Will Behind Tape Reader fix jerky motion caused by 3D tool paths?

[gmoushon]

Sorry if this has already been covered. I know I'm not the first with this issue, but I'm having a very difficult time finding the answer.

I'm struggling to eliminate, (or significantly reduce), the jerky motion caused when drip feeding a 3D tool path to a 1995 MyCenter-0 with a Yasnac MX-3 control. I'm using DNC4U software at 9600 baud which is the highest baud rate the machine is capable of, and I'm sending the data in 10 byte chunks.

It was suggested that a BTR might be the only solution. The tool paths are being generated by Fusion 360.

From my limited understanding, this jerky motion is the result of the many small liner moves generated by the CAM as well as the limited look-ahead buffer size of the CNC. If that is the case, I'm confused as to how a BTR will help.

Will a BTR get the data to the control faster than the RS232? And, if it does, will that be enough to overcome the acceleration/deceleration issues of the servos?

I'm not opposed to buying a BTR, I just don't want to throw money at something that will have little or no effect.

Maybe I'm asking too much for a 1995 vintage machine.

Thanks for any feedback.
gm

 

[5 axis Fidia guy]

We dealt with this on our older Kuraki, the only thing we did to help was to force arcs to all toolpath segments, get as many I's and J's in there as you can. If you are doing complicated machining, I think you will be in trouble.

 

[werkzeug]

i once ran a 88 supermax cnc. it was boggy and jerky like yours. memory too small to load anything in there, and buffer so small it was always jerky, either waiting for data or data overflow of the buffer.a balance act between how fast you are machining/executing the code lines and getting more flowing into your buffer.i had a hassle just 2D ing paths. Doing 3D paths sounds more painful.i dont think you can help that situation. sorry.old slow machine.

 

[Mud]

Will a BTR get the data to the control faster than the RS232? And, if it does, will that be enough to overcome the acceleration/deceleration issues of the servos?

Probably not, and it very well might make it worse. The BTR communicates with the control at the speed of the tape interface which is probably slower than 9600 baud. Someone else here might know what it is. I used to use one and had to slow my engraving to 8ipm when going through the BTR.

 

[gmoushon]

Thanks for the reply Mud...was hoping you would chime in.

I guess another question would be: Would this 1995 vintage machine handle a complicated 3D or engraving tool path if the program would fit in its on-board memory? If not, then maybe I just have to limit it to basic 2D machining and simple tool paths.

gm

 

[Mud]

Thanks for the reply Mud...was hoping you would chime in.

I guess another question would be: Would this 1995 vintage machine handle a complicated 3D or engraving tool path if the program would fit in its on-board memory?

gm

Will it cut it? Yes, even a '70s tape machine will cut 3D. How fast it will cut and how much of the program will fit in memory are questions for the control maker or someone who knows that control very well, that's above my pay grade.

For comparison, I've seen a 1989 Dynapath Delta 20 cut small complex engraving toolpaths on a compound curve at 32 IPM, the machine was way too heavy to bash around like that but the control didn't break a sweat.

 

[Bill in PA]

I am kind of surprised you are having jerky movements. I drip feed a 1989 Fagor 8020 with 32K when 3D surfacing (at 9600 baud). It runs smooth as long as I am running G5 (round corner). G7 (square corner) pauses at the end of each move.

If all your code is straight lines, then the control does not need to look too far ahead anyway. Is the resolution of the code generated too fine? Are the straight lines overly short?

If your control already has an RS232 port, then a BTR would be redundant. I did not know that a 1995 machine would still come with a tape reader.

The BTR on my lathe is 2400 baud, but there are ones available now that are much faster.

Bill

 

[gmoushon]

I am kind of surprised you are having jerky movements. I drip feed a 1989 Fagor 8020 with 32K when 3D surfacing (at 9600 baud). It runs smooth as long as I am running G5 (round corner). G7 (square corner) pauses at the end of each move.

If all your code is straight lines, then the control does not need to look too far ahead anyway. Is the resolution of the code generated too fine? Are the straight lines overly short?
Bill

The jerky motion happens most often when I'm doing an engraving routine, (although it is present in other routines to a lesser extent.) The code that Fusion 360 is producing has sections with a TON of straight short lines. I've improved things significantly by adjusting the Tolerance and Smoothing settings, but I still am struggling with it.

gm

 

[Bill in PA]

If you take a section of code that will fit within the machines memory and run it, how does it do?

Bill

 

[gmoushon]

If you take a section of code that will fit within the machines memory and run it, how does it do?

Bill

Just loaded as much of the engraving portion that would fit and it acted very similar. Maybe it seemed a little smoother, but it wasn't very different.

 

[gmoushon]

If your machine can do 3D contours native to the control,

Not sure what you mean about "3D contours native". It can do all the basic CNC manipulations of the axes for instance, (G17, G18, G19), but it only does arcs, not splines.

That seems to be the source of another of the frustrations I'm experiencing: I do my modeling in Solidworks then import it into Fusion for the CAM. It looks like Fusion is seeing some of the arcs I've created in Solidworks as splines and I don't know why. Specifically the engraving. I purposely modeled all the engraving as arcs but Fusion doesn't see it that way.

I know this is another issue I have to work through, but for now, I'm still trying to fully understand what I can expect from this machine.

gm

 

[Mud]

Just loaded as much of the engraving portion that would fit and it acted very similar. Maybe it seemed a little smoother, but it wasn't very different.

That sounds like you are just exceeding the ability of the control to process the blocks of code fast enough.

 

[gmoushon]

Yeah...that's how its looking to me.

 

[Mud]

If you slow down th efeed with the override knob does it smooth out?
How fast is the programmed feedrate?

 

[gmoushon]

Override does not make any difference and the feedrate is 15 ipm.

 

[gmoushon]

can it do helical interpolation?

can it do arcs?

ISTR there are questions for both in the post setup in fusion

If you do not answer them it will turn a helical entry ramp from say 6 lines to 100


I managed to get a 2 1/2 axis multi contour part for my Heidenhain into less than 200 blocks.

Yes. This machine is meant to be a drill and tap center. It does helical interpolation and I rigid tap on it all the time.

I was not aware of that setting in Fusion but it makes perfect sense! If I do not cut the Tolerance and Smoothing way back, it shows a lot of points on the helical ramping portions of the tool path.

I'll look for those settings. Any idea where they're found?

Thanks!!

 

[BugRobotics]

+1 for gustafson's posts

Yes. This machine is meant to be a drill and tap center. It does helical interpolation and I rigid tap on it all the time.

I was not aware of that setting in Fusion but it makes perfect sense! If I do not cut the Tolerance and Smoothing way back, it shows a lot of points on the helical ramping portions of the tool path.

I'll look for those settings. Any idea where they're found?

I have included a screenshot of the post processor output window and highlighted the "allowHelicalMoves" property. These properties are defined in your post processor and if you don't see these then you may have to edit the post.

To give you an idea of what that property does for a simple helical boring op...
With it set to "Yes" my code is 49 lines and uses Heidenhain's helical interpolation cycle
With it set to "No" the same operation is 3600 lines and uses linear moves in X,Y and Z

 

[gmoushon]

Thanks. I found that setting in mine and it was already set to "Yes".

Maybe the bigger question is how do I reduce the number of points in the CAM output before it gets to the post-processor? That is where I seem to have the biggest issues. I guess I'll just have to keep playing with the Tolerance and Smoothing to minimize as much of these points as possible.

I still don't get why Fusion is treating geometry as a spline when I purposely drew as a pure arc. Maybe its because I'm drawing in Solidworks and doing the CAM process in Fusion.

gm

 

[BugRobotics]

Hey gm,
I also do all of my modeling in SolidWorks and import to Fusion for toolpaths. I don't recall ever having that issue. Can you share an example so I can test?

Here is some info regarding smoothing: Solved: Understanding Smoothing - Autodesk Community

 

[Mud]

Override does not make any difference and the feedrate is 15 ipm.

If you can turn the feed back significantly (50% or more) and the jerking does not change, I'm not convinced it's processor speed. Is there a parameter for absolute stop that can be set, or something like that that's causing it to halt unnecessarily along the toolpath? Is there a part of the toolpath where it's more noticeable?