Preview Build 6852
I managed to get everything pieced back together enough to release it as a preview. You can get it at http://www.grzsoftware.com/file/MeshCAM2-6852.exe . The finishing doesn’t have progress bars yet; that’s going to require a whole lot more code because of the new workflow and the multithreading. It does seem to work though and I’d appreciate the feedback. I’ll warn you now though, I made lots of changes so you are guaranteed to find bugs. Be sure to run everything through CutViewer.
I still have several outstanding bugs to address that have not been fixed yet. I want to get this in a stable place before I move on.
Robert,
When attempting to use “Machine Whole Stock” I can not get a finish pass. Everything seemed OK with a “Machine Geometry ”
Jeff
Robert,
I’m having the same problem. Also, when loading a large image file (i’ve been cutting lithophanes) I’m pegging both processors. duo core T2500 @ 2.0GHz w/2gig ram.
Although, I think theres a way to desiginate just one processor to an app, but I have to figure it out.
I do like the new background and the supports are great!
Bob
Whoops. I’ll get a new version posted tonight.
The image loading should not be multithreaded so I’m surprised at the pegging of both cores. How big was the image?
-Robert
It turns out that I can’t duplicate the problem- it works fine for me. Can you give me a process using the sample files that causes no finishing path to be generated?
-Robert
Robert,
The image was big, close to a meg. It loaded ok, but it pegs when you try to generate the toolpath. I just retried it and my task manager went to 100%
Also, I did a test with a dxf about 500k.
3 axis: Machine whole stock dosn’t crash but dosn’t generate a correct toolpaht.
2 sided: Same thing as 3 axis:
Indexed: again same thing.
When I switch to machine geometry only it works fine, and very quickly!
I’ll send you the dfx.
Bob
Robert,
One more thing. I tried the image file using the last version 6738 and only pegged the one processor when I tried to generate the toolpath.
Bob
Found the problem- it’s something related to the tool units. It looks like you work in MM and I know that Jeff does as well. It worked for we with the tool in inches and failed with the tool in mm. This should be an easy fix although I don’t know why it’s happening.
The multithreading was only added in this release so the last release will not peg both of cores. Also, you may look at a program called “Process Lasso” if you don’t want everything pegged. It will monitor the system and reduce the priority of any process that causes the system to become nonresponsive. To me the task manager at 100% is a beautiful thing; it means that I’m finally using all of my computing power I paid for.
-Robert
Robert,
I chose inches when I when asked to select in mesh cam.
Perhaps there is something the conversion that makes it look like mm. The dimensions were large because I just drew it up and exported it without paying any attention to dimensions. I was testing the cut.
Thanks for the tip on lasso I’ll check it out. Can’t stare at a screen waiting for something to happen, have to multitask, else I’ll explode.
Thanks again,
Bob
OK- I found it. It seems to be related to the toolpath layout code. Choose something non-even for the stepover like .51mm instead of .5mm. I’m not sure what you used, my virus scanner marked your email as having some type of autorun virus so I didn’t get to open it.
This my be bigger to fix but I expect this at some point.
-Robert
Robert,
Glad you found the problem. I was unable to duplicate the problem using samplerhino.stl (in mm and scaled 5X). Very strange, still reliably missing the finish pass on my file. If you need more files to play with I can post the stl on my site or email direct. Even zipped it is still 4 meg, got to use all that disk space I paid for.
Jeff
Virus! oh no, something got disabled. Thanks. Just updated and scanning everything. ;(
Thanks for letting me know,
Bob
First, let me state the MeshCAM ROCKS!
I am guessing this is a bug
http://www.batbuilds.com/~adam/cnc_rev5/untitled.PNG
and my test part
http://www.batbuilds.com/~adam/cnc_rev5/Part1.stl
This is just a simple test part, “Enable Finish Pass” and only the “Waterline” is checked, step down at 0.25, 0.5, and 1. Min Surface Angle at 10, 60, and 120. That was what I tested to make sure this was a bug.
Offsetting Surface take a long time? Is this because of my big part, or do I need to get a faster computer? (Dual core, Dual proc Xeon @ 3.0GHz, 4G RAM
Hi Rpbert!
I imported an .BMP file and only run finish in Y-direction. If i use 1mm as stepover it will not make any cut. If i use 1.1mm or 0.9mm there was no problem. Another thing is that when i ask for mm when loading an BMP the arrows in the zeropoint get wery big. So i need to load it one more time and then the arrow are normal. It’s not a big deel but still a bug.
Like your program wery mutch so i look forward to see the boundary clipping get finish.
Gunnar
Robert,
6852 is looking pretty nice overall. One thing I’ve noticed is that parallel finishing doesn’t seem to respect the maximum angle argument. If I use parallel finishing without waterline and specify a maximum angle of 1 degree, shouldn’t parallel finish just the horizontal surfaces and leave all sloped surfaces alone?
But I am impressed with the path linking in this version. It seems to be very well behaved at the edges of machining regions.
Randy
I’ve found the no-finish-path bug and another that Jeff found that led to huge gcode files.
The bad waterline paths have been improved- I’ll be curious to see what you think on those.
I thought the max angle parameter was being respected but I’ll have to test it again. I agree that the edge conditions seem to be improved and they will also be more accurate as you increase the tolerance.
Adam- what tolerance value did you use? That’s the number one factor in increasing offset time.
I’ll try to get a new version out over the weekend or early next week.
-Robert
What a difference 0.0001 vs 0.001 makes for the tolerance! Looking forward to the new release this weekend.
-Adam
I’m looking forward to the new version. I have one model where Y finishing would or would not work depending on where I placed the origin (much of the part is at “top of stock” and if the origin was in the non-machined area the Y finishing would not work) but I could not reproduce the problem with any simplified test STL’s (or with a slightly changed version of my original STL)
Randy
Robert,
Polishing up the old bug stomping boots.
New features are really nice and much appreciated but usually some there are bugs which have have to die in the process. The hunt is much better than any game I’ve ever seen. Progress sure beats a highest score.
To all reading this, even if you think you are crazy, delusional and making dumb mistakes , send some feed back to Robert. More often than not I have been of some help even with weird remarks. Robert is trying to write a program for all of us, the least we can do is try to help. Some of us use the program in ways Robert can’t be expected to imagine.
Jeff
Jeff
OK, I will.
Robert
Looking for, what I hope is a simple addition to the software. Once upon a time in a version long ago. In Roughing mode, the bit would only cut right side, then head back, advance and cut the next path, but only on the right side of the bit. I think this was controlled by “Mill Direction” To cut in conventional or climb mode. That does not seem to work any more. This could be a bug, or just me being dumb. Regardless of that, my CNC is not as strong as some of those all steal rings. If I am cutting on the left side of the bit (climb mode?), the bit will catch and cause the cnc to jump and that part comes out crappy. It there are way to have the bit only cut on the right side, in both roughing and finishing mode?
My cnc, rev 5, just got it running last weekend.
http://www.batbuilds.com/~adam/cnc_rev5/dsc05361.jpg
Projecting Toolpath takes a VERY long time. My parts are big. I am cutting out wings for RC planes. Is there a way to speed this up?
Different approach to cutting around an object. A simple test peace, a rod, on end. A water line cut will do a lap around the peace, then step down, then another pass. Why not a constant dive? The final path would look like a screw. So no longer cut in layers, but cut down in a spiral shape. Just and idea. (Yes, I understand how hard this would be from a coding standpoint
But I figured you were up to it)
-Adam
Adam –
The one direction only roughing was removed some time ago when I changed to the more efficient algorithm I have now. Normally people want to machine in one direction to reduce backlash problems. Since roughing toolpaths generally don’t worry a great deal about surface finish I removed this. Although some will miss it I don’t think this option will be making a return. Hopefully new features will offset this loss.
With regard to projecting, it’s a big job. I would say that you should not use an more tolerance than you need since that can add a lot of calculation time. I would think that .001″ or greater would be good.
I see that you know how to work with engineers, give them a difficult suggestion then stroke the ego a bit. I think I’m going to resist the urge though- I created enough bugs in the last few months to last me quite a while. If I create any more I’ll chase even the most diehard users away.
-Robert
-Robert
Take your time, kill those bugs. I will make my CNC play nice. Thank you for all you great work.
-Adam
[OT warning!] Adam, great looking machine! Are you running the plastic bearings right on the extrusions? I have used Delrin AF on anodized aluminum and it is an excellent combination. And the OSH bag! You must be somewhere in California.
[back on topic]For a gently curved, convex shape like a wing core use the largest diameter ball-end bit you can fit in your spindle, and as short as possible to still reach as deep as you need to cut. The bit stiffness makes an enormous difference in cutting stability, even on a big iron machine like my (still mostly new) Tormach mill. Try a stepover of about 1/8 of the bit’s diameter, and use a tolerance of about 1/5 the stepover. That should put you in a ballpark for having a reasonably smooth surface without excessive computing time, if you’re willing to do a little hand-smoothing afterwards. Those are parameters that work well for me for general stuff.
Randy
-Randy
I am using real bearings, Ball Bearings R8 ZZ 1/2″ x 1 1/8″ x 5/16″ off of ebay
The X and Z have single bearings. The Y is using two in an end block to help provide strength to the screw and stop wobble. The slides (extrusions) are using what every 8020.net puts on there slides, I think is is some form of delrin. The bolts, and most other supplies are from the local OSH (Livermore, CA). The 8020 stock (X, Y slide rails) are from 8020.net, but, I must admit that the 8020 was from my rev 4 CNC, that never worked.
Rev 4
http://www.batbuilds.com/webfolders/CNC Rev3B7/Images/main.PNG
Rev 3 was used to build rev 4. No plans needed for rev 3, I got board one day
.
http://www.batbuilds.com/webfolders/wing_cnc/dsc04401.jpg
Rev 1 was a dual X,Y hotwire foam cutter, rev 2 was that, but much bigger.
Cutting tool is a RotoZip
. It has a 1/4in chuck and I have a good selecting of bits, 1/4~1/32 all carbide. I also use a few router bits 1/4~1/2 for roughing.
http://www.rotozip.com/Shop/ProductDetail.htm?IID=69281
So, very fast 15,000~25,000RPM. Great for most plastics, and wood, but I can not cut metal
Thank you for the info on cuts. I have been cutting negative wings from polyethylene (Cutting material of choice), then laying up the molds with fiberglass. Currently computing time is no issue. When I am going for a final cut, I will give MeshCAM as much time as it needs.
I will try using one of my big roughing bits and see what the finished product looks like. Just looking at the output from CutViewer, all ready looks better/faster then my normal output
. Just as long as I keep up with the shop vac those new settings should be easy to work with.
http://www.batbuilds.com/webfolders/wing_cnc/dsc04392.jpg
At some point I will post this info on my website, instead of loading Roberts comment’s section.
Thank you for the input.
-Adam
Adam-
OK, dsc05361.jpg that you linked up in comment 19 was your Rev 4 and dsc04392.jpg is your Rev. 5? I am still really impressed at what you have built from scratch. And extra props for the channel-locks with zipties as hold-down clamps!
That’s funny though–I just stopped at “your” OSH on the way home from Fremont to Rio Vista this evening to buy a couple of liquid-tight conduit fittings.
And I’m glad if my suggestion has helped your machining. You’re welcome.
Randy
dsc05361.jpg is rev 5, the one I just finished. Rev 4 was a huge monster and my stepper were not powerful enough to make it work.