An Epic Update – Kossels past, present and future
I’m sure everyone’s eager to know what the status of the OpenBeam Kossel Pro is. Well, we have one heck of a surprise for you.
As you may know Johann and I started on the Kossel project and went down somewhat separate paths. Johann’s designs have always been elegant and rooted in the Reprap world, where parts are printed. I did not have a 3D printer for most of the earlier development process so I focused on the RepStrap route: I used a laser cutter and designed around laser-cut geometry with the understanding that I would later replace the laser cut parts with stamped metal components.
The original Kossel Legacy had a LOT of 3D printed parts. Earlier on in the design the extrusions were rotated 45 degrees from each other to simplify the motor vertices. From a design for manufacturing standpoint this was also a nightmare for assembly. Assembling the vertex required holding 3 different pieces in the right orientation and sliding them together. But the biggest problem came from extrusion wear. The original Kossel had custom designed sliders that held a set of 623 bearings at 45 degree angles. These bearings then rode up and down on the Kossel extrusions. Because ball bearings were typically case hardened steel, they wore down the aluminum extrusions pretty significantly.
Around this time, the Kossel Pro’s design looked very different. I’ve always favored building the linear axis with OpenBeam’s NEMA17 motor mount on one end, 608 bearings on the other side (mounted via our 608 to NEMA17 adapter) and a shoulder bolt for the shaft. As the OpenBeam system designer this makes for great business sense; the amount of parts tooled would be minimal and they can be approximated with lasercut parts for the most part. However the system wasn’t without it’s drawback; Johann quickly pointed out that we needed two long pieces of OpenBeam for each linear axis when he only needed one and on top of that there was no way to control the two pieces of extrusions being parallel and keeping even spacing throughout the travel span. (An extrusion that long can be expected to bow and twist, both under load and from allowed manufacturing tolerances).
Then there’s the question of what the carriages would ride on. We looked at IGUS N17 rails but we weren’t impressed with the amount of slop in them. Worse yet the cost was pretty significant, even with Pacific Bearings’ N17 slider clones. We almost purchased a die for a 1530 extrusion with integrated N17 bearing rail geometry, but around this time our old extrusion supplier started having major problems producing our regular extrusion to specification and we quickly cancelled the order over quality concerns.
In early January, (armed with a computer loaded with Chinese IM software), we located a supplier for recirculating ball bearing linear rails. We immediately arranged to have a small engineering sample set sent over with one of our regular plastic part shipments. Initial testing was very positive. The rails are made of high carbon tool steel, so hard that they had to be wire-EDM cut to length when we order them. Bolting this to the OpenBeam extrusion greatly stiffened the printer’s frame and guarantee that the printer frame will be straight. The recirculating ball carriage works like a charm – there are no eccentric spacers to adjust, no tedious assembly process to worry about. With the linear rail in hand, we focused on the tri-wing / Y shaped design, rotating each of the linear axis 90 degrees. The result was the printer you saw in the video.
I also provided Johann a set of linear rails and asked him to consider rotating the vertical members 45 degrees back into the “normal” configuration. Armed with these rails, Johann redesigned the Kossel’s vertices by combining a few of the motor brackets into one. This design later evolved into what became the “Mini Kossel”. The redesign addressed a lot of my concerns with regards to ease-of-assembly, but the shape of the 3D printed vertexes posed a challenge to retool into injection molding. We had designed some stamped metal parts to replace the Kossel’s vertexes, but they lack the good looks of the 3D printed parts.
The breakthrough came when I started modelling the Mini Kossel into Solidworks to help with assembly instructions. I was driving to MicroRAX to cut some aluminum extrusions one night after work, when it hit me like a ton of bricks: The shape of the lower and upper vertices are so similar they may as well be aluminum extrusions cut to length with holes drilled into them. It turns out that the answer was in front of us all this time!
A frantic few nights of late night CAD and a few request-for-quotations later I am happy to present to you the newly redesigned OpenBeam Kossel Pro:
Not surprisingly this looks just like a Mini Kossel scaled up. It maintains the same 250mm diameter x 250mm height build envelope and yet the foot print of the printer is slightly smaller than the original Kossel Pro with more usable space inside the machine frame for electronics (although the printer did grow a little bit in height). The parts will be 100% compatible with the Mini Kossel (except in extrusion and linear rail length). The Reprap Mini Kossel’s frame is exceptionally stiff and we expect this full metal frame construction to improve in stiffness over the Reprap counterpart.
As a design engineer who had sunk 6 months into prototyping / R&D of my own design, there had to be darn good reasons (and lots of pride swallowing) to mothball a design six months in the making. So, allow me to walk you through my reasoning.
As a crowd funded project and an open source hardware company we feel that we have a responsibility to the community to do what is best for everyone. In this case what evolved into Johann’s Mini Kossel is clearly the more elegant and minimalist of the two machines. Early on we had made the decision to try to make as many parts interchangeable between the Mini Kossel and Kossel Pro; the availability of well produced injection molded parts that can be reused in various configurations will help bring quality of the machines up and the cost down.
Today we are happy to announce that by merging the Kossel and Kossel Pro’s design tree we have achieved 100% sub assembly level compatibility between the two branches. To support the community we will supply BOTH a mechanical “vitamins only” kit to the Reprap community (requiring 3D printed parts) as well as a turnkey mechanical kit. We aim to release the Reprap Vitamins kit within the next month, and we will make the turnkey mechanical kit available for purchase after our Kickstarter rewards have been fulfilled. Builders will be able to start with a Reprap Kossel and upgrade various components as needed with injection molded parts (available from our web store and fulfilled by Amazon.com). Our injection molded components will be slightly more expensive than 3D Printed parts but will offer more advantages thus still representing a good value proposition.
We also believe that Johann’s Mini Kossel design is more elegant and minimalist. If we were to bring the Kossel Pro as presented in the Kickstarter video to market, that design will likely face rapid obsolescence upon release. By taking the efforts to redesign now and making the design 100% compatible to the more open Reprap Mini Kossel branch, we stand a very good chance of establishing the Kossel / Mini Kossel as the benchmark printer for Delta 3D printers. As 3D Printer designs come and go we certainly want ours to be produced and used 2-3 years down the road (which is an eternity in the Reprap world but then again people are still building threaded rod Prusa Mendels).
It has been a mad frantic crunch to redesign the printer; we are releasing manufacturing files to our manufacturing partners this week. With regards to the schedule we had padded the original schedule considerably but this safety padding is now gone. Any additional major delays will now result in day-per-day slip in the manufacturing schedule. That being said we are hopeful that we can still honor the original delivery dates (before Christmas) and we will advise in a few weeks when our manufacturing partners have had a chance to plan their operations around the new parts.
So, let’s take a closer look at the CAD screenshots!
Here is the redesigned end-effector with the auto-deploy probe. The end effector is only 10mm bigger than Johann’s 3D printed end effector for the mini kossel, but it packs a LOT more features. A ducted fan system provides uniform cooling to the part and a 3rd fan provides dedicated cooling to the hot end’s thermal barrier.
Here is the auto-touch probe. It is adjustable for deployed length. In Johann’s design you set the probe off and measure the offset in height to the hot end tip. Here you have the option of doing things the other way around: Shim the hot-end tip up with a piece of material of known thickness then adjust the probe until contact is made. I think this is an easier way to go.
The auto probe now requires just a simple bump to the side to deploy, greatly simplifying the deployment process. It is modular and can be mounted onto Cartesian style printers as well. We will be making the injection molded parts available to other 3D Printer kit manufacturer at reasonable prices to speed adaptation of this crucial improvement.
There are more CAD files; but for now these are the first two parts that we are going to be prototyping (as they require the most TLC to get correct). We expect to have prototype parts back in hand this coming week and we’ll be assembling it into our test bed printer.
We apologize for being behind in our updates; we needed to focus on getting the redesign completed so that we can alleviate any schedule slip. Now that we are over the hump with the design work, we will be updating more frequently as prototype and first article parts start coming in.
-=- Terence, Rachel and the furry monster puppy