[Alexa Flame]

Dear Tom,

Saw close topic, but not the same problem, so opened a new one.

I need to flatten the space for PCB engraving. I scan the physical PCB surface and fill the results to Geo Correction Table. The table's offset is set to 0. The machine's coords are also set to 0 when I apply the table and try to use it. It looks like only 1st cell of the grid is used and interpolated in all direction but not the entire table... I did one simple experiment: I set my surface as "V"-shape, with X=0, Y=0 at one of the egdes, while Z falls from 0 to -5mm, then rise up to 0 again. Then when I manually or by G-code drive the machine from one toward the other edge of "V", Z automatically goes down... and only down, despite the slop should go up again.

If you need more details, I'll give you. Perhaps I don't know somthing or there is real problem in geo correction?

Regards,
Alexa Flame

[TomKerekes]

Hi Alexa,

What is the range of travel that you wish to correct over? What is your Grid size? What units are you using? (The Geo Table is always in Inches). How much does Z change (as observed on the KMotion Axis Screen) for each inch of x motion. Please Post the table you are trying to use. Are your CAD Coordinates such that (x,y) = (0,0) is at the bottom left corner of travel?

Regards

[Alexa Flame]

Dear Tom,

My test grids were not big, like 5*8 or even 2*4.
But my assumption was it works in the current unit system. I worked in millimeters...
Yes, initial point of the table was also 0,0 of my PCB. Z-travel was big for clear testing (0 to 10 mm) but real expected deviation should be within 1 mm and less.

Here is one of my tests with "bridge" shape (I expect it is along Y axis). For convenience, I always shift the whole table by Z such that Z=0 at X=Y=0. So, (0,0) becomes my reference point, but I don't think it is really needed.

6, 2
5.0000, 4.0000
0.0000, 0.0000

0, 0, 0.0000, 0.0000, 0
0, 1, 5.0000, 0.0000, 0

1, 0, 0.0000, 4.0000, 4
1, 1, 5.0000, 4.0000, 4

2, 0, 0.0000, 8.0000, 5
2, 1, 5.0000, 8.0000, 5

3, 0, 0.0000, 12.0000, 5
3, 1, 5.0000, 12.0000, 5

4, 0, 0.0000, 16.0000, 4
4, 1, 5.0000, 16.0000, 4

5, 0, 0.0000, 20.0000, 1
5, 1, 5.0000, 20.0000, 1

Thank you.

Regards,
Alexa

[Alexa Flame]

Dear Tom,

I've done the test in inches. It works. Z axis follos this curvy space! Your idea that problem can come from unit system was correct. Thank you. The problem is solved.

4, 2
0.2, 0.2
0.0, 0.0

0,0, 0.0000, 0.0000, 0.2
0,1, 0.2000, 0.0000, 0.4

1,0, 0.0000, 0.2000, 0.4000
1,1, 0.2000, 0.2000, 0.2000

2,0, 0.0000, 0.4000, 0.2000
2,1, 0.2000, 0.4000, 0.4000

3,0, 0.0000, 0.6000, 0.4000
3,1, 0.2000, 0.6000, 0.2000

Geo Correction Table must be in inches!

Regards,
Alexa

[TomKerekes]

Thanks for posting back Alexa

[bhurts]

I know this isn't my thread but for the sake of easy searching later I'll ask here.

I am interested in geo correction but have not tried it in any way.
What I'm wondering is can it be used to correct for out of square using a minimum amount of points without affecting the travel distances significantly.
By this I mean by using just 4 points. One at each end of travel in x and y. I'm guessing you could use just 4 points to adjust for sqareness of travel? Let me know if this makes any since. And just for my awareness do you know of a place where a accurate grid can be purchased to do a full geo correction across the entire table spaced in 1 inch cubes.

Thanks
Ben

[TomKerekes]

Hi Ben,

Yes the Geo Correction Table can be just 4 points.

Attachment 255820

4 points allows for correction of rotation, non-orthogonality, trapezoidal, scale. See here for an example:

http://www.dynomotion.com/Help/KMoti...Correction.htm.

No I don't know of where an accurate Grid can be purchased. There are lots of places that can print patterns on Mylar Sheets with accuracies on the order of ~ 0.001inch which I think are used for PCB manufacturing. I think printing a 24inch x 24 inch pattern of your choice might be ~ $100.

I've heard of some users stretching piano wire to somehow set straightness of travel.

[bhurts]

Thanks Tom great news.

Ben