CNC Making Discussion

[scordatura]

I have access to a CNC machine. It uses VCarve as the software to import the .stl files. From there it generates the G code. I am thinking about going down the dark path of CNC. It would be cool if we could share info about the CNC approach to violin making. I have read posts by Mike Molnar and others. I realize that there is not an easy path with CNC. I was a full time programmer at one point so I am not afraid of working on programming. 

My first dilemma is how to generate a good .stl file. Are people creating a 3d scan of existing violin parts using a camera, using a probe to create the points, or drawing it in a CAD program? I have found a few files but would like to do my own. The outline of the top or back does not seem to be a big deal but getting the arching correct is the challenge to me.

Here are two files that i found.

Violin_XIII.Bottom.stl

Violin_XIII.Top.stl

[Nick Allen]

So these are not actually compatible with VCarve Pro. It doesn't recognize the file type even when trying to import the 3D model into the modeling tab. 

Is the file corrupted or correctly formatted?

[Nick Allen]

Update: Im just and idiot and couldnt figure it out. Got it working lol.

[scordatura]

No worries Nick. It happens! From what I recall do you work where you CNC?

[Nick Allen]

47 minutes ago, scordatura said:

No worries Nick. It happens! From what I recall do you work where you CNC?

I have several CNC machines at my work to use. I'm running a Shark with a water cooled spindle. 

About the model. It would probably benefit to change the geometry by squaring off the edges and removing the f holes. Also, retainer tabs need to be inserted. 

Another thing, I think that it could benefit from having both the top and back be two different files, one for the outside, and another for roughing out the inside. 

Other than that, it's something to work off of in Blender/Maya/3DS Max. 

[scordatura]

In a very quick look at the file, I'm not sure I am happy with the arching. I need to look closer at the files. If anyone is interested the source of the file is here:

http://community.foundry.com/discuss/topic/61185/chiralsym-s-3d-violin-released-under-a-creative-commons-license?mode=Post&postID=551026

When I looked at the scroll, I knew this is not necessarily the be all end all as the scroll is not accurate.

Yes retainer tabs. 

Sounds like we need to talk over this next time I am in Pittsburgh. Over a beer of course 

[Michael_Molnar]

I am traveling in Patagonia and will get back to this in a couple weeks.

[Don Noon]

17 hours ago, scordatura said:

The outline of the top or back does not seem to be a big deal but getting the arching correct is the challenge to me.

I will be getting into plate modelling in the next few weeks, and it worries me a bit at the moment.  I am using Fusion360, a very powerful program (and it's FREE!), which I have heard has the best CAM  there is for violinmaking.  I have been extremely impressed with it thus far, although I have also heard that the 3-D modelling for plate arches isn't quite so good.  I'll let you know when I get to that point.

[Fiddlemaker5224]

Hey guy's interesting thread. I have been using the Carve wright to arch the tops and backs of some of the instruments I make. I just upgraded to the Shark HD4, I was able to scan the top with the probe, then smooth it with the software. One thing I noticed with the file that was posted, is it is missing the cove. This changes flexibility less action means less propagation of the sound waves. Many of the instruments do not have a back with the same arching as the top. As for the Carvewright machine, well I have upgraded. Enough said about that. I will be scanning a back plate in the next few days. The top scan did not come out well enough to use. The next top I make I will hold off cutting the profile until I do the scan.

One thing I learned is that the CNC machine is good for removing the bulk of the material but still needs the finishing touch by hand. Same goes with the interior of the plates. Each time I tried the interior roughing, I ended up doing extra work for the tuning of the plate. On these plates, the tone always seemed to be off. So I went back to the hand method of removing the interior of the plates.

This is the scan of the top stl. Feel free to take a look but I suggest the you do not use it until it can be modified with at least a 1/2 inch larger border.

Scanned with the digital duplicator on a Shark HD4 DIG_0004.stl

Scan was done with 1/32 step over and took 27hrs to complete.

I just do not have the software to remodel it the way I would like.

[scordatura]

6 hours ago, Fiddlemaker5224 said:

Scanned with the digital duplicator on a Shark HD4 DIG_0004.stl

Very cool. What violin was the source of the scan? Looks like a Del Gesu outline.

[scordatura]

6 hours ago, Fiddlemaker5224 said:

One thing I noticed with the file that was posted, is it is missing the cove.

Yes I agree that the file that I posted above is not a perfect file. I am going to look a little further to see if it is usable as a starting point.

 

6 hours ago, Fiddlemaker5224 said:

One thing I learned is that the CNC machine is good for removing the bulk of the material but still needs the finishing touch by hand. Same goes with the interior of the plates. Each time I tried the interior roughing, I ended up doing extra work for the tuning of the plate. On these plates, the tone always seemed to be off. So I went back to the hand method of removing the interior of the plates.

This is what I have read from others. My feeling is that this process will always need some scraping at minimum to finish. I want to use CNC as a way to remove say 98% of the wood. As far as the interior is concerned, if you cut the arching first it seems that you would have to support the top or back (particularly the top as it is softer) to prevent vibration or flexing when clearing the inside wood. You could create a negative of the arching and cut a support cradle with the CNC to remedy this. 

Call me a rebel, but I have no remorse in using a CNC machine to replace the rough wood removal. I know this goes against the tradition of hand crafted violin making. While carving wood can be fun, the time it takes is not how I want to spend my time. To me it is the design, finish and optimization of the individual wood and model that separates the Strads or the Sam Zs from the rest of the pack. There is also enough things like edge work and other details that will keep the work from being too perfect or sterile. I am also not trying to create a huge volume of instruments. Luckily I am doing this for fun and not relying on it for a living. Just my 2 cents. 

[Don Noon]

7 hours ago, Fiddlemaker5224 said:

One thing I learned is that the CNC machine is good for removing the bulk of the material but still needs the finishing touch by hand.

47 minutes ago, scordatura said:

This is what I have read from others. My feeling is that this process will always need some scraping at minimum to finish. I want to use CNC as a way to remove say 98% of the wood.

If you have a common 3-axis machine, there will always be two problems:  1) the exact center of the cutting tool, which can not cut, therefore ends up compressing the grain (or if you use a non-center-cutting tool to avoid that, then you can't exactly cut a concave shape), and 2) there will always be some areas of the plate where the cutter will be going at an extremely shallow angle against the grain, and give micro-tearout and/or compression damage.  The only way to avoid this is with a machine that can tilt the spindle axis as required.

47 minutes ago, scordatura said:

Call me a rebel, but I have no remorse in using a CNC machine to replace the rough wood removal. I know this goes against the tradition of hand crafted violin making. While carving wood can be fun, the time it takes is not how I want to spend my time. To me it is the design, finish and optimization of the individual wood and model that separates the Strads or the Sam Zs from the rest of the pack. There is also enough things like edge work and other details that will keep the work from being too perfect or sterile. I am also not trying to create a huge volume of instruments. Luckily I am doing this for fun and not relying on it for a living. Just my 2 cents. 

I have posted before that my motivation for CNC is primarily for self-preservation, to prevent further decline to wrists that have accumulated a lot of damage from multiple fractures and sprains, as well as carpal tunnel surgery, on both wrists.  The more work I can have a machine do, the longer I can keep going.

Roughing maple is a major pain, but there is also the problem (with spruce as well) of getting from the lumpy shape of the big gouges to the final arching.  For me, this is the most tedious and slow part, and frequently I have to make corrections or adjustments if I have gotten too aggressive in roughing.  I hope the CNC will get much closer to my intended arch, and leave less for me to do.

On thing not yet mentioned is probably the strongest feature of CNC:  making tooling and forms.  Since these are usually 2-D, things can go very fast, and with extreme precision... at least much better than I could ever do by hand.  I have been thinking of tools and assembly methods I'd like to use, but haven't had the ability to make things precisely enough.

I hope that CNC will help in producing more instruments, as my historic pace of 3/year has been slowing down, and not keeping pace with clients.

 

 

[Fiddlemaker5224]

4 hours ago, scordatura said:

Very cool. What violin was the source of the scan? Looks like a Del Gesu outline.

That was a Modified LaMassie shape. The upper bout was increased to allow more volume, this resulted in a better tone when the finished instrument was played. I also build 8 string Hardanger style instruments. This added width makes a difference in the projection. The Del Gesu instruments were slightly shorter. So I guess you can say this is a blend of the Strad and Del Gesu design.

[Fiddlemaker5224]

4 hours ago, scordatura said:

Yes I agree that the file that I posted above is not a perfect file. I am going to look a little further to see if it is usable as a starting point.

Yes it seems to be a good starting point, I used to just scrape a cove in the top and back. Arthritis has pretty much ended that practice. Now I use a tool I made based on Edwin J. Ward's idea from the The Strad Facsimile book. Works out well for me. I can scrape the Final edge height, add the purfling, then scrape the cove. That is why I shaped the plane blade to have a cutting edge at both ends one flat and the other is a radius of the cove.

Edited March 4, 2019 by Fiddlemaker5224
Added Photo

[scordatura]

I like the above tool. Always good to be able to use a tool in more than one way!

[scordatura]

1 hour ago, Fiddlemaker5224 said:

That was a Modified LaMassie shape. The upper bout was increased to allow more volume, this resulted in a better tone when the finished instrument was played. I also build 8 string Hardanger style instruments. This added width makes a difference in the projection. The Del Gesu instruments were slightly shorter. So I guess you can say this is a blend of the Strad and Del Gesu design.

I am working on a Kreisler del gesu copy. I have stretched the length to bring it to a more “standard” length. Luis Bellini told me that he did this with all of his Guarneri models.

[Nick Allen]

So I imported the file generated with the probe into VCarve Pro. The program can't generate any vectors off of it. I was thinking about using it as a sort of trial run demo to show customers at my work how the CNC works. Was the model generated with the VCarve software, or the CarveWright?

[Fiddlemaker5224]

9 minutes ago, Nick Allen said:

So I imported the file generated with the probe into VCarve Pro. The program can't generate any vectors off of it. I was thinking about using it as a sort of trial run demo to show customers at my work how the CNC works. Was the model generated with the VCarve software, or the CarveWright?

Vcarve desktop was used, That is the same problem I ran into. Do you need the vector dfx?

[Fiddlemaker5224]

The Carve wright machine is history due to lack of support from the manufacturer. I lost all my pattern files with it. Now it is just spare parts, I sunk enough money into it that I could have bought the Shark to begin with. Lesson learned.

 

[Marty Kasprzyk]

I make my plates flat so I don't have do any of this s...

 

[Nick Allen]

2 hours ago, Fiddlemaker5224 said:

Vcarve desktop was used, That is the same problem I ran into. Do you need the vector dfx?

Yeah probably. 

[Nick Allen]

On 3/2/2019 at 5:39 PM, scordatura said:

 

Sounds like we need to talk over this next time I am in Pittsburgh. Over a beer of course 

Let me know next time you're in town!

[Michael_Molnar]

On 3/4/2019 at 2:17 AM, Fiddlemaker5224 said:

One thing I learned is that the CNC machine is good for removing the bulk of the material but still needs the finishing touch by hand. Same goes with the interior of the plates. Each time I tried the interior roughing, I ended up doing extra work for the tuning of the plate. On these plates, the tone always seemed to be off. So I went back to the hand method of removing the interior of the plates.

I agree that the CNC is great for getting the plate "roughed out", running the purfling channel, shaping the plate, and cutting ff's. I do not use sandpaper for finishing spruce. In fact, I never touch spruce with sandpaper because it destroys the annular ring figure. My finishing scrapers are burless, just sharp.  It is very important to remove enough wood to take away the compression tracks sometimes left behind by the endmill center. For this, I rub pencil over the entire plate to make sure I get down to soft uncompressed spruce. Maple is easier, usually. I also wet the wood to raise the compression tracks for scraping.

The compression tracks from center tips are minimized with fishtail end mills. I am experimenting with bullnose end mills. Ballnose  endmills are unpredictable for compression tracks. 4-axes CNC's tilt ballnose end mills to eliminate the track from the tip.  In any case, use sharp carbide end mills.

I thought I showed in my workbench thread how I flip the plate to hog out the inside. When I return from traveling, I will check and post those photos if needed.

[HoGo]

I've done some vector modelling of F-5 mandolin in RHino but I don't think it would be too diffrent in Fusion. I had to learn on my own mistakes but the result was good. What I found is that the result is more in the technique used to create the model than in the tool (Rhino or Fusion).

First you must decide what is your input: mesh of points from scanner or set of crosssections or topographic map... (IMO best is to use set of crossections with simple topographic map).

Second you need to decide what precision you want which would rule out some tools (I won't name them as they have likely diferent names in various SW) (I personaly wanted to stay as close to input as possible and wanted the surfaces go right through the drawn lines)

I f the arch is symmetric you'd be better creating just one half and then mirror it but you need to keep the eye on tangency at centeline so you won't end with a "crest" or ridge along the centerline - create dummy surface for the transition to the other side that will be used for modelling the arch. You will probably need to create the surface in more than one step and be prepared to go back and adjust the input curves if they won't work out together to create nice flowing arch.

I believ egood set of crossections (upper, lower, center and long arch together with edge shapes should be good start, perhaps crosssections at points) and few number 8 shaped topo- lines would allow to create good model easily.

(On mandolin, the shape is not symmetric so I had to create the recurve area with blending into scroll and points separately and then create the center  part of arch to be tangent to the edge surfaces)

You can get results like this: (feel free to contact me if you want to ask more)

[scordatura]

On 3/10/2019 at 7:34 PM, HoGo said:

believ egood set of crossections (upper, lower, center and long arch together with edge shapes should be good start, perhaps crosssections at points) and few number 8 shaped topo- lines would allow to create good model easily.

 

This is what conceptually I want to do. For me with the Kreisler model that I am pursuing, I have cross sections and long arches but no figure 8 topographical lines. I am trying to find the time to get in there and work the CAM software.

One idea I have is to get my hands on the plaster cast of the top that Morel made that was at my last visit at the Library of Congress. I would live to make a positive mold of it. After that I would make arching corrections to create a symmetrical arch. That is if the cast did not incorporate arching correction. I saw it years ago and am a little foggy on the instrument that it was taken from. I think it was the Kreisler DG.

Another route is to use the CAT scan images that I know exist and bring those data points into the CAM software.