Going From Printed to Production

I understand the value of rapid prototyping and how 3D printing has revolutionized this ability. But once you're done prototyping and need to produce high tolerance, great fit/finish, and high quality parts on a medium to mass production scale you need machining.

My question is. What is the effort to get your printed part to production? Do you start from scratch, can you re-use the 3d instructions, etc, etc? Please give me your experience in this process.

I'll ask this. How are you going from STL or AMF files to machining?

You print a part, prove it out, make sure it is what exactly you want, and then begin the manufacturing process. Material? Work holding? Machine type? Tooling? Can it actually be machined affordably? I.E. no radii or features that can't be machined using off the shelf tools, or feasibly. Can you hold onto it? Is it a lathe part or a mill part? Horizontal mill? Live tooling Y axis lathe? Is the material readily available and affordable for your price point?

Obviously we know little to nothing about the part you are talking about. You can't make your price point $3 when you have $4 worth of material in the thing. Add in addition costs for tooling, fixturing/workholding, and labor, these are all factors in the manufacturing process as well. It is a bit more complicated than just sending a file to your 3d printer, and then coming back a few hours (days) later to a finished part.

While it depends on what your building. For the most part building a prototype is a very small fraction of the work required to take a product to production.

We design an assy in 3D cad
Build prototypes (Machined)

Once a design is final. Detailed specs for each part are determined. The tolerance stack must calculated to insure parts will fit and function as required.

Then you will need to figure out how to make the parts the most efficient way based on the quantity of parts being built. Build fixtures, Develop process's, program machines etc.

Edit: bigjon61 beat me to the submit button

What if we start very simply. Here is the AMF for a prism (I took this from wikipedia):



<?xml version="1.0" encoding="utf-8"?>
<amf unit="inch" version="1.1">
<metadata type="name">Split Pyramid</metadata>
<metadata type="author">John Smith</metadata>
<object id="1">
<mesh>
<vertices>
<vertex><coordinates><x>0</x><y>0</y><z>0</z></coordinates></vertex>
<vertex><coordinates><x>1</x><y>0</y><z>0</z></coordinates></vertex>
<vertex><coordinates><x>0</x><y>1</y><z>0</z></coordinates></vertex>
<vertex><coordinates><x>1</x><y>1</y><z>0</z></coordinates></vertex>
<vertex><coordinates><x>0.5</x><y>0.5</y><z>1</z></coordinates></vertex>
</vertices>
<volume materialid="2">
<metadata type="name">Hard side</metadata>
<triangle><v1>2</v1><v2>1</v2><v3>0</v3></triangle>
<triangle><v1>0</v1><v2>1</v2><v3>4</v3></triangle>
<triangle><v1>4</v1><v2>1</v2><v3>2</v3></triangle>
<triangle><v1>0</v1><v2>4</v2><v3>2</v3></triangle>
</volume>
<volume materialid="3">
<metadata type="name">Soft side</metadata>
<triangle><v1>2</v1><v2>3</v2><v3>1</v3></triangle>
<triangle><v1>1</v1><v2>3</v2><v3>4</v3></triangle>
<triangle><v1>4</v1><v2>3</v2><v3>2</v3></triangle>
<triangle><v1>4</v1><v2>2</v2><v3>1</v3></triangle>
</volume>
</mesh>
</object>
<material id="2">
<metadata type="name">Hard material</metadata>
<color><r>0.1</r><g>0.1</g><b>0.1</b></color>
</material>
<material id="3">
<metadata type="name">Soft material</metadata>
<color><r>0</r><g>0.9</g><b>0.9</b><a>0.5</a></color>
</material>
</amf>

What would it take to turn this into a production piece?

mulepic said:

I understand the value of rapid prototyping and how 3D printing has revolutionized this ability. But once you're done prototyping and need to produce high tolerance, great fit/finish, and high quality parts on a medium to mass production scale you need machining.

My question is. What is the effort to get your printed part to production? Do you start from scratch, can you re-use the 3d instructions, etc, etc? Please give me your experience in this process.

Click to expand...

Its called engineering.

Some people do the engineering before 3D printing. Then there are the people that 3D print before engineering, and that can cost lots of $$!!

Tim

behindpropellers said:

Its called engineering.

Some people do the engineering before 3D printing. Then there are the people that 3D print before engineering, and that can cost lots of $$!!

Tim

Click to expand...

So I gave a simple example of a prism. Let's focus on that and give concrete information. What would it take to get that prism to production from those 3D instructions?

What if we start very simply. Here is the AMF for a prism
What would it take to turn this into a production piece?

Click to expand...

Ok simply, All of the above answers.

You have an AMF file that defines the shape of the part you want. That's it, everything else needs to be figured out.

DesertCoyote said:

Ok simply, All of the above answers.

You have an AMF file that defines the shape of the part you want. That's it, everything else needs to be figured out.

Click to expand...

There are tools that will convert that AMF to G-Code and tool paths. What are the steps before and after that?

We would design it in some CAD software , Solidworks in our case, and when we were happy with the 3D printed version, which we make by exporting STLs, send the actual CAD file or a parasolid or whatever to the guy who's going to do the CAM. Also, the part should be designed with machining in mind in the first place.

We just did this with some small vacuum manifold type parts that come in a left and right hand version. We've got it so most of it's done on a CNC lathe, then there is a second drilling operation to make two luer tapers with further holes extending from them and finally the left and right parts are defined by drilling to connect the taper holes to one or the other of two concentric manifold regions. So 90% of it can be done as one part and the symmetry is only broken in the third operation.

This was a great example where 3D printing was useful as it took some experimentation to get the luer (commonly seen as a hypodermic syringe) taper right as the luers are a bit of a mystery depending on whether the parts are metal, plastic or glass. 3D printing small parts (20mm diameter here) is exponentially faster than large parts so you can get your answer in the hour. Now we've got it the parts have to be made from acrylic so it's off to CNC.

On thing about engineering: use real CAD software.

mulepic said:

There are tools that will convert that AMF to G-Code and tool paths. What are the steps before and after that?

Click to expand...

It doesn't work that way. Machining is totally different than RP. You can't just covert to g-code.

As stated above...

You need to define the purpose of this part
Select a material
Define dimensional tolerance required for the part to function.
Surface finish and coatings
Quantity to made

All the above and more will drive the process required to make the part.

rcoope said:

We would design it in some CAD software , Solidworks in our case, and when we were happy with the 3D printed version, which we make by exporting STLs, send the actual CAD file or a parasolid or whatever to the guy who's going to do the CAM. Also, the part should be designed with machining in mind in the first place.

We just did this with some small vacuum manifold type parts that come in a left and right hand version. We've got it so most of it's done on a CNC lathe, then there is a second drilling operation to make two luer tapers with further holes extending from them and finally the left and right parts are defined by drilling to connect the taper holes to one or the other of two concentric manifold regions. So 90% of it can be done as one part and the symmetry is only broken in the third operation.

This was a great example where 3D printing was useful as it took some experimentation to get the luer (commonly seen as a hypodermic syringe) taper right as the luers are a bit of a mystery depending on whether the parts are metal, plastic or glass. 3D printing small parts (20mm diameter here) is exponentially faster than large parts so you can get your answer in the hour. Now we've got it the parts have to be made from acrylic so it's off to CNC.

On thing about engineering: use real CAD software.

Click to expand...

Thank you for the real-world example. That helps explain what's involved.

Further to the real CAD thing, we run an institutional shop for biomedical device development and maintenance. People who submit jobs are engineers or engineering trainees, all of whom are very teachable in the art of submitting makable parts to a shop. The 3D printing thing in particular has brought some groups out of the woodwork who are trying to prototype using free/open source CAD software and so far the results have not been encouraging. In the case above for example, I ended up getting the student down to my office and doing the design together. He is actually a clinical trainee and not an engineer so I'm cutting him some slack but he had a lot of trouble modelling what he wanted and then even more trouble getting models exported to anything we could use.

If there's one thing that drives me up the wall it's real engineers using toy tools, be they software or mechanical.

I'm starting to come to the conclusion that 3D printing does NOT ease the transition from prototype to production. It simply makes rapid prototyping more rapid.

mulepic said:

I'm starting to come to the conclusion that 3D printing does NOT ease the transition from prototype to production. It simply makes rapid prototyping more rapid.

Click to expand...

Yes and no. Sometimes it might be unclear on a feature or two exactly what you want/need/can live with. In this case a 3d printed part can help. But I think to your point, ya, it does very little to get to production. A printing process is completely different than a machining process, as everyone else stated. You have a part you want to make so the printer takes your model (be that in whatever format they need) and does their thing (which might include building 'supports', adding clearance between features so they print correctly, etc) and sends you the printed part. You can't take the printed part to a machinist and say "I want 200 of these, here's your information" and hand them the part and your amf file or whatever it is.

Mike1974 said:

You can't take the printed part to a machinist and say "I want 200 of these, here's your information" and hand them the part and your amf file or whatever it is.

Click to expand...

Sure would be cool if you could!

mulepic said:

Sure would be cool if you could!

Click to expand...

Let me clarify before I get beat up about this

You could certainly take a printed part to a machinist and say I want to make these out of xx material and I would like 200,400,1000 whatever pieces. They would then likely ask about tolerances and material specs, then they would need some time to quote the job and get back to you. Also things like lead time, will you accept over/under percentages (since we seem to be talking production quantities), surface finish, etc etc. The files and processes the printer used will (most likely) be worthless to said machinist, except the solid model or cad file, presuming you have something mainstream ( .stp .iges .prt .x_t ). So, yes the part can help, but whatever money and time you invest in the printed part does not "carry over" to the production end of it.

Presumably the designer or engineer would have to be aware of the manufacturing technologies available to them for mass production if that is the ultimate goal and design the product around those capabilities. And then the 3D printing can be a quick send-off to get the part back and see how things look and fit before giving the go for mass production.

On the flip side, you can't design or engineer a product of complex geometry that only a 3D printer can produce, and expect this part to be manufacturable with subtractive manufacturing after the fact. For example, if you designed a part with internal geometries and structures that only a 3D printer can produce, then only a 3D printer can produce this. A machinist cannot make complex internal geometries on a mill or lathe.

So I guess the design of the product must have been designed with full consideration for the manufacturability and mass production in mind. And 3D printing in that case is merely a complement to that process to complement the design process. On the other hand, in the future, certainly products will be designed with the full intention of relying only on 3D printers and nothing else. And that is the beauty of additive manufacturing -- geometric freedom without manufacturability constraints.

But in the future, as 3D printers improve, there is no reason you can't 3D print the final part to tolerance without need for finish subtractive machining. It is only a current conundrum that 3D printed parts need finish machining to achieve acceptable surface finish and tolerance. This need not be the case in the future. For example, we can already do some extremely precise (down to subnanometer precision) 3D printing for a very long time already. Less covered by this recent "additive manufacturing craze" but focused ion beam (FIB) systems or electron beam systems in special scanning electron microscopes (SEM) with gas feeds have been used to selectively sterolithographically grow nanometer scale 3D structures for a long time in the applied physics community. This is the kind of precision that even the best mazak or moriseiki mill/lathe cannot achieve with a mechanical contact cutting process.

There is no technology barrier to the precision that additive manufacturing can achieve.

In addition to the possibility of making features for 3D printing that can't be machined, there is also a problem where people who only have access to consumer FDM printers get used to compensating for the lack of support material by designing special geometry like arches instead of flat roofed slots for example. So when using 3D printing to check a prototype, the printer should be capable of reproducing the desired part without it's own special geometry.

I don't think it's been said explicitly above but while you can't give a shop a 3D printed part and say "make this", you absolutely can give a shop a CAD model with a few tolerances and they absolutely can make that. The important thing these days is not particularly 3D printing, it's CAD. Say it with me: CAD.

rcoope said:

In addition to the possibility of making features for 3D printing that can't be machined, there is also a problem where people who only have access to consumer FDM printers get used to compensating for the lack of support material by designing special geometry like arches instead of flat roofed slots for example. So when using 3D printing to check a prototype, the printer should be capable of reproducing the desired part without it's own special geometry.

I don't think it's been said explicitly above but while you can't give a shop a 3D printed part and say "make this", you absolutely can give a shop a CAD model with a few tolerances and they absolutely can make that. The important thing these days is not particularly 3D printing, it's CAD. Say it with me: CAD.

Click to expand...

Ha! Provided there are not closed internal geometries, or other geometries even a 5 axis can't produce. Hence the beauty of additive manufacturing.