Welcome! Log In Create A New Profile


New printer concept, looking for collaborators; hybrid deposition-milling

Posted by Adouglas89 
New printer concept, looking for collaborators; hybrid deposition-milling
November 24, 2016 03:34PM
Ok, I want to launch a new 3d printer project. I am pretty serious about it, and have put lot of time and money into it already. I am trying to find a worksite and tools to pursue the development of the printer.

Basically the concept I currently see as the most promising is one that produces casts make of the same material commonly used for shell casting of steel, in investment casting. Then some ancilliary equipment to make the casting process practical. Basically, a form of lost wax casting.

But I should say first, that the fundamental process of depositing a whole layer of build material rather sloppily, then milling, then filling all extra space in the build volume below the top of the latest layer, with support material and it can be sloppily, milling again to level the surface etc., deposit more build material, mill again, etc. and then remove the support material, then fill the void with a castable material has a lot of promise compared to a lot of other options, for our purposes:

a)it only takes relatively basic, available technology, which does not require a ton of capital to produce to begin with. The mills etc. like a sherline are a bit costly, but not like high powered lasers. This improves independence and the constriction that the ruling class can impose. We can use a much flimsier and cheaper structure than a sherline, btw.

b)The problems that printing without support material imposes on geometries are eliminated. Like droplets on powder printing or laser sintering, literally any geometry you can draw on a computer can be produced, within the minimum internal radius of features etc. because the milling bit has to be able to fit in there.

c)Better than the powder based approaches, this acheives high dimensional accuracy and smooth surfaces, both of which are critical to produce truly useful stuff. Lets face it, there is very little useful stuff on e.g. thingiverse, and this is due to the fundamental limitations of printers as we know them. The steps on the sides of curved surfaces are a big problem if you are making moving parts, fluid seals, etc. In worst case scenario, you could just print in zero contraction plaster (used in dentistry), followed by zero expansion wax or similar, then near zero expansion thermoset polyurethane polymers to get top quality plastic parts. Or even just plaster and polyurethane, then skip the casting. They take time to set at each layer, but the layers can be relatively thick, depending on the geometry of the part. There are surely better materials, that e.g. set faster so you can get better build rates, and the heart of the system, this Shape deposition modelling indiscriminate deposition-mill-deposit support-mill process gives geometries, accuracy and smoothness that cannot be achieved by any other printer approach I have heard of.

d)It can have an output which is 100% strength, fully dense, normal engineering materials.

It does have some limited capacity to produce fully assembled parts, except that they would need sprues attached to each part, which could be cut off later.

But ultimately the reality is that we live in a world that is a hell of a long way from scifi land of multimaterial printers, and there is a lot of room between there and where we actually stand with the availability that people have of the tools and equipment to influence the built environment or realize our inclination to invent. A serious improvement in the state of available 3d printers would be worth a ton. With this project, I am proposing leapfrogging many of the problems of the printers we see around us, even high cost cloud printers, such as the stairstep sides and flimsy materials and poor accuracy. This system would cost way less than a $250,000 fortus 3d printer, and be much more available, and I think produce more valuable parts.

The cost per unit produced and speed, and possible geometries, could also be a very tangible improvement over ordering out to a machine shop. I have some back ground in 4 axis and 5 axis milling, so I know this relatively well. I also considered putting my time into a low cost 5 axis mill system, but in the end it gets quite complicated and expensive, and you are still stuck with a lot of limitations. This could really overcome a lot of those.

I have one collaborator, Michael, in Toronto, who can help as a software engineer. But fundamental to the whole thing, the only part that is really very uncertain in terms of being doable, is the materials part of it. And even then, if the system is designed to make use of different materials practical, then as mentioned above, known workable if less ambitious combinations, can be implemented.

The build material has to be zircon, or ceria shell material particles bonded together, or I think alumina might also be an option. The binder is probably best a gel like colloidal alumina, ceria or zircon with long molecules. There are some patents on using such binders with such powders to produce unsintered solids by 3dp, the droplet on powder approach, with strengths that look suitable, in the 2000 psi tensile stregth range, which of course is very low, but plaster of paris is about 200, and that works for aluminum casting.

The support material might be a powder with a water soluble binder, I think the gel stuff might not be water soluble any more after initial drying... uh actually come to think of it, it might be. Ok, that would be an issue. Perhaps some other solvent could be used to dissolve the binder.

They both need certain properties, I am sure there are more, every time I try to list them I seem to have to decide all over on some stuff...

a) to stick to each other

b)to be millable ( the ceramics of course are highly abrasive and tend to crack on impact, so a ceramic end mill like a finishing burr (kind of like a rotary wood rasp) would probably be in order, or even a ceramic bonded abrasive type bit. A tool changer to switch at least once between roughing and finishing bits is probably needed, and to get good feature sizes more will be needed.)

c)to have reasonable strength

d)the support material has to be removeable from the build material without damaging the build material.

e) Critically, they need to not contract much or ideally at all upon deposition. Such contraction is one of the major problems in layer by layer or additive manufacturing. If each layer is deposited, and ends up with a contractile stress, the force ads up layer after layer, causing at the very least elastic distortion of the object, preventing accuracy from being as easily achieved. With some materials, some degree of computer compensation could be employed, although this is not an easy thing to do, I see systems all over where they could do this, but no one ever does, so I am betting it might be harder than it looks. One of the questions is if this slurry of particles can be deposited with little contraction. There are some methods of reducing any contraction, like adding filler so the fraction of material that was a liquid to begin, and so the net bulk contraction with goes down....

In normal industrial investment casting, the shells are made with a colloid that is short particles or chains, not the long chains, and this leads to a lower strength. The shells, are, importantly, also sintered, which greatly increases strength, of course. It reduces the accuracy of the process, though, greatly, due to distortion of the object when sintered, a nearly universal problem with sintering. Although with extremely small (micron sized) particles and low void volumes (gel casting), the distortion can be quite small and relatively predictable, I read one paper that described making ceramic drill bits that required no post processing (except maybe sharpening, I don't know) acheiving roughtly a =/- 50 micron tolerance (over the size of a dril bit severla inches long I guess).

It also makes it way harder to remove the material from the final, cast object, maybe impractical in cracks etc.

So the goal is to avoid sintering.

There are various means to work with weaker materials. Suppose that instead of a shell around the void into which the casting material will flow, there is a block of material, for instance, with the void embedded in it. In other words, the void is surrounded by a great deal more material. Most forces will be withstood by such a structure.

Ok, so if we can have a way to deposit support material and build material layer by layer. This is when the milling comes in. There are many hybrid milling/printing systems, shape deposition modelling is one of the most interesting ones. In this case, it seems like a fundamentally easier thing to try to print the cast rather than trying to print the metal directly, with a similar material, the zirconia ceramic shell material, as the support material, but with a different binder. But the milling of the steel entails substantial forces, shot or electropeening of the metal layer too, so the material demands on the strength of the ceramic in that role are more substantial etc.... nah, this approach of printing the cast then casting seems substantially more reliably attainable, and to have certain advantages in relaxation of the demandingness of the requirements on the zirconia material, that's the main perk. Although really if the metal layers are quite small, like a few thou, then the forces could be so small on it it could be fine even with the unsintered ceramic, the method of depositing the metal withough heating leading to expansion problems is not trivial.

We can think of milling as a magic step that makes material go away wherever the bit can reach mostly, and it does take some time, although the pounding of the blades is one complication, and the wear of the bits etc. But we know that totally works, as milling has been around, and I have some subtantial background in milling. The path planning for a simple 3d tool path is relatively practical. For five axis, it has a bad reputation, but I think crude automated five axis toolpathing is probably practical, to get like 90 percent of the benefits of the extra two tilting axis. A tool changer is surely required here, because tool wear during the roughing stage will probably be substantial, so you switch to a tool of known shape that is very little worn out, in order to do the last couple thousandths of an inch. That way the bit replacement cost is quite low over time.

Also the origin of the forces during casting need to be nailed down better. I have some idea. As the material, suppose steel as that is the most interesting material, cools, it goes through a slushy region, and then enters the plastic region, then gets progressively stronger. The volume changes that occur after the solidification starts and I guess also dissolved gasses coming out are the origin of all forces, I think. The dissolved gasses could be I assume mostly or all removed by pre treating the material with several freeze melt cycles, though. Which is getting more complicated, but it is good to know it won't be a total dealbreaker.

The solid liquid transition entails about 50% of total volume change, and the thermal contraction of the solid the rest, apparently.

The build material can be adjusted by combining different powders, some with a higher cte than steel, some with a lower, to produce a solid with the same bulk cte as steel. Except cte changes some with temperature, so that is another problem. It has to be close enough that elastic distortion can occur, instead of cracking or delamination of the steel from the side of the cast.

Another approach to helping to avoid delamination from the side of the mold might be to do the casting under vacuum, so that voids in the steel region can form without any force, actually. If they were really small or later flowed together, that might be ok. Hopefully the steel sticks to the build material well enough. That is one of the requirements of the build material.

Oh, I should say, too, that resorting to other alloys that melt at lower temperatures etc, is also interesting, but there are so many different possibilities of how to do things at each stage, and the uncertainty so high, that I think it might actually make a lot of sense to just leapfrog straight to trying to work with steel. The temperatures are higher, but there may also be some things that make it easier, notably the sheer amount of literature to draw upon when it comes to how to cast steel, and the ultimately higher value makes it more justifiable to invest a lot of time and effort.

Ok, so we have our mold of material, which would take a bit of developing maybe, to get the combination of all properties to a sufficient degree, then preheat everything, put things under a vacuum to eliminate the complication of air bubbles, then caaareeefully pour in the steel, then carefully and slowly bring it back down in temperature. Ideally in an automated machine. I think a top to bottom temperature gradient might be a better idea than an inside out gradient, so that if the sprue, which is the stem where the steel flows into the void, is located at the exact top of the cast remains unsolidfied until the last. That allows more liquid to flow in as the solidification occurs, in contrast to the usual inside out gradient, which leads to sprue cloggage. A relatively small gradient is I think probably desireable, because otherwise I don't know if a progressive contractive stress would build up, similar to how it usually does in 3d printing. I don't think so, or not much, because soon after melting it is hot enough for plastic flow of the solid. Still, it is one hazard, and knowing what options might be there to mitigate it is important.

Then we got to the stage of the metal in a solid state, surrounded by ceramic material of the same cte, and we cool it carefully. Interestingly, the steel could be hardened by bringing it through the right temperature regions and holding it there for suitable times.

Then the build material needs to be removed, I think either high pressure water jets or maybe ultrasound, like those in sink dish cleaning ultrasound things, might work ok. As long as a sound wave of suffient amplitude went through the solid such that the expansion phase is enough to break apart the particulates the ceramic is made from, that might work pretty good actually. Plus the material can be recycled.

Then you manually remove the sprue, and you have your object, hopefully of good material properties, fully solid and dimensionally accurate. A vibratory polishing step in which the object is submersed in a bath of ceramic spheres and vibrations cause pounding of the surface to smooth it out, could be used to improve surface finish without changing the average surface boundary dimensions much at all. The premium investment casting research I did revealed companies that advertised reliably achieving +/- 80 microns over 5 centimeters accuracy, but that was a process that entailed photolithograpy 3d printing a master mold, followed by I think wax models, that get covered in the shell material, the material sintered, then the casting poured. Thats quite a few steps, so the loss that occurs between the void shape in the cast and the final metal object is hopefully pretty small.

At this point the biggest point that concerns me is the forces that might arise during casting, and successfully resisting them.

I tend to be located in Toronto these days, anyone here from Toronto? We could further advance the concept, and proceed with getting a workspace and stuff to prototype this.

One of the advantages I think is that this hybrid deposition-mill-cast approach can have different materials substituted in, so the development effort is almost inevitably going to lead to a major improvement
Re: New printer concept, looking for collaborators; hybrid deposition-milling
November 26, 2016 09:27AM
hello? Such are the barriers to getting stuff like this done, explaining partly why it had not been done for do long, despite the reason to, and resources available.
Sorry, only registered users may post in this forum.

Click here to login