3D Printing -- Pretty Close to Ready

[pickelhaube]

From start to finish. NOT COUNTING SANDING AND FINISHING.

How long does it take to print out one unit ?

[gburch]

From start to finish. NOT COUNTING SANDING AND FINISHING.

How long does it take to print out one unit ?

Estimate: 20 -24 hours total printing time.

I'm going to be working on trying to cut this more (it's already down somewhat), but I can't imagine right now that I could get it below, say 12-15 hours with the kind of optimization tricks I've been using.

That makes the current design basically a no-starter for any kind of "mass production" with a single printer. Although the printer is pretty reliable and doesn't need to be "tended" while it's working MOST OF THE TIME, it does require manual starting and stopping of each piece (the "automated build platform" that Makerbot included with the current version of the printer, and that I used at first, has basically been a flop -- not ready for prime time. In theory it allows creation of a "build queue," that would allow the machine to build one part after another on it's own. But the version they've released now isn't really ready for prime time, and I switched to the earlier model simple build platform.

If I (or anyone else) were to try to go into serial production of kits for limited sale, I think they'd have to get multiple machines. At c. $1,500 per machine, though ... well, you can see the economics don't really make sense ...

[pickelhaube]

That is a lot of time.

The use of these machines may not have it's advantages in production.

But prototyping may be the niche.

It takes soooo much time to make prototypes but after the mould making is done , the copies can be made relatively fast.

But you do have the additional materials which are quite pricey.

How much would you say material cost would be in your S1 ?

I would say it would cost about $30-$50 worth of resin for one unit.

RTV cost about $300. You could probably get about 20 units before your moulds start breaking down unless you use the slow set RTV.

But I could pour one unit in about 4-6 hours.

All that being said if the machine makes a unit by itself . With little input by technician. This could be the way to go. You just need to allow for the time involving process.

[gburch]

Rocket Factory update:

Having discovered my dimensioning error mentioned above, I've sucked it up and gone back and reprinted a full set of fins, and then continued on with including as much printed-in detail as possible to the bottom body section. Here are some pics of the fin assembly and body printing process in progress:

Here are fin components in various stages of assembly. On the right, you can see the top part of the fin, still attached to the "printing armature" -- elements of the print which are cut away and serve to hold the part straight during the print as it tapers upward and would otherwise not have sufficient mechanical strength while the plastic is still semi-molten. Next to it is the top fin part after having been cut away from the print framework/armature. To the left are fins in various stages of assembly from the three parts that make them up. You can see the pretty big gaps between parts that have to filled in the finishing process.



Here's a close-up of the top fin part, as it comes out of the printer. This shows detail of the support structure I've designed into the part:



Here's the bottom body section about 1/4 way into a print. You can see the external support structure for the O2 vents built up, and the first layers of the O2 vents themselves being laid down. On the right is the smaller steam vent, which is fully printed (and was enlarged slightly over the last pic I included of it, as it also suffered from the same error in reading the blueprints):



The part a little further along in the print:



The finished part, offered up by the printer:



The part after initial removal of support structure, but no fine clean-up (and, obviously, no surface finishing):



You can see the fully assembled fins, awaiting their first rough sanding and filling.

That is a lot of time.

The use of these machines may not have it's advantages in production.

But prototyping may be the niche.

It takes soooo much time to make prototypes but after the mould making is done , the copies can be made relatively fast.

But you do have the additional materials which are quite pricey.

How much would you say material cost would be in your S1 ?

Basically, cheap as dirt. So cheap, I haven't bothered to even estimate it but I'd swag it at well under $10 for the plastic in the bird -- probably more like $5.00.

I would say it would cost about $30-$50 worth of resin for one unit.

RTV cost about $300. You could probably get about 20 units before your moulds start breaking down unless you use the slow set RTV.

But I could pour one unit in about 4-6 hours.

All that being said if the machine makes a unit by itself . With little input by technician. This could be the way to go. You just need to allow for the time involving process.

Huh -- I had no idea pouring took that long. While the print process *IS* long, and you *DO* have to add in surface finishing, I'm thinking there MAY be some kind of rough cost/effort equivalence point in comparison to traditional methods. I haven't gotten there yet with the technology and techniques I'm using now but, as I said a million years ago when I started this thread, we may be getting close.

[gburch]

I'm being more systematic about trying to keep track of how much time and effort goes into each step of what I'm doing. Here's a pic of the current fin set after three of the four have been rough sanded after the first big, sloppy coat of acrylic modeling paste filler:



This is the biggest sanding stage of the process I'm currently using and takes about 40 minutes to one hour per fin to get to the level shown here. The body sections should go quicker as there are fewer details and printing imperfections to sand around and less changes in contour that have to be managed. If I had to guess, I'd say the whole rough sanding for the complete bird would be a six hour job or so. Buttering on the putty with popsicle sticks before this stage takes just over an hour for the whole bird.

[Bar]

What kind of filetypes does the machine use?

[gburch]

What kind of filetypes does the machine use?

The machine is run by a program called ReplicatorG. It wants the standard CNC .stl input file type. Getting from the kind of output the surface modeling tools I use to .stl was one of the MAJOR challenges I had to overcome in the early stages of this project, and is documented in excruciating detail in some of the early pages in this thread.

The bottom line is that I finally ended up finding a pretty slick pathway from the modeling tool I am most adept in -- Anim8or -- to the printer by first outputting a Wavefront .obj file from Anim8or, then opening that with MeshLab and saving it as an .stl file type. ReplicatorG then accepts the .stl file and automagically generates a .gcode file, which is the actual instruction set for specific step by step moves by the bot.

Along the way, I have pretty much mastered the art of faking the difference between solid modeling (which the .stl file standard requires) and surface modeling (which I know). This is done by maximising "closure" or "manifoldness" as much as possible in the surface model and then using the MshLab "fix holes" utility to clean up anything I didn't catch in Anim8or. By now, though, I've gotten "mesh closure" down pretty much to an art, and so I'm not needing to "fix holes" any more.

Meanwhile ...

[graycap]

You might be interested to check out the following. As you can see 3d printing of AFVs might already be here. SeeRPG Manufacturing featured on the following site http://www.miniaturefigurines.co.uk

As you can see they are offering models of the Cromwell and PzKw IV in 1/72, 1/60, and 1/100.

Evidentlly these models are made used asystem called "selective laser sintering. The process entails the fusing together of layers of very fine powder in this case nylon) into a 3D model by a computer CO2 heat laser. Whatever that means"!

[Bar]

What kind of filetypes does the machine use?

The machine is run by a program called ReplicatorG. It wants the standard CNC .stl input file type. Getting from the kind of output the surface modeling tools I use to .stl was one of the MAJOR challenges I had to overcome in the early stages of this project, and is documented in excruciating detail in some of the early pages in this thread.

Nice.
That all sounds very good.
Are you making your own files, or have you tried printing out any other files?

[gburch]

Nice.
That all sounds very good.
Are you making your own files, or have you tried printing out any other files?

Everything seen in this thread after the first pic I posted is all my own modelling work. I have downloaded and printed a few files, but since my primary interest in this technology is making 1/18 models, I'm almost entirely concerned with creating my own stuff.

[Bar]

Nice.
That all sounds very good.
Are you making your own files, or have you tried printing out any other files?

Everything seen in this thread after the first pic I posted is all my own modelling work. I have downloaded and printed a few files, but since my primary interest in this technology is making 1/18 models, I'm almost entirely concerned with creating my own stuff.

What size is your build chamber?

[gburch]

What size is your build chamber?

10cm x 10cm x 11cm (high) -- thus the fact that large parts are made up of components that have to be glued together

[gburch]

Scenes from the Rocket Factory ...

Components for a rocket body laid out and numbered, with the Rev 0 and Rev 0.5 A4s in the background:

One of the workers at Kleine Peenemunde:

[aae83]

Admit it: the bird with the Apple II is the real mastermind for this project... or perhaps the tiger with a "thing" for CRTs.

[Bar]

What size is your build chamber?

10cm x 10cm x 11cm (high) -- thus the fact that large parts are made up of components that have to be glued together

I understand.
Looking good!

[gburch]

Rocket Report 80 bazillion and one:

I'm now deep into applying the surface finishing techniques I've been refining since the Rev. 0 bird, described above. Here are a couple of pics of the first fin to go through the full "rough filling" process and at the stage of sanding after the first coat of filler primer (thanks, Mathew, for that tip -- it's a real time-saver!):


You can clearly see the three layers of surface here: the plastic, the acryllic paste and the filler primer.

Here are pics of that fin mounted on the bottom body section, after a second coat of filler primer:


The other fins still need final spot putty and sanding before the first coat of filler primer. The body section is probably three steps like that away from the first coat of filler primer. Each of those steps only takes a few minutes, with each one being shorter than the one before.

Admit it: the bird with the Apple II is the real mastermind for this project... or perhaps the tiger with a "thing" for CRTs.

The whole project is bird-brained, so I vote for the cockatoo.

[gburch]

All parts of the Rev 0.5 A4 now have at least one coat of filler primer:



I am now a complete convert to this great primer.

[pickelhaube]

[gburch]

sanding, painting, sanding, painting ...




I don't mind, though, because I can tell I'm going to get a finish better than the first bird.

Meanwhile, I've revised the fin design yet again, with an improvement in the fillet contouring that should decrease filling and sanding time.

[Bar]

Beautiful.
And what would you say the cost would be for one of those things striaght out of the printer?

[gburch]

Beautiful.
And what would you say the cost would be for one of those things striaght out of the printer?

"Cost"? well ... depending on how you look at it, $5.00 (the cost of material) or $10,000 (everything I've spent, plus the value of the time I've put into it) ... or something in between.

Seriously, I'm not close to figuring out what I might charge IF I ever do decide to try to sell a kit. There's three reasons for this. First, I haven't gotten to the point where I have a DESIGN I would call "final." Second, I haven't figured out what a reasonable production process would be for a kit. And third, I haven't even figured out what the parameters that should go into pricing should be, much less how I'd balance them against demand -- assuming there would be any ...

As things stand now, I'll probably have a "final" design within the next couple of months. I had hoped to be able to get a better handle on the other two factors in the same time frame, but now I'm not so sure. The key element (I THINK) in this exercise is getting a second machine to use for "production," building it, calibrating it and translating all the work I've done into a form the second machine can build. This last factor will be necessary, because there have already been three major upgrades in the Makerbot product, and translating the designs I'm using from ones tailored to the machine I've got now to the new machine will take some work -- how much, I don't know.

Beyond this, before I would ever "go commercial" even in a small way, I'd want some very experienced modelers to build some kits and give me some feedback about feasibility. Given that I really shouldn't spend the money for a second machine for a few months, at least (to say nothing of the time -- my wife is already starting to have "3D printing widow syndrome") ... I think it's not reasonable to think anyone other than "beta testers" might get their hands on a V-2 kit before ... Christmas (?)

... but then, I might have an attack of sanity and ditch the whole idea ...

[Bar]

I see what you're saying.
I love the design of that rocket.
And i'd love one to pop into a diorama.
And i am an experienced model maker, lol.

[gburch]

Rocket Report:

Final surface finishing on some parts, while first coats of white gloss lacquer paint go onto others:

A full set of the latest iteration of fins in the foreground:

This new design has slightly fewer print problems with the outside edge of the fins, and the new, smoother fillet on the bottom.
A view showing the fin joint in close-up:

A couple of notes here. The finishing technique I've used on this Rev 0.5 rocket can produce a surface as smooth as one has patience to work with. Once again, I can't stress enough how much of an improvement Mathew's suggestion of using auto filler primer was. My wife is an artist, and she often says that the difference between an amateur and a professional is just a big enough collection of little tricks. Well, that is one little trick that makes a BIG difference!

Second, the fin-body joint is quite visible in this pic. When I look at it and it bothers me, I think two things that cause me to stick with this design. First, there are lots of joints like this on "XD-type" models that are the result of manufacturing and "buildability" compromises. Second, super-scalers could completely deal with this by permanently joining the fins to the body and running a putty/paste fillet along the joint and sanding it before final finishing. I MIGHT do that on my next full build if I feel confident enough that I've gotten to a "final" design.

Three major issues that haven't been addressed at all so far are the vanes that extend inward from the inside of the bottom of the fins into the rocket exhaust, the rocket nozzle (an area not shown in any pics I've posted for a while), and panel lines. I have ideas for addressing all of these, but I'll save that for future posts.

[gburch]

An example of rapid prototyping made possible by 3D printing:

On the right is the nose section I've been using up to now, finished and ready for painting. On the left is a refinement in which the warhead section is a separate, removable piece with a snuggly-fitting connector. The new pieces have the first coat of filler drying before sanding.

I made this change because it will ease painting (the warhead section was often a different color), and will also allow modular "upgrades" like the White Sands "Bumper" flight.

[gburch]

Warhead production:



The removable warhead work described in the post above was just a proof of concept, using parts I already had printed out. The "cut" in the 3D model for the top-most part was too high to really be a scale warhead, so I went back and re-modeled the last three segments of the body so that the dimensions of the warhead would be accurate to scale. I improved the design of the coupling, as well. The result is the nose section and warhead on the right, filled, primed and ready for painting to be incorporated into the Rev 0.5 bird.

The warhead in the center is having its first filler at the tip. This is a dip-and-drip process for the very top-most part, since the printer does a poor job on this because it comes to a fine point and the deposition process for such single-point elements inevitably causes deformation. On the left another warhead has its first dip-and-drip coat drying in a cradle made up of scrap parts. The coupling parts are in the foreground, still on their print bases.

Meanwhile, I'm working on refining the body segments where the fin section joins the section housing the oxidizer and fuel tanks. In my original rush to produce a proof of concept, I glossed over important details in this part of the rocket and am working on trying to create a maximally accurate depiction of the details in this part of the model.