“What are the limits on the material that can be used. I would assume, possibly wrongly, that there is a relatively limited range of organic based plastics that can be used because they have the right mix of properties. Is this limited selection of materials a restriction on what can be achieved ?”

and that’s it in a nutshell.

You can go to the corners of the physical/thermal envelope, but you always have to give up something.

Bloke in Spain spoke (for example) about the possibility of making patterns for investment casting. This has been done semi-successfully, but the problems of making and controlling a feedstock that’s suitable for the process have not really been overcome. This material would be more like a wax than a plastic, hard to make into a consistent filament or powder feedstock.

Better results have been obtained by using 3D printing to make the molds for casting investment-cast cores, but the (relatively)-poor surface finish limits design and does not allow for near-net casting, which is one of the goals of modern investment casting in the first place.

And so it goes.

I’ve had a fair amount of sucess usuing the process to make tooling for processes like drape-molding and vacuum-forming, and I don’t see why it wouldn’t work for blow-molding and similar processes although I’ve never heard of it being done.

And so on. If you want the process tuned to some corner of the material envelope, you have to give up something else. The makers have already produced a process which is pretty-optimal for most applications.

I think you have to see the process for what it is. As Bloke in Spain describes, people like Distributed Defense and these car-making guys are just hypnotized by the process, and thye are applying it to everything whether it makes sense or not. Making the heating ducts of your car out of $5/cubic inch plastic by a process that takes 10 days on a $40,000 machine, instead of using 9¢-a-foot flexible hose, definitely falls into the category of ‘because I can’ and not ‘because it makes any sense’. All this talk about all-in-one manufacturing being more-compact and lighter, and therefore ‘better’, is mostly hype and/or self-delusion, because it ignores the real world that the product is supposed to be aimed at. Something built like this rapidly becomes impossible to assemble, finish, service or repair.

IBM once famously-designed the chassis of a business machine using this principle – a single weldment made out of fully-tooled sheet metal parts. It all tabbed together like a jigsaw puzzle and was robotically-welded together. It was incredibly strong, incredibly light and very cost-effective. Boothroyd/Dewhurst run amok. This was in the early days of CAD, and everything was still 2D. But the design absolutely worked and fit together – on the screen.

But it was impossible to assemble in any realistic way. At one point, they were assembling drive belts to the chassis before welding it together, and wiring a spare belt in place for future repairs. And the service costs were completely out of control – you couldn’t get to anything to fix it. Service techs were cutting access holes in the chassis. It was a complete fiasco.

This approach works well for a product that is essentially-disposable. The printer engine at the heart of many HP printers is made this way – it can only really be assembled, it can’t be disassembled, and it certainly can’t be repaired beyond some very simple attentions. Maybe not the best approach for a costly product like an automobile that’s filled with failure items.

llater,

llamas

I think that over the long term, printing an AR-15 lower the way it’s done now is *necessarily* a transitional process. The few people I know who are doing this is not because they need an AR-15 lower – the reason is ‘because I can’ – it’s as a political statement. As of today, anyone who wants to be a firearm owner under the radar will have obtained a number of “80% lowers” and will be hard at work with a bench drill, but the very act of fabricating something that The State considers dangerous – however flimsy – out of a bag full of plastic stock is a subversive act.

Sure, at the moment, it’s “Cargo Cult Manufacturing”, but the patterns for those lowers is already different from the engineering design that was released a few weeks ago.

The fanbicators have modified their designs to deal with stress points where the falures were, and they turned around many iterations of a design in less time than it would take a cast-maker to set a new mold.

It wouldn’t surprise me to see by this time next year, a functional lower that was component-compatible with a traditional lower, with maybe laminated metal reinforcement and a profile noticeably different from current designs – I’ll defer to the people who actually do 3D fabrication on that point – but the issue is that not just designs, but techniques are going to evolve very quickly.

And I’ll bet money that casting will become a cottage industry. I can’t think of a *better* way of making molds than with a thermoplastic, that can be formed by a computer that can calculate casting shrinkage along with optimal locations for spigots and vents. I wouldn’t want to fire a lower cast from pot metal, but I’d expect more variety in the casting materials than just bronze.

And thanks to Llamas for a genuinely informative series of posts, so helpful when one has an actual practitioner to comment.

There are three physical/kinematic limits to the speed of 3D printing.

The first has to do with the speeds with which you can transit the print head in X and Y

(note that ‘print head’ is merely used as a generic marker for ‘fused deposition device’)

while still supplying feedstock at the required rates.

I do not think this has any limits that would be of concern. How fast you accelerate something is merely a function of how much force you are prepared to apply.

The second concerns the chemistry of the support material. Many people don’t understand that 3D printers almost always print 2 materials – the material of the part being built, and then a disposable support material, which is like scaffolding for the part under construction. The support material is removed when the part build is complete, either mechanically or (most-often) chemically, by dissolving it in a chemical bath. This is a major factor in the speed of delivery of the finished part – how long it takes to wash out the support material. A lot of the clever, patented technology in this area deals with minimizing the mass of the support material and making it wash away as quickly as possible.

Again, this is merely a problem of chemistry. The process is generally slow because the chemicals used tend to be quite benign. If you want the support gone in minutes, that would be easy to do, although perhaps rather more complex than an open bath of water-soluble chemicals. I believe there are systems available which reduce wash time from ‘a few hours’ to ‘many minutes’.

The third limit is a thermal limit. This is the big hurdle.

Fused-deposition is a delicate balance of temperatures – the feedstock is delivered in a continuous stream of liquid material, which has to fuse almost-instantly with previously-deposited materials. It must be hot enough to stay liquid as it exits the deposition device, but not so hot as to overmelt the previously-deposited material. The already-built material must be hot enough to allow the deposited material to fuse with it, but also cold enough to provide support and not slump or re-melt. This means that there is a zone of remelt bettween the new material and the old which is only 2 or 3 thousandths of an inch deep, being consistently maintained while the print head travels at speeds of 20-30 inches per second.

It’s like arc welding – there is a limit to the speed with which you can lay down weld metal that is defined by the rate at which you can selectively-remelt the base material. The required temperature gradient has a third variable – time – which is very-narrowly delimited by the melt temperature of the material and its thermal conductivity.

That’s where the problem lies.

Moore’s Law appears to apply relatively accurately to computing power because at the time it was postulated, memory and processor technologies were incredibly-coarse when measured with a molecular yardstick. But as devices approach the molecular level (within a few orders of magnitude), the rate of change will inevitably slow and diverge more and more from the Moore’s Law prediction.

The functionality of FDM technques is already at a pretty-fine level – a good balance has been found between granularity of build quality and speed that stays in the possible thermal regimes of the materials in question. To deposit material faster, you must deposit it in a coarser stream (sacrificing accuracy of the part) – the thermal parameters will not be denied. So a Moore’s Law-type progression will not apply unless/until materials can be discovered which have lower melting points and lower thermal conductivity and yet are still acceptable for finished models.

The future of rapid-prototyping methods may lie with processes which do not have thermal limitations. There are some very-interesting possibilities in laser-cut-and-fused laminar manufacturing techniques, which do not rely on melting feedstock and then fusing it together. But these techniques have limitations on size and fine detail. They work wonderfully for cylinder heads and turret-press frames, not so much for gear-trains and jet engine impellers.

Pardon my ignnerance – how do I post a photograph here? That doesn’t involve me posting it to some other website first?

llater,

llamas

I am not clinging to any wrong paradigm. I mention 2D printers because the cost of the physical inputs seems relevant to the cost of 3D printer physical inputs. 2D printer inputs have not fallen much in price, even as 2D printers have got dramatically cheaper. This would suggest that 3D printer input costs won’t fall that much either. But, maybe that’s wrong.

USB sticks now carry a lot of data and cost very little. etc. I know this. We all know this. That’s Moore’s Law in action. What I am asking is if Moore’s Law applies, as Dale Amon asserts, to 3D printers, the way it clearly does to USB sticks and the like. And I would truly like to know. I suspect Dale Amon is being somewhat optimistic. Insofar as the current limitations of 3D printing are in computer power and hardware cleverness, yes, increased computer power will improve 3D printing dramatically . But if physical input costs remain approximately where they are now, Moore’s Law will not apply.

This seems to be the key point. If something like Moore’s Law does apply to 3D printing, then the important thing is to get some particular application working, however expensively, so that it can be sped up later. A 3D printed car now is insanely expensive and time-consuming to make thus. But when Moore’s law starts to do its thing …?

But does Moore’s Law apply? The difference would appear to be, as llamas said in the very first comment above, in the cost of the raw materials. These loom very large, and will surely continue to do so. 3D printing will presumably get much quicker. So, will the over-all costs nosedive the way computing costs have? Will 3D printing improve not only in speed and cleverness, but also, in particular, in its cleverness at using cheaper physical inputs? With tiny objects, input costs count for less. But for things like cars …?

It’s noticeable that (regular 2D printer) toner cartridges haven’t dropped in price nearly as much as printers or computers.

I’d welcome yet further comment on this. Dale? Anyone?