--- In DynoMotion@yahoogroups.com, Tom Kerekes <tk@...> wrote:
>
> Hi Albert,
>  
> Yes you are the Master at isolating complex subtle bugs.
>  
> Sorry for the delay but this took some thought.
>  
> The problem was triggered by these GCode lines:
>  
> N22 G01 X-12.87638 Y-2.86441 Z-0.84131
> N23 G01 X-12.88663 Y-2.8691 Z-0.84127
> N24 G01 X-12.87638 Y-2.86441 Z-0.84131
>  
> These are two microscopic line segments at the very end of a longer path forming an exact direction reversal.  The two segments are only 0.01mm in length which is less than the Collinear Tolerance.  When combined together they form a total length of exactly zero.  This zero length vector causes problems when computing the slow down based on direction changes.  Another condition for this to be a problem is that previous path and Collinear Tolerance has to be just so that these segments are not combined with previous substantial length segments.
>  
> I think there existed another possibility where multiple microscopic segments might combine together to form a non-zero but infinitesimal segment that might still cause numerical problems.  So the following change was made:
>  
> Whenever the result of combined collinear segments results in a segment that is less than 1/2 the Collinear Tolerance in length it is discarded.  The end of the previous segment (if there is one) is adjusted to make up for the difference.
>  
> There should be a pre-release version 4.29a very soon (hopefully today).  Please verify it resolves the problem.
>  
> I'm greatly indebted to you for isolating these problem cases so well.
>  
> Regards
> TK 
>
>  
> From: albertplatek <albertplatek86@...>
> To: DynoMotion@yahoogroups.com
> Sent: Sunday, April 8, 2012 1:30 AM
> Subject: [DynoMotion] Re: Trajectory Planner Bug 3
>
>
>  
>
> Hi Tom,
> Can you duplicate this error?
>
> Best regards
> Albert Platek
>
> --- In DynoMotion@yahoogroups.com, Albert P³atek <albertplatek86@> wrote:
> >
> > Hi Tom,
> >
> > There is some problem with execution BUG3.ngc gcode with Collinear
> > Tolerance 0.0006 in
> > Smaller value or higher value of Collinear Tolerance fixes the problem.
> >
> > Acceleration of X axis jump up to 670 and motor aslo stall.
> >
> > I use:
> > KMotionCNC 4.29z
> > + "Fix5milRaduis/DSPKFLOP.out" (multiple motion segments are created less
> > than one servo sample time)
> > + "FixTP429/GCodeInterpreter.dll" (forgot to fully optimized the neighbor
> > segment)
> >
> > Settings like in screenshot + Tau Low Pass Smoothing 0.008
> >
> > 2 files in attachment
> >
> > Best Regards
> > Albert Platek
> >
>