3D Printing -- Pretty Close to Ready

[CW4USARMY]

Cool! Cant wait to see your creations!

[gburch]

So, below are a couple of large images to give you an idea of what my Makerbot can create NOW. I include a 1/18 fig for comparison.

A few notes:

-- The roll of ABS filament you see is the raw material the machine is using.

-- These prints are a product of the first usable set of parameters for the machine I've managed. I think quality will improve somewhat with tweaking.

-- The surface texture resulting from the layering is probably not going to go away. I have some ideas for dealing with that, but it will take time to develop and implement them.

-- The little "stringies" you see can be decreased somewhat with tweaking, as can the gaps. But that will be a major job of trial and error.

-- I have a LOT to learn to get to the stage where anything that might look good with existing 1/18 model tech can be produced.

[PanzerArm]

Very cool! We have one of these machines at my work that I have been thinking about asking the engineers to see if they can make certain parts of custom vehicles that for me are rather tedious.

-Kevin

[gburch]

an update ...

The machine is working well, although I've identified one issue that may require major disassembly to correct. The belt that moves the x-axis is very tight and the means of adjusting it that is supposed to work doesn't. This causes a deformation of parts along the x-axis (so that, for instance, something that's supposed to be a perfectly circular feature along that axis has a slightly oblate shape). Unfortunately, that mechanism is buried deep inside the machine and I'll have to suck it up and spend some hours to get at it.

Another issue that's even more time consuming is that I've discovered that something I'd hoped would be easy isn't. I'm a pretty accomplished 3d modeler for computer graphics and simulation purposes. You can see some of my work here:

http://www.orbithangar.com/advsearch.ph ... er=default

Unfortunately, I've learned that something I always knew about is a MAJOR issue: there are two fundamentally different kinds of computer modeling; surface and solid. Graphics and simulation modeling is done with surface tools. The ten years of learning I've got invested in 3d modeling is all in surface applications. CAM work like 3d printing HAS to be based on solid modeling. I had known this in a superficial way when I dove into this project, but I had assumed that there would be some relatively easy was of exporting finished models created with surface apps to solid model file formats.

There's not. The differences between surface modeling and solid modeling are so fundamental that it turns out that the only programs that have been written for this kind of thing are clumsy, hard to use, error-prone and not really satisfactory. I've gone through all the stages of grief on this (anger, denial ... etc.) and finally admitted to myself that I'm going to have to suck it up on this one, too and basically start from scratch. I did a lot of research to see if there are simple, inexpensive solid modeling tools so I could minimize the learning curve and try to apply as much of what I already know about surface modeling to solid modeling.

There's not. So last night I bought the cheapest and allegedly most simply "pro-sumer" solid modeling package I could find -- Alibre Personal Edition. I've had a couple of hours of fiddling with it since, and can see it will be some time before I've mastered even the basic tools (shaping, extruding, cutting, etc.)

Bottom line: just to maintain my sanity, I may continue to report my progress here, but I think it'll be a while before I can make something that even begins to fit with existing 1/18 stuff ...

[Jay]

Gburch, I use Solidworks at work (so yes, it's a solid modeler). If you need any parts drawn (and have have drawings/references/parts to dimension from), I could do some up for you. What type of files does the machine/software use?

[gburch]

That's a kind offer and very tempting, considering what I've been learning about solid model application interfaces. I'm so used to the grab/push/pull/etc. interfaces I've been using in surface modeling software for years and years. The CAD-ish paradigm that Alibre, SolidWorks, etc. are based on is like returning to my high school drafting class in 1973, compared to the way the programs I've been using work. I've looked at hours of video tutorials for Alibre now and am just amazed that you can't just drag a face to identify it and yank it around with the mouse to put it where you want. I understand why they do this -- the idea is to be precise so a CNC tool can deal with your work, but I can't believe a better basic modeling interface for these programs can't be developed.

Anyway, I may come back to you if I get to the point where I feel like I can't master the Alibre tools, but I've got this hard-headed idea that I SHOULD be able to do everything myself. I keep thinking that if I'm smart enough to have made simulation software models that have thousands of fans, and smart enough to build a 3D printer from a thousand fiddly little parts, then I ought to be smart enough to learn how to use Alibre.

[Jay]

Haha, I think the same way about surface modelling! I did Product Design at uni so we were pretty much schooled on solidworks from day one. We also used Rhino as a surface modeller. I started to get the hang of it near the end but I find Solidworks childs play.
It's all about sketching shapes first, then extruding the sketch to make a 3d shape. Then (like hand drawing) all you really do is add or subtract shapes to the first shape to make more complex shapes. But yeah, I guess it's easy for us to say "you just do this, then that and voila! You have an engine modelled....etc"
The other thing is that i've been using Solidworks for over 10 years now...but i'm STILL learning things. Especially when they come out with newer versions/tricks each year.
Sometimes when it comes to this type of stuff, you have to unlearn what you have already learnt......if that makes sense

[gburch]

Very well said. It's about unlearning what I already know.

The program I'm struggling with -- Alibre -- seems to work pretty much like SolidWorks -- you make a flat 2d figure and extrude. I've figured out that a curved tube (which would be an obvious necessity for my first major target, making something like a drop tank) is done by creating a series of circles on planes spaced along the long axis and then "lofting" an extrusion along the circles. I can model a shape like that in a surface modeler in under a minute. In Alibre, not so fast ...

Anyway, I'll struggle on for a few more weekends and see if I can get the hang of it.

Another option I'm looking at is Blender. It's a surface modeler, but it looks like it's got solid exporting capabilities. On the other hand, it too has a pretty kludgy interface ...

[gburch]

... at the risk of trying forum members' patience, I'll keep using this space as a sort of diary of my on-going efforts to put together a 3D printing capability that might someday support at least a little 1/18 work. If anybody else ever tries to go down this road, maybe they can learn something from my trial and error process ...

I've achieved a breakthrough on the software side. It turns out that Blender does indeed export to the .stl file format that is the standard for moving between the design stage and the building stage. Blender, being an open source, "hacker-ish" creation that seems to be attempting to satisfy too many user communities (modeling, rendering, animation, CAD, etc.) has a deplorably messy interface and I couldn't see using it as a true modeling tool.

So this morning I tried an "assembly line" approach. I exported a mesh I'd made in Anim8or (which is one of the simplest and slickest mesh-modeling apps I've ever used -- super intuitive for the basic job of creating a shape) to a Wavefront .obj file, which I imported into Blender for the sole purpose of translating into an .stl export.

Eureka!

ReplicatorG, which is the app that imports .stl and automagically generates the "gcode" file that is the actual machine code that tells the printer which specific moves to make to achieve the print, accepted this file without a hitch, and I've done a successful test print.

So I've now established a workable workflow system for creating 3D models and turning them into an instruction set for the printer.

Ahead lie many challenges. My online research in the Makerbot community indicates that there is no easy solution to the mechanical issues that underlie the x-y distortion I'm getting in my prints. The probable cause of this lies deep inside the machine and addressing (and hopefully fixing) it will require a major mechanical disassembly to get access to the problem area. That will have to be a job for some weekend in the future.

[gburch]

... another update in my little online diary as I work to try to make current-tech hobbyist-scale 3d printing into something that might be usable in the 1/18 hobby ...

-- I continue to get better at translating my 3D models into the format required to print. I'm now using MeshLab to convert the meshes created by the modeling software I'm most comfortable with into the .stl files that is standard for CNC machine input.

-- I've also run enough calibration prints through the machine to achieve acceptable scaling of the model to a final print size. Thus, getting to 1/18 (or any other) scale is just a matter of doing the kind of arithmetic any scale modeler is used to doing.

-- I picked the V-2 as a first project to play with because it's basically made up of very simple shaped that are relatively easy to model in 3d software, and also because those shapes are easy to "close" into solid (rather than surface) models that CNC work requires.

-- because I'm still struggling with some asymmetry in the output of the X and Y axes (the plane parallel to the "ground"), and also to be sure I can fit the pieces in the build volume, I've cut the V-2 body into quarters down the Z axis, i.e. from top to bottom.

-- right now I'm printing the first vertical inch of the body in four quarters. Given that a 1/18 V-2 is just about 30 inches tall, I doubt I could do a full production model this way. That would mean that the main rocket body would be made up of almost 120 quarter sections -- more than any modeler would likely be willing to fiddle with to build a kit.

-- I'm beginning to think that only the most avid fan of this technology would be willing to do the post-printing work required to produce a good looking finished product, with the output from the kind of printer I have. I've just begun to experiment with how to deal with the imperfections and surface texture left behind by the printing process. Right now, I've achieved somewhat satisfactory results by covering the whole outer surface with putty, buttering it on with a flat blade, and then sanding it down to a smooth finish. This works, and I think some day I'll have a decent-looking V-2 doing this, but I have doubts about whether anyone else would be willing to do all the work that this entails.

[gburch]

here's a pic of the very raw first stab at the very first parts of a V-2 body. This is the "business end" hollow to the depth of the nozzle, in two layers of quartered sections. They haven't been glued together -- they're just sitting on top/next to each other, so the gaps between parts are greater and the alignment between parts is poorer than they will be when they're fastened together. And of course, there's no sanding or puttying of the outer surface.

[gburch]

Between work and a bout of the flu, I haven't had as much time as I'd like to continue trying to master my 3D printer, but I have made some progress. I diagnosed the problem that was causing round objects to not be round, which required opening the machine up and rebuilding part of it. But the results have been pretty good:



That shows a couple of test prints of the the first two sections of the v-2 rocket body with circular precision good enough to do whole sections in a single print. The large body section on the left is the tallest I can get out of my machine. That a considerable improvement over what I was originally achieving, and makes the V-2 project actually realistic -- to me, at least.

Here's a pic of the latest of a bazillion test prints of one of the lower sections, showing how I'm building the fin fillets into the body:



I'm pretty close to having solved all of the problems with both the printer, and the model-specific problems on the V-2. At the rate I'm going, maybe by summer I'll have a 1/18 scale rocket!

[gburch]

Another update. The real world continues to intrude on my efforts to produce usable product from my printer, but this weekend I have a few hours to work on version nine bazillion of my V-2. Each major component is presenting design and production challenges, but I'm making progress, finding at least acceptable solutions to the issues as they crop up. One major benefit of this technology is the ability to engage in a fairly quick design-build process that allows continual tweaking and improvement.

Here's a pic showing various versions of the components that I've given the most attention to:



In front of the German officer figure you can see a test piece for the fin from a couple of weeks ago I've been using to develop surface finishing techniques on. As I'd hoped, covering a relatively flat surface like this with putty and then sanding produces an acceptably smooth surface for painting. You can see a little putty on one of the body sections on the left as I'm beginning to do the same kind of thing on compound curves to see how hard getting a smooth finish will be on that kind of surface.

Over on the left, you can see my very first attempt to make a 1/18 scale human figure. It's very rough and was intended as just the simplest proof of concept. I can see that with a good deal more work, I could develop techniques for making at least body parts for figures in this scale.

[gburch]

Here's the latest in what seems to be the developing pattern of weekend updates.

I've had a number of breakthroughs resulting in reaching what seems like critical mass in developing my own knowledge and skills with the 3D printer and combining what I'm learning into design tweaks in 3D models and minor changes in machine parameters The visibly most significant of these in terms of the V-2 model is what seems to be a workable set of methods for making the main body structures at or close to the limits of the bot's precision and size, that have the added benefit of taking the least time of any of the methods I've explored. Instead of taking four or more hours to make one of the body sections, with cracking and delamination along the way, I'm now cranking out one of the sections in just over an hour, and with no flaws in the printing process. Thus, the top three sections you see in this pic were made in a half a day, instead of (with false starts and total print failures) a full day per section for the bottom three sections:



(Note that production examples from testing and tweaking have taken over more than half the space of my WWII diorama, which has been dusted off and moved to at least temporary storage piece by piece as work on Herr Professor von Braun's little toy has grown to take over more and more table area.)

Here's a pic of the body sections laying flat, with last weekend's product in the foreground:



(I think the guys in the jeep are part of Operation Paperclip.) My current plan for the development of the body is to go on building "up" from where I am now and then move back to apply all I've learned in the current version to the bottom three sections.

Work on the fins moves more slowly. I honestly think a CNC mill would be a better tool for flat-ish objects like this. One that would do the job pretty well is very close to being released at the same stage of technology development as the Makerbot:

http://www.mydiycnc.com/

Based on the status of that project, though, I'm thinking it won't be at a stage I'd try until probably mid-year, at the earliest.

[gburch]

The breakthroughs keep coming. The pic below shows what is getting close to an acceptable "Version 0" or true "Beta Version" of the V-2:



I woke up the other day with a whole new approach to the fins clear in my mind, rushed to create some test pieces and, within an hour, had a full test version completed. This is the fin on the left in the pic above, which I've been using for continuing development of surface finishing techniques. The fin on the right is a tighter, cleaner fit that represents what will likely be the Version Zero printed version.

I'm now making progress on faster, better surface finishing. Currently I'm refining use of Liquitex "Artist Acrylic Modeling Paste." A test piece I have drying right now uses a 50-50 dilution of this with water, and looks like it will yield the best results yet, in terms of smoothness, speed and minimization of sanding (but there will still be LOTS of sanding involved in production of a finished product). I've painted a test body section with Tamiya flat black laquer spray can paint that was finished with Liquitex modeling paste, and it looks quite good -- very smooth.

Body section printing and joinder is getting better and better, although I was somewhat disappointed to conclude that the very thin-walled printing that works well on the upper two-thirds WON'T work on the lower section with the fin fillets. Because of the complex, compound curves, more internal support structure is required to avoid major deformation and delamination of print layers in the lower sections. Thus, even with a somewhat refined design and print method, those sections take 2-3+ hours to print for each section.

In the pic above, you can see the blue painters tape I'm using to tack the internal section mating collars, as I work to refine those before gluing them in.

I'm now thinking through some of the finer details, such as the guide vanes below the nozzle and the various bumps and fairings on the fins and elsewhere.

I may post some thoughts this weekend about where all this is leading in terms of producing a true "Version One" model.

[pickelhaube]

I am following you on this .

Is it possible for you to print ribs or runners as you layer up ? That should give you some rigidity to hold the " tubes "

Would it be a big deal to program that into the ploding ?

[gburch]

I am following you on this .

I'm glad -- although your interest more than anyone else's on the forum makes me nervous about the possibility of ever offering anything I might create for sale ... You set what may be an impossibly high standard, I'm afraid.

... but maybe more about that later this weekend...

Is it possible for you to print ribs or runners as you layer up ? That should give you some rigidity to hold the " tubes "

Would it be a big deal to program that into the ploding ?

The simple answer is it's no real problem to design essentially anything into the shape that gets fed into the printer. And the printer will faithfully TRY to print just about anything you tell it to. The hard part comes in figuring out designs that the printer will create WELL. For instance, there's some fairly complex thermal problems that come with large shapes. The build surface is heated and that heat level can be set and regulated. And the plastic comes out of the extruder at a defined and controllable temperature (and rate).

Problems come with plastic that's been laid down well above the build platform that will cool at a different rate than the plastic that's closer down toward the build platform while the build is still on-going. After the build is complete, all the plastic will cool down to the same temperature and shrink slightly -- DIFFERENTIAL cooling (and therefore shrinkage) during the build is one of the primary causes of the delamination and deformation I've been discussing in my "progress reports" above.

Beyond this is the REALLY tough problem of "toolpath." This is the precise ballet of movement of the x, y and z axis stages that defines how the build progresses from moment to moment and layer to layer. I'm using a really good tool that automates toolpath generation, BUT it makes some assumptions that I have yet to really even begin to master. Instead, what I've done is try to figure out what priorities the toolpath generator uses, and design the parts so that things don't get crosswise during the build. So far, this has worked fairly well, but I'm trying to get through all the various parts to an acceptable "Beta" version of the V-2 as a baseline from which to continue to make improvements in the future.

To make a long story short, I'm DEFINITELY going to investigate more complex "internal" structural design to decrease the amount of plastic that's being laid down in the more complex shapes that will minimize build time (which will have the benefit of decreasing the impact of differential cooling), and also have a workable toolpath.

It's all very much a work in constant progress. I sincerely appreciate your interest.

[gburch]

Weekend Update ... (what is this, Saturday Night Live?)

Progress continues into a stage I'm calling "Fins and Finishing." Here's a pic of Version Zero as she stands now:



Note a couple of things. I haven't stopped my frenzy of designing and building long enough to fab a stand that can support the bird's weight with stability and also allow a close fit of the developing fin prototypes, so she's sitting on an upturned body bottom piece (from the early prototyping era) that happens to be flat and as close as possible to the right dimension. Since I've ended up making the parts of the fin fairings that extend BELOW the body as part of the fin assemblies, and these parts also extend inside the outside diameter of the lowest body section, for now the fins can't fit tightly onto the registration tabs that are part of the lowest body sections. Thus, the fit of the fins in this and pics below isn't as good as it will be when they are mounted so they can support the rocket's weight. Figuring out what stage of the final assembly process the fins should be glued onto the body is something I really haven't devoted too much thought to.

Here's a closer shot of the fins:



The fin on the left is the version currently closest to a Rev. Zero version. Looking at the earlier version on the right, you can see I've been having fits with the outside "mold line" (that's an archaic term in 3D printing, but I can't think of a better one) of the fins. This is because the fins taper to a point in cross section, and such super-thin elements go beyond the printer's resolution. The result is a ragged edge that's neither straight nor solid.

The fin on the left represents the solution to this. I ended up dragging the fin edge outward a little, so that all those ragged edges are outside of the final finished line, then, after assembly, I sand them straight. With the version on the left, I think I've found a workable solution to this issue.

Note that all of the fins are in relatively early stages of surface finishing, so the rough surfaces aren't representative of finished product.

The next two pics show body section in close to finished form as they should be before painting:




The top pic is the three-section nose assembly that has been glued together with more stout coupling rings that make it completely rigid. There's' coat of filler / putty drying on this, which accounts for the gnarliness (to use a Sheenism) of the point on the nose.

The bottom pic shows the mating line between the body mid section and the lower section. The mid section has also been glued together with stout Rev. 0 coupling rings and is also quite rigid. It has been filled with acrylic paste and sanded to a paint finish. You can see the contrast with the lower body sections, which have had no surface finishing work.

SOOO ... some thoughts on where this is leading. I could have a Version Zero example of the V-2 ready for painting in two weekends. (FWIW, I'm going to paint this one, appropriately, in the big black and white checkerboard pattern that the early Penemunde test A-4s used). By the time the final coat of paint is dry, the number of hours I will have in this project is truly scary.

Could I ever create something that others might accept as a kit? That's the tough question. I don't feel like I would have the courage to offer even a tentative product until I've designed, modeled, printed, finished and painted a "rev. 1" myself. (Again, FWIW, I'm thinking this would be in the simplified production 3-color splinter camo scheme that most of the birds that were fired off at Holland and Britain used.)

Even what would be a very "finished" kit by the standards of the process I've been engaged in would be a major modeling challenge. I'm thinking it wouldn't be something that anyone other than a pretty accomplished and PATIENT builder would want to approach.

An option I'm thinking about: I MIGHT consider selling kits to the folks on the board here who do custom work for others, for resale by them. This might avoid people getting kits and being unhappy with how much work there is to create a finished product.

But that lies well into the future -- probably mid to late summer at the very earliest. One thing I'd have to figure out would be what would be a fair price for a "master builder's kit." The current version probably represents 20+ hours of printing time if I add up the time for all of the components in their final form. (That doesn't account for prototyping -- that's well over 100 hours of print time at this point, I'm sure.) None of this takes into account the time I've spent designing and modeling in the computer, or development of finishing techniques.

When I think about that, I realize there's no way this endeavor could ever be "profitable," in any meaningful sense of the word, to say nothing of the fact that while the printer would be churning out pieces for sale, I can't use it for my own projects. I MIGHT consider buying another machine just for production of parts for sale but, again, that adds a whole new economic dimension to this craziness.

Bottom line -- I just don't know ...

[Jay]

......maybe this would be good for making model mould masters. So you would just finish up the surface as usual with some bog/filler and sanding, then use it as a plug to take moulds off. I don't know about you, but I like the idea of designing the object in CAD first to get it as accurate as possible, then transferring it to the 3D print (taking into account the slight variances in plastic deformation during the printing process).

[Black_Dragon_One]

Here's the latest in what seems to be the developing pattern of weekend updates.

I've had a number of breakthroughs resulting in reaching what seems like critical mass in developing my own knowledge and skills with the 3D printer and combining what I'm learning into design tweaks in 3D models and minor changes in machine parameters The visibly most significant of these in terms of the V-2 model is what seems to be a workable set of methods for making the main body structures at or close to the limits of the bot's precision and size, that have the added benefit of taking the least time of any of the methods I've explored. Instead of taking four or more hours to make one of the body sections, with cracking and delamination along the way, I'm now cranking out one of the sections in just over an hour, and with no flaws in the printing process. Thus, the top three sections you see in this pic were made in a half a day, instead of (with false starts and total print failures) a full day per section for the bottom three sections:



(Note that production examples from testing and tweaking have taken over more than half the space of my WWII diorama, which has been dusted off and moved to at least temporary storage piece by piece as work on Herr Professor von Braun's little toy has grown to take over more and more table area.)

Here's a pic of the body sections laying flat, with last weekend's product in the foreground:



(I think the guys in the jeep are part of Operation Paperclip.) My current plan for the development of the body is to go on building "up" from where I am now and then move back to apply all I've learned in the current version to the bottom three sections.

Work on the fins moves more slowly. I honestly think a CNC mill would be a better tool for flat-ish objects like this. One that would do the job pretty well is very close to being released at the same stage of technology development as the Makerbot:

http://www.mydiycnc.com/

Based on the status of that project, though, I'm thinking it won't be at a stage I'd try until probably mid-year, at the earliest.

can u do funtion jet wheel like the flyinhg wing jet wheels?

[gburch]

......maybe this would be good for making model mould masters. So you would just finish up the surface as usual with some bog/filler and sanding, then use it as a plug to take moulds off. I don't know about you, but I like the idea of designing the object in CAD first to get it as accurate as possible, then transferring it to the 3D print (taking into account the slight variances in plastic deformation during the printing process).

I know exactly nothing about the techniques of mold-making that we see masterwork examples of here on the forum. In theory, so long as the mold-making didn't subject the printed parts to heat that would deform the ABS, then, yes, I suppose this is possible.

Maybe over the next year or so I'll try to team up with one of the masters of molding techniques we see here and try to develop a good work flow between the techniques I'm developing and the established ways of the molding gods.

To make that a useful undertaking, I'd have to learn enough about the mold-making process to create designs and printing functions that produced parts suitable for mastering.

can u do funtion jet wheel like the flyinhg wing jet wheels?

If I understand your question, then again, in theory, yes, I could make working landing gear. But I'm certain that this would present challenges I haven't even thought about yet, as the size and shape of parts like that is completely different from what I've been working on with the V-2. I picked this project because the shapes involved are RELATIVELY simple and fit well with what I knew about the design and printing process at the time I started.

One of the things I've learned with great effort is that the level of hobbyist 3D printing I'm exploring isn't at all at the stage where you can feed it any design and the machine figures out how to build it. A very significant part of what I've been doing is figuring out how to design parts that are as accurate to sale as possible, while at the same time being able to fit through the many steps between 3D design in the computer and finished product that comes out of the printer.

[margerisons]

Any update on this?

I looked at the exact same printer for work and hobby.
What's it like for making 1/18 stuff? any good for vehicles?

[gburch]

What's it like for making 1/18 stuff? any good for vehicles?

That's what this thread is about. I suggest you review it in detail if you're thinking of doing this. The bottom line, though, is that there's a lot of work involved. You can't expect to just download or create a neat 3D model of your favorite tank and expect to push a button and have it pop out of the machine.

[gburch]

I've ordered the black and white paint for the Rev. 0 version, which I will have before next weekend, so I'm on track to start painting then. Meanwhile, I continue to work on fine finishing:



You can see that I've now printed a stand that holds the rocket firmly while I work on it, and that I can use for fitting the fins, and will also serve as a painting platform.

I've also begun design and prototype work on some of the parts for Rev. 1. These include fins with a refined, closer-to-scale cross-section, and experiments with improvements on the bottom body sections that are aimed at reducing print time.

[CW4USARMY]

I admire you persistence and patience. Already looking great and will really be something when its all complete!