Drag Knife cutter and DXF2GCODE

[Timothy March]

Hello Christian
I have been using DXF2GCODE to create gcode for a drag knife cutter. The features you have with this program are great for this. To be able to change start points, travel direction and reorder cuts are key elements to using a drag knife. Was wondering if you would consider adding as a option, a drag knife tool with offset input info into your program. I am including a link to a download package that I have made available on other forums on just how I use your program and a few others programs too. To achieve the needed swivel action needed for such a tool.
http://ubuntuone.com/41QpyBkkTxyKiMY5ImKYWz
Keep up the great work.
TJ March

[Timothy March]

Just found a plugin for CamBam for drag knife, I know nothing about programming but thought this might be of some help with my request. Attached file is wrote in Python

Please take a look
TJ March

[JP]

I will implement it for you; consider it as a birthday present ;)
But you have to give me a while - not yet sure when I will do it.

But just give me a while

[Tim March]

JP
I don't know what to say
Thank You, Thank You, Thank You

Tim March

--
--
You received this message because you subscribed to the Google
Groups-group "dxf2gcode-users".
To post a message, send mail to dxf2gco...@googlegroups.com
To unsubscribe, send mail to dxf2gcode-use...@googlegroups.com
See http://groups.google.de/group/dxf2gcode-users?hl=en for more options
and the dxf2gcode project page at http://code.google.com/p/dxf2gcode/
---
You received this message because you are subscribed to the Google Groups "dxf2gcode-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to dxf2gcode-use...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Christian Kohlöffel]

Hello Tim,

happy birthday to your 55 aniversery. I'm very happy that JP also buyed
into this topic and finds the time to do it for you. Otherwise i would
dig also a bit more into this topic since it is very interesting in
terms of maths. You know i also started as an maschinist programming NC
maschines and now I'm more or less far away from that but it's a good
balance to my other hobbies to dig a bit into programming in combination
with CAM. Keep on learning new things, thats a very good approach and i
think dxf2gcode is living from users like you having good ideas and
supporting in realizing it ....

JP, Tim,
i also did already made up my mind how to integrate the feature in the
best way into the code and GUI. I think it's not required to have G40;
G41 and G42 options anymore for gcode using the drag knife option right?
Therefore we can just reuse all of that realted functions to implement
the drag knife action when it's selected. Tim what is you opionion, if
this option is selected (lets say in the config file) it will
automatically generate the offseted path without any addition click or
is there the need to enable or disable this action? However i would use
the existing variable with list of geos: shape.stmove. What's your opionio?

best regards
Christian

Ps: JP what is about the sourcefrog account for the download of
dxf2gcode source and exe files. Is there a possiblity to also get
account to that server to upload updated etc.?
https://groups.google.com/forum/?hl=en#!topic/dxf2gcode-users/cDtt5yO0WKU

[Tim March]

Hello Christian
You guys are fantastic! What you suggested sounds great G40, G41 and G42
would not be needed, being able to see paths for swivel action would be
a plus if that is what you mean. The CamBam.py has place for input for
final cut depth I think this should be handled by Z Final mill depth in
main interface. Values for Offset, RetractDepth, SwivelAngle will be
variables that will have to be input by user. After everything is done
and I know how to use it, I will be happy to write a document describing
proper use of new funtion.

Tim

[Timothy March]

Hello again JP and Christian
Was thinking on my drive from work that a couple of other variables will be freed up in the config.cfg file, Tool_Parameters, diameter and speed. Will not need a tool diameter and spindle will not be running. As far as start_radius I am not sure how this works in DXF2GCODE, but to let you know with my drag knife I am limited to a turn radius of the tickness of the material, up to 1/4 inch. I will be flying home Friday to Michigan, currently working in Kansas City looking forward to working on my CNC that I will be using drag knife with. My wife told me the last shipment I made for parts came in ;) Attached is a picture of my design.

Tim

On Saturday, March 8, 2014 3:11:58 PM UTC-5, Timothy March wrote:

[JP]

It's certainly an option to disable the g40 g41 g42 when using swivel knife, if you insist. I pretty much already have it implemented (there is really hardly any math needed for this option), but not in the GUI only when exporting to file. With the GUI it's too much of hassle, because of the set nearest starting point, and change of direction.
I will upload it in the weekend, I think, so feel free to put it in the GUI.

Also, I will look into adding you to the sourceforge Christian

[JP]

All right I have implemented the function, but it’s not yet finished (check the svn and uncomment line 862 in dxf2gcode.py).

If it needs to do a swivel action with an angle bigger than dragAngle then I set a flag (drag in ArcGeo) such that the Write_GCodefunction knows that it needs to retract to a sufficient “drag-depth”. But ATM I do not check this at Write_GCode function in Shape.py, because the source code would be a big mess (at least how I did it :P ). So this part still needs to be implemented.

 

ATM I have placed the function in dxf2gcode.py (updateshapewithswivelknife loc 885) such that you can see directly the result of the swivelknife option, but this code can be placed almost everywhere. If you now change the starting point of the shapes, then they are not updated, what it should if you want this option in the GUI. Because the starting point and the end point should be update such that your knife stays in the same direction each time.

Also keep in mind in the current situation you cannot mill with different infeed and final mill depth (because now with non-closed shapes it wants to cut backwards, what your knife can’t do without changing the end and starting point, or it needs to do an exact copy of the first run)

 

So still a few points that should be fixed/implemented. If someone wants to do that or has an nice idea please tell/go ahead

[Timothy March]

JP & Christian 

When I used the program in past after getting all directions and start points set, I would export without optimizing. 

Tim
Sent from my iPhone

--

[Christian Kohlöffel]

Hello Jp,

just had some time to test it. For me it works on several simple lines
etc. However i very often get the followin error message:

File "D:\MyWorkspace\DXF2GCODE\trunk\source\Core\ArcGeo.py", line 89,
in __init__
lo = sqrt(pow(r, 2) - pow(m, 2))
ValueError: math domain error

Looking in it in higher detail shows problems with the following arc:

ArcGeo
Pa : X ->171.793 Y ->155.000; s_ang: None
Pe : X ->173.207 Y ->155.000; e_ang: None
O : None; r: 0.707106781187
ext : None; length: None
Core.ArcGeo __init__ 87 : - r: 0.707106781187; m:
0.707106781187

this arc seems to have a too small radius in order to find a valid
center ....
Hope this helps a bit.

Best regards
Christian

[Christian Kohlöffel]

With another file i had a failur with too short line (0 length). I added
the following to Point.py correct it:

def unit_vector(self, Pto=None):
"""Returns vector of length 1"""
diffVec = Pto - self
l = diffVec.distance()
if l==0.0:
return Point(1,1)
else:
return Point(diffVec.x / l, diffVec.y / l)


Can you confirm this failures with more complex shapes? I used test.dxf
from svn.

regards
Christian

[JP]

Well yes and no. I would say this is not the fault of my program but the fault of the dxf file. Many of those dxf files on the svn are terrible created. If I reproduce those shapes with autocad I got no error at all. So I can really create al kind of difficult shapes without a single error (if you need some example files, let me know).
The point.py "fix" you shouldn't use. That is not a fix but a try to circumvent the true problem. How the hack can you have a geo with length zero... That should not be possible, i.e. it means the dxf file is wrong.

The only true error you get is when you have points in the dxf file. but yeah.. how should deal with points if you use a knife. But again this does not give traceback error, but it will just give a path that is not correct.

So honestly, I cannot reproduce the errors at all (if I have created the dxf file decently, i.e. use autocad).

[Christian Kohlöffel]

Hello JP,

i understand what you mean, so you prefer not to catch such failure and
trust on the user making correct dxf files, right? Thats also OK for me.
Someday i may try to catch such failures during import in order to solve
such issues too.

regards
Christian

[JP]

Hiya,
Yup that's correct, but if you really want to catch all those errors, then when importing is certainly the place for it.

BTW I did found an error with 2pi>abs(angles)>pi, but I have fixed that.

[Timothy March]

Hello JP & Christian 

I know you guys busted butt on this, this weekend. You two are very talented. Not trying to push but what happens next in the process.

Tim
Sent from my iPhone

[Christian Kohlöffel]

Hi Tim,

to be honest, i did nothing just was testing a bit. Everthing was done
by JP, so thx. a lot for that. As a next step i may integrate the code
from JP in the way i explained it (If JP is OK with that), but first of
all i think next step will be that you test is :-) ....

regards
Christian

[JP]

I think that would be great if you could do that Christian. Do remember that every time you change the direction or the start point of the shape you have to update the complete shape (which could be quite time consuming if you have to deal with many shapes – e.g. layer select etc).

And that the mill depth has to be changed every time the flag, drag (in ArcGeo), is set. This is needed because when this flag is set the knife makes a strong turn, which is not possible when it is too deep in the material; but it still has to “hit/cut” the material, because it has to drag, otherwise the knife will not change direction.

Thanks for finishing it

[Tim March]

JP and Christian
Great work JP, I would be honored to test it when you are ready.
Tim

--

[Christian Kohlöffel]

That may be the task for next weekend if wheater is not that good.

regards
Christian

[Timothy March]

Sounds good to me I have a file ready.

Sent from my iPhone

On Mar 19, 2014, at 2:06 PM, "Christian Kohlöffel"

[Timothy March]

Hello JP & Christian
Just wanted to make sure you understand that I will be testing DXF2GCODE with the control software that I will be using to run my CNC
in simulation mode. My CNC is not finished yet. I am in Kansas City working at this time and only go home every other weekend. Last weekend when I was home I was able to get a good start on putting CNC together. Next step is to build vacuum table base.

Tim

[Christian Kohlöffel]

Hello Tim, hello JP

i think it's fair enought to use the simulation for now. I did put the Gui together (see attached figure of my last export). The basic functions are implemented. I would suggest you try it and report. With the more advanced functionallity i prefer to wait until your maschine is working. In order to test it you need to have the python environment installed and get the most updated version from the trunk. For more details how to do that see links below. (Dont hesitate to ask if there are any problems with installation)

https://code.google.com/p/dxf2gcode/wiki/DevelopmentEnvironment
https://code.google.com/p/dxf2gcode/wiki/DevelopmentEnvironment_Win7
https://code.google.com/p/dxf2gcode/source/checkout

Please find some parts of the config options i used below (in blue :-)). You need to change maschine type to drag_knife. the other options are partly not implemented (e.g. dragDepth).

regards
Christian

Ps: Please be kind with my englisch, i'm not a native speaker as you are ...

[General]
    write_to_stdout = False
    live_update_export_route = False
    default_SplitEdges = False
    default_AutomaticCutterCompensation = False
    maschine_type = drag_knife

[Drag_Knife_Options]
    # dragDepth is not used by now as well as dragAngle. May be done later.
    dragAngle = 20.0
    dragDepth = 0.0

[Tim March]

Greetings Christian & JP
I need a little help. I have installed all requirements for python environment no problems. In your previous message you referred me to link https://code.google.com/p/dxf2gcode/source/checkout
 On that page it says:

# Non-members may check out a read-only working copy anonymously over HTTP.
svn checkout http://dxf2gcode.googlecode.com/svn/trunk/ dxf2gcode-read-only

I have installed TortoiseSVN when I try SVN Checkout the Non-member URL http://dxf2gcode.googlecode.com/svn/trunk/ dxf2gcode-read-only
dose not work.

Regards
Tim

[Timothy March]

Christian & JP

It might be the firewall here where I am working. Will try again when I get out of work. 

Tim

Sent from my iPhone

<drag_knife1.jpg>

[Tim March]

Christian & JP
Seen what I did wrong after viewing my post on google groups pasted to much in,(dxf2gcode-read-only) got it downloaded.

Tim
Feel like a dumb me

[Timothy March]

JP & Christian

The GUI looked great and work well for selecting start points and changing directions, I thought screen updated very well. There are a couple of problems  though, the first DXF file I tried with DXF2GCODE was the shape of a star. The GUI showed where I had selected the start point but in the g-code start point started after the end of the last arc which was at the end of path. This closed shape should not have had a swivel action at start point or at the end of path, which is the start point. Also there needs to be a place for inputting of drag knife offset,( distance from center line of spindle to tip of blade), which would determine the radius of arc for swivel action. The drag knife offset would be different if a different size blade was used or from one drag knife brand to another. The path for swivel action looked very good, just that arcs were very large, JP I think you nailed the math. Need to add bladeOffset, dragAngle and dragDepth. Second file was of some scallops results looked the same as first file with same problems. The third file is of a model airplane fuselage more complex with different layers. I have attached archive containing nc and DXF file for all tested. In addition nc files that I created using the old method to get swivel action so you could compare paths, take a look at them on a simulator.

Regards

[Christian Kohlöffel]

HI Tim,

Just checked the issue with the different starting points. I can not
confirm that. Just attached 2 parts of the exported files for same dxf.
So the starting point is there like i selected it.

First export with selected starting point at ~ X-10 and Y-15
(* SHAPE Nr: 1 *)
N70 G0 X -9.000 Y -15.000

Second export with selected starting point at ~X50 and Y10
(* SHAPE Nr: 1 *)
N70 G0 X 51.000 Y 10.000

Wheater there is a swivel action or not, this is part of the algorithm
provided by JP. I'm not sure that i did fully understand it up to now
(didn't look into detail). Maybe JP can support here. This function is
implemented now in the scipt source/Gui/StMove.py function
make_swivelknife_move.

The drag knive offset is defined by the already existing parameter of
the tool diameter. The offset is defined by diamter/2, which is the radius.

hope this helps.

Best regards
Christian

Attached please find the function:

def make_swivelknife_move(self):
"""
Set these variables for your tool and material
@param offset: knife tip distance from tool centerline. The
radius of the
tool is used for this.
"""



offset =self.shape.LayerContent.tool_diameter/2
dragAngle = self.shape.dragAngle

startnorm = offset*Point(1,0,0)
prvend, prvnorm = Point(0,0),Point(0,0)
first = 1


#start = self.startp

#Use The same parent as for the shape
self.parent=self.shape.parent

for geo in self.shape.geos:
if geo.type == 'LineGeo':
geo_b = deepcopy(geo)
if first:
first = 0
prvend = geo_b.Pa + startnorm
prvnorm = startnorm
norm = offset*geo_b.Pa.unit_vector(geo_b.Pe)
geo_b.Pa += norm
geo_b.Pe += norm
if not prvnorm == norm:
swivel = ArcGeo(Pa=prvend, Pe=geo_b.Pa, r=offset,
direction=prvnorm.cross_product(norm).z)
swivel.drag = dragAngle < abs(swivel.ext)
self.geos.append(swivel)
self.geos.append(geo_b)

prvend = geo_b.Pe
prvnorm = norm
elif geo.type == 'ArcGeo':
geo_b = deepcopy(geo)
if first:
first = 0
prvend = geo_b.Pa + startnorm
prvnorm = startnorm
if geo_b.ext > 0.0:
norma = offset*Point(cos(geo_b.s_ang+pi/2),
sin(geo_b.s_ang+pi/2))
norme = Point(cos(geo_b.e_ang+pi/2),
sin(geo_b.e_ang+pi/2))
else:
norma = offset*Point(cos(geo_b.s_ang-pi/2),
sin(geo_b.s_ang-pi/2))
norme = Point(cos(geo_b.e_ang-pi/2),
sin(geo_b.e_ang-pi/2))
geo_b.Pa += norma
if norme.x > 0:
geo_b.Pe =
Point(geo_b.Pe.x+offset/(sqrt(1+(norme.y/norme.x)**2)),
geo_b.Pe.y+(offset*norme.y/norme.x)/(sqrt(1+(norme.y/norme.x)**2)))
elif norme.x ==0:
geo_b.Pe = Point(geo_b.Pe.x,
geo_b.Pe.y)
else:
geo_b.Pe =
Point(geo_b.Pe.x-offset/(sqrt(1+(norme.y/norme.x)**2)),
geo_b.Pe.y-(offset*norme.y/norme.x)/(sqrt(1+(norme.y/norme.x)**2)))
if not prvnorm == norma:
swivel = ArcGeo(Pa=prvend, Pe=geo_b.Pa, r=offset,
direction=prvnorm.cross_product(norma).z)
swivel.drag = dragAngle < abs(swivel.ext)
self.geos.append(swivel)
prvend = geo_b.Pe
prvnorm = offset*norme
if -pi<geo_b.ext<pi:
self.geos.append(ArcGeo(Pa=geo_b.Pa, Pe=geo_b.Pe,
r=sqrt(geo_b.r**2+offset**2), direction=geo_b.ext))
else:
geo_b = ArcGeo(Pa=geo_b.Pa, Pe=geo_b.Pe,
r=sqrt(geo_b.r**2+offset**2), direction=-geo_b.ext)
geo_b.ext = -geo_b.ext
self.geos.append(geo_b)

else:
self.geos.append(copy(geo))
if not prvnorm == startnorm:
self.geos.append(ArcGeo(Pa=prvend,
Pe=prvend-prvnorm+startnorm, r=offset,
direction=prvnorm.cross_product(startnorm).z))

self.geos.insert(0,self.geos[0].Pa)

[JP]

Hi, again :|
Well it precisely reacts how I wanted it - I have compared all your provided files. With the star: it starts at the correct point. The testOld actually starts too early - remember your knife "lags". And it should end with a swivel action! Because otherwise your knife does not end in the correct way... i.e. if I just let it end as in the testOld it will not be in the correct position to directly do another shape - now in this case you only have one shape but maybe I want to process right after it another gcode file... (as far as I could see cambam does it the exact same way, as I do it)
I think you should really test it on a working machine, and I'm pretty sure you will come to the same conclusions - with all your files.

Offset is already implemented check Christian's comment. (dragAngle, and dragDepth are partly implemented - still needs to be finished - now you have to add that little part of code manually)

So honestly I'm not gonna make a change... sorry. But I truly believe it's working correctly!!

[Tim March]

JP
After further testing  I agree with you I was always trying to figure out cut order from the drawing end. The fuselage drawing I provided in last email was not drawn correctly. I fixed mistakes that were in it and DXF2GCODE worked perfectly. Now I also know about tool setup in config file after Christian posted. Just as a note the manufacture of drag knife suggested always start cuts in a plus x attitude with blade. Maybe that is the way they have to do it with the cam software they use and sell. The drawing of the fuselage has a couple of lead in and lead out cuts I thought were needed with the old way I did swivel action to get blade ready for next cut. That effort is not needed any more with the way you set this up. You are a genius just put the blade in the attuide of first cut and let DXF2GCODE do all the work. Attached DXF file I used in my last test, nc file is results. Swivel action moves are just barely visable but with my drag knife .140" offset is all that is needed.

Thank-you for the birthday persent
Tim

[JP]

;)

The drag knife is now fully implemented and should work perfectly - remember to use it set maschine_type to drag_knife.

I have updated the gcode write function to support the drag depth change, also made small change with the feed rate (for all write functions) – now it no longer duplicates the previous rate, in the exported file

"Z infeed depth"/axis3_slice_depth is changed to "Z Drag depth"/axis3_slice_depth if the maschine_type is set to drag_knife.

On Tuesday, March 25, 2014 2:51:58 AM UTC+1, Timothy March wrote:

JP
After further testing  I agree with you I was always trying to figure out cut order from the drawing end. The fuselage drawing I provided in last email was not drawn correctly. I fixed mistakes that were in it and DXF2GCODE worked perfectly. Now I also know about tool setup in config file after Christian posted. Just as a note the manufacture of drag knife suggested always start cuts in a plus x attitude with blade. Maybe that is the way they have to do it with the cam software they use and sell. The drawing of the fuselage has a couple of lead in and lead out cuts I thought were needed with the old way I did swivel action to get blade ready for next cut. That effort is not needed any more with the way you set this up. You are a genius just put the blade in the attuide of first cut and let DXF2GCODE do all the work. Attached DXF file I used in my last test, nc file is results. Swivel action moves are just barely visable but with my drag knife .140" offset is all that is needed.

Thank-you for the birthday persent
Tim

On 3/24/2014 6:44 PM, JP wrote:

Hi, again :|
Well it precisely reacts how I wanted it - I have compared all your provided files. With the star: it starts at the correct point. The testOld actually starts too early - remember your knife "lags". And it should end with a swivel action! Because otherwise your knife does not end in the correct way... i.e. if I just let it end as in the testOld it will not be in the correct position to directly do another shape - now in this case you only have one shape but maybe I want to process right after it another gcode file... (as far as I could see cambam does it the exact same way, as I do it)
I think you should really test it on a working machine, and I'm pretty sure you will come to the same conclusions - with all your files.

Offset is already implemented check Christian's comment. (dragAngle, and dragDepth are partly implemented - still needs to be finished - now you have to add that little part of code manually)

So honestly I'm not gonna make a change... sorry. But I truly believe it's working correctly!!


On Monday, March 24, 2014 9:04:16 PM UTC+1, Christian Kohlöffel wrote:

HI Tim,

Just checked the issue with the different starting points. I can not
confirm that. Just attached 2 parts of the exported files for same dxf.
So the s

...

[Timothy March]

Thanks again JP

[Timothy March]

Hello JP
Have a question, in the Config file it has the following

[Drag_Knife_Options]
    # dragDepth is not used by now as well as dragAngle. May be done later.
    dragAngle = 20.0
    dragDepth = 0.0

Is this still to be done?
do swivel with dragAngle greater than value
swivel height at dragDepth value
Or am I missing something to get these values to work.

Tim

...

[Timothy March]

Hello again JP
As fare as the Drag_Knife_Options in Config file if not implemented yet. You have the angles all worked out very well dragAngle might not be needed. But the dragDepth value is very important because without it you could only cut very thin material. If you were to cut wood venier with blade at full depth during swivel action you would brake a blade. Or what I will be cutting most foamboard 4mm thick, you might not break a blade but you would damage the material.
Just something to think about.
Tim

[Timothy March]

JP 

I will be going home next weekend I will post some pictures of my drag knife to show how it freely pivots on spindle centerline. What I indent to show you is if tip is too deep into material during swivel move it will not pivot to desired angle. The harder the material the less the depth into material would have to be. I wish I had my machine put together and running so I could show you what I mean. The work you have done is very impressive.

Tim

Sent from my iPhone

--

[jan.iemhoff]

Hello dxf2gode world,

Yes , a picture (foto) could say more than words.

Rg.
Iemhoff

Vom Handy gesendet.

Timothy March <timoth...@gmail.com> schrieb:

[JP]

Well apparently you didn't get the idea of fully implemented :P


but yes that is implemented just import the file with the maschine_type set to drag_knife and you will see a label called drag depth. Just fill in a different depth and you will see that it goes to that depth when the angles are greater than the dragAngle. I forgot that Christian already had some variables for this declared in the config file. So I might make a config file change. But it is already fully implemented, and should work perfectly.

[Timothy March]

Thank You for your patience and doing all the work. I figured out what I was doing wrong I used a bad Sd: label in layer name was not getting any change in Z height.
Tim

[Bill Trondsen]

Hey guys, really nice program and the drag-knife should be a helpful addition.

I got Rev514 running and plugged in a simple test part to see how it works.

The cut path shapes look correct, and I checked the G-code - the start and end points are calculated correctly - nice work.

I tried using the layer name to set tool diameter "Td: .375" and the FeedXY/FeedZ values and it did not work for me - maybe I formatted the name wrong?

All of the values written to G-code come from the input boxes, which defaults are set by the config file.

I tried the 2 (diameter) tool and verified a 1 (radius) offset in the G-code, and using the 10 tool gave a 5 offset.

This seems pretty straightforward to use.

I do suggest that you might want to preset the velocity mode to G61 (Exact Stop) instead of G64 (Constant Velocity) - this should prevent the machine from rounding the corners.

This value is set in the postpro_config file.

Cheers - BillT

[Timothy March]

Thanks for checking this out Bill

Tim

Sent from my iPhone

--

[JP]

The reason why the layer name did not work is because you have not written it in capital letters (MILL:), checkout the http://code.google.com/p/dxf2gcode/wiki/LayerControl


Setting G64 to G61 might indeed be a nice change to apply - always,- although everybody could change it to whatever they want (postpro_config.cfg)


Nice pic BTW, probably will use this to write a Wiki.

[Bigman74066]

Hi Guy's,

I can't find the latest Windows build with the drag-knife feature in it.

Timothy's link to the package is not working anymore.

On sourceforge there is an older version that doesn't have the feature yet (?)

Could you please make it available for download again ?

Thanks !

Op zaterdag 12 april 2014 11:39:25 UTC+2 schreef JP:

[Timothy March]

This is not the newest build but it has drag knife option. The reason the link did not work is because Ubuntu One no longer is available where i made it available. This site has link to DXF2GCODE download that I have made available. Be sure to report back to user group here to report how DXF2GCODE is working for you.

http://www.openbuilds.com/resources/

Tim

--