PDA

View Full Version : CAD Mandolin



grsnovi
Mar-21-2004, 8:50pm
A week or so ago, I was discussing the difficulties of modeling a top using a CAD system. Well, I've been fooling around learning Rhino (to do the surfaces). I've figured out how I'm going to do the scroll (and have built some scroll surfaces) but here is what I have so far:

http://hometown.aol.com/gsmith9810/top.jpg

grsnovi
Mar-21-2004, 8:56pm
BTW - the above image was a screen shot of the top after I worked on it inside of another CAD package (not Rhino).

mandomood
Mar-21-2004, 9:03pm
that's pretty wild....

interested to see more...

grsnovi
Mar-21-2004, 9:30pm
Here it is inside of Rhino:

http://hometown.aol.com/gsmith9810/top2.jpg

Bill James
Mar-21-2004, 10:00pm
Gary,
What are you using for dimensions or are you just working on a good procedure for building the surfaces?

grsnovi
Mar-21-2004, 11:59pm
Hi Bill,

I'm sort of winging it, but not quite.

I am trying to get in the ballpark, but as far as worrying about getting a dimensionally correct "Loar" top, I'm not.

I'm using a mix of: measuring my mando (for overall width and point locations) and simply convenient geometry creation for other aspects.

As far as the curve and recurve (which in the example is identical both outside/top and inside/bottom, I just eye-balled the approximate elevation above the binding on mine, estimated the width of the recurved area and a depth.

I'm not working from anyone's drawings or a set of cross-sections (but I do know that the outside and inside surfaces wouldn't be done this way nor likely would the recurve simply trace the outside profile.

For the f-holes, I snagged a good front view photograph off of a dealer's web page and then "slipped" that into my model using the outline on the photograph to locate the end ellipses and the mid-points. I then sort of connected those key points using a couple of tangent three-point arcs.

I am not a "surface modeler" by any stretch of the imagination.

Surface modeling with CAD is tricky.

Keeping all of your surface patches tangent and all of your blends looking nice is hard.

I have been involved in CAD for close to 20 years (mostly machinery design oriented).

I'm using this exercise to teach myself a little about surfacing (which my software doesn't do well, but which Rhino does very well - I'm just a total novice with Rhino).

My real interest was to get a good looking scroll and to get that blended well with the rest of the body. I'm less interested (at this point) with getting the rest of the body "perfect" but I needed a pseudo-correct body to blend the scroll into.

Bluemando
Mar-22-2004, 6:06am
very nice Gary, cant wait to see the scroll results. The stuff I am using obviously cant compete with this. It sure beats the point cloud we discussed before.

HoGo
Mar-22-2004, 8:27am
Nice work Gary. I wanted to put my own plans into 3D and left that idea, but now I'm considering to give it a try again. The surface you showed looks good, but the topo-lines (lines connecting points with the same height of arch) of a real mandolin should approach an ellipse as you go higher. It is almost elliptic already half the way up the hill. and the highest point is usually under the bridge, which is not at the widest part of the body. And another curve-ball is that the recurve doesn't go around the entire top. It gets shallower and vanishes somewhere at the "waist" area.
Can you deal with these "oddities"?

Tom C
Mar-22-2004, 11:34am
That would make a great sink. For a toilet seat i guess you need to use an oval hole instead of F-holes.

grsnovi
Mar-22-2004, 1:24pm
Thanks HoGo,

Like I said, I am not a surface modeling expert and while virtually every product we see in the market has something akin to "organic" surfaces, products themselves actually make up less of what engineers design - hence surfacing doesn't represent the lion's share of the software market. That said, our customers want more surfacing than we currently offer and I figured it was time to give myself a project or two to help myself learn.

I am 100% confident that a 100% accurate surface model can be developed using CAD.

The hard part in the CAD industry is separation of the difference between designing something new (from scratch) vs re-creation of an exact thing that was created using other more traditional techniques.

So, for me, the hard thing is looking at something (in this case a mandolin top) and deconstructing it into all of the various techniques needed to do it "right". Right now, I'm just going for "simple" (ie: close enough for 80% of the folks).

Also, as I mentioned, I am not working from a set of plans. I thought about buying a couple of sets (and maybe I will for VMF#2 - virtual mando F number two).

There are so many different ways of arriving at what (at first glance) appears to be the same end result but each technique is easier/harder and each offers pros and cons as far as supporting future edits and/or incorporation into bigger projects.

Three completely different (basic) ways of getting a "top" would be:

1) starting at the tail and proceeding towards the neck, create profile curves every 1/2 inch and then "connect" them all with a surface. Or...

2) looking down at the face, draw your elevation curvatures on a series of planes each a 1/16th inch apart (for example) and then "drape" a surface over those curves. Or (the technique I used for the above)...

3) draw a path curve and a sweep curve and revolve the sweep around the path. I drew a basic "A" sort of "pear" shape using an ellipse, a small circle and two tangent lines. Perpendicular to that, I drew the curve I was going to sweep by using two tangent arcs. The system takes that curve (stretches it uniformly as needed) and runs it around the path.

This was the least "work" for me - and until I command the full arsenal of techniques in my head, I'm going with what is easy.

Part of the trouble is that the teaching of how to use various tools is often left to the guys who wrote the computer code and they are generally the least capabale when it comes to explaining the practical applications.

HoGo
Mar-23-2004, 6:55am
Gary, what about combining 1&2. I'm quite good at drawing 2D. That's a result of 15 yrs of math (I liked geometry) studies, plus violin drafting training. Most people can not see a difference between Strad and Guarneri. But there is really great difference in Loar look vs. most new madolins (even vs. Master models). If you can see the difference you are on the right way.
You are possibly the only "virtual builder" on the cafe. I think every builder should spend some time drafting (paper or CAD) and designing instrument. It's the best way how to get into the aesthetics of musical instruments. Most builders just buy plans and follow the numbers and tracings. I heard Steve Gilchrist and John Monteleone have never used bought plans...
I can provide you some topo maps plus some crossections. I tried to draw maps for a scroll too, but I don't know how good they are.

grsnovi
Mar-23-2004, 12:28pm
HoGo,

The "three" methods I described above were only three of possibly twenty different approaches and of course, combining any/all of them is what surface modeling guys do on a daily basis.

Once you get a surface (and depending on your software) there are various ways of increasing its "order" and pulling/pushing points to achieve local deformations, etc...

Creation of simple surfaces is quite easy - you can easily run a surface through a network of curves that describe the "U" and "V" profiles at various locations. The hard part is then: getting all of the surfaces that you created to: line up (smoothly) and subsequently "stitch" together. The problem is what each (separate) surface building technique leaves along each edge.

Very simply the scroll could be defined by three surfaces: a flat ellipse, a three-sided patch which runs in from the body sweeping up into the inside point and a four-sided patch that sweeps from the body around and into the elliptical flat.

The trick is to get all of your guide curves correct and obtaining aesthetically "pleasing" free-form curves in CAD software is often quite difficult!

Anyway - FYI, the recurve on the (outside) top surface of my new Collings MF is very constant around the entire perimeter. There is quite a "sharp" delineation at each of the treble-side points. The recurve doesn't really disappear until you are into the last inch or so up in the neck area.

As far as your offer for topo maps and cross sections, thanks, but I'm getting where I want with what I have so far. As I said, my goal here is not to end up with a CAD model that somebody would take and feed into a CNC to cut (although that is sort of where my head was when I started). This was all just triggered by a thread Bluemando and I kicked around a couple of weeks ago.

I happen to be on a paid sabbatical from work at the moment and I'm just doing this to amuse myself.

grsnovi
Mar-23-2004, 4:28pm
This scroll is three surfaces: a flat circle, the inner face created by creating a surface bounded by three edges and the outer sweep along two "rail curves" with two profile curves.

http://hometown.aol.com/gsmith9810/scroll.jpg

c3hammer
Mar-24-2004, 1:54am
Nice Gary! Looks like you have too much time on your hands, LOL! I modeled all the surfaces for something like a Rigel G110 at one time. It is fun to mess with though.

Cheers,
Pete

Michael Lewis
Mar-24-2004, 2:09am
Gary, this is interesting. The F style design is a VERY involved and complex surface. I'm sure you can appreciate that the actual contours effect the actual tone. I know you said this is just for fun, and that you are not trying to model a perfect top plate, but once you have done this much work it seems you might be tempted to go a bit farther. Do you find this as compelling as playing?

grsnovi
Mar-24-2004, 12:21pm
Hi Michael,

As I mentioned, the above modeling is a learning exercise for me and (as such) there is no comparison to playing as (for me) playing is purely relaxation.

I certainly wasn't there when Lloyd decided to make changes to existing designs, but I sort of doubt there was a lot of pencil and paper "engineering" involved. Also, while one can probably get an "ideal" geometry for a carved top in a generic sense, the actual finished piece has to depend on the piece of wood used. A piece of spruce with tighter grain would need to be thicker/thinner in places based on the specific piece of wood - not based on some idea of "perfect geometry".

Knowing where and how to make those sorts of judgement calls is what separates an experienced, master builder from a wood butcher.

Although the "temptation" to go farther might exist, the question would be what form would it take? I'd be hesitant to output a file and send the file to somebody to cut (for fear of finding mistakes at the milling stage). I already know that I have a tendency to purchase tools that I'll only use once, so I could see me buying a CNC mill to "play" with. In the back of my mind I have always thought that being set-up to do various lutherie repairs and such after I retire would be a possibility. Owning a CNC mill would allow me to get into pseudo-production on reproduction parts (allowing me to make two or more exactly the same).

Bottom line is that I've never even carved a top and were I to build one, I'd probably manually carve the top - just as a carving exercise.

Marty Jacobson
Mar-25-2004, 9:44pm
Good stuff, Mr. Smith. Rhino's rendering capabilities leave something to be desired, and if you're doing it purely for fun, I'd definitely play with Maya. There's a "Learning Edition" which I believe is free for non-commercial use, and that includes MentalRay, the whole shebang.
Now there's a fun program.. and I think you might find Maya's NURBS controls more intuitive than Rhino's approach. Just an idea.. Take care.

twombo
Apr-25-2004, 1:28pm
how about a neck in Rhino?

grsnovi
Apr-25-2004, 4:01pm
Nice twombo,

I sort of lost interest after attacking the scroll. Not being a Rhino wiz, I wasn't sure if I should just add my scroll work to my previous top work and go from there, or start over clean. In any event, my play time is over and its back to work tomorrow.

I have built a complete (CAD) guitar which was double bound, fretted with working tuning machines.

I think that some may have thought that my efforts were strictly for "fun" or for "show" - not quite. My models were accurate and I assume had I the equipment, I could output files and cut wood via a CNC set-up (hence MAYA was never even an option).

G