Circle pocket, via adaptive clearing, cutting off center

[CNCwoodMan]

Hello,

Bit of a strange issue here and looking for advice on how to troubleshoot. I am using a custom built CNC and have not run into issue with cutting 2d contours of circles or any other shapes for that matter. I have done testing on a 1" circle cutting on the outside then on the inside and comparing spacing and it was very accurate. My issue here, pictures attached is when I mill out a circular pocket it seems to throw it off center. From there, it seems to throw off the contour cut of the circle creating a slant on the contour cut as you can see. However, it throws off the contour of only the two circles that have pockets in them. The third solid circle is a perfect circle shape with no issues I have tried two different cam programs to generate GCODE and the same thing happens so I am reluctant that it is the CNC machine and or the MACH3 software interpreting the GCODE. I have attached the GCODE as well. Anyone with any suggestions or help it is appreciated!

Similar Threads:

• Can RhinoCAM do adaptive clearing?
• Saturn, cutting trivets with Fusion 360 3D Adaptive Clearing
• 3D Adaptive clearing question
• Understanding HSMWorks Adaptive Clearing
• Adaptive clearing algorithm video

[awerby]

It looks like you're losing steps in X or Y. But it could also be a mechanical issue, like a coupler that's slipping when stressed. Check to see if it returns to the same place you set your X and Y zeroes.

[routalot]

Not enough information to really get to the root of the problem.Pictures of the machine would help as it would show more of the drive mechanism and spindle mount.I agree with awerby that a mechanical issue is very likely at the root of the problem.It could be flexing at the spindle mount,drive belts stretching,couplings,gantry flex or a combination of any.It might also be the workpiece moving.

[BBMNet]

Hello.

Are your circles perfect circles? It might be possible that the parts flex when they are machined thus distorting the geometries.

That problem looks rather familiar to me as I retrofit textile equipment and when the proper procedures are not followed fabrics get bent or stretched.

As a test I suggest you try doing that figure on a piece of wood.

Regards.

[CNCwoodMan]

It looks like you're losing steps in X or Y. But it could also be a mechanical issue, like a coupler that's slipping when stressed. Check to see if it returns to the same place you set your X and Y zeroes.

It does seem that way now that you mention it. Looking at scrap cut offs it seems to constantly loose steps in the belly of an arc. I have recalibrated the motor steps in mach 3 and it is dead on using a ruler. I know it's not a good method but it is all I have now. I am hoping to generate a test cut file and post results. It does seem to return to same x and y zeroes. I would have to test that differently as the home is in the middle of a pocket post cut.

Sent from my SM-G960U using Tapatalk

[CNCwoodMan]

Not enough information to really get to the root of the problem.Pictures of the machine would help as it would show more of the drive mechanism and spindle mount.I agree with awerby that a mechanical issue is very likely at the root of the problem.It could be flexing at the spindle mount,drive belts stretching,couplings,gantry flex or a combination of any.It might also be the workpiece moving.

Here are the pics of the machine itself. Let me know if you see something that sticks out. I have done flex testing on gantry etc...all seems pretty secure. I guess I could look into replacing the belts....

Sent from my SM-G960U using Tapatalk

[CNCwoodMan]

Hello.

Are your circles perfect circles? It might be possible that the parts flex when they are machined thus distorting the geometries.

That problem looks rather familiar to me as I retrofit textile equipment and when the proper procedures are not followed fabrics get bent or stretched.

As a test I suggest you try doing that figure on a piece of wood.

Regards.

The thing is the contour is perfect on circles that do not have a pocket in them. I am cutting from plywood and it isn't moving as the pockets are done first then I cut out the contour. So, it is secure within the workpiece at all times.

Sent from my SM-G960U using Tapatalk

[awerby]

If the tool returns to the zero points that rules out lost steps. Look for mechanical issues like the ones Routalot mentions above.

[CNCwoodMan]

If the tool returns to the zero points that rules out lost steps. Look for mechanical issues like the ones Routalot mentions above.

Alright, so I think I am going to bring feed down to like 5ipm and let it run to see if it cuts as it should. That would prove mechanical then, right?

Sent from my SM-G960U using Tapatalk

[routalot]

I doubt that it is the spindle mounting to the Z axis as that looks very secure and better thought out than many.I would begin by looking for a tiny bit of play in the rollers against the rails and if no adjustment seems necessary on any of the moving parts I would try a couple of test cuts.We can't see how you have been holding the workpiece in place and we don't know whether you are climb or conventional cutting.Could you try a comparison of climb and conventional cutting,part way through a plywood or MDF sample.As you have been doing it is the way I would also go about it with the inner shape cut first and then the outer.With both cut types for comparison a bit more should become apparent.The sample in the original post looks to be a composite of some sort and these are not the most friendly of materials for your tooling so I would recommend a very close scrutiny of the edges to check for sharpness.

[cyclestart]

The cut order is:
1) inside features
2) top piece climb cut in z-0.125" step downs
3) bottom pieces climb cut in z-0.125 step downs, alternating by depth.

There's no evidence to work shifting, the step downs would give it away.
The tool is dull or the material is ugly.
It looks like plywood but does it have some elasticity/compressibility ? It looks different than the plywood in my shop.

[CNCwoodMan]

Alright, so to answer a few questions it is birch veneer plywood from the big box stores. I decided to take on a larger pocketing job and it seems to loose steps only on pocketing jobs which is super weird. I can cut contour cuts all day long at a feed rate of 50 and .125 z depth no issues and accurately. When I star doing pockets with circular geometry game on. The step loss sound like a clunk and typically almost always pauses in the x-for a brief moment then continues. At first I thought it was depth and speed but brought it all the way back to a feedrate of 25 and it did happen. The pause in the axis movement seemed more noticeable. I then put a new but on as I thought dulling could be an issue, nope...happened again. I am going to try air cuts next to see if it happens as well. When cut is finished I home it and I am definitely off. I tried to video the x axis jump but barely noticeable on phone and sound of spindle drowns the clunk. The pictures show how off home is in the software and on piece. Driving me crazy! Could it be the stepper boards?

Sent from my SM-G960U using Tapatalk

[CNCwoodMan]

This is an example of the contour cut, in a circular pattern with no step loss.

Sent from my SM-G960U using Tapatalk

[CNCwoodMan]

Finally got the jump on camera. Notice at around 5 seconds in the x axis hesitate, clunk sound and you can see the belt in bottom right corner jumpTapatalk Cloud - Downlaoad File 20200724_165202.mp4

Its super hard to pick up the clunk but you can see the hesitation if you focus on the gantry.

Sent from my SM-G960U using Tapatalk

[routalot]

I'm pleased to learn that you found the problem,even if I couldn't view the video.Good luck with fixing it.

[CNCwoodMan]

Alright guys, in the past few days I have accomplished and learned alot! So, as mentioned before I noticed that the x axis motor was slightly stalling at different intervals certainly effecting cut quality. So, I embarked on a journey to replace all my probotix side step controllers with leadshine dm542e controllers as it was suggested as a superior upgrade. Had some troubleshooting to figure out but got them installed successfully and let me tell you, so much better. The motors step so smooth and quiet! It has completely eliminated clunky and random motor stalling. After all that headache I proceeded to do another big pocketing precision test cut and FAIL! Same issue as before. However, one thing I noticed is before the drive controller upgrade x and y was off when o would home the spindle during or after crappy offset cuts and now it is bang on in the y axis and consistently off on the x axis at home to the left by 1/8 inch!?!?! The interesting thing is as my z axis stepsdown while cutting it continually shifts cuts to the left as shown in the picture. How can this be so consistantly inconsistent LMFAO!!! I have tightened all the extrusion rail wheels, checked and tightened belts, ensured the proper amps are being fed to motors, checked pulleys on motor shaft for slipping (there was none), went through pin input and output config, tunes the motors for steps, acceleration and velocity, tested multiple feeds and speeds with different depths and air cuts, different materials to cut with, new bit, I have even created a new profile in mach 3 with default settings and set everything back up again, all with no luck. At this point issue is either in the computer, 25 pin connector (which I have a replacement one coming tomorrow) or the breakout board controller.

What else is there or could it be ? Any thoughts?

Sent from my SM-G960U using Tapatalk

[BBMNet]

Hello.

If possible post both the codes for you initial circles and that of your wooden sign. I´ll try to run them on a machine we are currently retrofitting.

Please repost the picture of your wooden sign drawing arrows showing the axes.

By the way the ovals surrounding the sign, are they really ovals or should they be circles?

Regards

[CNCwoodMan]

Hello and thanks!

As for the design in the 39 picture they should be ovals/elipses. I have attached the picutre showing axis directions and I am attaching the GCODE (zipped) for the circle cut outs and for the 39. Appreciate you testing this out for me ! I hope its not the GCODE, I have tried multipel Post Processors and went through the code manually......

Hello.

If possible post both the codes for you initial circles and that of your wooden sign. I´ll try to run them on a machine we are currently retrofitting.

Please repost the picture of your wooden sign drawing arrows showing the axes.

By the way the ovals surrounding the sign, are they really ovals or should they be circles?

Regards

[BBMNet]

Hello.

The fact that your machine does not home correctly at the end of the cycle even when cutting air tells me there is some sort of configuration problem. That is also reinforced by always being the same distance.

Yet it could be possible that there is some odd mechanical problem like a damaged timing belt in at least one section that makes the pulley jump over at least one tooth. Hence the distance error is also always the same.

One thing that puzzles me is that for the ovals the left side appears to coincide at every pass yet they appear elongated on each pass on the right side.

I just downloaded your file. Let me run them and will report back.

In the mean time write if possible a program to exercise the X axis all along it say some 10 times.and then return home. That is just to confirm how the axis moves looking for any anomaly along the path. If there are tag the segments so that the belt can be carefully checked in those sections. Also confirm if the homing distance error keeps being the same as you mentioned.

While I don´t suspect the drives as the source of the error It may be worth to interchange the X & Y units to check if the problem is not transferred to the other axis.

Regards.

[BBMNet]

Hello.

In case you haven´t seen the other post here is what I have found so far.

First I have to warn you about some apparent distortion in the Y axis of the posted pictures as the screen resolution is EGA (640X350) rather than VGA (640X480). From them you will notice that apparently there are no errors in your programs.

However I must pinpoint the fact that I had to resize them to my machine proportions and the circles program had to be shifted so that no negative coordinates are requested.

My program works on all four quadrants (X & Y axes) but the workpieces and hence the screen are kept on quadrant 1.

A note: due to memory limitations in my postprocessor I could not convert the whole sign program but as you can see most of it got converted. I´d say enough to confirm that your program is ok.

It is important that you notice that neither the circles nor the ovals overlap at the left side as in your previous pictures. Yet it is interesing to see that the 39 sign as done on my machine is a mirror image. The reason lies in how the movements of the machine are defined.

The next step is to analyse how your machine behaved when running that test I mentioned before. Was there any anomaly?

To find out if some tests I usually perform can be done on your machine please confirm if your computer has a parallel port. It may be possible that I will suggest to perform some test using the parallel port and a small interface card as shown in the picture.

I have one question. If you mill concentric squares, does the shift show and if it does, would it be the same distance?