CEB 4 design planning
Abstract: This is the design rationale for the 4th prototype of the CEB Press built at Factor e Farm. The design requirements are: (1) Design-for-Disassembly; (2) Lifetime design; (3), modularity; (4), integration with the rest of the GVCS; (5) Design-for-fabrication; and (6), OSE Specifications.
General design is intended to be one where we assume the availability of a CNC Torch table for rapid fabrication. The machine is intended to be producible as a kit. It is designed to be collapsable into a 3x3x6 foot crate for ease of shipping. Welding is minimized to allow quick fixing via disassembly - with modular, easy-to-source parts. The intent of bolt-together structure is to facilitate fabrication. We are determining whether the fabrication time can be brought down from about 50 hours per machine, excluding electronics, to approximately 25 hour per machine. To achieve this, we aim to cut most parts on a torch table, followed by bolting together, so that a user can receive a kit and put it together like IKEA furniture. Since bolting is quicker than welding - under the assumption that a CNC torch table produces accurate pieces - bolts are favored over welding wherever possible. This is different than standard custom fabrication practice, but it is intended to produce superior results in terms of lifetime design and ease of fabrication. We are exploring the feasibility of production straight from a CNC plasma torch table + welding to a finished, replicable product that can be sold directly as a kit without the producer needing to assemble the machine at all. To achieve this, we are aiming for strict quality control to allow all components to fit together with 99.9% success rate for each machine - such that only 1 of 1000 components of the ~100 components per machine needs to be recut.
For the electronics, we are streamlining controller production to approximately 3 hours per machine starting from raw circuit boards, followed by milling on a CNC Circuit Mill, followed by populating with components and assembling a controller box with sensors, in a time under 6 hours per controller. This time should be reduced to under 5 hours per controller with ready-made controller boards.
BK: Bolt-together doesn't necessarily facilitate fabrication. I would argue bolting often takes far more time than welding. You have to make a minimum of 2 holes, often in difficult to access locations, line them up, and then insert a bolt. Welding will often take less time and will make it easier in Kit form, as less asssembly will be required. For Example: In mounting the hopper, P3 had bolt on supports on the back of the hopper. This was time consuming because the bolt needs to go close to the bottom, and you are bolting to a large piece. If the supports are welded, you can simply put the hopper onto the frame without any extra work.
BK: The XM Design Rationale page states that reduced time and complexity of fabrication process takes priority over modularity.
 Press Foot- P3 vs. "piston" type
BK 6/22: Both James Slade and Dan Schellenberg have decided to use a "piston" type of press foot, which mounts on a clevis type cylinder. I've heard replicators from both sides say that theirs is superior for pressing bricks and the wear on the cylinder; however I have seen no proof either way. So, from my perspective, here are a list of proven benefits of using this design over the P3 version:
- Enables use of clevis ($258-$362) cylinder vs. the more pricey cross tube ($520).
- The cylinder can be recycled with this version; in P3, after the press foot is welded to the cylinder, you can't use it for anything else.
- Welding on the cylinder risks melting the seal at it's entrance, which would cause leaking. Also, the welding is very tricky, as you can't warp the plate you are welding. and with that much welding, boy, is that tough!
- A normal cylinder is serviceable by machine shops; after one has been welded on, many shops will refuse to service because of liability.
- More parts to fab
- supposedly longer fabrication time (according to marcin)
- Does not follow industry standard for best proprietary vertical machine on the market (AECT)
 Hopper Changes
Problem Statement: In Prototype 3 (P3 for short), the hopper had gaps at the top, and the seams didn't go together well; the 4 sides ended at different heights, making it difficult to mount the grate.
Solution BK 6/19: Dan Schellenberg's CEB used the same general shape of OSE's prototype, but cut off some of the top section, and a portion of the sides. Link to hopper model. This saves material and simplifies mounting the grate. No disadvantages are seen.
MJ 6/19: For materials use efficiency, P3 requires one 5'x10' sheet of 1/8" steel. If this requirement is still met, then we should continue.
BK 6/20: After more investigation, I realized Dan's hopper was made from the original hopper design. Marcin said this design had too shallow of a slope, and that soil tended to sit on top rather than roll into the hopper. So, it became necessary to re-design shape from dan's.
BK 6/20This afternoon, James Slade and I discussed issues with the hopper, and came up with the following:
- Hopper interface plate lip causes soil to load up at those spots this is a real issue, as user must climb on machine to clean out clay with a stick.
- Walls should be as close to vertical as possible, as clay tends to stick to sides if they are too horizontal.
- The grate was not as steep as James would have liked it to be.
Some possible solutions we came up with are:
- The grate to be mounted on a hinge, which can slide up and down on the hopper mounts. This would allow user to easily change steepness of the grate as necessary to fit their soil type.
- We can make a hole in the bottom of the hopper, and keep it capped, either with threads or some sort of plug. This would be for when "bridging" happens, so the user doesn't need to climb up the machine.
- The lip should be eliminated as much as possible. If the hopper can be fitted directly to the input hole on the frame, this is best.
- Welding plates between the hopper and the frame is an option, so that the soil will flow better between the two.
BK 6/20 Some issues I've run into with designing, is that it is virtually impossible to design so that the side pieces don't need to twist a bit to match on both sides. To have the least twisting possible, either the top side, or the bottom must be out of square.
 Mounting, on sides
Problem Statement: The grate supports were time-consuming to mount, and used a lot of material. Also, bolting to the hopper is time consuming.
Back StoryIn P3, we mounted using 4x4 angle supports which bolted to the primary legs and hopper, and 2x2 tubing which bolted to the hopper and into leg-holders on the frame.
Discussion: Dan's machine welded the tubing to the CEB instead of bolting. This uses less material and takes less time. Dis-advantage is that you can no longer easily disassemble tubing from hopper. Arguably, there is no reason you would need to disassemble them anyways. It can be stitch welded, since there's not a lot of force needed to support hopper. This would make disassembly by cut-off wheel feasable.
For the angle supports, He replaced them with more tubing with a bent 1/2" plate welded to it. These attached to the frame via leg-holders. There was 2x2 angle welded to the front of the hopper, and the hopper sat on top of the bent plate. Advantage is easier assembly, cleaner appearance. Disadvantages- can only support force coming downwards, not upwards. We could bolt them together to solve this.
 Hopper Interface Plate
Problem Statement: The Hopper interface plate of P3 is arguably obsolete.
Back Story: The initial reasoning was to make hopper removable according to design for disassembly. There were nuts welded to the top of the frame to make that gap. In testing, OSE realized that it let soil leak out thru the sides. So, for the prototype release, the nuts were removed from between the plate and the frame, and the plate simply bolted to the frame, with nuts welded underneath the C-channel.
Discussion: The reason the plate was necessary initially was for shaking, and preventing soil from leaking out around the bottom. Since the shaking is ruled out, the only purpose is preventing leaking. This can be acheived by welding thinner plate (probably 1/4") directly to the frame U channel. This lowers part count, and at least an hour of labor in drilling and torching holes. No disadvantages are seen.
MJ: Interface plate is for purposes of design-for-disassembly, an essential component for lifetime design. Also, that allows entire CEB to be packaged in a 3x3x6' crate.
 Assembly Mechanisms
Problem Statement: The previous assembly was time consuming and allowed soil to leak from sides.
Back Story: It was originally built with a few door hinges welded to plates, which attached to the hopper to allow the whole thing to shake more easily, and to be easily disassemblable. Assembly and testing showed that it was very difficult to get the plates close enough to not allow soil to leak out. The seams were duct taped to prevent this.
Discussion Dan's CEB showed that even with the entire hopper welded together, the hopper still shook readily. This proves hinge/shaking theory invalid, allowing us to find easier ways of assembling.
My proposal is for the attachment of side to back, we weld some 2x2" angle, 1/8" thick if available, or 1/4" if not, to the back piece. There will be holes in this which bolt to the side pieces.
Since the joint between the front and side pieces is not a right angle, angle will not work. For people who have a press brake available, Simply bending a piece of 1/8" plate and welding to the front, bolting to sides, would work. Since OSE doesn't have one, other possibilties are using piano hinges, or welding 2 pieces of 1/8" together. If we stitch weld them tightly enough, no soil should leak out, and it won't take as long as welding the whole thing.
- The steel for these connections would cost roughly 5.82, assuming 2 pcs of 1/8"x2" flatbar for length. Piano Hinges are listed as 9.98 at Lowes.
- Since the hinges are only marginally more expensive, wouldn't need to be welded, and already have holes drilled in them, this is the more cost effective choice.
 Grate Mounting
Problem Statement: In p3, the grate mounted by bolting the grate to the hopper supports and the grate mounts. It was very difficult to install, as it required a person to hold it up, while another person installed bolts. Also, removing the grate is difficult if you need to remove rocks or something from the top of it, as you would need to remove it entirely.
Solution: Dan's machine used hinges welded to the top of the grate which were also welded to the back piece of the hopper. Initial installation would still be a little tricky, but not as hard as getting the bolts in. This then allows the user to tilt up the grate if necessary. It also eliminates the need to mount it from 2 sides, as the hinge supports it, and it sits on top of the hopper.
MJ: Design for disassembly should be maintained.
Problem Statement: The old shaker took a lot of time and materials to make.
Solution: BK 6/19 Dan's CEB used an entirely different shaker process. The motor attached directly to a circular plate, with a hole drilled off center. It attached via some special adapter for lawn mowers. Photos to come. The circular plate was enclosed between a sandwich of 3 plates, which sheilds the user from any shrapnel. This considerably lowers part count and labor. The only possible disadvantages are that it may not provide as much shaking as the previous model. However, it seemed to shake plenty when I watched it.
BK 6/19 Spoke with Marcin, and he informed me that since this is not a "wheel" motor, having the eccentric attached directly to the motor will ruin the bearings inside. I need to investigate the cost difference in the motors.
I propose to change the arms from 4x6 angle to 4x4 angle. There is really not much weight on any of the legs, and the 4x6 is overkill. I have not done math to prove it, but even if the entire weight of CEB were on it, its only 2000 lbs. The only downside I can see is that the sensor holders will need to be made longer to read the sensors.
MJ: Practical considerations of carrying with forks and bumping with tractor when loading make 4x6 channel desiarable, and I would make it