Newbie - Difficulty getting smooth motion on simple contour pass

[rsnewmanRFVUW]

I'm brand new to cnc, working with a 20yr old Techno Davinci stepper router, and I'm having trouble running a simple elliptical contour path.  I get jittery motion, it feels like the machine is pausing at each line of code.  If I try a circle it runs smoothly.

I create the ellipse in Illustrator, import to Fusion as .svg, and create the body by extruding .25".  Then in CAM I do a single pass contour, tolerance .0001", smoothing .0004" as others have advised.  I get about 140 lines of code, alternating G1's and G2's.  

Is there more that I could be doing with the CAM settings?  The Techno stepper controller is widely reviled, I'd be willing to blame it, but I want to make sure this is not my error before I throw more time and $$ into the machine.  

[LibertyMachine]

It's not the machine, per se, but the controller. It sounds to me that you are data-starving it.

A circle will run smoothly because it's only a few lines of code. 

An elliptical shape doesn't really have a radius, as it's constantly changing. CAM software will output 1000's or small little linear moves to accommodate the shape.

My suggestion for you would be to: Set the Tolerance to .001" (or .0254mm) and turn on Smoothing, also to .001"/.0254mm. Both of these are found on the "Passes" tab of the operation.

[rsnewmanRFVUW]

Thanks Seth, I just tried that, still got the bumpy path.  Fewer lines of code tho,

When you mention data-starving, is that the inability of the controller to process the data fast enough, or is it the computer feeding it too slowly?  I'm using an IBM Thinkpad with WinXP, going out the serial port, which is what the Davinci requires.

My intention is to use this for cutting and inlaying pearl in banjos, those tiny endmills would never survive such a bumpy feed.

[Trigg3r]

Did you plot the circle the same way, in inventor or in fusion ? Does an ellipse drawn in fusion do the same ? Just trying to eliminate poor geometry as a cause. Data starving normally means greedy machines waiting for code from not being fed fast enough, although by the sounds of things you could be suffering the reverse, as the controller struggles to process the information fast enough. The effects can look similar iirc.

[rsnewmanRFVUW]

I get the exact same code and results with the ellipse drawn in Illustrator or Fusion.

If it is data starving, I don't think its the computer - I tried 2 different machines, the Thinkpad and a desktop also running XP.  Same results.

[LibertyMachine]

Help me understand this:

Fusion > Laptop > Motors

Is there controller software on the laptop, something to interface between the motors and the laptop? Or are you drip feeding the program on the laptop into another control box?

[rsnewmanRFVUW]

There is software on the laptop, the "Techno G-code Interface".  It takes the the .ncd file and translates it into  .run and .out files.  To run the program it uses the .run file.  The .out file is for downloading to the router's memory, which is tiny and useless.  But the machine can run a downloaded program without being connected to the computer., using a start button.

So it seems there is a controller in the machine, but it only accepts Techno's .run files.  There doesn't seem to be any way to access parameters for it.

[LibertyMachine]

Next question:

Do you know if there are G or M codes you can call that will activate a "smoothing" internal to the machine?

For instance, with my DMG Mori vertical, I can call a G05.1 Q1 and it turns on a buffer that smooths out the accel/decel over the next several hundred lines. The allows it tohandle large amounts of code without data-starving.

What are other forums saying? I assume there are Techno specific sub-forums, perhaps on CNC-Zone or Practical Machinist?

[HughesTooling]

Can you export and upload your f3d file here so we can experiment with it. I have seen this before on splines and part of the problem is corner rounding using the cutter radius. I guess you're using a 2d profile, try using 3D contour using the same tolerance and smoothing settings.

Mark

[LibertyMachine]

Thanks for chiming in Mark. I knew I recalled someone else experiencing similar issues but I couldn't find the thread where it was discussed

[rsnewmanRFVUW]

I'm not aware of any G code for an internal smoothing.  This is a pretty old and simple machine, I'm attaching a brochure describing it so you don't overestimate it's abilities.

I posted on the cnczone Techno forum, specifically asking about the controller.  Only got advice to look for G61 and replace it with G64.  Nothing about the machine itself.  There are threads about people replacing the Techno contollers with Geckos or Leadshines, but no explanation why

I'm attaching my code for a 1" x 2" ellipse, tolerance .0005", smoothing .001"

Thanks for the assistance

Richard

[HughesTooling · Accepted solution]

Here's an example that shows 3d contour give a smaller more efficient toolpath than 2d contour, 2.3kb vs 5.3kb

The 2d contour produces OK code for the ellipse but the offset is a spline and the code is like yours, made from lots of line and arc moves. The 3d contour gives all arcs.

Sample file attached.

Mark

[rsnewmanRFVUW]

Thanks Mark, but I can't figure out how to open that .f3d file, either in Fusion or on my Macbook.

How embarrassing!

Wait a minute, I think I figured it out

[HughesTooling]

---

@rsnewmanRFVUW wrote:

Thanks Mark, but I can't figure out how to open that .f3d file, either in Fusion or on my Macbook.

How embarrassing!

Wait a minute, I think I figured it out

---

The quickest way to open the file is with New Design From File on the File menu. The other way is upload to a project using Fusion's data panel.

[rsnewmanRFVUW]

Yes, I had uploaded it to the data panel, nice to know how to do that.

I post processed using the Davinci post, and got a very different looking code from the one you show, I'm attaching it.

I've been using this post   7437_rs274.cps ‏30 KB

I found it here - http://forums.autodesk.com/t5/hsm-post-processor-forum/techno-isel-davinci/m-p/6092215/highlight/tru...    see post #12

[HughesTooling]

There are 2 ops in the file, try right clicking the second op and selecting post from the right click menu and you'll see 3d contour gives a nice toolpath. The first op (2D Contour1) shows 2d contour doesn't work well with the offset ellipse.

Mark

[rsnewmanRFVUW]

Thank you Mark, now I see that.  I'm at home, will try running that 3d contour on the router when I get to the shop this morning.

Thank you for being so patient with me.  

Richard

[xander.luciano]

Hey!

I'd recommend looking through this thread for some more information on data starvation, smoothing, and tolerances.

http://forums.autodesk.com/t5/computer-aided-machining-cam/understanding-smoothing/td-p/6636189

Best,

-Xander Luciano

[rsnewmanRFVUW]

Thank you Xander, that is a great thread!  Always nice to see numbers to confirm theory.  

I had found the NextGenCAM page you referenced when I first learned about smoothing a few days ago.  I'm confused about the language in this paragraph - does a "higher tolerance" actually mean a lower deviation?   

Recommended Settings

Smoothing works best when the Tolerance (the accuracy with which the original linearized path is generated) is equal or greater than the Smoothing (line/arc fitting) tolerance. In other words, the more accurate the data sent to the Smoothing "filter" the better job it does –up to a point. A good rule is to set the cut Tolerance between 1 to 4 times greater than the Smoothing tolerance.

Also, having just been introduced to 3d contours by Mark, I'm wondering why it gives a smaller, more efficient toolpath than an otherwise identical 2d contour?  I tried it on a complex curve and the difference was impressive.

[xander.luciano]

No problem! Glad you found it informative!

And you are correct, "high tolerance" = low deviation = smaller tolerance number. Think of it like, "The aerospace industry has a lot of high tolerance parts." Just guess on this but it might be because .001 would be 1 thousandth, while .0001 would be 1 ten thousandth of an inch? Could be totally wrong on that though.

Now for your second question, this might be a little technical but we'll see.

So The reason 3D contour is working better is because it is receiving a different set of information. 2D contour will receive a long list of points for each contour you select. It will then create the toolpath off this list, and attempt to smooth it. 3D contour is only given the outer contour as a boundary condition, and being told to just "find a valid toolpath inside this curve. So all it is doing is avoiding the center island which is given to it as an approximated 3D model. Because it is "avoiding" the center island, it offsets from the first curve, then offsets again, until it collides with the center island. In this way the smaller pass will be smoothing in the 3D contour than the 2D, because the 2D is generating the smaller curve off the direct geometry, while the 3D contour is generating off the outer curve, and avoiding the inner island.

I hope that makes some sense haha, it's hard to put in words. Let me know if you have questions though,

Best,

Xander Luciano