Welcome! Log In Create A New Profile

Advanced

Team Open Air

Posted by rocket_scientist 
Team Open Air
March 02, 2010 01:28PM
I doubt that I will ever come close to being the prize winner, but I would like to declare a team and base design. I don't have a blog yet, (or even a working RP!), so I will add that later.

The basis of my design is a cubical structure, with a fixed bed and 3 axis gantry for the extruders. The fixed bed will allow the largest possible piece to be only the width of the extruder and gantry mount hardware smaller than the RP outside dimensions, instead of less that half with a moving X-Y table.

I call it open air because two sides and the corner between them will be open to allow pieces to extend outside the RP. I intend to experiment with growing parts. By this I mean making the first half of a truss column, then moving that piece out over the edge of the bed and hold it in place with arms/clamps that swing out from the sides. The edge of the bed would have holes to put registration pins into, and the first piece made would have holes for the registration pins to keep the alignment and positioning correct. Panels can also be made in quadrants, so that pieces larger than the largest dimension of the RP can be made.


The Z axis will be the main stage, and run off three vertical lead screw drives in the remaining 3 corners. The Y and X stages would use truss beams of plastic webs and steel strap on the top and bottom for prevent creep. The cross truss would be open on the bottom and back side so that the extruder head can be automatically swapped out.

There would be multiple heads, each with its own plastic feed, power and data cables along the back top. The Gantry would move to the storage location of the current head, a servo would unlatch it and move it into its storage position, where it would snap into spring clips to hold it in place and insure that the extruder does not drip/leak. Then the gantry would move in front of the next head and the servo would grab that head, pull it out of its spring clips and lock it into position.

The accuracy after exchanging heads would be improved by having a pair of laser pointers cross their beams at one spot in the range of motion of the head, with photodetectors on the far side to see when the beam is blocked. The extruder would then be moved slowly across one beam to fix X, across the second beam to fix Y, then positioned exactly over the intersection and lowered to fix Z. When a milling head is used, this process would fix the tip of the milling point.

The multiple extruders would contain different types of plastic, and different colors of the main plastic. One or more milling heads would be included for subtractive 3D prototyping. The milling head might also have a range of bits that can be swapped in and out like the gantry exchanges heads. This would allow small bits for detail work, larger ones for fast work, ones for metal and others for plastic.

The plastic extruders would have 3 or more orifices that can be slected by sliding a plate of the end of the extruder. This would allow very fine filaments to be used for details and edges, then after several edge layers have been built up, the coarser opening would be used for much faster fill.

A multi-color quad head might be used to provide more colors without using up lots of head slots. This would have four smaller stepper motors feeding separate plastic feeds into a single heater-extruder head. Each stepper, feed, control would be rotated 90 degrees from the others so that all the parts would be made in quadruplicate. It might even be possible to have all liquefied feeds go into a single extruder orifice and mixed at different rates to obtain a rainbow of colors.

The bead would be selectively heated to reduce warping and power consumption. A pattern of 16 by 16 small aluminum blocks would be bonded onto a thin stainless steel bed. The gaps between blocks would be filled with a high temperature epoxy or other stiff filler material to increase the bed's stiffness. Each block would have a 10 ohm carbon comp resistor bolted onto the underside, and wired in rows and columns. The heater control board would have 16 power MOSFETS for each row, and 16 for each column. Heat would be applied by powering an entire row with 12V, then allowing only the desired blocks to heat be grounded by the coresponding column MOSFETS. With a 0.1 ohm precision resistor in series between the columns and the column MOSFETS so that an analog mux would allow the controller to measure the resistance of the carbon comp resistors and use that to measure the heat of each block. The thin stainless steel will prevent heat from moving sideways, but still allow the heat to flow upwards into the plastic parts. The pattern of parts being made would determine which blocks are heated, so that the plastic deposited becomes insulation from the air to reduce the power needed to keep the plastic warm.

That's all I can think of off the top of my head. When I actually get a Mendel working to make parts, I will see about exploring and testing these ideas.

Mike


Team Open Air
Blog Team Open Air
rocket scientists think LIGHTYEARS outside the box!
Re: Team Open Air
March 02, 2010 08:34PM
To build a truss or panel bigger than the working surface, it is necessary to add new material onto already made pieces that have been moved to extend of the edge of the bed (secured by movable arms/clamps and positioned with registration pins). This means that the extruder has to build right up to the side of something taller than the current piece under construction. Normally, the extruder head would bump into the higher piece and prevent getting very close, so a second extruder head is needed. This one would have a long thin extension or pipe, like a hypodermic needle, to add plastic in the tight spaces between the existing piece and the new material. To prevent the pipe from plugging up, it will need to be heated. This can be done by having an inner core that is somewhat conductive, like nicrome wire, surrounded by an insulating layer, and then a good conductor on the outside like copper, and run a current through to heat the interior of the needle. Or needle is made of conductive metal and a nicrome wire runs up the middle to heat the plastic. Since this will be filling in between a previously made piece and the bulk of the new material, this needle would also be fairly large in diameter so that plastic can be extruded quickly. The length of the needle will set an upper bound to the thickness of a previously made piece that is being extended. Even though fresh hot plastic sticks well to previously laid layers, to get a good structural bond between the old piece and the new, the interface between them should interleaved or dovetailed. This will provide more surface area between the old part and the new, and it will allow better heating of the old part in the thin fingers between freshly laid down plastic to allow it melt together better.

If circuit boards are to be made from scratch, they need to be stronger than just fused ABS. Current circuit boards are FRP, fiberglass reinforced plastic. To make this we need to add another extruder to supply fiberglass thread or yarn. This will be harder to extrude with a pinch roller, but with rubber rollers it should still work. To tack it down, something sticky has to extruded at same time. The best choice would be a monomer of the plastic being used so that it will bond with the rest. It might also work to extrude a little bit of melted ABS along with the fiberglass yarn to hold it down. At the end of each length of fiberglass yarn, a knife will pass over the extruder to cut/break off the piece just laid down. It will be easy to lay down alternating layers in X and Y, but I see no way to easily add fiberglass fibers in the Z direction. The best bet will likely be to make other pieces rotated 90 degrees and glue them on after fabrication. It may also be possible to make tufted fiberglass that is only secured at the bottom to make fiberglass insulation. This would be laid down like a carpet, then a layer of plastic to make a bed for the next layer of fiberglass.

If a fiberglass extruder has been added, it also becomes possible to make belts. The belt would be laying down on its side and the fiberglass would be wrapped around and around to provide the tensile strength of the belt, and the plastic/rubber built up around it to hold it together and provide traction. And of course if the belt is toothed, harringboned, or otherwise structured, that can be laid done by the 3D process as well. Raw, unvulcanized rubber can be extruded with pump or auger, and the heat cure it as it is laid down.

One obvious solution to the conductive material extruder is to use solder paste. This is already commonly used for surface mount devices. The paste would once again be extruded by a pump or auger. Once on the surface, a separate heater would be used to locally cure the solder paste. This could either be a resistively heated metal surface on the extruder, a separate head with a heated metal surface, or an intense infrared lamp and reflector to concentrate the heat in one spot. If a heated bed is being used, that would also help to cure the solder paste if it is heated up to the safe working temperature limit of the plastic. A higher working temperature plastic would be good to use for the board body so that the bed can be heated up to closer to the 220C needed to reflow the solder paste. High temperature epoxy, requiring an extruder that delivers the two components in liquid form, then using the bed heater to post cure, would make an almost identical PC board substrate to that of normal PC boards. Using this approach, SMD ICs can be added to the board after the solder paste is laid down but before it is set. If an additional head (getting quite crowded on the top!) used to pick and place the SMDs with the precision of the 3D prototyper. A tape of parts in order that they are placed would still need to be made separately. However, such tapes may become popular 'vitamins' for standard builds such as making another reprap.

To automate the building of parts bigger than the reprap, one or more robotic arms can be added to the open sides of the reprap to move the completed part and hold it in place while the extension is being made. This also allows for more flexible placement of the built piece, such as holding it at a 45 degree angle to the bed. Since the arms could also remove finished parts and place them near the reprap, and the grab the pieces later to do some assembly, a useful first build would be a parts bin with a motor to rotate a carousel of parts. Designing this to hold all the parts to make another reprap, the reprap can first build and fill the carousel, rotating it as needed to fill all the bins, then pick the parts back out of it to assemble a new reprap. This would require drill/driver and socket wrench like functions on the arms in addition to the grasping ability. Vitamins needed for the next reprap would have to be placed in the carousel by hand by the operator, but this would still make the assembly of a new reprap much easier for the non technical, even if only sub assemblies could be made this way. Since the robot arms would be made mostly of reprap buildable parts, the accuracy and repeatability would not be great, especially if the reach was long enough to build sub assemblies. To improve on the accurate positioning of the arms, LEDs would be placed at each joint, and 2 or more CCD cameras added to the reprap. When the arm reached a critical position, each LED would be flashed in turn, the cameras would take two exposures, a bright with the LED on and a dark with it off to subtract off the room lighting, and then find the angles from each camera to each LED to compute the arms actual position. The servos in the arms would then receive correction signals until the camera tracking placed each one exactly where needed.

Adding copper to a plastic part can be done in one of two ways. Hot copper wire will remelt the plastic and stick in place. The traces of a circuit board can be built up from multiple wires. Like the fiberglass, a cutter will be needed on the end of the copper wire extruder to break off the current piece without pulling it off the board. Since the wires will not be well connected electrically, especially where a trace breaks up into several lines, it would like still need the solder paste to be added and heated to connect all the wire together. Or as each piece contacts another trace, the outer ring of the extruder can run an electric current through the exiting wire to where the new wire just touches it to weld them together. Another possible approach is to pass a very thin wire through an electric arc (there goes the 200 watts limit!) to melt the wire and deposit the droplets on the surface. This may suffer from lack of control and produce a very blobby. This might be improved by going over the copper after it is deposited with a milling or polishing bit to smooth it out. Another possibility is to build a layer of copper like a copper clad board by pressing hot ribbons of copper over the surface, then welding the edges together. Then the layer would be miller out with a milling tool. This would be easier to get the complex lines and geometry of a PCB than trying to place, bend, and attach copper wire ribbons of individual trace dimension.


The current stepper motors used in extruder is likely over kill. To get the required torque to push the plastic filament through the heater and out the tiny orifice with a direct coupled motor you need a large motor. But unlike the X, Y, and Z stage motors, the extruder motor never turns very fast, the plastic is extruded quite slowly. This is a case where a much smaller motor that is geared down to the low rotational speed but high torque of extruding plastic would work well. There seem to be few stepper motors in this smaller size available, compared to the ubiquitous NEMA 17 motorS, so it might be necessary to switch to a normal, small DC motor. To control the rate at which the plastic is extruded, a current sensor to maintain constant torque should work well. Even with the gear reduction, this should allow for much smaller extruders, and the quad-color extruder I mentioned above. This would also further decrease the cost and mass of non-RP parts.

Mike

Edited 1 time(s). Last edit at 03/03/2010 06:24PM by rocket_scientist.

Team Open Air
Blog Team Open Air
rocket scientists think LIGHTYEARS outside the box!
Re: Team Open Air
March 16, 2010 08:23PM
To make an extruder that can build between a new piece on its first layer, and an existing piece at least an inch thick, we need a heated, narrow and long extruder tip. Since even the modified extruder tip with cut-off valve has trouble keeping warm enough to keep the plastic flowing, a long, needle like extruder will be even more difficult. One way to keep the plastic liquid is to run a thin strand of high resistance nichrome wire down the center of the needle, and then connect it electrically to the needle barrel (if the needle is made of metal). This way a current can flow through the nichrome wire and heat up the plastic around it. If the needle is made of stainless steel, it will have a lower heat conduction than copper or brass, but across the thin walls this may make little difference. Or the needle can be made of glass, with a coil of nichrome wire on the inner surface. This would be made by winding the nichrome wire around a thin steel rod. Then standard glass bead making supplies and tools would be used to build up a thick layer of glass on top of the nichrome. The insulating qualities of the glass will keep much of the heat in, decreasing the amount of power needed to keep the needle warm enough to work. Since the glass is more fragile than copper, brass, or steel, it would help to make the whole needle extension on a pieces of threaded copper so that replacement needles could be screwed in as needed. With a piece of copper wire going up the outside to carry the electric current back, or the return nichrome wire going up the center is the glass has fully insulated the coiled wire, the two electrical connections can be made through the threaded copper, and motor brush like connection so no separate wiring is needed. Since the glass will be frequently going from room temperature to 250C, Pyrex or other low thermal coefficient of expansion glass should be used.

To make it easier to fill between an existing piece and a new one, the extruder tip can be tilted to 45 degree angle so that it clears the old and new material better. It the junction between old and new is dovetailed or interdigitated, this will not help as much filling into the gaps. Since this is the junction of a structural piece, the extruder can have a larger orifice, and thus thicker, stronger needle, and fill the space more quickly but with less precision.

We need to add an SD or other flashdrive storage device to the mother board to store one or more pieces or projects so that the reprap can function offline. It also needs an LCD display or other human interface device to select from stored files, display progress, estimate time remaining or filament used.

If we add a force gauge to the extruder tip, we can tell when the extruder has 'crashed' into the work piece and needs to be reset. This message can also be sent over the USB, serial, ethernet, etc bus that is providing the g-code instructions. Adding a short range infrared range finder would help keep the extruder head the exact distance above the part for each layer, decreasing the chance of a 'head crash'. If a scanning LED or laser and simple monochrome CCD camera are attached to the Z stage, it would also be possible to to stop from time-to-time and scan the surface of the part to verify it is correct. This scan could also be requested from the host computer, especially over the internet, to check the status of the build remotely. This same feature would allow objects to be placed in the build stage and 3D scanned into the computer to convert into a buildable model. Since the 3D scanner can only see the top surface, it might be needed to make several scans and combine them in software to make a more accurate model of the part to be replicated.


The reprap should report the status of all detectable functions like extruder temperature, extruder stepper motor/geared motor drive current (is the filament jamming?), overall power consumption, heated bed temperature, out-of-filament sensor (another opto-switch). It might be useful to add a pressure sensor to the extruder tip to monitor flow rate, or even control the extruder temperature to maintain a constant back pressure. This would also be another good way to detect out0of-filament. All these values should be sent back to the hosting computer to monitor the status of the build.

Mike


Team Open Air
Blog Team Open Air
rocket scientists think LIGHTYEARS outside the box!
Sorry, only registered users may post in this forum.

Click here to login