Circle not accurate

[rwebcon]

I have a problem which someone may be able to shed some light on.
I’m using a new StapCraft Q404 with a tool changer. Software is UCCNC UC100 motion Controller. I use Aspire 11 with the Stepcraft UCCNC ATC Arcs(mm) port. My issue has to do with cutting a circle 3.2 inch in diameter. Using the circle tool, I create a 3.2 circle and duplicate it having two circles the same size. I add nodes to the second circle in each quarter of the circle. I toolpath a Pocket for both circles and run the job. The problem is the first circle will come out stretched, not a precise circle. The second circle with added nodes cuts a perfect circle. Here is a link showing the first circle with the problem.
https://www.dropbox.com/s/m1xnjvfs115o0 ... 0.mp4?dl=0
I have run Backlash test on both X and Y axis, and I get readings of zero to .001 mm, all most perfect. I have run tests on both X and Y to make sure they both travel at a set distance and physical measure the distance to make sure. I’m working with Stepcraft to try and figure out why this is happening, but to date no definitive answer.
I think the problem is in the “interpolation mode selection” under Configuration, General Settings.
Does anyone have any ideas or suggestions as to why this is happening and how I may go about fixing it? Is there some resource you could suggest?
Thanks, I appreciated any ideas.

[spumco]

Since nobody else has chimed in, can you post the G-code for both tests, as well as your profile?

You may be right about the interpolation mode, but it'd help rule out a program issue if I/we can see what's working for you and what isn't.

[cncdrive]

There is no inerpolation mode which would make a circle look like yours.
It could be 2 things:

1.) The circle in your circle is not a circle but a shape as on your photo.
Or
2.) Mechanical problem of your machine.

[rwebcon]

Attached G Code file. For some reason I can not attach Aspire file???
I can not see how it could be a misalignment. If I cut the Aspire circle using the circle tool an it comes out distorted and take the same circle an add four nodes to it an it cuts perfect how can you
account for this. This is my first experience with UCCNC so I'm a little surprise at the results. I ran an Axion for over 7 years an cut circles all day long with no problem.

Thanks for taking the time to look at this.

[cncdrive]

As I wrote the UCCNC does not make oval circles in any cases, even if the circle is optimised it can come out as a smaller circle or a diagonal, but never oval shape.

[spumco]

Code looks OK. I'm getting normal circles in demo mode.

1. Please post the code which works for you (quadrants?) Need to see if there's something seriously different.

2. Can you post the profile you're using please? It's in the UCCNC/Profiles folder on you computer and is named "xxxx.pro"

[rwebcon]

Here are files requested.
I really appreciate all the help.
Double checked all alignments an all looks good. Marked my spoil board with vertical and horizontal lines so I can place material on board.
They are 47 inches long. They are all perpendicular over the 47 inches within .03. That should be good.

[spumco]

It's hard to determine where the problem is because you've uploaded two different programs. Your CAM (Vectrix) is not outputting complete circles, but is doing arc segments on for both programs.

The other problem is that your machine and UCCNC are set up for metric units and your model is in inches. Vectrix CAM post processor is (automatically) outputting the G-code in metric to be compatible with your machine settings, but it makes deciphering the g-code miserable.

Would you please create a couple simple programs for testing? First - use metric units for your model and for your tool (you won't be cutting, just trying to generate 'clean' g-code that's easier to read). Pick a simple dimension - like 50 or 100mm diameter, and a 10mm dia tool. Center of circle at 0, 0 - that means the 10mm cutter should be at X55 Y0 for a CCW circle (if the start point is at 3 o'clock when viewed from above).

The program should just move to a start point and make a circle. Second program should do the exact same thing, but with the 'nodes' added.

Put a pen in the spindle and paper on your work surface and you should be able to jog down a small amount and touch the pen to the paper. Obviously, don't turn on the spindle. If you can't get a pen in the spindle, then just tape it firmly to the side of the spindle and leave it there for both tests. The programs LOOK fine in the toolpath preview, but we need to see what the machine is actually doing - especially when there's no cutting forces on it.

The intent is to have the pen trace the program (circles) on the paper. This should simplify the program to the point where we can identify if this is a program issue or a UCCNC settings issue.

So, step one is to create simple metric-unit model and have CAM output g-code that's easy to read, and run the test and see what happens.

Step two will be to hand-write a simple G-code file - removing CAM from the variables - using G3 arc moves.

Step three will be to do the same as above, but turning off Constant Velocity mode in UCCNC to remove that variable.

NOTE - Your profile has very a tight linear addition length (used for constant velocity mode), but that shouldn't matter for arcs, and both your 'working' and 'not-working' programs are using the same Constant Velocity mode & settings.

Good luck & report back when you've had a chance to draw some lines and see what happens.

***EDIT*** - maybe you should do the hand-coding first. Four G3 moves (circle quadrants) using the pen and paper to ensure your machine isn't flexing or doing something strange.

[rwebcon]

Converted Aspire to metric for both tool and vector. Files are much smaller now. Just did a profile cut. Running the test with pen maybe a little more difficult trying to attach pen.