Re: [CAD_CAM_EDM_DRO] Re: Measuring accuracy

Ok, I was a bit upset when I saw what was happening before.


Fred, since you answered I'll pose these questions about Desk CNC. Why
is it that with my parts I need to leave 3mm of stock on roughing to
avoid gouging the part? Why does it seem that the software will not cut
a vertical wall? I actually cut a part with code DeskCNC generated. I
had a lot of variation in wall thickness that wasn't nearly as obvious
when cut with another software. I had set .25mm z steps and .25mm step
over for finish. To say the least progress was extremely slow. A $100
software cut it much faster, 5% of the time DeskCNC or most programs
would take. Actually, I didn't even wait for it to finish, 24 hours was
enough and there was a good 40 to go. My DeskCNC demo is now expired
but unless the newest version does waterline finishing, I don't think
I'd consider purchasing. Will DeskCNC write a G02 or G03 command if I'm
working from an STl file?

Anyone who wants to see what I'm making, an iges file is available here.
www.edfinfo.com/temp/test.igs

I understand the accuracy of the machine does not have anything to do
with the accuracy of the files, but why bother with CNC if the code is
wrong to begin with? I might as well pay someone else to make it to my
print, at least if it's wrong it won't cost me anything. Sort of kills
the hobby.

I looked at BobCad, didn't like their sales people, the software has a
steep learning curve. I had the above issues with DeskCNC. Millwizard
doesn't save code so it's a worthless demo.

The one I've played with the most is MeshCam. Now, I realize that it is
capable of generating code and parts I'll be happy with, but processing
time is long. Radii would still be segmented but I can write the code
for most of the things I'll need. Machine time with this software is short.

I'm going to look at DeskProto and OneCNC. Both are much more expensive
but I'll see what they can do.

Thanks, for the input and thanks for not flaming me for this question.

Greg





Fred Smith wrote:

> --- In CAD_CAM_EDM_DRO@yahoogroups.com, Gregory Kamysz
> <smsupply@a...> wrote:
>
>>Doug, that's what I thought, but which software under 1000 USD
>
> actually
>
>>does this? I've tried several and running a G02 is much more
>
> accurate
>
>>for cutting a circle than any of the CAM software I tried.
>
>
> Greg,
>
> You are confusing two things here.
>
> 1) The accuracy of CAM software, as it is applied to roughing
> surfaces with radiused cutters. Vector, StlWork, and DeskCNC will
> produce 3D code as accurate as the beginning model. They all allow
> you to back off from this utopian position and make a more realistic
> calculation for gross roughing of material with finishing stock left
> behind. They also will let you process the surface data at a finer
> resolution than the original data in the case of .stl files which are
> triangle approximations. In the case of Vector, it will evaluate a
> nurbs mathmatical surface with whatever interpolation accuracy you
> desire. The model is defined numerically and there is by definition,
> no measurable granularity in either the original design, or in the
> resulting toolpaths. All our commercial software will produce
> accurate programs, I cannot speak for other programs. What program
> are you using that has this problem?
>
> 2) The other thing you need to consider is realistic accuracy
> achievable with the cutting tool available. A ball nosed end mill or
> router bit is a spinning lever with the spindle bearings being the
> fulcrum, and the force applied at the cutter tip. Ideally the
> spinning cutter will shear material with sufficient ease, that the
> deflection from the lever will be within expected tolerances on your
> machine. In reality, the cutter has a "dead zone" right in the
> center where you have programmed it to do all it's cutting. This
> dead zone is caused by the decreasing surface speed as you approach
> the center, as opposed to the programmed speed out by the full sizeed
> edges. This slower cutting area will cause the tool to scrape and
> plow material away, instead of the nice planned shearing. The
> resulting surface is rougher than the programmed accuracy, simply
> from the leveraging of the slow moving cutter through the material.
> A longer bit will produce even worse results. If instead of using a
> ball nosed cutter, you use a radiused corner (bull nosed) or a flat
> ended end mill or bit, there will be no attempt to cut with the "dead
> zone". (unless you plunge vertically along the Z axis). This will
> make for much better finishes and for more accuracy on the sides and
> bottoms of the cut. You will however, not be able to cut concavities
> smaller that your cutter end flat, and a flat end mill is not
> productive for smooth curved surface. The flat end tools however are
> much better at roughing as they cut full depth at a higher cutting
> speed than a ball nosed cutter.
>
>
> I suspect your perception of inaccuracy is operator error in creating
> your code, and that when you get around to making parts, you will
> find that mechanical cutting has a much greater liklihood of causing
> inaccuracy, than the worst CAM program.
>
> Fred Smith - IMService
> THE source for low cost Cad-Cam
> $75 sale, still on, http://www.cadcamcadcam.com/hobby
>