Official Luthiers Forum!

Anyone know of software that will convert a GCODE 3D profile to a solid model or point cloud?
Thanks
Nelson

You can convert the toolpath to a DXF, but it is very difficult to convert a 3d toolpath (such as one for a carved top electric body) to part geometry. It is not worth trying to do this, your best bet is to make a model and try to tweak it to be as close as possible to the part cut by the toolpath.

I've spent a good amount of time on this issue...from what I saw, it is not worth pursuing this.

There is a Rhino plugin that purports to do some similar things, but it is not up to snuff.

Trev

_________________
http://www.PeakeGuitars.com

Sounds like something Vericut or the Metacut utilities might do. They'd spit out an STL or somesuch, possibly, and you can grab points from that.

That said, I also agree with everything Trevor said.

_________________
Bob Garrish
Former Canonized Purveyor of Fine CNC Luthier Services

Nelson,

This may not help much but the other option is to run the tool path on some low grade wood and then send that blank out to have laser scanned. A buddy of mine does this work, mostly for ships, but he is keen to do some guitars for me when I am ready...for some trades! But I am sure that there are other people all over the place that have his technology.

Shane

_________________
Canada

STL's are the pits!...I highly recommend learning to model the surfaces and not settling for STL's or other mesh formats. They are a pain to deal with and are inherently "crunchy".

In addition to the software Bob mentioned, MasterCAM will also output an STL of the machined part.

Surfacing takes some practice, but it is well worth the trouble to learn. Or, if you don't have the time...check around and i'm sure you'll find someone to give you a hand.

_________________
http://www.PeakeGuitars.com

Also, the stuff you get back from a laser scan is still probably an STL. Even if you ran some other software to convert the scan into a more friendly solid or surface it will still not be "perfect". The solidworks reverse engineering stuff is not worth messing with by the way...the results you get from something like that are not pro quality in my opinion.

This isn't to say that this stuff doesn't work per se, but that it just is not the best way to do it....

Trev

_________________
http://www.PeakeGuitars.com

I guess this is the process of "reverse engineering..."



Yes, there are programs out there that will do this, but they are not cheap, and they have very large learning curves. Also, the quality of your model will obviously depend on the the number of points that happen to be in the g-code.

Once you reverse this, you will have something similar that resembles (basically is) a "point cloud." You can then, with the aid of software, make this point cloud into a solid model.

Look at XOR2 (rapidform) and geomagic studio to get an idea as to what I mean...like others have said here, not an easy process at all!

Hope this helps you....

http://www.rapidform.com/portal/default ... gn_XORSub1

http://www.geomagic.com/en/products/studio/index.shtml

_________________
-CyborgCNC
http://www.cncguitar.com
https://www.facebook.com/CncGuitar

Thanks for all the feedback, guys!
I had a friend develop this part in Rhino a couple years ago from a point cloud that I had sent.
He then "CAMMED" it in MasterCam if I remember correctly.
It is the exterior surface of an archtop topplate and I would like to develop the interior surface also.
He worked on it for a while but got busy with other stuff and I've just been doing the interior with pin router setup.
The problem with the interior is that it is not "parallel" with the exterior as it varies in thickness from the center to the outside.
This makes for a complicated surfacing challenge.
Nelson

I guess your friend doesn't have the surface he made? Or, you could remodel the surfaces if you still have that old point cloud...

_________________
http://www.PeakeGuitars.com

He had a hard drive crash and lost the model.
I probably need to re-digitize the plate and have one of you guys remodel it from the point cloud.
The digitizing process is slow as it takes several hours to "peck the points" accurately.
Yeah, I need to learn Rhino. I use ProE at my day job and it's not very "contour friendly".
Nelson

Edit: My friend did a great job of tweaking the contours in Rhino which is the Gcode that I'm now using and would be reproduced by digitizing.
The contours are "smoother than a baby's behind". The original contours were my hand carved plate.
Kind of demeaning to think that Rhino can "carve" a plate geometrically smoother than I.

In retrospect, and given the amount of data you have, there should be a way to do this in Rhino.

If you import all the points from the G-code, offset them to get the ballnose centre, then offset that to get the probable contact surface--you should end up with something close enough that a quick clean-up will give you a workable copy of your surface (within a very small tolerance of identical). You'll need to write an import/translation script.

_________________
Bob Garrish
Former Canonized Purveyor of Fine CNC Luthier Services

Bob, That's a bit of a hefty translation script. I have done some work along those lines (using VB) and figured in the end, it was easier just to remodel the part. The tricky part is that in order to determine what part of the spherical ball mill is touching the part, you have to take into account the points immediately surrounding it. It's ultimately the slope of the path across these points in all 3 coordinates (x, y, and z) that determines what part of the bit is actually carving. It's certainly not impossible, but it is not easy either, especially considering that the points from the existing toolpath are not inherently arranged in a nice evenly spaced grid....not an easy issue to get around.

Trev

_________________
http://www.PeakeGuitars.com

It's not trivial, but it's easier than you think. That said, it's one of those solutions that only makes sense if it's something you already know how to do. I have access to Vericut and can reverse engineer it through that route, which'll provide a good model, but it's a professional service.

_________________
Bob Garrish
Former Canonized Purveyor of Fine CNC Luthier Services

Guys, I normally digitize with the same diameter probe as the ball nose cutter to be used.
Would that not eliminate the need for the math?
Thanks
Nelson

Nelson that takes care of it if you intend to just run code from the points, and, they need to be fairly close together. On the other hand if you ever want or need to use the model for something else, you are right back where you started.

Mike

Mike, I see what you're saying.
It would seem that a surface developed from a digitized point cloud could be "offset" in Rhino the radius of the cutter to generate the actual surface.
But I guess that "offset" doesn't apply to a complex surface like it would to a curve, for example.
I'm sure you guys wouldn't be talking about the complex math above if that were the case.
Nelson

Hi Nelson,

Don't include me on any kind of complex math. I can barely spell it. That is why a cad program is so wonderful.

Nelson here is a screen shot showing what I mean by my statement. I hope you can see if clearly enough. If not, and if you have Rhino, send me a pm and I will send you the file.

Attached is a set of three arcs. One black, the model, one blue, a .060" digitizing tip, and the red represents a tip .500" in diameter, such as you would use to represent your cutter. If you will notice, all three arcs converge at the apex but both your digitizing tips are sitting on top with their tips touching the top of the model, while the two ends of the arcs you can see the two tips are using the side of the ball. All the digitizing I have ever seen, and that is not all that much, use the bottom of the tip for the top of the model surface and center of the radius for the side surfaces.

Even if you use a tip that matches your cutter, you will still get a point file that is shown like I have pictured here. Therein lies the need for the high dollar software, or a really good math person!

I hope I have explained that clearly enough, and keep in mind, with my limited experience there may well be many other options out there for obtaining a perfect point file, but that hasn't been the case for me.

Mike

Bob, I might be able to point you towards some work for your Vericut package.

The digitizing software that I have dealt with in the past does all the necessary calculations and gives you a corrected point cloud for the surface you've digitized (I've used Renishaw software).

Digitization is a great tool, but it is not a necessity. All this stuff can be done using "manual" tools (profile gauges, digital pictures, etc.) in conjuction with the surfacing tool of your choice. Most surfaces are created using a basic framework of curves..and if you can prove out the individual curves in that framework, you go a long way in making sure you have a good surface. One way to do it is to CNC cut profile gauges...make sure they fit the part in the appropriate place...and surface away...

Trev

_________________
http://www.PeakeGuitars.com

Mike--Thanks for your pictorial.
I'm thinking there's a difference between a scan done with a CMM, for example, running software such as Renishaw and a scan done with Mach3.
Mach3 simply records the point cloud that represents the center of the probe sphere rather than the actual part surface. Yes, the point of contact does change from the bottom of the probe to the side as it progresses "over the horizon".
That is why I was wondering if the "offset" command, or whatever jargon a particular CAD package uses, would do the calculations based on the radius of the probe sphere point cloud and then generate the actual part surface. Perhaps it takes something like Veracut to do this as Bob mentions above?
It's entirely possible that I'm not understanding the issue.
Nelson

Vericut creates a solid or mesh model based on your actual CNC code.

Renishaw, and other types of scanning software, do the math to calculate and display points that would actually lie right on the surface of the object being scanned.

This type of offset is kind of an oddball type of process and isn't something that most cad systems are setup to do.

Trev

_________________
http://www.PeakeGuitars.com

Trev, what would the process be for creating in Rhino the inner surface of an archtop plate after the outside surface has been modeled?
I was thinking that there would be some type off offset option that would generate the inner surface if given the thickness desired.
What I'm getting at is it looks like this would be a similar process to going from a point cloud generated at the center of the probe sphere to generating the actual model surface. I'm asking because I'm not very familiar with Rhino. I'm thinking that ProE has a "shell" command that would do this but not sure.
Thanks
Nelson

Nelson the command in Rhino is offset surface and it works exactly like you described. Select the surface to be offset, choose direction and input the amount of offset. Rhino does not have a good shell command that I know of, but MOI does. In fact, I am thinking of buying MOI just for that and a couple other features. You can import back and forth between Rhino and MOI with no loss of data.

Mike

Trev:
I'm a big fan of the physical process, as well (as I advised Sheldon Dingwall in a thread he posted on doing his necks). The only time I bring out the laser scanner or 'grab points' is if something is -really- curvy (PRS-style body) or needs to be super-precise (as in machinist precise). Someday I'm going to machine myself a really big profile gauge, though I'm also a fan of sawing up analogues (casts, seconds). Getting a precise piece of 3D reverse-engineering equipment is expensive, but $60 will get you a very accurate piece of 2D reverse-engineering equipment at any computer store. If you have a big enough flat screen, you don't even need the scanner!

PM me about that VC stuff, I'd like to look at it for sure!

Mike:
IAAM (I am a mathematician), so my perception of the scripting problem is probably highly skewed by that most of a decade of doing university math all day . Rhino is MAGIC for solving many of those same problems without needing to remember a formula (and it doesn't make mistakes!)

Offset is your friend, for sure, on getting a uniform-thickness plate. If you have a proper model (with profile curves) then you can also use Rhino to quickly make some modified contours (like an inside contour with a thick middle and thin edges).

_________________
Bob Garrish
Former Canonized Purveyor of Fine CNC Luthier Services

On a rough carve for an archtop back or top plate, you can actually get away with just copying the outside surface down a certain amount. Most archtop plates are composed of a number of surfaces...and it can actually get a little squirrelly when doing a true offset and trying to make all those offset surfaces match up cleanly. It's a lot easier to just copy that top surface down by 3/16" or whatever your desired thickness is. You will end up with some areas that are slightly thinner than desired, but it will be a pretty minimal amount on a typical archtop shape. If you want, add about a 1/16" to compensate for this (plus any allowance for material removal by sanding).

I hope this helps you out!

Best,
Trev

_________________
http://www.PeakeGuitars.com

Hi Parser,

I am by no means a Rhino expert, not even close. but I have on my computer a arch top model where the top is a single surface. If you set up the guide curves correctly and use the patch command, you can actually get a single surface.

Now on the other hand, after reading your copy suggestion, and I haven't tried this, but I would bet that will yield the same result as offset. I think if you use the copy command and copy it .2" lower or offset it .200" lower there would be no difference.

Mike