[725franky]

Hi,

I have a 2019 Taig ER16 mill that I CNC'ed with a Xylotex Stepper/Driver kit. I am, by trade, a CAD Technician. Other than what I have recently (last three months, on the weekends) learned I don't know anything about CNC or Machining in general. So, I need some help.

I am working on making a part that has several rounded vertical surfaces. Like a square with rounded corners and cylindrical protrusions. I have noticed that my mill is making a flat edge on a round cylinder. The last part I made has a .35" Dia. cylinder toward each end of it, which are 2.75" (in the X) apart. The one cylinder is coming out with a flat spot while the other seems to be coming out fine. I am using Gsimple for my G-code and it is using G3 codes to cut the cylindrical profiles. What could be going wrong? Is it me or is something not right with my mill?

I have attached some pics of the cad files of two parts that I made that had this problem.

Thanks for the help,

Frank

[djh82uk]

not sure exactly but sounds like a software issue to me

DJH

[725franky]

Do you mean software as in the controller software? I use EMC2.

[djh82uk]

more likley the Gcode generation i would have thought.

Basically something is telling the machine to do what it is doing

DJH

[725franky]

In EMC when a program is opened all of the tool paths can be seen. When one portion of a tool path is clicked on (selected / highlighted) the line of g-code is shown for that particular portion. I did click on the tool paths on the D_4 part and the 180 deg. radius was completed using two G3 codes so I don't think that it is the G-code. I may try writing the G-code by hand just to make sure; I just don't really want to mill another piece because I don't have that much aluminum to work with. Thanks for the suggestion; I may try writing it by hand just to make sure.

-Frank

[djh82uk]

Heya

I had problems with circles, they kept coming out as squares with rounded edges, it turned out to be that my cad and Cam programs were not compatible, changed to another cad/cam and works great.

unless there is a strange problem in emc2 i don't know.

The machien just does what it's told, unless you have amazingly bad backlash or something

DJH

[project5k]

i wouldnt think that it would be back lash, or all of the round pegs would have flattend sides, not just that one... i mean i guess it might be possible that one part of the screw is looser in the nut that the other, but seems unlikely to me..

[Stepper Monkey]

Remember that in most toolpath generation, it will fail safe and not allow you to cut a shape that would cause the bit to "crash" into other areas of the part or to otherwise incidentally remove any area not specified to be removed. For instance, it will not cut all the way around a complete circle spaced 1/8" away from a wall if you are using a 1/4" bit - this would cause unwanted removal of wall area in order to complete that circle. This is far different from hand coding where commands are absolute. Not being able to think for you, the program will wuss out and take the safe way out almost everytime.
Also remember that the program doesn't know what is beyond the boundaries of its work area. If it won't allow the bit past that boundary, and your work it defined right up to the boundary, it can't be cut. Hence the shape you wound up with.
Try putting your part in a work area defined somewhat larger and see what happens.

[725franky]

I am having difficulties with my controller and once I get those sorted out I will write the G-code by hand and see what happens. It appears that my Xylotex controller has died so until I get it fixed no more milling for me.

Thanks to all for the suggestions!

-Frank

[neilw20]

i wouldnt think that it would be back lash, or all of the round pegs would have flattend sides, not just that one... i mean i guess it might be possible that one part of the screw is looser in the nut that the other, but seems unlikely to me..

If you have backlash compensation turned on, then when each an axis reverses during a circle, it becomes the slowest axis (it is stopped) so this can cause the cutter to pause while the screw reverses.
I leave the back lash compensation turned turned off. Usually the small error it is intended to compensate for can cause not so small features to appear.
I only turn on if I need some really precision position for drilling, etc.

[725franky]

I do have backlash compensation turned on. And I wondered if that might be the problem. But If I use no backlash compensation wont my parts be off by quite a bit? My backlash on the axis in question, the X axis, is .0055" wouldn't that add up to be substantial? how much backlash is it ok to just leave?

Thanks for the suggestion.

-Frank

[neilw20]

I do have backlash compensation turned on. And I wondered if that might be the problem. But If I use no backlash compensation wont my parts be off by quite a bit? My backlash on the axis in question, the X axis, is .0055" wouldn't that add up to be substantial? how much backlash is it ok to just leave?

Thanks for the suggestion.

-Frank

I think that is the problem. Do you get the flat at change of x direction?
You need spring loaded nuts or similar. .0055 is too much.

[725franky]

Where can I get spring loaded nuts for a Taig mill? Yes the flat spot begins at the change in the x direction. I know that .0055 is too much and that is why I had the backlash compensation turned on....I thought that it would keep that from happening. I can tighten my lead nut (it's a split one) I will try that for now but I would like to have something different to get rid of my backlash.

Thanks,

-Frank

[Stepper Monkey]

You do need to fix the enormous backlash as well, but I don't think your problem with the part in question is backlash. For one its too severe, and second it only happens on some areas of the part and not others. I still think it is a toolpathing issue in the program or something similar.

As for backlash, backlash compensation actually only corrects for some things, and it slows the machine down a LOT. There are some other nasty side effects as well. It is definitely best to run the machine without backlash comp, get the actual backlash of the machine as low as possible, and just live with what is left.
.0055 is way too much. You should easily be able to get a Taig down to a very liveable .002 or so with no trouble. If you need it much more accurate than that, a Taig likely isn't your answer and you need a lot more expensive of a solution than hobbyist gear!

[neilw20]

You do need to fix the enormous backlash as well, but I don't think your problem with the part in question is backlash. For one its too severe, and second it only happens on some areas of the part and not others. I still think it is a toolpathing issue in the program or something similar.

As for backlash, backlash compensation actually only corrects for some things, and it slows the machine down a LOT. There are some other nasty side effects as well. It is definitely best to run the machine without backlash comp, get the actual backlash of the machine as low as possible, and just live with what is left.
.0055 is way too much. You should easily be able to get a Taig down to a very liveable .002 or so with no trouble. If you need it much more accurate than that, a Taig likely isn't your answer and you need a lot more expensive of a solution than hobbyist gear!

Next step up (in my opinion. I am biased. I have one)
X and Y very accurate.
For the price, you can live with Z performance with careful programming
to allow for dynamics of column slide. I can expalin if you wish.
I got mine down to below 0.001" with a bit of care.
SIEG Super X3 CNC Mill.
http://www.syil.com.au/product_X3.php

[725franky]

I wish I could have afforded a CNC'ed X3 or an Industry Hobbies CNC! But that was not the case so I must work with what I have. Does anyone know what I can do to remove the backlash other than just tightening the lead nuts? Like a Zero backlash upgrade for the nuts or something? I guess the next part I make I will have to try using no backlash compensation and I will hand write my G-Code.

But Before I can make any more parts my controller has to be fixed...sending it in today (as long as nothing urgent comes up).

Thanks for all the help guys. I'll let you all know what happens when I get mill back up and running.

Thanks,

-Frank

[under-dog]

Remember that in most toolpath generation, it will fail safe and not allow you to cut a shape that would cause the bit to "crash" into other areas of the part or to otherwise incidentally remove any area not specified to be removed. For instance, it will not cut all the way around a complete circle spaced 1/8" away from a wall if you are using a 1/4" bit - this would cause unwanted removal of wall area in order to complete that circle. This is far different from hand coding where commands are absolute. Not being able to think for you, the program will wuss out and take the safe way out almost everytime.
Also remember that the program doesn't know what is beyond the boundaries of its work area. If it won't allow the bit past that boundary, and your work it defined right up to the boundary, it can't be cut. Hence the shape you wound up with.
Try putting your part in a work area defined somewhat larger and see what happens.

True but from what I have seen that typically leaves an unmilled area of material between the two "close proximity" geometries. This results in additional buffer goemetry being left in between your desired geometry. From the sounds the mill is fully cutting out the geometry but it is removing additional material and leaving a flat spot.

[under-dog]

Were the cylinders in the CAD files created with ellipses or a circle. I guess this could depend upon the program as well too.

I know i have had issues when importing certain geometries into DeskCNC(the cam I use).

I use Solid Works for design

Ellipses can cause issues especially if they have been offset. I remember when I used to use ACAD. There were two trypes of ellipses. One worked well for toolpaths and the other did not. True ellipses and polyline ellipses.

So the way your CAD files were created can have an effect on what the gcode converter "sees" when it is imported.

[725franky]

The Circles in the CAD files were circles not ellipses. I use Alibre Xpress for my CAD. However I was using Gsimple for the g-code that I used and my CAD files were not used.

I am still waiting on my controller. I sent it in but have yet to hear from them. Once I get it back I am going to try writing my g code by hand and re-running the parts.

Under-dog,

What does your Taig have for backlash? Do you just leave it or do you compensate for it?