If you want to form a piece of sheet metal into a shape, you’ll probably think about using a die. That’s certainly a great way to do it, but it presupposes you can create or purchase the die, which may be a showstopper for small projects. [Dardy-7] has worked out how to use a lesser-used technique — incremental sheet forming — to get similar results with a CNC machine. The idea is to trace out the form on the sheet metal with a round blunt tool.
He got good results using an inexpensive dapping tool, although he’s seen other use heated titanium ball bearings. In addition, he’s worked out how to adapt existing tool paths, like the ones you might download from the Internet, to use with this technique. You can see a video of the workflow below.
The technique requires a jig to hold the workpiece that allows the tool to push the metal down. In the examples, the jigs are just quick plywood assemblies and the workpiece is 0.6 mm aluminum. Some of the jigs have support underneath to help create parts that have parts poking up (that is, a positive curvature).
We couldn’t help but think that a 3D printer could easily create a backing plate that would serve the same purpose as under support in the current jigs. [Dardy-7] notes the process is slow and not as accurate as some other methods. On the other hand, if you have a CNC machine it should be an easy addition to your repertoire.
If this makes you want to run out and buy a CNC machine, you might like to read this first. If you want to dive into the tool paths, we can help you there, too.
>”We couldn’t help but think that a 3D printer could easily create a backing plate”
That’s what they did. They took the 3D model of the shape and printed out a matching solid to place under the sheet.
That’s cool but it would take lots of advanced physics to figure out how the sheet metal will react to certain pressure in certain places. Factors like work hardening and temperature would have to simulated and the exact characteristics of the material would matter. Seems a lot easier to machine a die or you could 3D print that backing plate and hydroform with better surface finish.
You’d think so, but that’s kind of the point of the academic research papers that have already been done: to establish a simple model of how metal behaves under incremental forming.
It turns out that there’s a bit of a radius at the top by the fixture plate, but outwit light gauge aluminum of the right type goes and stays basically where you put it. The tool creates a very localized increase in stress, so you have localized yielding. Further, you’re mainly working the area that hasn’t been work hardened, so the hardening and new 3D shape of the completed area helps prevent further deformation.
This is 1940’s tech though. How sheet metal behaves under strain is very well understood. If you can’t find it online for free you can probably find a used textbook for near the cost of shipping.
I don’t see this for mass production but for one off’s or hobbyist level production it seems sufficient.
If you’re being serious about it, they go through all that stuff for die stamping too. They analyse for thinning of the sheet, risk of tearing, etc.
Ron Covell would tell you you’re wrong.
Also, as others have said, people have been forming sheet metal since any of us were born, it’s pretty well understood and working this stuff out is pretty well catered for. Plus, unless you’re using exotic materials, you’ve got a lot of leeway. There’s a reason steel is as popular as it is.
What exactly is he making? Is it just a proof of concept? He mentions furniture at the end of the article, but that thing looks uncomfortable.
Ford used two tools, one either side…
https://youtu.be/iNQ40MYwZqw
That was an awesome video .. Thanks!
I understand the rapid prototype draw for this but at a production level it can’t be faster than traditional stamping methods.
Obviously 500k for a set of dies will make most people balk, but the beauty of mass manufacture is everything uses the same setup with a few smaller pieces making up the difference between an economy car and the luxury version.
It’ll likely be faster and cheaper for small runs. It’s like 3D printing to injection moulding.
Cool. So it’s intended for rapid prototyping of stamped metal parts
Came here to say this!
It’s a very cool technique and an awesome way to potentially reproduce sheet metal parts such as wings etc. for old classics.
Doing it “single sided” with a CNC mill definitely makes it more accessible.
This is pretty cool, but they probably could have achieved similar results with type of hydroforming. Basically you make a die, shouldn’t be a big deal with a CNC machine, and put it in a box with the material, then place a thick rubber mat on it followed by a metal lid, once everything is in place it goes to the press. You apply enough pressure to force the rubber and goes into plastic deformation, when it does the workpiece is forced on to the die. Experimental aircraft people have been doing it for years, here’s an example: https://youtu.be/xw5yEMsDxR8
This does the same thing but stops at that cnc machine you mentioned in step one. How would that be more advantageous for rapid prototyping?
In a pure one off situation there is no advantage. If you have to make a few dozen parts, hydroforming the way shown in the video would be much faster, many more parts would justify a multiple hit method with several different dies in a big press.
Every method has its place and this one is pretty cool for one off parts. Hell if I had known about it before this article I would have used it for many parts and I may use it in the future.
At the end of the day it boils down to how many parts are needed and if it’s a hobby or if the job is keeping the lights on.
That’s bo Hydroforming in the Video.
bo =no
It is hydroforming by definition…using a fluid to form metal. By exerting enough pressure to cause plastic flow and form the metal onto a negative form. I imagine you’re familiar with the positive process involving carefully designed near net parts welded together and pressurized to make the final part…it’s common on parts like expansion chambers for 2 stroke engines. Think outside the box a little bit.
I don’t know your background or career but here is mine. I work in a shop that ranges the gamut of one off prototypes to full blow production. I’ve never thought of the method described above, but have used the rubber method. For low quantities it works great, if I had to make more then a few dozen parts I would be designing dies to stamp them in a press in multiple hits for efficiency. My point being that there is more than one way to skin a cat and you shouldn’t dismiss a new process.
I been in collision and resto for 27yrs. Pretty sure I can get you 95% of that in about 2hrs by hand. CNC machines are like 3d printers, they have their niches they’re really good at. So just stop trying to make them do something they aren’t good at putting undue stress and wear on the machine. Besides, you might actually learn a fucking thing or twine besides sitting on your arse and clicking on the mouse.
The benefit of a CNC machine is that it won’t take you 27 years of experience to get to 95%. A bonus is that it won’t leave snooty remarks on Hackaday about its work.
Let alone that needing just one machine makes it possible to do a wider array of projects in a smaller space.
Yeah better get off our asses and stop trying to make progress in robotics. No way that whole silly automation fad is really going anywhere, right guys? Guys?
Once worked in the sheet metal business (15 years). Forming over a large area and doing it quickly with accuracy requires tonnage. Minster machines do a good job of it but the dies are $$$$$$$. The turret punch presses? Not so well and only if you are looking for knockouts or very specialized things like hinges. YMMV. Whole businesses are built around this.
very interesting stuff, hope to hear/read/see more of this
Yeah if we could form our own steel car bodypanels it would be amazing
Hi, can I know what software you have used to model the toolpath? I am using siemens nx and its hard to model tool path for complex shapes.