A fork of the ground breaking Kossel 3D Printer, with all 3D Printed parts injection molded for ease of assembly and cost reduction.
High Resolution shots of the printer's output here:
About a year ago, I designed the OpenBeam Construction System as a low cost way to rapidly build mechanical prototypes. One of the application I had in mind was the construction of 3D Printers. Up until OpenBeam, the traditional method of building 3D Printers was with 3D Printers.
3D Printers are great, and we have been involved in the 3D Printing community for a very long time. However, when used in a production setting, it quickly becomes an artificially created bottle neck in work flow and a possible support issue as well due to variances between parts. So, with the OpenBeam Construction System, we chose to do things differently. We designed components that form the fundamental building blocks of 3D Printers and other small scale automation machines. These components are designed to be as modular and reusable as possible. This way, individual components can benefit from the economies of scale of modern mass production techniques while still allowing designs to iterate and evolve.
Design Philosophy of the OpenBeam Kossel Pro
After Johann developed the ground breaking work behind the Rostock 3D printer, I donated some material to him to work on the next generation deltabot, code named: "Kossel". The team at Metrix Create:Space have been working on and off on the Kossel and it's bigger variant Kossel Pro for the past six months, and we are now confident enough in its design to be launching this Kickstarter.
We will be designing our parts for maximum interchangeability between the Kossel and the Kossel Pro. We expect that parts such as the extruder body, the end effector, the ball joints, and the carriages to be reusable between the two machines. We will also be intentionally over designing our parts - for example, we are using a commercial linear ball recirculating rail, and we will be designing a fully ball bearing arm joints. Such features may not be necessary for a machine the scale of the Kossel / Kossel Pro, but by over designing, it allows for future expansion - larger printers, more aggressive printing speeds, multiple extruders, etc. These features are beyond the scope of this current kickstarter, but we hope that our contribution will allow others to build off our work.
Build Area: Cylindrical Build Area of 250mm Diameter x 250mm Height
Linear Actuator resolution: 0.2mm / full step, 1/32 Micro stepping
Extruder: OpenBeam designed Dual Use extruder body. Includes parts for both standard NEMA17 stepper mount and geared pancake motor. Extruder can be configured for both direct drive and bowden drive.
Hot End: Testing so far had been done with genuine J-Heads from Brian Refsnyder; if stretch goal is reached, we will be testing our own full stainless steel hot end design.
Swing arms: Carbon Fiber swing arms with OpenBeam designed full ball bearing ball joints.
Linear Bearings: 12mm Recirculating ball linear slides.
Electronics: Metrix Brainwave
Heated Bed: Optional.
Specifications are only part of the story. If you will be at the Bay Area Maker Faire, come by to the Metrix Open Hardware Alliance Booth to see the printer in action and say hello to some of its designers!
OpenBeam is committed to being a good member of the Open Source Hardware community. We donate a percentage of our proceeds towards open source hardware causes; we are a member of the Open Source Hardware Association, we sponsor the development of Slic3r and other slicer programs with monetary and hardware donation, and we are sponsoring our local mini Maker Faire in Seattle. For this project, we will be working closely with the folks at Metrix to grow the Metrix Open Hardware Lab. We are going to do something really cool that will change the way people fundamentally think about prototyping electronics. We can't tell you all the details yet, except that it will be something really, REALLY awesome.
Acknowledgement and Special Thanks:
The OpenBeam Kossel Pro is 100% Open Hardware. We wouldn't be able to achieve what we've done without building on the works of those who had come before us. Specifically, we would like to extend our thanks to:
* The RepRap project.
* Johann Rocholl, for his pioneering work showing the world that delta-geometry robots are possible for 3D printer work
* Matthew Wilson, for his work on the Brainwave board, and challenging conventional wisdom of what's required for 3D Printer control electronics.
* The development team behind Marlin and Repetier Host* The guys on the Deltabot Developer List.
* Matt Westervelt, Marc Goodner, Matthew Wilson, and Richard DeLeon - the Alpha Prototype build crew. Their feedback had been invaluable in shaping the development of this printer.
Additionally, I would like to thank the following folks:
* Martina and Karm Ning Tam (my parents). My father especially, who has been helping me with the supply chain issues.
* Rachel Flamm, aka, future Mrs. OpenBeam, for not throwing me out of the house when I haul home a piece of antique electronics or a boxes of aluminum bits and leaving them around the house.
* The staff at Metrix:Create Space, for running one of the most awesome hackerspaces on the planet, for fostering the Seattle 3D Printer community, and for all the prototyping work that they've done on the Kossel Pros.
And lastly, we like to thank our supporters, for your continued support for Open Source Hardware. Thanks for reading,
-=- The team at Metrix, Terence, Rachel, and the furry monster puppy
Risks and challenges Learn about accountability on Kickstarter
First of all - a disclaimer. We endeavor to make the Kossel Pro an easy to build 3D Printer, but please understand that 3D printers are not household appliances like a microwave or TV. Both the building and operating of a 3D printer requires careful attention to details and a certain level of mechanical skills. This is true with ANY 3D printer that is currently on the market. The Deltabot design represents cutting edge, state of the art in Open Source 3D Printers - and being on the bleeding edge comes with the normal trade offs that early adopters will face.
We've mitigated against a full blown, "sh!t the bed" sort of failure (ie, the thing not working) on this project by building and testing multiple Kossel Pro 3D printers over the last 6 months to validate the design concept. We did not simply run to Kickstarter when we got our prototype running - we built multiple machines with multiple members at Metrix, tested different configurations, and experimented with different assembly methods. We will be systematically replacing all the laser cut and rapid prototype parts with mass produced parts (either via injection molding or metal stamping) that will be more accurate and repeatable. The result should be a vastly improved design than what the alpha testers have been building, in terms of machine accuracy and stability.
The biggest risk is a schedule slip. We have learned from our mistakes on the OpenBeam Kickstarter (which delivered majority of the rewards early, despite being overfunded by 300+%) and will be applying what we have learned onto this Kickstarter project. Specifically we are doing two things: 1) We are hiring someone whose day job is to create assembly documentation to do the actual assembly documentation, and 2) we are limiting the number of kit configurations and the choices people can make on their machines to streamline our fulfillment.
When a 3D Printer kickstarter project like this launches, people inevitably asks for more and more upgrades - dual extrusion, printing with nylon, etc. While it is great to be so enthusiastic, the harsh reality is that good execution of a design project requires guarding against scope creep and focusing on the priorities and tasks on hand. As a result, we have already laid out all the stretch goals for this project, to prevent scope creep from taking over the schedule. The schedule that we've posted is one that we are confident to meet or exceed. At times these measures may seem heavy handed, but we would rather under promise and over deliver instead of promising the sky and then being a year late.
We designed our machine to run 1.75mm filament. 1.75mm filament is easier to handle, easier to bend, and requires less driving torque, which helps us in the direct extruder drive configuration of this machine.
Our testing so far have all be done with PLA. We have a hexagonal heated bed designed that we are waiting for funds to prototype and put into production. As a heated bed is required for ABS printing to prevent part warpage, we have not done much in terms of ABS testing. All testing so far had been done with either the UBIS hot end from Printrbot or Brian Refsnyder's J Head hot end, so it would definitely work with ABS filament.
We have an all stainless extruder design in the works. We have not made the final extruder selection; we've structured the cost of the printer in such a way that we can include an all-stainless hot end that is currently on the market (about $100.00) in the full printer kit. Obviously, we like the design that we did, or we wouldn't have put the efforts into designing it. Hot end design, and what drives the hot end design, will be a future project update - we intend to offer it as a no-cost upgrade as one of the stretch goals. Once the stainless steel hot end is in place, we expect that we can print with PC or Nylon, although we will not officially support these materials.
We have been working with http://MatterHackers.com and they have been great at providing us filament samples, including the new Laywood filament. We will publish our test results and print settings of each of the filament that we test. Likely, for the full kit, we will ship the printer with a sampler pack from Matter Hackers so that builders can start printing right after they have finished building their machine with a filament that had been tested against their build configuration and known to work. Of course you are welcome to try whatever Reprap community standard 1.75mm filament that you can get your hands on - but understand that we cannot possibly test every filament out there for compatibility and optimal settings.
We've designed the extruder body to be able to handle both direct drive and bowden cable drive, with 1.75mm filament. All our testing so far have been with a direct drive extruder, as it is easier to tune. In the coming weeks and months we will be testing our bowden drive extruder as well. The kit will ship with parts to build either configuration.
The Kossel Pro uses standard open source 3D Printer controller and 3D Printer firmware - specifically, we use a modified version of Marlin in all the prototype machines. Any of the Reprap software will therefore work with the printer. We are a big fan of Repetier Host (www.repetier.com - available for Mac and Windows), and on the Raspberry Pi OctoPrint is a great way to enable wifi printing (www.octoprint.com).
The beauty of this implementation is that the firmware does all the heavy lifting / coordinate translation. Therefore, the tool chain and software required to drive the machine is the same as any reprap style machine. We feed the machine cartesian coordinate G-code and the firmware takes care of the rest.
For myself personally, I use Solidworks - but just about every 3D modelling package out there will generate an .STL file (Google Sketchup, FreeCAD, OpenSCAD are all good low cost / free versions to try). You can also download objects from Thingiverse (www.Thingiverse.com) where people share out their designs.
Once the .STL object is procured, a slicing software is used to generate the machine code. the software takes an STL file and slices it into layers - and generates the tool path for each successive layer. Common slicers include Slic3r, Skeinforge, Cura, and KissSlicer.
As mentioned earlier, we are a big fan of Repetier Host. It integrates Slic3r and allows objects to be dragged and dropped into the build envelope, rotated, scaled and manipulated. Once the objects are "plated", Slic3r can then be called to generate the G-code to send to the printer.
We will be including setup instructions for Repetier Host.
What is the printer's accuracy? XYZ printer claims they can do 0.1mm resolution / 0.05mm resolution / etc...
I always explain to people that common FFF (Fused Filament Fabrication) 3D printer is essentially a computer controlled hot glue gun - it operates by melting a plastic noodle and squirting out molten plastic from a hot tip. There is a BIG difference between putting plastic and hoping that it stayed where you *thought* you put it, and the actual part result. The actual part is a function of the sum of all the errors and variances in the additive fabrication process, with a bit of warpage, shrinkage, and entropy thrown in for good measure. A more correct way to specify the machine's accuracy would be to specify a percentage variance based on the part's size - bigger parts will have a bigger tolerance in the size of the part. This is due to the fact that more tool path passes are required to build a bigger part and each tool path incurs its own tolerance. Therefore, to claim that a printer can hold 0.1mm accuracy is misleading at best.
A common mistake 3D printer folks make is to associate microstepping on the motors to actual commanded accuracy. This is simply not true. Microstepping is merely a way for the motor driver to vary the proportion of current between the two coils of a motor to smooth out the motion between discrete steps. On the OpenBeam Kossel Pro, for example, we run a 20 teeth GT2 timing belt pulley that is 2mm pitch. The motor is a 200 steps per revolution motor. Therefore, per revolution of the pulley the belt advances 40mm (20T x 2mm/T pitch), and per step the belt advances 0.2mm (40mm / 200 steps). This is the theoretical drive resolution on the motor axis. Slack in the belt, belt stretch, etc can all affect this accuracy.
Delta robots further complicate this by the way that motion is achieved - by mixing input from all 3 towers. Thus, the resolution of the printer is actually the lowest in the middle of the bed - as it requires the least amount of motion on the tower to achieve motion in the middle - and the areas in front of each tower has the highest resolution, as the arm it is directly in front of has to travel the greatest distance to eek out a small amount of motion. Here's a great thread on the Deltabot Group on the variable resolution results of the Delta Robot:
That being said - they say the proof of the pudding is in the eating. The Totoro figurines and all the test prints were printed in the middle of the build platform, where the printer has the lowest accuracy. The output quality of the printer is definitely on par with the best of the open source FFF machines and I would argue that for the right geometry (one that does not require support material) it is capable of beating a $20,000 Stratasys Dimension (smaller nozzle size and step height).