Problem with Mach3 postprocessor?

[austexjwlry]

Hello!

Last year about this same time we tried the trial version of Cut3D using TurboCNC, but our old Windows 98 CPU cratered, so we decided to upgrade to Mach 3 with an XP CPU. We are now also using the full licensed version of Cut3D, as well as Mach 3, etc.

The first problem we had was using Mach 3 set to absolute instead of incremental, causing the Z axis to try to run off the table, following the large extra circles outside the toolpath. We corrected that.

Now we seem to be having a scale problem, in that the model in Rhino and Cut3D is the same size. But our machine cuts the model thinner than it should be in the finishing path. It cuts about right in the roughing path, except for a fin from what seems to be a missed step. It mis-cuts the model the same with both the Mach 2/3 Arcs (inch) (*.txt) and Mach 2/3 ATC Arcs (inch) (*.txt). We made a new model of the same piece, almost identical in size, and it did almost the same, but with the fin in a different place and again cut the model too narrow in width in the finishing path. It seems if we were having a problem with our machine, the scale problems would exist in the roughing path also. The missing/misaligned step causing the fin is reproduceable, also leading us to think it's not the machine missing this step at the same exact place each time.

The motors seemed tuned pretty well, actually running about a third below the velocity where they would stall. And the acceleration is set so that the motors stop and start well without excess travel or stressing them with excess G's upon starting and stopping. They run quietly and smoothly; they are warm to the touch but not excessively hot.

We don't believe the scale problems are coming from our machine either because we used a steel rule to get our X, Y, and Z axes close, then used Mach's internal calculator to refine our jog distances to within .007 accuracy, checking travel with a dial indicator. It's pretty close.

What we are trying to do is cut an ankh shape out of a green carving wax tablet. We made the .stl file by opening a .gif in Inkscape, autotracing the design, assigning vectors to the corners, etc. We then saved this as a .dxf to our desktop, then opened it in Rhino. We then grouped and joined the curves, ending with 2 closed curves. We extruded these curves to the thickness desired. When a NURBS model is saved from Rhino as an .stl, it converts NURBS to mesh automatically. We then opened this .stl in Cut3D and assigned material size, etc.

Everything looks fine in the roughing toolpath preview, finishing preview, and cutout preview. Everything is saved to a floppy using Mach 3 processor, and sent to our IBM Vista computer, which is running Windows XP with a Pentium 4, 2.4 Ghz, 512 MB RAM. This CPU is not connected to the internet. It is a fresh, clean install of XP with no other programs added except Mach 3.

We are not sure what other info would help you in diagnosing our problem. Thanks in advance for any suggestions and help you can give us.

Sincerely,

Diana

[austexjwlry]

I forgot to include the information about our mill. It is a small 21" long by 15 1/2" wide by 17 1/2" tall kit built 3 axis mill. It uses an Alien CNC 3 axis simple step 1/8 micro stepping driver, which seems identical to the Xylotex board specs. The power supply is a Panja 12v
DC, 6.5 Amp. The stepper motors are 1.8 degree, 2.0 Amp, weighing .7 kg. It has a 5" square table.

Thanks!

Diana

[austexjwlry]

Hello again,

This is a picture of my little mill and a new wax celtic cross design we generated which we just got through roughing with no problem! This time with no missed steps causing a little wax fin where we missed a step. On the wax tablet is a partial rough cut of an ankh which I aborted because of a missed step.

The original size of our model was 1.09" length x .6821" wide; the length of the actual finished rough cut is 1.060" lengthwise x .644" wide - pretty close. We are using a .034 ball bur with cutting surfaces on just the ball not a ball-mill with cutting surfaces up the shank. Using 3d roughing enables the use of a ball bur which costs us 50 cents each not $10 to $50 each like the little ball mills. This roughing took 1 hr 46 minutes.

Now we have to try the finishing cut.

Diana

[kilrabit]

I have not had any dimensional troubles with CUT3D I also use rhino, and no problems as of yet.

You say that is is only off one direction, If the scale was off it would be off in all three directions. It sounds like you have a tuning problem, or losing steps on the on the axis that is off. Try setting the speed slower on that axis to see if there is a change. Check the stepper connectors at the shafts for loose allen screws, or anything else mechanical. Rotate the piece 90 degrees and see if the same dimensions are off, or if they are off on the same axis as before.

If this does not help someone else will have some more ideas, usually the forum at Vectric will fix the problem with a little patience.

[austexjwlry]

Kilrabit,

Thanks for the suggestions, but for the most part our problems have cleared up with the last model. Now I don't think it was a problem with cut3d or the Mach3 post processor but as you say we need to check our mechanical and electrical connections. Getting rid of everything we don't need in Windows XP might help Mach3 a little.

We did notice the bulb in a lamp on the same circuit dimming, almost blinking, during the day; perhaps there is a problem in the circuit feeding the little mill & dedicated computer. I'll check that tomorrow!

Anyway with a little tweaking of bur size etc we should be able to produce some nice waxes for casting and molding. The finished size of our Celtic wax was 1.037 long by .642 wide and the original size in cut3d was 1.09 long x .6821 wide. The difference is only about .06"; we can adjust the difference in cut3d before we send it to mach3.

We included a picture of the wax after finish cut using the same .034" burr as roughing. Also a picture with a quarter for scale. We have a lot to learn!

Thanks so much,

Diana

[kilrabit]

OK, and good luck to you.

[austexjwlry]

Kilrabit,

You were right. I had about 1mm play in both the X & Y axis. Calling it backlash would be an understatement. While we had it apart we did a complete machine setup using aluminum foil folded into little squares for shim stock . It's only .0005 thick per sheet and might hold up for awhile. We recalculated the steps per inch on each axis and it really made a difference!

The new rough cut of the same simple Celtic cross cut so nice you'd think we had polished it after cutting! Please see picture.

Thanks again,

Diana

[austexjwlry]

Hello,

This time I tried the same simple cross. It looked great through the rough and finish cut. Then it went most of the way through the cutout cycle and instead of raising up to go to zero or whatever, it rammed the bit through the wax model destroying it. See attached picture. I don't have a clue what went wrong!

Diana

[ManitobaKeith]

Hi Diana,
I don't have Cut 3D so can't really say for sure, but it looks like your safe Z height has been changed to be too low or your Z axis could be missing steps. Is it easy moving after you made the adjustments to it? You could try an air cut and see what happens. Did it raise at all when it was trying to go home? Try posting the question in the Cut 3D section. Some folks that have Cut 3D and might have the answer wouldn't necessarily look in this section.

[kilrabit]

Sorry it took me so long to check back on you, but I was not close to the internet.

It does not look like the Z axis plunged down, It looks like it did not raise up properly when it returned to zero after the cut.

Did you shim it any? Since it was cutting so well did you increase the speed of cut? Did you use the same g code for this cut? You might look at the connector from the stepper to the ball screw and make sure the coupler is not loose or has not moved. You might take the stepper off and turn the screw by hand to be sure there is easy movement. Possibly under prolonged use there was heat build up and a binding occurred, so check the lubrication of the screw also.

Are you running a full crosshatch, or rough and finish in two directions?

[austexjwlry]

Hello everyone,

We're sorry we didn't get back to the forum sooner! We've been running our little mill 12 to 15hrs a day every day with great results. We've produced around 50 wax models using only the roughing and finishing tool paths without using a cutout tool path as its simple to clean up the waxes by hand with files. The tool paths created in cut3d have been flawless. I don't even preview or run a simulation of generated g code. I just start Mach3 and blow the cut wax off the model once in a while if I think about it! We still have a lot to learn. Now I want X3 mill with a 4th axis, a big router, a plasma cutter, waterjet or laser cutter etc... etc. Ha Ha!


Just wanted to touch bases and thank everyone for their help!

Diana