Announcement

Collapse
No announcement yet.

Dangit, why did it do that! (Mach3)

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • Dangit, why did it do that! (Mach3)

    I spent all morning aligning the pencil grinder, creeping up on just barely touching the ER32 spindle taper with the grinder at reduced speed so I couldn't hurt any thing. I was running a MDI command to make the grinder run the 8 degree path, moving the X-axis a few tenths at a time and re-zeroing it until the stone was barely touching at the end of travel.

    I then wrote this simple code and loaded it.
    G01X-.174Z-1.2378F1
    G01X-.018
    G00Z.028
    G01Z0
    G00X.028
    G01X.0002


    The idea was to run the 8 degree angle, move the X slightly inward, move Z back past zero by .028" then move it back to zero to take up any slack, do the same with X but stop at .0002" short of zero when I would manually zero it and start over, giving a .0002" increase in cut.

    It worked fine initially but somehow re-zeroed the X with it out of position a bunch (.100" or so) in the negative direction. I'm no code wizard for sure; is there some limit or setting that made Mach reset the X axis on its own. My own ignorance I'm sure.
    Milton

    "Accuracy is the sum total of your compensating mistakes."

    "The thing I hate about an argument is that it always interrupts a discussion." G. K. Chesterton

  • #2
    You need to pay attention to the modal commands. In this case G90 and G91. G90 sets the interpreter to consider all movements as absolute. This means that zero is wherever you reset it to be and all movement commands are referenced as absolute locations relative to that zero point. G91 is incremental mode. In that case positions are referenced to the last position. If you give two positive x axis moves in a row they are added instead of referenced to the zero location.
    Free software for calculating bolt circles and similar: Click Here

    Comment


    • #3
      That code seems a little odd to me. Looks like the first line moves along your contour, then pulls away in X, moves back out of the bore (OK so far), then the next three lines move the tool back and forth for no easily apparent reason. I guess you're moving it to X.0002 Z0, then intending to re-zero Mach 3 while it's in that position, which would shift your .0002 increase in X as you explained.

      One thing I can think of is that Mach is a little klunky when dealing with fixture offsets. It does what it's supposed to do, but it's not exactly intuitive at first. Sometimes it will add a previous entry to a later entry rather than simply resetting the value to 0 like you want it to do. Spend a little time on the offset page just resetting things and make sure it reads what you intended it to read. There may be something there glitching things up for you. Just experiment there and get it so maybe you can get all the displayed coordinates to read 0. Even if you don't figure out completely how Mach is functioning on the offset page, if you can at least work out a series of keystrokes to zero out all the displays there then you will have eliminated a possible source of glitches.

      Having said that, if I remember correctly, zeroing the X and Z values on the "Program Run" page when the machine is sitting at what you want X.0002 Z0 to be should accomplish a .0002 X increase, just like you want. Hitting the "Ref All Home" button isn't what you want to click, so maybe that might be causing your problem?

      Not sure if this is coming out jibberish or not and I'm not sure I would understand me if I were you reading this, but basically just fiddle around with the program for a bit and make sure your keystrokes are actually doing what you expect. One thing I find helpful is to exaggerate things in experimentation so nothing is too small to see, i.e. try writing a similar program but make all movements to go no less than .1" and speed feed rates up so it doesn't take forever, then let it "cut air" so you can check things out. If something unexpected or unpredicted happens, it should be a little easier to at least figure out exactly when/where it's happening.

      Comment


      • #4
        try writing a similar program but make all movements to go no less than .1"...
        No need to do that. Just multiply the axes by a sufficient amount in the scaling factor to the right of each axis DRO.
        Free software for calculating bolt circles and similar: Click Here

        Comment


        • #5
          Thanks Evan. So I should have put a G90 in at the beginning of the code? I realize I have a lot of knowledge to assimilate.

          Tyrone the seemingly no reason moves were simply to move the axes past zero a bit then back to zero to cancel the little bit of backlsh that's in the machine.

          I've since reset everything and went back to MDI commands. I guess that eliminates whatever made it change the X zero as I've just about got the taper straightened up now without any wierdness..

          Only problem now is that I ran all the passes with a bit of a reduced grinder speed and on the finishing pass I bypassed the regulator and plugged it directly into the airline for max speed and about halfway through the shaft of the stone loosened in the grinder collet, the stone wobbled and wore a bit lopsided. I've now redressed the stone and have to creep back up on a finishing cut at a slower speed to be done with it.

          Other than that, it's worked pretty close to plan.
          Milton

          "Accuracy is the sum total of your compensating mistakes."

          "The thing I hate about an argument is that it always interrupts a discussion." G. K. Chesterton

          Comment


          • #6
            Are you sure your machine is not losing steps? On one of the mach3 screens, it shows the absolute machine axis positions. Also, there was a bug in earlier version of mach where if you used a usb device to increment the steps, it would somehow jump back and forth between g90 and g91.

            Comment


            • #7
              I'm pretty sure it's not loosing steps b/b. I'm now doing it with manual inputs and it's working great although a bit tedious. I slowed the lathe spindle down to 50 rpm, (thank goodness for the KB controller!) reduced the pencil grinder speed to (I'm guessing) around 1/2 speed and am taking only .0002" per pass with every 3rd pass a re-run. The grinder & stone I'm using doesn't like any abuse AT ALL or it goes all whippy and quickly wears out-of-round.

              I've got it all flowing pretty smooth now and am half way there. The finish is great but the runout is only down to .001". Better but not where I want it. I'm learning that these kinds of jobs with mickey mouse tooling take a lot of time & patience. If I had a proper T/P grinder I'd be done & sipping a cool one by now.

              The pencil grinder has been howling for a long time now and still feels tight but that skinny little 1/8" shaft has got to be flexing under load. I'm just hoping that I can creep up on a half thou or better. Originally I figured it'd be a piece-o-cake to get it perfect but now I'm not so sure.

              I think the little machine's toolpath is accurate & repeatable but the grinding tool is lacking.
              Milton

              "Accuracy is the sum total of your compensating mistakes."

              "The thing I hate about an argument is that it always interrupts a discussion." G. K. Chesterton

              Comment


              • #8
                If the machine is losing steps it will be on the direction reversals. To test this write a long series of back and forth moves for each axis, one axis at a time and just a few steps each way. Set up an object for it to touch at one end of the move and see if it either moves the object or develops a gap when it should be touching.

                If it is losing steps then you need to adjust the parameters in the motor output pins that controls whether it steps on the rising or the falling step pulse. I can't tell you which because I know nothing about the driver you are using but there are only 4 combinations of the step and dir pin settings for each axis. Trial and error will show you what works. Do it one axis at a time.
                Free software for calculating bolt circles and similar: Click Here

                Comment


                • #9
                  To add to Evan's advice, u can also look into the acceleration settings and use g64 in your g code (constant velocity smooths out the motions)

                  Comment


                  • #10
                    Just as an aside, if writing G code manually, many G codes are modal, you can save a bit of time of only writing G00 or G01 once, until a change is required.
                    Instead of G01 etc in front of every line it applies to in sequence.
                    Max.

                    Comment

                    Working...
                    X