Another approach to assembly solver (A2plus)

[Kunda1]

@kbwbe perhaps you may be interested to use the Github releases/tag features so it's easier to follow the evolution of A2plus + create appimages for it for testing (just like realthunder does for asm3 ?

[Turro75]

Hi Manuel,
your assemblies are always useful to find strange problems. I was hunting a problem with axial constraints within your motor, but found a problem within your piston part.

Please find 3 files attached:
- modified piston
- stable assembly
- instable assembly

Open the assemblies and hit the solve button. you see what happens.

Stable assembly:
I modified the piston "Kolben" and inserted an axis. This axis is used for an axialConstraint

Instable assembly:
Instead of the newly inserted axis, the surface of the drilling hole in piston is used for same axialConstraint.
The problem i found is that the center of surface is 5 meters !!! away from piston part. Is there anythin special with that drilling hole ? Is it a cone with very small angle ?

Please test the files...

instable still having a lot of errors even with last getMovement.

EDIT:

during solving a2+man I got exception on vec3.normalize()

Code: Select all

#project this vector to the own axis
        #This marks the beginning of the moveVecotr
        dot = vec2.dot(ownAxis)
        vec3 = Base.Vector(ownAxis)
        vec3.normalize()  -->EXCEPTION DUE TO NULL VECTOR
        vec3.multiply(dot)
        #Sub both vectors to geht the movevector shortest way between both axes
        moveVector = vec2.sub(vec3)

[kbwbe]

instable still having a lot of errors even with last getMovement.
during solving a2+man I got exception on vec3.normalize()

Code: Select all

#project this vector to the own axis
        #This marks the beginning of the moveVecotr
        dot = vec2.dot(ownAxis)
        vec3 = Base.Vector(ownAxis)
        vec3.normalize()  -->EXCEPTION DUE TO NULL VECTOR
        vec3.multiply(dot)
        #Sub both vectors to geht the movevector shortest way between both axes
        moveVector = vec2.sub(vec3)

Hi Turro, thank you for feedback. I will have a look.

[kbwbe]

@kbwbe perhaps you may be interested to use the Github releases/tag features so it's easier to follow the evolution of A2plus + create appimages for it for testing (just like realthunder does for asm3 ?

Hi Kunda1,
with next master-update, there will be release tags.

Edit: Appimages are making no sense for this python WB. A simple git pull updates the WB to new versions.

[Kunda1]

Hi Kunda1,
with next master-update, there will be release tags.

Edit: Appimages are making no sense for this python WB. A simple git pull updates the WB to new versions.

Awesome.
Re:appimage
Great that it's pure python so testing is easier. Says the word and ill add it to the add-on manager under alpha status for others to test as well.

[manuelkrause]

Hi Kunda1,
with next master-update, there will be release tags.

Edit: Appimages are making no sense for this python WB. A simple git pull updates the WB to new versions.

Awesome.
Re:appimage
Great that it's pure python so testing is easier. Says the word and ill add it to the add-on manager under alpha status for others to test as well.

Simply: nice!
Given the simplicity to add, the convenience to work with and the proved results, it's a worthy alternative to asm3, that many FC users would like to have! And so far we're only talking about ~3 months of development, on top of hamish2014's work of course.

I again want to express my thankfulness @kbwbe for having taken this job!

[kbwbe]

@Turro75, @manuelkrause, @wandererfan

Hi Guy's,

i got axialConstraint stable now. It solves Manuel's problematic motor assembly now without problems.
I switched back to old calculation of axial.getMovement(), but adjusted the refPoints during loadSystem().
Now the refPoint of axial is set near to the BoundBox.Center of the referenced surface.

@Turro75: there is a new function inside axial dependency:
.

Code: Select all

    def adjustRefPoints(self,obj,sub,refPoint,axis):
        if sub.startswith("Edge"): return refPoint
        face = getObjectFaceFromName(obj,sub)
        bbCenter = face.BoundBox.Center
        if bbCenter.distanceToLine(refPoint,axis) < 1.0e-12:
            return bbCenter
        v1 = bbCenter.sub(refPoint)
        v2 = Base.Vector(axis)
        v2.normalize()
        dot = v1.dot(v2)
        v2.multiply(dot)
        refPoint = refPoint.add(v2)
        return refPoint

.
which has to be called here:
.

Code: Select all

	.... Dependency.Create()....

        elif c.Type == "axial":
            dep1 = DependencyAxial(c, "pointAxis")
            dep2 = DependencyAxial(c, "pointAxis")

            ob1 = doc.getObject(c.Object1)
            ob2 = doc.getObject(c.Object2)
            dep1.refPoint = getPos(ob1,c.SubElement1)
            dep2.refPoint = getPos(ob2,c.SubElement2)
            axis1 = getAxis(ob1, c.SubElement1)
            axis2 = getAxis(ob2, c.SubElement2)
            if dep2.direction == "opposed":
                axis2.multiply(-1.0)
                
            dep1.refPoint = dep1.adjustRefPoints(ob1,c.SubElement1,dep1.refPoint,axis1)    
            dep2.refPoint = dep2.adjustRefPoints(ob2,c.SubElement2,dep2.refPoint,axis2)    
                
            dep1.refAxisEnd = dep1.refPoint.add(axis1)
            dep2.refAxisEnd = dep2.refPoint.add(axis2)

.
Hi all,
because of this important bugfix, v0.0.2 has been released.

[Turro75]

@Turro75, @manuelkrause, @wandererfan

Hi Guy's,

i got axialConstraint stable now. It solves Manuel's problematic motor assembly now without problems.
I switched back to old calculation of axial.getMovement(), but adjusted the refPoints during loadSystem().
Now the refPoint of axial is set near to the BoundBox.Center of the referenced surface.

@Turro75: there is a new function inside axial dependency:
.

Code: Select all

    def adjustRefPoints(self,obj,sub,refPoint,axis):
        if sub.startswith("Edge"): return refPoint
        face = getObjectFaceFromName(obj,sub)
        bbCenter = face.BoundBox.Center
        if bbCenter.distanceToLine(refPoint,axis) < 1.0e-12:
            return bbCenter
        v1 = bbCenter.sub(refPoint)
        v2 = Base.Vector(axis)
        v2.normalize()
        dot = v1.dot(v2)
        v2.multiply(dot)
        refPoint = refPoint.add(v2)
        return refPoint

.
which has to be called here:
.

Code: Select all

	.... Dependency.Create()....

        elif c.Type == "axial":
            dep1 = DependencyAxial(c, "pointAxis")
            dep2 = DependencyAxial(c, "pointAxis")

            ob1 = doc.getObject(c.Object1)
            ob2 = doc.getObject(c.Object2)
            dep1.refPoint = getPos(ob1,c.SubElement1)
            dep2.refPoint = getPos(ob2,c.SubElement2)
            axis1 = getAxis(ob1, c.SubElement1)
            axis2 = getAxis(ob2, c.SubElement2)
            if dep2.direction == "opposed":
                axis2.multiply(-1.0)
                
            dep1.refPoint = dep1.adjustRefPoints(ob1,c.SubElement1,dep1.refPoint,axis1)    
            dep2.refPoint = dep2.adjustRefPoints(ob2,c.SubElement2,dep2.refPoint,axis2)    
                
            dep1.refAxisEnd = dep1.refPoint.add(axis1)
            dep2.refAxisEnd = dep2.refPoint.add(axis2)

.
Hi all,
because of this important bugfix, v0.0.2 has been released.

Just made a quick test, now it works perfectly.
One more step to the first release

[kbwbe]

Says the word and ill add it to the add-on manager under alpha status for others to test as well.

Hi Kunda1,
have a little bit patience, we are very near to this point, but not completely.

[Kunda1]

Says the word and ill add it to the add-on manager under alpha status for others to test as well.

Hi Kunda1,
have a little bit patience, we are very near to this point, but not completely.