLC and macros for Strathclyde & EDMsolutions cnc > edm's > i wrote the software that generates all the cutting tables ( etables > ) > for them also > now I've re-written all the user macros for OKO's too > circular vectoring spherical in any axis, subgates and back tapering, > copper/graphite in many material combos > > blah blah blah i knows a S^&%&load about cnc edm > > i think the F2V is overload and too slow, yes the process is slow, > but > the intelligence is fast. > you can advance slow but you better run away NOW and FAST > > i've posted it all too many times, schematics, code, ini's , hal's > nuff said > just seems like a old story that nobdy listed to anyways > > tomp > tjtr33
tomp, It will eb a little while before I get a chance to tare into this. I had a major mechanical failure in the truck on the way back home and need to deal with that over the coming week(s). I would be interested in links to the schematics, ini's, etc. you mentioned for reference if not implementation. EBo -- ------------------------------------------------------------------------------ Everyone hates slow websites. So do we. Make your web apps faster with AppDynamics Download AppDynamics Lite for free today: http://p.sf.net/sfu/appdyn_d2d_jan _______________________________________________ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers