Art Plotter

[Fozzie55]

Hi

This is a 'need help' thread but I thought I would show it as a new machine build.

I have built a 1 metre square plotter for drawing on canvas with marker pens. My first attempt, I have used routoutcnc software and driver boards.

My problem I think is in 'backlash'. If you look at the photos, when I draw a sample grid of squares you can see from the top left one which is perfectly alligned down to the bottom right one which is off considerably.

In the software I am using there is a box to input a backlash compensation figure. When I put a positive number in, it magnifies the error, so I presumed a negative number would do it but the software won't let me input a negative.

Do you helpful people know if this is the problem (software) or have I got something wrong in the build? e.g Toothed belts being pushed or pulled?

Thanks in advance.

This is a link to a Youtube Video of it working... "http://http://uk.youtube.com/watch?v=Ji65HalG0D4"]http://http://uk.youtube.com/watch?v=Ji65HalG0D4

[ger21]

Backlash is not cumulative. If you're starting at the top left, and drawing the squares from left to right, one row at a time, I'd guess you're loosing steps when traveling back to start the next row. If it's a G0 move, try it with a much slower G1 move and see if it still does it. Or just cut your rapid speed in half and try it.

[Fozzie55]

Thanks for the suggestion Gerry. I have tried it now very very sloooooow and it plots identical I am afraid.

Now I don't know about this really but I imagine that if it plotted a line 100mm long and on returning to the beginning of the line it lost 1mm in slack, so back 99mm then drew another 100mm line it would be 1 mm out. If it repeated this move it would add 1mm each time it did it, no?

[ger21]

Thanks for the suggestion Gerry. I have tried it now very very sloooooow and it plots identical I am afraid.

Now I don't know about this really but I imagine that if it plotted a line 100mm long and on returning to the beginning of the line it lost 1mm in slack, so back 99mm then drew another 100mm line it would be 1 mm out. If it repeated this move it would add 1mm each time it did it, no?

It can't lose 1mm each time unless something mechanical is loose and is actually moving each time. Make sure you're pulleys are tight and that they are not spinning.

[ger21]

One question. Did you draw the grid first, and all the squares last? I'm assuming that you did.

[Fozzie55]

I was just using 1mm and 100mm as an example for the theory of backlash and compensation. 500m losing 0.01mm everytime, wouldn't they still add up or am I being stupid

Yes the grid first which was more or less accurate then 77 squares. The plot in the video was done big and red just for the video. I have done some A3 smaller plots with a fine pen to test.

Cheers

[Khalid]

Please check for your step/unit in Mach3 or whatever else...

[ger21]

Think of a nut on a screw that has 2 turns per inch and 1/2 turn of backlash. With the backlash taken up, and the axis at zero, make 4 turns to the right, and it moves 2 inches. Now make 4 turns to the left, and it only moves 1-3/4"because the first 1/2 turn takes up the backlash. Now turn 4 turns to the right again and it again 2" from the zero point. It still takes up the backlash and only moves 1-3/4", but it was already at the 1/4" mark because it doesn't return to where it started. Does that make sense? If not, how about this.

Same scenario as above. Tell the machine to move 2" to one way, and 2" back. Do this 10,000 or even 1,000,000 times. If the error was cumulative like you think, sooner or later the machine would eventually crash into the other end as the error accumulated.

[Fozzie55]

The software I am using is not G Code based. It has it's own FL codes, MoveXY, MoveZ etc. http://www.routoutcnc.com

Thanks again.

[Fozzie55]

Think of a nut on a screw that has 2 turns per inch and 1/2 turn of backlash. With the backlash taken up, and the axis at zero, make 4 turns to the right, and it moves 2 inches. Now make 4 turns to the left, and it only moves 1-3/4"because the first 1/2 turn takes up the backlash. Now turn 4 turns to the right again and it again 2" from the zero point. It still takes up the backlash and only moves 1-3/4", but it was already at the 1/4" mark because it doesn't return to where it started. Does that make sense? If not, how about this.

Same scenario as above. Tell the machine to move 2" to one way, and 2" back. Do this 10,000 or even 1,000,000 times. If the error was cumulative like you think, sooner or later the machine would eventually crash into the other end as the error accumulated.

Thank you Gerry, I understand now, backlash either way. So not backlash, not speed, could be the software? or Kahlid's suggestion?

[Fozzie55]

Thanks again for the explanation of backlash, silly me

I have tried all the parameters in the software, slowed it down, changed accelaration, points / mm etc. but it still drifts off the same amount, about 14mm X and Y after the whole plot (and how little the Z axis moves that drifts too)

It is working great apart from this major point.

Could it be the stepper motors? I have tried them really slow so they won't be missing steps and it is still the same.

Any more ideas please?

[Fozzie55]

I have tried turning microstepping off and do full steps (thinking the motors are not as accurate doing microsteps) and it's still the same...