Geometric inaccuracies.

Discussion in 'AutoCAD' started by SAM90, Aug 15, 2003.

  1. SAM90

    SAM90 Guest

    I have rotated the UCS by 20deg on the z-axis.

    When I draw a line [horiz. or vert.] then check the angle it's slighly out e.g. 20.00000066deg.

    Also when I drag a block in then check the angle in props it is 20.00000066deg.

    When I explode the block the lenghts are wrong. How can this be? - Kev
     
    SAM90, Aug 15, 2003
    #1
  2. SAM90

    R.K. McSwain Guest

    I also rotated the UCS 20° and drew a line, then listed it to the max:

    (angtos (angle (cdr (assoc 10 (entget (entlast))))(cdr (assoc 11 (entget (entlast))))) 0 16)
    returns "20.00000000000000"

    Is your SNAPANG set to ZERO. I bet it's off causing your ortho lines to be off also.
     
    R.K. McSwain, Aug 15, 2003
    #2
  3. Sorry, I reread, and it looks like the length must relate to that line,
    while I implied that there was not way to know what length you were talking
    about.

    As the kids say these days, "My bad."

    -R
     
    Randy Richardson, Aug 15, 2003
    #3
  4. SAM90

    SAM90 Guest

    Thanks for the replies, chaps. SNAPANG is zero RK. I think Randy's reference to the calcs and your reference to "state plane coordinates" could explain it. You see, the job is 355 miles fromm 0,0 [I think the architect works on the ordnance survey national grid].
    The problems caused are when trimming or extending and the errors are compounded by grip editting. - Kev
     
    SAM90, Aug 18, 2003
    #4
  5. SAM90

    OLD-CADaver Guest

    >The problems caused are when trimming or extending and the errors are compounded by grip editting. - Kev <<

    So, the problem is not geometric inaccuracy but improper editting techniques??

    We frequently work with USGS coord's, and haven't experienced that problem.
     
    OLD-CADaver, Aug 18, 2003
    #5
  6. SAM90

    SAM90 Guest

    Yes, grip editting may exacerbate the problem, but I think the problem is down to accuracy. How come I can insert a carefully made simple block, explode it and then when I check the length of a line it is incorrect?
     
    SAM90, Aug 18, 2003
    #6
  7. SAM90

    OLD-CADaver Guest

    I dunno, but it seems to isolated away from our systems. We frequently work at large coord's and do not experience that problem. If the block is 10" sq., it is still 10" sq. after insertion.

    Are the dimensions of the block accurate prior to exploding?
     
    OLD-CADaver, Aug 18, 2003
    #7
  8. SAM90

    SAM90 Guest

    I've worked on other jobs where 0,0 is miles from the job and hav'nt had the problem, maybe it's something the this job's set up. No, the dims of the block are not accurate prior to exploding, I just exploded to check. Could it be anything to do with limits? - Kev
     
    SAM90, Aug 18, 2003
    #8
  9. SAM90

    OLD-CADaver Guest

    I'd check the block and MAKE SURE it is accurate to the same level of decimal places the target drawing is using. Limits shouldn't cause this problem.
     
    OLD-CADaver, Aug 18, 2003
    #9
  10. SAM90

    SAM90 Guest

    Apologies, I should have checked this first but I took the blocks from the symbol library and assumed they were ok. However, it doesn't explain dragging blocks off Design Centre and their rotation is 20.00000066 deg for example (that's in the 20 deg. UCS, in world they go in at 0 deg. - Kev
     
    SAM90, Aug 18, 2003
    #10
  11. SAM90

    OLD-CADaver Guest

    The block insertion angle lists as 20.00000066 or the line from the block lists off?
     
    OLD-CADaver, Aug 18, 2003
    #11
  12. SAM90

    SAM90 Guest

    The block insertion angle [Rotation property].
     
    SAM90, Aug 19, 2003
    #12
  13. SAM90

    OLD-CADaver Guest

    My AUPREC is always at 8, LUPREC=8; LUNITS=4; AUNITS=0.

    I finally got an error to creep in to the angle measuremnt. Using Architectural units, at about 92,719 miles (489,555,852'-0" or 5,874,670,224") the angle measured 22.00000031. (92,719 miles is roughly 4 times the circumference of the earth)

    Using the same units
    at 900 miles (~dist across Texas) the angle is still 22.00000000

    at 2000 miles (~dist across USA) (or 10,560,000' or 126,720,000") it was still 22.00000000

    When working at those distances it is probably undesirable to inches as a base unit. The magic number seems to be 16 significant digits, more or less.
     
    OLD-CADaver, Aug 19, 2003
    #13
  14. SAM90

    SAM90 Guest

    In your test did you rotate the UCS about the z axis by 20 deg before inserting the blocks? In mine the error starts at about 5 miles from 0,0. I couldn't create an error in world UCS. My LUNITS is 2. Still experimenting...
     
    SAM90, Aug 19, 2003
    #14
  15. SAM90

    OLD-CADaver Guest

    Tried it both ways, same result.
     
    OLD-CADaver, Aug 19, 2003
    #15
  16. SAM90

    SAM90 Guest

    Thanks, that should be enough to satisfy most users [earth based, of course], but it seems to confirm I have a problem in my drawing. I couldn't download the Solar.dwg though. - Kev
     
    SAM90, Aug 20, 2003
    #16
  17. SAM90

    Allen Jessup Guest

    Just in case you want to see it I posted Solar.dwg in Customer files.



    "SAM90" <> wrote in message news:...

    Thanks, that should be enough to satisfy most users [earth based, of course], but it seems to confirm I have a problem in my drawing. I couldn't download the Solar.dwg though. - Kev
     
    Allen Jessup, Aug 22, 2003
    #17
  18. SAM90

    SAM90 Guest

    Thanks
     
    SAM90, Aug 22, 2003
    #18
  19. SAM90

    Scribble Guest

    haven't tried it lately but we used to get strange results when using
    Australian Map Grid coordinates.
    Our location is around x=6,500,00 y=30,000 (metres) and this caused strange
    results with trim, extend and creating tangents to circles.

    --
    Cheers
    Steve



    Architectural units, at about 92,719 miles (489,555,852'-0" or
    5,874,670,224") the angle measured 22.00000031. (92,719 miles is roughly 4
    times the circumference of the earth)
    base unit. The magic number seems to be 16 significant digits, more or less.
     
    Scribble, Aug 26, 2003
    #19
Ask a Question

Want to reply to this thread or ask your own question?

You'll need to choose a username for the site, which only take a couple of moments (here). After that, you can post your question and our members will help you out.