Creating a large scale plotter using string

[cadcoke5]

I am hoping to design a large scale pen plotter. This is for a real-world application, where I want to print out some large patterns for kites I want to make. Since I have a Mindstorms EV3 kit, I though that would be the way to go.

Initially, I was going to make a type of plotter I have heard called a "Polargraph". Basically, it suspends a pen holder by two strings connected to winches. Those strings are wound/unwound to move the pen, and can create drawings on a wall. Here is a link to someone who created one, and even a supplemental video explaining the math.

Lego robot Spider Pen

Here is a forum dedicated to the Polargraph concept.

www.polargraph.co.uk/forum/

However, I came across another concept that I think will be better plotter style, called a CoreXY. The reason is that the CoreXY design is much more rigid, and will work on the floor. Which, for the projects I am currently working on, is a better arrangement. Here is a link to that design. I will probably change it so that there are two rigid pipes. They already have that horizontal one, but I will put one on the left vertical side as well, to keep that axis rigid. Also, the pen holder will hold everything; the Mindstorm "brain", the two main motors, and the pen holder, so that I don't have to run wires anywhere. The only thing outside of that will be cords, pulleys, and the rods. So, the motors in the theory link are simply replaced with pulleys.The pulleys on the other corners can simply be taped to the floor.

http://corexy.com/theory.html

For the moment I am focused on the physical design, mostly to determine if my idea is workable. I am currently stuck on one aspect.

Because of the quantity of line, up to 12ft, I doubt I can just use a winch system. I am concerned that the string i use would tend to bury itself in lower layers of string. So, instead of the winch, the far ends will be tied off, and the print head moved by using friction to pull on the strings. But, I really don't get enough friction.

Perhaps I should abandon that idea, and go back to using a winch system. Each motor would need a pair of spools, for a total of 4 spools. I had considered using plastic sewing bobbins for the various pulleys in the system, and they are wide enough to use gift wrap ribbon. I might have used them for the take-up spools, but I doubt they would hold enough length.

Any ideas about ways to increase friction if I decide to use a friction wheels at the motors? Ideally I would stick to using what is in a Mindstorms standard kit, so that others can easily copy my design. But, I am open to other possibilities. Perhaps there is some sort of string that has better friction? Paint rubber cement on the string, and/or the pulleys?

Obviously the software will eventually be needed. All the Lego plotters I have seen so far, that have the X and Y axis movements separated have failed to publish their software where I can find it. But, it may be within my capability to do. I imagine I would write it to accept Gcode, and just translate that into motor movements.

Any suggestion?

-Joe

[JopieK]

You'd better use timing belts. I have experience with it because of helping young high school students with some projects. If you use strings the pen will easily wiggle a little bit.

[allanp]

You could try wrapping the string 5 times round the friction drive roller before it leaves to increase the surface area of string in contact with the roller. Increasing the diameter of the roller will also have the same effect, and make sure you have a tensioner. Another method would be to pinch the string between 2 rollers, like in a mig welder. I think fishing fishing line would work quite well.

Edited September 27, 2015 by allanp

[cadcoke5]

Thanks for the replies. A timing belt is out of the question, because of the lengths needed. The cheapest I came across was $2.61 USD/ft.

I had been using an idler pulley to keep the string wrapped about 270 deg around the drive pulley. But, I did not realize that multiple wraps were possible.

I will try both the multi-wrap and pinch-wheel ideas as my next test.

-Joe

[cadcoke5]

In my Minstorms kit, have not found a good combination of wheel with rubber, and a groove that will successfully apply friction to a string. The thin wheels have rubber that is too wide to go into the grooves on those wheels, or on the groove on the edges tank tread wheels.

I just had another idea for rubberizing the pulleys. After failing to find a wide rubber band, with a small diameter, I though of using balloons. If they are cut into a loop, then I have a good variety of sizes to choose from. Certainly not very durable, but very cheap to use and easy to replace.

-Joe

[Saberwing40k]

I think you could use Lego link tracks instead of timing belts. The larger ones are quite durable, and inexpensive.

[quilkin]

  Quote

After failing to find a wide rubber band, with a small diameter

Would a bicycle inner tube (a thin one, racing bike not mountain bike), cut into sections, help with this?

[cadcoke5]

  On 9/27/2015 at 6:45 PM, Saberwing40k said:

I think you could use Lego link tracks instead of timing belts.

I am not certain what link tracks are. Is this a type of caterpillar tracks? If so, remember that I need about 24ft of cord, so if this is a substitute for cord, it would probably not be viable because of cost.

-Joe

[Saberwing40k]

What dimensions are you using? Since this is a practical project, you might want to make a robot that just drives around. Several have been shown on this site and others, which would eliminate several problems. Also, yes, link tracks are a type of Lego caterpillar tracks.

[cadcoke5]

The overall size will be 8ft x 4ft. A drive-around printer is not viable This is because when using odometry (i.e. keeping track of wheel rotation), the wheels invariably slip a bit, and the robot gets further off the intended path the more it drives. It requires either a way to know where it is, or a mechanical way to tie it to the outside world. in my case, the pulleys at the 4 outside corners are its anchors to the world.

-Joe

[Moz]

Have you tried using tyres for the friction rollers instead of plastic drums?

You might also be able to use string for positioning and wheels on the ground for drive - doesn't Mindstorms have rotation counters? Again, use tyres for better friction and calculate your position that way.

Finally, I'd be very tempted to hang the thing from a ceiling or frame so you're using cheap string for the structure instead of expensive Lego. I've only built CNC mahines but we gained a lot of precision on a bit router by making our effective beam depth 1.2m instead of the 100mm extrusions everyone else uses. Mechanically somewhat more complex, but very effective (in our case a carbon fibre tubing frame attached to the sliding cross beam - it only weighed about 1kg)

[cadcoke5]

In the Mindstorms set, there are two choices of wheels that are not the very thin ones. The smaller version has no way to drive the wheel, so it can only be used as a free-wheeling wheel. I looked again at the larger rubber tires. They have some parallel grooves, that could be used to run the string, and perhaps the small pulleys can be used to keep the thread aligned with the groove. Alternatively, I might drill some holes through some bricks, to make a more constrained hole for the string guide.

Yes, the Minstorms motors have encoders, which make them ideal for this sort of application.

The reason I went away from hanging from the wall, is that hanging the material that I draw on starts to become an issue. Gravity also helps with the force to apply pressure with the marker.

However, I did think of an approach to the Polargraph concept that would make it more accurate. The idea is to put tensioning spools on the bottom, that pair up with the strings that hang from top. The idea is to keep tension on the line, to reduce the wiggling. But, the geometry of the polargraph really requires that the string hang at a minimal angle. Otherwise the forces on the motor are greatly magnified. If the angle creates too little tension, that can create a problem as well. Thus, the total area of the polargraph for my 8'x4' drawing area, would expand to at least 12' x 8'.

I would also like to speak to someone who is familiar with the various softwares that run the gantry type of plotters. I imagine I can use one of them with no modification, as long as it can handle the large number of rotations my very large plotter will use.

I will keep you all updated on any progress.

-Joe

Edited September 28, 2015 by cadcoke5