CAM solids--the best of both worlds: solid models are even better in CAM.

One of the big changes over the past few years has been the ever-increasing use of solid modeling for designing parts. This is truly a great thing. You are probably familiar with the leading solid design software--products like SolidWorks, Solid Edge, Autodesk Inventor, CATIA, Pro/ENGINEER, UGNX, and a host of others.

All provide a mechanical design engineer the ability to work with a 3D solid model while designing his "part." His job is faster, easier, and less prone to mistakes and omissions. He can let the computer calculate stress, volumes, thermal, and mechanical properties for him. He can "render" the part and visualize it in 3D on his computer screen. He can even use a fancy 3D printer (stereolithography, etc.) to produce a physical prototype.

Design engineering has come a long way from the 1970s when I was a mechanical design engineer working at a drafting table with a pencil, T-square, and triangles. Twenty years ago, most parts were defined on a blueprint and sent to the machine shop for manufacture. Back then it was a big improvement when large Xerox machines replaced the ammonia-based true "blue" prints.

But we still programmed parts from paper drawings. I'd spend hours going through multi-sheet drawings, figuring out what the parts really looked like, checking dimensions, and frequently finding drawing and dimensioning mistakes or areas that couldn't be machined as drawn. One of my personal programming practices was to sketch an isometric view of the part from the three orthogonal views on the drawing. You had to know the correct Z-level of every step and pocket to get this right. 1 earned my B-2 Bomber project pin programming complex transmission casings this way.

Even better for CAM

As good as solid modeling is for a design engineer, it's even better for a manufacturing engineer/NC programmer/machinist. A part designed as a solid model helps CAM in a number of important ways.

First, it is a better part definition. If the designer created a valid solid model, then every detail and nominal dimension are available to the programmer without looking for drawing dimensions. The programmer can spin and inspect the part solid on the screen without trying to visualize it from orthographic projections.

There will be fewer design errors. There will be no undefined areas. The solid is itself a complete geometry definition. And since a modern CAM software should be able to open the CAD solid directly, there is no need to redraw the part or re-create the solid model geometry.

All this sounds pretty simple. A designer designs his part as a solid model using design CAD software. He sends his solid model file to the machine shop as a computer file. The NC programmer opens the solid file in his CAM software, hits a button, and a CNC program is spit out, right? Not exactly, although many people seem to have this opinion.

The big issue is that CAM software is a tool for an NC programmer to develop an optimal CNC program based on his own machining knowledge and practices. No one has software that replaces the machinist, any more than anyone has CAD software that replaces the design engineer.

Both CAD and CAM are very powerful tools that should empower their corresponding users to apply their knowledge more efficiently in the completion of their tasks.

Programming from model?

There is another important issue at a more detailed level. An NC programmer frequently doesn't program from the solid model he was given.

This is an odd statement. Allow me to explain. The end product of a design engineer's efforts is a solid model of the finish part. It is a model of the final part as he wants it produced and manufactured. He typically has little idea how it will be machined. Or in how many steps or in how many processes. The finish part model he provides is not the part as it will start or end any of the machining processes. If you produce the solid model you want to achieve at the end of a machining process, it is called a "process model."

Let's go back to my B-2 transmission housing and pretend I am doing it today from a solid model of the finish part. I open the solid model file and get a good look at the part. I plan my job. I'll rough machine side 1 on a vertical mill, all parts. I'll rough machine side 2 on another vertical mill, all parts. Then they go out for thermal stress-relieving. When parts come back, I'll finish side 1 and side 2 in two more setups. I'll then run them through a drilling center to do the hundreds of small tapped holes on side 1.

So far, this involves five setups all together. I decide to build a process model for each setup, using my CAM system's solid modeler (some good CAM systems like my GibbsCAM include a CAM-focused solid modeler, as opposed to using a design-focused solid modeler).

For setup 1's process model, I remove all the detail I'm not going to machine, like the backside, all the small details, and small holes. I add 0.100" of material to all surfaces. I add some tabs on the outside for future setup pinning and clamping. I orient this model the way it will sit on the machine for machining.

I do similar things for setup 2's process model. I can use the setup 1 process model as the initial material definition for setup 2. The process models for setups 3 and 4 look much more like the finish part model, as I only remove the small holes. Setup 5 only drills the holes. I program a different part program for each setup.

I'm almost done, when my customer calls with a revision to his design. No worries. Thanks to "associativity," I can open the new revision and have my process models and toolpaths updated automatically in minutes. I check the shop schedule, pick the best available machines, and post out G-code that will not need editing. I'm ready for setting up a machine and checking first articles.

Good option to have

You don't have to create process models all the time. They are an option. CAM software typically provides ways to "not machine" model details that you don't want to. But as the parts and processes become more complicated, creating process models is a nice option to have.

There are lots of other examples where the model machined is not the finish part model.

* Mold and die shops machine the tools that will later make the finish part. They need to model the molds and dies from the finish part model, adding draft, shrinkage, parting line surfaces, as well as adding and deleting numerous other details. They need CAM modeling from the finish part model.

* Production shops may need to model and machine "tooling" such as soft jaws or fixtures from the final part model. You can even "fix" a solid that isn't quite right.

These are some of the reasons why being able to create CAM solid models is so useful and just some of the reasons for parts that start as CAD solid models. If a part starts off as an IGES file (wireframe or surfaces) or even a drawing, it is very handy to be able to create your own solids to visualize and machine from.

There are CAD finish part solid models and there are the CAM solid models we make from them for machining. CAD solid modeling is certainly similar to CAM solid modeling, but models are typically created by different people for different purposes. If you are going to machine solid models, you are going to create CAM solid models as well. So why not pick a modeling tool intended for the CAM function and the CAM user? Gibbs and Associates, www.rsleads.com/505tp-188

Bill Gibbs President Gibbs and Associates