PDA

View Full Version : CNC help, software related.



Nolan
Mar-03-2004, 10:27am
I have access to a CNC machine and was thinking it would be cool to have a few tops and backs made up. I have been told that I need to create a cad file with the specs and then they should be able to plug it into the machine and make the tops/backs. We don't want to get a finished product out of the CNC just get the plates down to a point where only final graduations are needed.
I have never worked with a CAD program but I am a systems analysts and work with different programing langauges so I don't think the computer part of it should be a problem. My friend who builds mandolins is supplying the specs from a Loar in the area here.
Anyone have advice on what CAD software to use? Any advice on using a cnc in general?
Thanks!

Nolan
Mar-03-2004, 11:00am
Ok, shoulda done a search before I posted up... I found a really good thread on this.

douglas2cats
Mar-03-2004, 12:12pm
Don't have any personal experience using it, but I think AutoCad is a pretty widely used product in manufacturing industries. I believe a lot of other software makers make sure their files formats are compatible with AutoCad too, so you may be able to find something less expensive. Ask some questions of the normal operators of the CRC to find out what file formats they usually receive their data in, then make sure whatever software you use can output in that format. You're probably going to need some kind of holding fixture to orient your blank. Nothing fancy - just something with some reference points on it that can be used to zero the machine with respect to your blank and fix its position. I would definitely make some additional blanks of the same size out of some scrap wood and do some test runs to see if you need to tweek the CNC program. At least in the automotive sheet metal domain, it's not uncommon to scrap the first 2-5 pcs when tuning in CNC's/Laser cutters/etc. As much as you can, pick the brains of the people that normally use the machine you have access to.

c3hammer
Mar-03-2004, 12:38pm
Creating surface models of a mandolin top is actually very difficult.

The combination of eliptical outline, variable radius recurve and a the flatter neck side vs. tailpiece side, makes for one difficult job to model.

With all that said, I have a somewhat decent A5 top modeled already. This has nothing to do with gradations, simply the curvature of the outside shape.

If you are interested in seeing a file, just let me know.

Cheers,
Pete

GaryM
Mar-03-2004, 1:10pm
I was talking to a friend who was showing me his cnc machine. He said...somthing like.. "you make a model and and paint it with some sort of black finish. There is a way to have it scanned and then converted to G code for some sort of program" ...I started drifting and thinking of an old little house on the prarie episode.

Nolan
Mar-03-2004, 1:47pm
Gary, that was hilarious! Very interesting though. I'd like to learn more about that.

grsnovi
Mar-03-2004, 1:53pm
I design CAD software for a living. Starting from a completely empty file and modeling the carved top and/or back of a scrolled "F" would be "sporty" with any CAD system.

However, you can easily take an existing pre-carved piece and digitize it to end up with the G-codes needed to drive your CNC.

GVD
Mar-03-2004, 3:24pm
If you have access to a CMM it would be pretty easy to scan a top and save it to an IGES file. The IGES file could then be converted to the CNC program.

GVD

delsbrother
Mar-03-2004, 3:58pm
Gary and GVD are correct - easiest would be to digitize an existing plate.

AutoCAD, while an industry standard in 2D, is not particulary adept at making 3D surface models with organic curves (I teach it for a living, I should know). You could draw the outline/fholes, etc. in AutoCAD and then export the .DWG or .DXF file to a CAM package (like MasterCAM), but you're going to have to work on the surfaces again anyway. Might even be easier to just draft it in the CAM software.

You could get nice results in a good surface modeling 3D CAD package, but if you're just trying to get rough plates it may not be worth the learning curve.

pathazzard
Mar-07-2004, 1:13pm
With all that said, I have a somewhat decent A5 top modeled already. #This has nothing to do with gradations, simply the curvature of the outside shape.

If you are interested in seeing a file, just let me know.

Cheers,
Pete
I search these archives all the time and find something else I need to know every time. I too was wanting to digitize an F style top and back for close roughing my plates on my ShopBot CNC router. I just got RhinoCad 3D and also VisualMill and the learning curve will kill you. I even already had AutoCad V14 background. Anyway, I was goning to hand carve as close as I could and go from that to create my digitized profile. Here's my question. I used the Macrostie (stewmac)blueprint to create profile templates and carved my first top. After finishing it, the dished out 'recurve' area appears too wide and it doesn't seem to look like the mandolin I already have. The raised area looks too low. Has anyone else noticed that with those prints? I went with them because I was told (and noticed myself) that the first Siminoff prints had a different outline (points location, scroll size) than what seemed to be the standard. And was told a Siminoff mando wouldn't fit in a tight fitting case. By the way, what CAD software did you create your A Style in? If you ever do an F style, I'd be knocking on your door begging for a copy! I just hope I haven't wasted my time carving a top that will cave in the first time I string it up. -Pat Harrell

Michael Lewis
Mar-08-2004, 2:40am
Pat, your top should be 9/16" - 5/8" at the center of the bridge area as measured from the flat gluing surface. If you closely followed the arching provided by Don McRostie you should have a fine top. Don't worry about collapse unless you deviated from the plan or made the top too thin. There are many versions of exterior arching out there in the world, not all are great, but you can pretty well count on the stuff from Don.

c3hammer
Mar-09-2004, 4:31pm
Pat, my top design was from some goofy notes on a few Loars. It's not from any actual drawings or personal measurements. The recurve radius and location from the binding is a complete punt on my part as I didn't have access to a proper mandolin when I sketched that out.

I have Surfcam and SurfSolids that I use mostly for machining injection molds. I don't do all that much solid modeling and in that pic it is just 2d for the most part. I could create all the surfaces, but I didn't really have a need for anything but the top at this point.

I don't believe I'll be doing any F5 drawings any time soon. My whole project was based around doing a carbon fiber mandolin. That has gotten as far as making two molded tops. Who knows if I'll ever get anything else going with it.

Cheers,
Pete

Mark in Nevada City
Mar-09-2004, 5:30pm
There's quite a learning curve involved with 3d modeling of complex solids like an F5. I'm not sure I would even attempt it and I'm a fair hand at a number of commercially available CAD programs. If somebody approached me with the project, I'd probably start with ArchiCad or Vectorworks (both stronger than ACAD for 3d modeling and blending Nurbs solids) and possibly utilize DTM (digital terrain modeling) --setting "contours" at say .100" vs. 2ft. You may be able to generate a plate from McRostie's "sections" -- but that could be tricky. I don't have experience with CAM programs, but that would probably be the "right way." It would be fun to try, but you'd have to have a whole lotta time on your hands and may not get the results you want.

cheers, mmm

pathazzard
Mar-09-2004, 5:49pm
I don't believe I'll be doing any F5 drawings any time soon. #My whole project was based around doing a carbon fiber mandolin. #That has gotten as far as making two molded tops. #Who knows if I'll ever get anything else going with it.

Cheers,
Pete
Whoaa! Man what an idea!! I've picked one of the Rainsong carbon fibre guitars and that thing had tons of tone! A carbon fibre mandolin..hmmm You could do the back and sides one piece, Rigel already does that with wood. And Rainsong does it that way with the guitars. It took a while for the traditionalists to accept Rigels designs but you could do a carbon fibre mando with the scroll and everything. Structurally I don't know how you'd approach the neck attachment. Pete are you up for that? I could really get interested in a project like that. - Pat

c3hammer
Mar-11-2004, 11:47am
Mark:

Digitizing an F Style top and back plates would be easy. You would then have an ASCI text file of all the X,Y,Z coordinates at what ever resolution you scanned at. You simply import this "point cloud" into any of the NURB modeling programs and you can generate surfaces quite easily.

I've always wanted to model a surface that I defined. Something that I know what the geometry behind it is and that I can change it in any way I choose down the road. Digitizing doesn't give you this information. It is very difficult to change the recurve radius, for instance, on a surface generated via a point cloud.

Pat, I have already done quite a few models and even a tool. I keep thinking I'll do something more, but I know how making a business out of your hobby can ruin a good thing. Right now I'm not to interested in ruining the fun I'm having learning to play a mandolin. But yes, I have all the equipment and background to make any thing I want http://www.mandolincafe.net/iB_html/non-cgi/emoticons/smile.gif

Cheers,
Pete

pathazzard
Mar-11-2004, 12:26pm
Pete, I have a digitizing probe for my 48x48 CNC and a top plate that I haven't gotten around to final tweaking before digitizing. I haven't worked with the RhinoCad very much (just got), but the software from ShopBot tells me that I can either generate a cut file directly or output a DXF for editing. I've done some of that in AutoCad so I was hoping I could play with the curvature there and then export the DXF back to ShopBot for cutting. I can see where manipulating a cloud full of points would be difficult so I'm hoping the DXF route is the way to go. I know what you're saying about the hobby/business thing. I did some art (pastel) for my self once and got talked into publishing prints from it. I made a chunk of money (long gone) but the crazy thing is that I haven't wanted to do art at all since then. But for now I'm consumed with the mandolin and after playing at guitar for 38 years and never bothering to learn the notes, I've been learning mandolin for a few months and I'm going over scales in my head while I try to sleep at night! After 32 years in my day job (machine and wood work)It's about time for me to try something else and mandolins pretty much tie together everything I'm interested in.

delsbrother
Mar-11-2004, 7:26pm
I don't have direct experience with those programs, but the idea of manipulating a point cloud DXF model in AutoCAD is harrowing at best.

grsnovi
Mar-11-2004, 8:19pm
A program like Rhino will allow you to define your boundary curves as well as the curve radii at critical sections. You can then edit the radius values and the resulting surfaces will update.

The problem with the whole "point cloud" approach is also its advantage: scan a FINISHED piece and the software creates a surface as defined by those points.

Subsequent editing can get tricky - depending on what tools you use.

I've never been a "surface" modeling kind of guy but I have been "playing" some with Rhino (R2 - which is an older release).

delsbrother
Mar-12-2004, 1:33am
Rhino is fun. And the demo version (25 saves only) can be dl'ed for FREE from the Rhino website. I tell my frustrated AutoCAD students about it all the time. (Slight financial interest as my school has ties to one of the distributors, but it really is good product - cheap too, by CAD standards)

I guess I'm confused as to what the end product was supposed to be - a rough plate that would be tuned some more with palm planes, etc. ? Then I would go the point cloud route and digitize a carved sample. If you want to do a lot of tweaking (changing the recurve area slightly, etc.) I would suggest a surface modeling program (i.e. a CAM program).

pathazzard
Mar-12-2004, 8:59am
The problem with the whole "point cloud" approach is also its advantage: scan a FINISHED piece and the software creates a surface as defined by those points.
If I wasn't doing any changes to the profile I could just generate a cutting file for my CNC directly, so no creating point clouds for editing. Making a DXF should let me bring into ACAD or Rhino and delete points and simplify the line curves. My understanding of a CAM program (I have Visualmill) is that it is just the interface that creates the post-processing file that the CNC reads for cutting, not editing. Am I wrong?

grsnovi
Mar-12-2004, 11:03am
I am not a CAM guy, but many of the CNC hardware packages include very powerful software packages that INCLUDE decent editing capabilities.

Bluemando
Mar-12-2004, 11:21am
I am the foreman, at a machine shop here in Texas. I deal with CNC programming all day along with 3 other counterparts. If it is at all possible that you could send a drawing (print) to exact specification, then I could e-mail you a program that would correspond. Not bragging but I just do this all day.

Bluemando
Mar-12-2004, 11:25am
Also let me add that I inquired for the thoughts of the other people here when thinking of machining a mando from scratch. I was told that Mr. Gibson thought of this as well and abandoned this method early on, ....I now see why. It is a difficult process. I would be willing to provide a M and G code (eia) program that would rough out the plates like you mentioned, not a complete instrument

Mark in Nevada City
Mar-12-2004, 1:20pm
Sorry, Pete, maybe my post wasn't clear. I think digitizing the plate WOULD be the way to go. I was talking about drawing from plan and section (still, I think, a tough proposition--but maybe not for somebody like Bluemando).

BTW Bluemando, I see how once you've got a print to "exact specification" you can generate a model of the plates, but knowing what you do about the contours of an F5 plate is it really possible to describe that form "exactly" in dimensioned, orthagonal drawings? I guess I deal with much more simplistic forms and the thought of modeling the scroll area gets my head a-spinnin'

cheers, mmm

grsnovi
Mar-12-2004, 5:51pm
Although difficult, creating exact "engineering drawings" of a completed spruce top is certainly possible. Modeling the "solid" (from scratch) including the carved arches, recurves and scroll is certainly also do-able (using any number of CAD tools).

Bluemando
Mar-12-2004, 7:41pm
I dont think it would be to terribly difficult. You could achieve fairly good "rough out" results using helical interpolation, with say a 3/8 or .5 end or ball mill. You would wind up with something a lot like the plates that come from stew-mac rough carved, with all the feed lines in it, but depending on the size of the cut you could make them blend to a point where it is almost smooth. Since rough out is all your going for, this would be ok, the line would come out with minimal sanding. Obviously the smoother it is machined, the longer the run time.

The programming language I use now is Compact 2 which is a little out dated, but I feel it gives me better control of the finished product.

As for the drawing, you could obviously do without it if you had a plate to take measurements from.

grsnovi
Mar-12-2004, 9:52pm
Getting tangent continuous blending in a surface modeler can be a challenge thats all I'm saying. It has been quite some time since I have attempted any serious surface modeling, but modeling in most CAD environments is not like carving wood or sculpting clay. So, to re-phrase my earlier statement, it would be fairly difficult for me given how infrequently I do surface modeling. In any event, I have the next six weeks off work and one of my goals is to surface model an F5 top - just for grins.

Bluemando
Mar-12-2004, 10:09pm
I would really like to see your results if you dont mind posting them when youre done.