Prints, Prints and more Prints.

Just a little blog post showing off some of the prints i've done so far, some are objects of my own design, some are from thingiverse.

One of my first upgrades was this print, its a lm8uu bearing holder i use with my y-axis. Works quite well and is on thingiverse:

Next, the obligatory Pink Panther woman from thingiverse:

One is at 50%, with infill, the second is at 100% with no infill, both are at 0.1mm resolution and they come up brilliantly. Im not sure what type of PLA the blue is, but its got a very opalescent quality to it that makes almost everything quite pretty.

Then some more printer upgrades:

This is my Custom designed y-idler mount, which wasnt around very long as i realised since im going to make a design with a base, i might as well make one with a tensioner that would fit on my own printer, so it was replaced with this (another of my own design):

This one works very well, very adjustable and comes together very easily.

Next, at the other end of my Y-axis i created a motor mount:

Again, this is my own design, works well but not much different from what you might see on other people printers.

My Next print was one of my my more complex and ambitious models. Its a box for the control board on my printer and its quite complicated in design, but its finally come together nicely:

The print itself didnt come out so great, but its quite good and functional, and heres the lid:

Then, the rest of my prints are mostly from thingiverse for friends:

There are quite a lot more prints then this, but these are the highlights..

My next post will be about the incremental upgrades i've made to my printer... from rap to strap!

The Software - Design and Control.

Control

3d Printers obviously require some software to drive them. Currently this is a pretty straight forward stack. Most printers use an arduino for hardware control and most arduino are using usb (as a usb-to-serial device) to talk to the printer.

I use linux, you plug an arduino into it and it'll come up as serial so theres no driver problems. Windows folk sometimes have issues, but generally those work with externally provided drivers.

The software stack you'll need for driving a 3d printer falls into a couple of areas

1. Drivers
2. Firmware
3. Firmware updaters
4. Model Slicers
5. Printer control

As a general rule this will mean:

1. Arduino IDE (for re-uploading the firmware - and you'll need to do this)
2. Slicers - Generally people are now using Slic3r, but sometimes sfact/skeinforge
3. Printer Control - Generally this is Pronterface (or printrun)
4. Firmware - there are several, Marlin, Sprinter, Repetier, 

The arduino IDE is used really only to update calibration data for your printer and then flash the firmware onto the control board. Current control boards all allow you to do this over the same usb interface you print with, older ones needed a programmer. Its pretty easy and im not going to go into that in any depth.

Slicers are software that will take a model (generally an STL) and convert it into "movements" that your printer will then perform in order to print your object - also known as gcode. There are several options for slicers. The most popular was Skeinforge/Sfact, but this is slowly being replaced with the simplier and easier driven Slic3r, but there are other decent alternatives such as Cura.

For printer control, theres really only 2 options, the almost all-consuming pronterface/printrun and repetier host also has its own control stack (which can drive both normal g-code based printers plus it has its own protocol for which you would need the customised repetier firmware)

For firmwares, theres generally 3 that are in use, Marlin (probably the most popular), Repetier and Sprinter (one of the original ones). Marlin and Repetier are both "step ups" from sprinter, they provider a few little extra's that you may or may not notice. They're all capable of being controlled by the same gcode though so its really quite simple to use them all.

And thats it.... Everything you see above is available equally on Linux, Windows and Mac OS and is FOSS. You can print for free!

However, there is one other component to the software stack that is almost a requirement. Often when you get models (stl's) they'll be "broken" in some way. Very common on thingiverse and even CAD software sometimes outputs broken stl's. As such this has led to the use of a bit of software called NetFabb Studio (or netfabb cloud) which can repair these broken models. Its available for free on all platforms in the form of its basic studio, and hopefully it stays that way, but its the one component of the stack that remains in a hazy proprietary license.

Design and CAD

This bit is optional and only required if you want to design your own parts. Again, im going to focus mainly on the free options out there (or at least those that available to linux-sporting folk such as myself).

As an x-engineer im quite fluent with AutoCAD (and rhino as it happens). As an x-3d graphics guy, im quite fluent with just about every form of software out there, Maya, 3ds, lightwave, Blender, this list is long.

Both blender and Maya are available in linux formats though since AutoDesk aquired Maya and they seem very anti-linux so god knows how long it will be before that linux port starts to disappear entirely. 

Blender is the only FOSS one in that stack with a largish community. And Blender is quite amazing - quite competitive with the Prioprietary 3d stacks.

For designing printable parts, you'll generally want CAD style software however. While 3d (artistic-focused) software is capable of generating objects you can work with, they lack alot of precision and often dont even care much about units. So heres a list of some of the software that falls into that category:

1. AutoDesk AutoCAD or Inventor (not cheap)
2. Rhino 3d (not cheap either)
3. FreeCAD (FOSS, GPL)
4. OpenSCAD (FOSS, GPL, parametric modeller)
5. Netfabb (never used it for this purpose, but i believe it does this?)
6. Heeks Cad
7. Google Sketchup.
8. Solidworks.

If your on a linux desktop like myself then there are a few options, but i've settled on two mainly, FreeCAD and OpenSCAD. These are both very CAD oriented applications with very very different methodologies.

FreeCAD

FreeCAD aims for an experience most CAD sporting folks (such as autocad users) will find more appealing. Its not hugely like autocad, its just a similar design paradigm. Its much simplier then AutoCAD or inventor, and there are features i really miss from both those but you can do quite well with it. If you browse on over to my model viewer, you'll see some quite complicated models i've designed in FreeCAD. It is very functional, but it has some serious pain points for an inventor/autocad user:

1. Rotations are a nightmare - I cannot stress this enough, this is my most hated part of FreeCAD. The co-ordinate system they settled on for applying rotations to a model makes me want to kill myself.
2. No xrefs - if you know what these are you know how wonderfull they are. Cant do this with FreeCAD
3. The tools you have available to model with are quite simple - AutoCAD has a tool for almost any user, often these tools greatly overlap. There is almost no overlap in tools in FreeCAD
4. No Snapping in 3d - this is generally another painful problem.

But there are also some plusses to it

1. Unlike autocad its been designed with 3d in mind from the beginning so it does 3d quite well (in fact you could say its no use as a 2d CAD in reality). AutoCAD on the other hand isnt great at making 3d models, it has the tools, but they're sparse and inventor is a much better product for this.
2. Quite parametric. I initially hated the lack of a command line (that autocad has) on freecad. It has a python console, but its really nothing like AutoCAD's command line. However, every model is really designed via dimensioning when you create it and as you build up a model it forms this heirachy of sequences that end with some complex model. The beauty of this is that you can then go back and modify parts inside your model on the fly. This is a very good part of FreeCAD and should not be under-estimated - its the most useful component of the FreeCAD stack.

Thats a brief view of how I see FreeCAD stacking up. Its probably the best CAD option for users coming from a CAD background.

OpenSCAD

The premise behind OpenSCAD is VERY different from CAD. Basically you design by programming. It has its own language which you use to generate models and its like designing models by writing software. Its a very powerful piece of software and not to be underestimated. As a traditional CAD user, unless you wrote software for AutoCAD in Lisp you'll be completely at sea with it, but if you persist you'll understand its power. Its available (FOSS and GPL) under linux, windows, mac.

I really only use OpenSCAD to generate gears, but heres what I see are its pluses and minuses, first the minuses.

1. As a traditional CAD user, its a steep learning curve
2. As a programmer, its a steep learning curve (as i am both, its not so bad, but still quite hard)
3. Compiling models can be a pain, it doesnt do incremental compiles as such so as your model gets more and more complex the length of time to show your model gets longer and longer - that is irritating.
4. Lacks some generic modelling tools (such as filleting).

The pluses:

1. If you've coded and used CAD, you'll "get" it pretty quickly
2. Very modular design - much like writing code you can grab modules other people have written to perform tasks for you (such as generating gears) and then simply add them to your model. This is very cool.

However, I think the way OpenSCAD has been written is its biggest downfall. It is entirely self contained, its an IDE, a renderer, compiler and everything all in one. It uses no external components and as such I personally think thats a big drawback. What I think OpenSCAD should have been would be quite simply summed up as "eclipse plugin" with a java-centric design. Some may laugh, but consider what this would give you (keep in mind, im not a lover of java, quite the opposite):

1. Eclipse has substantial tools for compiling (dependency generation, incremental compiles, the works - it is very complete and very powerful)
2. Eclipse can deal with project modularity very well - i.e. I want to make an openscad project that uses Greg Frosts gear generation tools - eclipse would be brilliant for this.
3. Eclipse has a very functional IDE - I dont really like openscad's ide
4. Eclipse has a very functional set of debugging tools.
5. Theres no reason to ditch openscads own language.

Personally, i think eclipse *IS* the tool that openscad should have targeted and its a shame its a very independent design. I would love to re-write it as an eclipse plugin, but i hate coding java and i know i'd not get very far.

The 'Strap...

Going all the way back to the beginning i'd say I've been interested in the whole idea of 3d printing (and more importantly the reprap) for some time. However, for about the last year i've started being interested in building and owning one. So for about that long i've been hanging out on the IRC channel just watching what goes on.

Being an engineer of course the first thing I thought was "im going to build one of my own design" which is something alot of people must think (when considering the types that the reprap will attract). Having watched it all for about a year and been designing my own for about that long (well, about 4 different models so far) was a good way of actually avoiding buying and building.

Eventually though (bout 4 months now) i finally decided to take the plunge. But decided to build something that wasnt too different to an M90 (a design that is inherently easy to 'Strap). I bought an extruder and hot end and was planning on building the rest from what I either had in the house or I could get at the local hardware store.

So what parts did I have available, here's the list:

1. Rods (smooth and threaded M8's) from the hardware store.
2. An old ATMEL CPU.
3. Some little metal brackets (also from the local hardware store).
4. Stepper motors and pulleys (from various old printers and computer parts).
5. Cotton String (oh yes).
6. Some bearings (from a local hobby store, lm8uu's and 608 bearings)
7. Control line from printers (instead of timing belts).
8. Power Supply (std atx computer PSU).
9. Everything else from wood (including carriages and mounts), cut myself from MDF (12mm, 6mm) and plywood.

Well, it didnt really turn out that way. The ATMEL controller just didnt have everything I needed to control a stepper (plus my knowledge of electronics is rather week) and i had a hardware failure. What I did end up building was by replacing some of the above with this:

1. Printrboard controller (mildly regret getting this rather then RAMPS).
2. Propper NEMA17 Steppers, T5 pulley's and timing belts (from a local store).
3. Propper X Carriage printed parts (Thanks Greg Frost) for a standard Mendel 90.
4. A Standard MK2 Heatbed (which i actually dont use)

So here's how that all came together:

Yes, alot of my printer is held together with cotton string (though it has been treated to be VERY taught and strong) at this point and you'll notice that initially i was printing without the z-plane connected (added the z-plane after the second print). The little metal bits that everything is sitting on are little cheap brackets from a local hardware store. I also (initially) used a Mendel 90 Vertical Z axis (which wasn't a good choice). Please do laugh if you wish, however one of my first high-res prints (at 0.1mm layer heights) was this (it was about my 8th print):

The full image is available on the thingiverse item from the same object, and a link to my dropbox high-res image for the same object. Not bad for an 8th print? Thanks to the guys on the reprap irc channel and just sitting on there, while reading the reprap forums and pages that my printer worked quite well from day 1. But lets go back to my first prints:

So theres a few issues there but the sizing was correct (second one was stopped mid way as I realised i forgot to actually push a new firmware onto the thing), but thankfully they were easily solved (literally minutes), attached my z-plane properly and here was my third print (this was actually at 0.1mm layer height):

The damage on that print comes from the fact i couldn't actually take it off the bed (blue masking tape that point). Now my fourth (at 0.2mm layer height):

Im pretty proud of the fact that my printer worked so well and from the beginning, and that i could solve my own issues pretty easily. But moving on...

The only problem I had that took me a while to solve was a lack of a working bed. I'd either get prints not sticking, or prints I could not remove. Im printing with PLA, I have a heatbed, i tried kapton tape, blue masking tape, and many many other suggestions i'd read here and there... none worked... The thing that solved this for me was switching off the heatbed and printing onto diluted PVA directly on the glass. Havent had any issues (with the bed) since then, and heres a video showing a PVA bed:

My final objective is to get here (my PJRun design - not complete as of this post):

But I have some objectives for my design:

1. Better bearing clip designs (im not liking clips-ins and cable tie mounts which are fairly heavily used on many models - including the M90).
2. Portability - it'll fold down easily
3. Open Design - its on github
4. Remote extruder motor (not on the Moving portion of the X carriage)
5. Better X-Carriage design (slimmer and will use a counterweight ultimately)

But to get there, I have a few steps first. To begin with, Im replacing the 'strap part of my rap with printed parts (that i've designed myself - see the temporary directory in github for those parts). I'll start documenting those in the next bog posts and show how this then comes together. The bits from my design are also viewable here, on my website (thanks thingiverse for their thingiviewer code).

At this point, i have also published a few of my own designs to thingiverse, my printrboard box and my LM8UU Y-Axis bearing mounts. My next blog post will be all about my software world (and how i live on the linux desktop with a 3d printer).