opensource cnc edge bender project

For discussions related to designing and making ski/snowboard-building equipment, such as presses, core profilers, edge benders, etc.

Moderators: Head Monkey, kelvin, bigKam, skidesmond, chrismp

ben_mtl
Posts: 583
Joined: Mon Mar 31, 2008 1:47 pm
Location: Sherbrooke, Quebec
Contact:

opensource cnc edge bender project

Post by ben_mtl »

Your help is needed here, thanks for contributing !

I've been thinking about a CNC edge bender for a while, I do have an idea on what I would do but I believe this could be taken to an all new level by making it a community project rather than a personal thing.

I will be sharing some ideas here, I just need some time to make a few sketches and try to explain how I envision the "software" part.
Of course those would just be my initial ideas, criticism (if constructive) and new/different ideas will always be welcome.

The way I see it, the main goal would be to be able to make machine for a couple hundred dollars. No need for fancy controllers or super powerful steppers. I'm convinced a well tought-out machine can be made for cheap.

Stay tuned...
A bad day skiing is always better than a good one at work...
ben_mtl
Posts: 583
Joined: Mon Mar 31, 2008 1:47 pm
Location: Sherbrooke, Quebec
Contact:

Post by ben_mtl »

OK here's a start :

Here's an Excel file I made :


Basically my idea is to feed the machine with 2 info (2 axis), the X axis would be the position along the edge, Y axis would be the radius the edge has to be bent to.

So in order to do that I used my CAD program, took a basic ski outline I had on hand, only kept one half of the ski (for test purpose) and put a point every 5 millimeters along the line.
A small script in my CAD program took care of exporting each point coordinates to an Excel file.

Then comes the Excel file whose link is at the beginning :
I pasted all my points coordinates in this file and what it does is it calculates at each point the radius of the edge.
The first and last point are special : the 1st point gets the radius of the 2nd one, and the last point gets the radius of the one before...

The 2 green columns give me the 2 coordinates I'm gonna need to run the machine : The distance on edge (X) and the radius at this location (Y).

The data for (X) can be used as is, the data for (Y) is not but it's a begining !

A few ideas :
- radii greater than 8 meters could be ignored, basically at that radius you don't need to pre-bend the edge anyways...
- The excel file doesn't specify the direction of the radius : it works OK to make radii in one direction (a circle for example) but it I want to make curves in both directions it's not gonna work (an "S" for example... or a ski shape if you include the traditional long non-reverse sidecut between the tips)... that's why I think we should start to just ignore long sidecut sections for now and concentrate on bending the tips (which is where the benefit is bigger anyways...)

For the machine itself : a 3 rollers system could work, with additional rubber rollers for the feeding maybe ?
My idea would be to run some tests in order to know which position of the rollers makes what radius : you put the machine in a position, run enough edge through it to make a circle then measure the radius... then you change the rollers setting, make another circle and measure.. and repeat for a few settings.
With that data we should be able to see how we could manipulate the radius info from the excel file so the machine actually bends to that radius...

OK I need to work now :p (my "real" day job... :/)
A bad day skiing is always better than a good one at work...
User avatar
skimann20
Posts: 345
Joined: Wed Feb 27, 2008 8:28 pm
Location: Taxachusets

Post by skimann20 »

their video:


did anyone else notice the size of the roach crawling on the floor. sorry my ADD kicked in while he was talking.

Ben, you logic seems to be correct.

the only issue I see with ignoring the side cut is the overall length maybe effected when the curve is not taken into consideration. so the total edge may be "short".
ben_mtl
Posts: 583
Joined: Mon Mar 31, 2008 1:47 pm
Location: Sherbrooke, Quebec
Contact:

Post by ben_mtl »

the only issue I see with ignoring the side cut is the overall length maybe effected when the curve is not taken into consideration. so the total edge may be "short"
That's why I'm measuring my length along the edge (curve) and not just a straight line between the tips, the sidecut won't be bent but the length will be there.

Every dimension on X will come from a measure of the length of an "arc", not the length of the chord.

I might have found a solution to consider "inverted radii", won't need too much work to make it work but I still think it's useless to try to bend to a 10m radius... the material is flexible enough to do it on the go when gluing.
A bad day skiing is always better than a good one at work...
twizzstyle
Posts: 2204
Joined: Tue Mar 07, 2006 8:25 pm
Location: Kenmore, Wa USA

Post by twizzstyle »

A big worry I've always had with this is slippage in the rollers. With a totally open-loop controller, if the edge slips at all it won't know, and you're hooped.

A closed-loop controller would guarantee success, you could use an optical sensor (like from a computer mouse)... but complexity goes WAY up.
ben_mtl
Posts: 583
Joined: Mon Mar 31, 2008 1:47 pm
Location: Sherbrooke, Quebec
Contact:

Post by ben_mtl »

Yes that's a concern I have too.
I think one way to greatly reduce the chance of slippage is to have a good "drive" : rubber rollers one each side of the edge material (top and bottom for more surface contact) + driven "bending" rollers should help a lot.
A bad day skiing is always better than a good one at work...
User avatar
MontuckyMadman
Posts: 2395
Joined: Fri Jun 20, 2008 9:41 pm

Post by MontuckyMadman »

I think that was a cricket or a beetle. We don't have roaches I the arid intermountain west.
Coulda beena cicada as well.
sammer wrote: I'm still a tang on top guy.
User avatar
skimann20
Posts: 345
Joined: Wed Feb 27, 2008 8:28 pm
Location: Taxachusets

Post by skimann20 »

MontuckyMadman wrote:I think that was a cricket or a beetle. We don't have roaches I the arid intermountain west.
Coulda beena cicada as well.
I didn't think it was a roach but that was the closest in size i could compare it too... whatever it was, it was big! baby mouse. okay, back on topic...
Richuk
Posts: 1146
Joined: Sun Jan 31, 2010 9:53 am
Location: The Duchy of Grand Fenwick

Post by Richuk »

Ben, about the design. Does this set of pipe benders and this video shows some promise?

http://www.holden.co.uk/displayproduct. ... de=095.164



Assume one motor driving two wheels in sync and a second adjust the aperture of the handles using a screw bearing.

Perhaps something like this would ensure traction:
http://www.normstools.com/images/normst ... drum.shtml, and a small spring loaded gate to stop the edge from being driven backwards. I understand retraction of the edge can be an issue when bending wire. Is it possible to use a large bearing as a form (the large wheel in the pipe bender example) around which the ski edge is bent?

Slippage .... if the program knows the actual length required and tells you, then you simply measure that length and mark it. If the program finishes with the edge held in place, then the finishing point should be the same each time. Compare your marked point with a premarked point on the machine. Then tell the machine to release the ski edge.

PS. I'm assuming you'll want to set parameters on the physical design and want us to run the tests?
ben_mtl
Posts: 583
Joined: Mon Mar 31, 2008 1:47 pm
Location: Sherbrooke, Quebec
Contact:

Post by ben_mtl »

I do have this pipe bender, that was my first tool to bend the edges and I still use it sometimes.
For a CNC I think it would be a nice starting point if mixed with the concept of your video : 2 fixed rollers (fixed position but they can spin of course), and the last one being mobile.

Actually on nice add-on would be a straightening/feed rig before the "bending" part... I'll try to make a sketch when I have a few minutes.

To ensure traction and prevent slippage I was thinking more of a couple rubber feeders/rollers (http://www.mcmaster.com/#roller-feeders/=p9nr0k) as it would not be abrasive on the edge material but I'm totally open to other ideas.

This schematic from user "threeninethree" shows what I think would be best : sandwich the edge flat between 2 soft feeders to maximize contact surface.
Image

You make a good point of preventing the edge retraction.. I didn't think of that !

For now my goal is to share ideas and come up with a commonly thought-out design and an effective approach on how to control that machine.
I'm ok to make a version of the machine and run some tests too but I don't want to waste too much money on mistakes that could have been obvious to other members of the forum. I'm OK to run some tests too, I have a small roll of edge material all rusted that could be used just for that !

One first iteration of the machine might well be manual to check the design and the parameters : is there slippage ? does it bend efficiently ? how do I adjust the bend radius ?..etc..

For example on the video link you posted, the bend radius is adjusted by sliding the 3rd roller along a straight slot. While it sure works, I think a circular motion of this last roller around the one before would be appropriate for the programming as a bending radius would be directly proportional to an angle.
A bad day skiing is always better than a good one at work...
User avatar
chrismp
Posts: 1443
Joined: Fri Feb 13, 2009 9:00 am
Location: Vienna, Austria

Post by chrismp »

How about using the principle of wire forming machines instead of the three roller design. I think this would be easier to build and program.

http://www.youtube.com/watch?v=SQIYZ-iKuG4

There's already an open source version of a small 3D wire bender out there: http://www.instructables.com/id/DIWire-Bender/
User avatar
falls
Posts: 1458
Joined: Tue Oct 27, 2009 5:04 pm
Location: Wangaratta, Australia

Post by falls »

When thinking of programming an edge bender I always perceived creating a profile as if it were just like a CNC router.
For a 3/4 edge wrap I really only have a few radii - one in the tip, blend to the sidecut, sidecut, blend to tail, tail radius.
From your CAD drawings you can see that radius x is 40 mm long, radius y is 50mm long, sidecut radius is 1200mm long and so on.
Rather than having lots of points 5 mm apart my idea was to make it bend to radius x then feed for x mm, then bend to radius y and feed for y mm and so on. For simple blended arcs this seems ok to me and the 5mm points approach misses the exact radius change locations. Maybe not that big a deal in real life? I just had a different idea.

In cad you can even draw your profile as a line drawing that represents the movement of your bending wheel and the movement of the feed wheels in 2D and then a CAM program should be able to create the g code as if it were moving a router in 2 dimensions. After all axes are just axes.

Just some thoughts anyway.
Don't wait up, I'm off to kill Summer....
User avatar
falls
Posts: 1458
Joined: Tue Oct 27, 2009 5:04 pm
Location: Wangaratta, Australia

Post by falls »

I also worked out an equation based on the 3 die type edge bender that specifies how far you need to plunge the bending wheel in (from a neutral point where the edge feeds straight through) to create a certain radius.
If you create a home at this neutral point all programs can run from that point.
As I said above a 2D cad drawing could have vertical lines of the length needed to plunge the wheel from the home point to create a set radius then a horizontal line that feeds the edge for the distance it needs to go at that radius then another vertical line to change to the next radius and so on...
At the end of the day you set up your x and y axis with their steppers so the distances they move/feed are equal to the lines in the drawing (exactly like they are setup for a CNC router). I am sure you can write a much nicer program that would do all this automatically. I just thought a visual representation allows me to understand it, plus I can't write a nice computer program!
In the end the equation defines the perfect geometry of arcs based on wheel positions, but you probably get some relaxation after bending so the equation would need to be "tuned " by your experiments.
Don't wait up, I'm off to kill Summer....
ben_mtl
Posts: 583
Joined: Mon Mar 31, 2008 1:47 pm
Location: Sherbrooke, Quebec
Contact:

Post by ben_mtl »

This Pensa DIWire bender has been on my reference list for a while, I think it's a great design but it needs some modifications to be used with edge material :
- get rid of the Z axis and maybe the solenoid
- replace the bending pin with a roller with a groove (not sure of the groove but it could be helpful).
- put a roller with a groove for the edge teeth at the end of the tube, just before the bending pin (now roller) so your sure the edge doesn't twist even slightly in the tube -» there you're very close to the 3 roller system with the 2 first rollers being fixed.
- rubber rollers instead of knurled knob for the drive, edge material is harder to bend than aluminum wire.. so it's harder to feed when bending

I'm not sure what kind of motors they're using but they might have to be upgraded ... can we keep the Arduino then ?

I only had time to look into the mechanical approach of the DIWire... the software is what I'm more curious about ! It looks like it puts points on the profile you want to bend, you choose the resolution and then it can compute what to do... now what it does with the point is what I'd like to check :p
It's on the to-do list !

Thanks !
A bad day skiing is always better than a good one at work...
ben_mtl
Posts: 583
Joined: Mon Mar 31, 2008 1:47 pm
Location: Sherbrooke, Quebec
Contact:

Post by ben_mtl »

Falls -» Yes to all you've said, running it as if it was a router is the idea behind my thoughts... that's why I'm trying to get an X and Y axis :) I have to investigate on the "Pensa" way of doing it.

I'm also with you on my test file, it's been designed with circles so it would be easier and I could have only a few points for the "real program". However for me it was more a validation than anything else : the concept has to be applicable to variable arcs and splines, seeing that for many points the radius is the same (and is a "round" number) just confirms that it's working and I'm landing on my feet.
A bad day skiing is always better than a good one at work...
Post Reply