Sandwich200
Release status: working
| Description | A fast, compact, folding, portable aluminum-framed CoreXY printer with a 200x200x190mm print area that folds into a 440x420x210mm box.
|
| License | |
| Author | |
| Contributors | |
| Based-on | |
| Categories | |
| CAD Models | |
| External Link |
The Sandwich200 is a folding, portable CoreXY 3D printer that uses a 1515 aluminum extrusion frame. Capable of printing at high resolutions and high speeds (currently tested at 350mm/sec print speed and 35mm3/s flow rate) owing to its CoreXY kinematics, it produces quality parts that are up to 200x200x190mm in size. In its assembled state it is 490mm high x 460 x 410mm (up to around 550mm due to the bowden tube); folded into its box it is 440mm x 420 x 210mm. The critical mechanical and electronics are separate from the box, so that protective foam reduces shock and vibration during transportation.
Contents
Goal
To design a compact folding, portable 3D printer using CoreXY kinematics that offers high speed and quality. The printer must be able to be built without specialized tools, must be completely rigid when assembled, and be safely freight-shippable in its own box.
Features
- Aluminum structure based on Makerbeam/Misumi 15x15mm extrusion
- Stacked variant of CoreXY that eliminates the belt crossing and its associated issues
- Folds up into its own box, dimensions 440x420x210mm
- Box halves are used when assembled to contribute to very good rigidity.
- Reduced risk of damage during transportation: Top CoreXY and Z Upright assemblies are totally separate from the box, which means that they can be packed in protective foam, away from the sides of the box.
- 200x200x190mm build area with heated bed (200x200x200 for a future revision)
- All hardware and electronics (except for filament spool) are integrated within the frame.
- Capable of 0.05mm layer height and 350 mm/s printing (so far)
- Emphasis on using 3D-printed parts and being easily hackable/upgradeable
- Geared bowden extruder and an E3Dv6 "volcano" upgrade, giving very high flow-rates (35mm3/s currently tested)
- Low moving mass — all motors stationary.
- No specialized tools or equipment necessary for assembly (accessible)
- LM8LUU linear bearings used throughout, except in the Z-assembly which is dual LM8LUU plus LM8UU for huge (75mm) vertical bed support. Longer bearings means less judder, noise and wear.
Hardware Specifications
- 12V 300W PSU (24v upgrade planned)
- 3mm Aluminium MK3 214x214 printbed
- Duet 0.8.5 electronics (high current stepper drivers, up to 30V)
- 84ozin (60Ncm) 200steps/rev 2.5A NEMA17 Stepper Motors (QTY 4)
- E3Dv6 3.0mm Bowden hot-end with "Volcano" upgrade and 0.4mm nozzle (up to 1.2mm available)
- 4.07:1 Compact Geared extruder with a MK8 Drive Gear at the centre
- TODO: Flex3Drive 40:1 extruder upgrade
Analysis
This section covers in-depth the decisions and driving factors behind why this printer exists. It was originally designed so that the EOMA68 15.6in Libre Laptop could continue to be developed whilst its developer is moving from country to country. Many of the parts of the 15.6in Libre Laptop are just over 230mm long and 12mm wide, so a 200x200mm printbed was essential, as is surviving freight shipping, as is minimising the amount of space that the printer takes up, both during shipping as well as operation.
However before beginning such a considerable undertaking, an analysis was carried out both of the feasibility and practicality of creating a 3D printer, finding a suitable existing base that could be easily modified, as well as checking if there was a pre-existing design that could already be used.
Comparative Analysis (or, "why another folding, portable 3d printer?"
There exists quite a number of folding 3d printers: shows approximately eight at the time of writing. Teebot, TeebotMax, Mondrian, Printrbot - all of them fold down; the PortaPrusa is especially noteworthy as it has a 200x200x200 print area yet folds down into a 37x37x8cm size.
However. None of these 200x200 printers is designed to be shipped in their own box: the electronics and the critical precision assemblies are all directly attached to or are integral parts of the outer frame. Thus it would be necessary to arrange packing material and an *additional* sturdy box for the shipping of the printer. That sturdy box effectively becomes part of the BOM of the whole printer. A portable printer is designed to be transported regularly: thus, the box may not be thrown away at each destination, as it will be required later on a regular basis. That box, therefore, needs to be stored somewhere when the printer is in operation.
Translation: the actual working size of every portable 3D printer has increased by approximately 5cm to 10cm in all dimensions during shipping. With the TeeBotMax] being a whopping 550x550x250mm when stowed, with the increased shipping dimensions of between 100 to 200mm on all dimensions, it would be completely impractical to carry, even without a packing case. Even the PortaPrusa would be increased in size to approximately 450x450x160 during shipping, and the box would have to be stored somewhere when the printer is in operation.
There do exist some printers that are contained in foam-packed metal briefcases: these briefcases are too small to fit a 200x200 printbed, so instead, they are forced to go with a smaller print area: 150x150mm is typical (only 6in sq.) - only one of the briefcase-style 3D printers has a 150x200mm print area.
For most purposes, 150x200 would normally be fine except until there is a requirement to print 230 to 260mm-long, narrow parts. These normally fit across the diagonal of a 200x200mm printbed, but would *not* fit across a 150x200mm. So, the briefcase-sized foam-packed 3D printer designs are also out.
Thus, with no existing 3D printer fitting the requirements to (a) be in its own box (b) have critical components and precision assemblies surrounded by protective foam during shipping (c) provide a 200x200 build area (d) be a practical compact size to both use, carry and ship, it becomes necessary to design and build yet another folding 3D printer.
Rigidity Analysis
A primary concern of a folding printer is that it is completely rigid when assembled, both the main frame as well as the printbed. Considerable design analysis was carried out prior to beginning the build, and, now that the first prototype has been made, it is confirmed that the design is as stiff as intended.
Frame
Rigidity of the frame - despite the design essentially being four main parts - was achieved through geometric analysis prior to the build, and tested in cardboard (see videos below). The two box halves are semi-rigid triangular wedges, with panels filling in the rectangular areas giving stability in two out of three directions. For assembly, the two halves are hinged (not rigidly joined) in two places at the bottom to create a pair of "jaws" that sits upright ("mouth" pointing upwards). On their own, these "jaws" are able to flex and rotate about the z-axis, but are stable in both X and Y, due to the flat panels in the box halves. However: the Top CoreXY assembly specifically uses very strong folded-metal stainless steel triangular corner-braces that allow virtually no flex at all about the z-axis, but, clearly, a stand-alone two-dimensional (flat) rectangle is still able to flex in both X and Y.
It is the *combination* of these two complementary assemblies that result in rigidity: each part stabilises the other. So when the Top CoreXY assembly (rigid in Z but flexible in X and Y) is attached in four places to the top of the box (rigid in both X and Y but flexible in Z), the entire assembly suddenly becomes as stiff as any regular non-folding 3D printer.
Printbed
The upright printbed assembly takes advantage of the stiffness of the assembledframe by being attached in four places. However, the printbed must also fold down, but must again be rigid when assembled. Several designs were considered and analysed: the preferred one is based on triangular underside hinged bracing, and split Z rail supports, with the whole arrangement being very similar to how trestle tables fold and stow.
The Z rail supports therefore take two linear bearings (four total: two left, two right). Long arms with the printbed are attached to the top pair of Z-rail supports, on 625 bearings so that the printbed rotates downwards. Short brace bars (again that rotate on 625 bearings) are attached to the bottom pair of larger Z-rail supports. A further set of 625 bearings are on the other end of the brace bars.
When assembled, the two Z-rail supports lock together with two wingnuts (each side) to create two whopping 75mm printbed supports. The printbed diagonal brace bars are long enough to reach almost half-way (100mm) along the underside of the printbed, so that the printbed is supported about its middle point. When stowed, the lower pair of Z-rail supports are allowed to drop downward: the brace bars rotate to lie flat against the main bars of the printbed.
There was some concern initially that the bearings would have some play in them, resulting in the printbed wobbling when assembled. However, it turns out that there is a huge amount of friction in the 625 bearing holders, which results in considerable stiffness, and the printbed does not flop about at all.
Analysis of Printbed "see-saw" issue
One small concern remains that needs to be corrected, which is a design flaw of the original Fusebox design as well. In the original Fusebox design, only a single Z-screw is used. The Z-nut is in the middle of a 140mm-long plastic part which extends over to and is attached both left and right to both Z-rail supports. There is **only one** such part, and there is no other bracing. Thus, when a downward force is placed anywhere on the left side of the printbed, a see-saw effect about the middle of the 140mm-long Z-nut part results in the right side going **up**.
This see-saw effect can be corrected easily, by arranging triangular bracing to extend from the bottom of the Z-rail supports (the long Z-nut brace goes between the tops of the Z-rail supports), all the way to meet as close to the Z-screw nut as possible (so as not to put stress on the horizontal brace itself). With two diagonal braces, any force applied to one side of the printbed will be turned into an attempted "rotation" of the upright 8mm Z-axis rods. As these are quite strong and are rigidly fixed to the frame, there will be no significant see-saw effect. In the original Fusebox design, the Z-rail supports are only 45mm long (the length of a single LM8LUU linear bearing) and the 140mm-long horizontal Z-nut brace between them is positioned half-way up. This leaves not really enough room - or leverage - to make good use of a diagonal bracing effect. However, in the Sandwich200, the Z-rail supports are an enormous 75mm high and have both a 25mm LM8UU bearing *and* a 45mm LM8LUU bearing (each side) which means that diagonal bracing would be both feasible and effective. The tricky bit will be to have effective bracing that can separate into an upper and lower half.
In practice, however, it turns out that although there is some flexing, it is not sufficient to disrupt printing. Perhaps if an extremely large, extremely dense part were printed as far over to the left or right as possible, it might become a problem. Printing parts off-centre is never recommended (especially on a heated bed), so in practice this see-saw effect is not a major concern.
Bill of Materials
The BOM is maintained in the source for the CAD model, and is auto-generated (including an accurate count of nuts and bolts required). A copy is here: http://hands.com/~lkcl/foldable3dsandwich200/sandwich200_bom.pdf
Printed Parts
The STL files are auto-generated from pyopenscad (included in the source). Recent copies may be obtained here: http://hands.com/~lkcl/foldable3dsandwich200/scad/stl/ and http://hands.com/~lkcl/foldable3dsandwich200/herringbone_gears/ (gear_large_46.stl and gear_small_13.stl). Full source is obtained with git clone http://hands.com/~lkcl/foldable3dsandwich200/.git/ Build requirements: install git, GNU make, python2 and a recent version of openscad. run "make download" followed by "make".
Tools
Recommended: obtain a small geared power drill or a screwdriver with a standard hexagonal toolbit adapter: there are an enormous number of M3 nuts, bolts and screws to assemble (over 300)
- 1/4in (6.5mm) M3 hex nut/bolt socket: narrow / long type. OD must be no greater than 8mm, extending for at least 10 to 12mm.
- Allen keys: 1.5mm, 2.0mm, 3.0mm, 4.0mm (these are for the MK8 drive gear, GT2 gears, etc. - get 2 of the smaller ones because they tend to wear out)
- Screwdrivers: electrical (for the EC), flat-head and/or pozidrive (for M3 screws)
- M3 spanner(s), adjustable spanner(s), or Pincer-nosed pliers with small eye for nuts (M3) (Qty 2 recommended)
- Another 3D printer capable of 0.20mm layer height (your own or use someone else's)
- Keyfile set (round, square, flat, arc, triangular etc.) and small Craft knife (for trimming 3D parts: keyfiles can also be useful for filing inside extrusion ends, so that M3 hex nuts fit properly)
- Method for cutting extrusions to length (see below)
- Wood saw (if choosing MDF, Plywood or Hardboard for the box sides). Obtaining the panels cut-to-order is recommended (most iromongers / hardware stores can do this for you)
- Soldering iron, solder, stand and sponge (usual deal for soldering)
M3 hex socket
Regarding the M3 hex nut/bolt socket, there are a huge number of nuts and bolts to assemble, so it saves a huge amount of time to have the right tools. However, if using hex-headed bolts (as supplied from Makerbeam), Makerbeam "cubes" which are recommended for convenience require a very specific M3 hex socket to get the bolts in. The access holes in the cubes are around an 8.5mm diameter, and to access the bolt through that hole it is necessary for the socket to reach down through that hole to a depth of at least 10mm, preferably 11 or 12. Thus, a standard (strong, larger OD, shorter barreled) 1/4in socket will not fit, but can be modified with an angle-grinder in about 10 minutes.
If preferred, however, the relevant upright extrusions in the box may be extended by 30mm, an allen-key style bolt used, where to tighten it a hole is drilled in the end of the extrusion. The technique is very effective but does require a metal drillbit and preferably a workbench-mounted drill[1].
Another alternative, which may be simpler, would be to use flat-head (not countersunk) M3 pozidrive screws into the cubes. However: ensure that the screws are the self-tapping type (or be prepared to tap the ends of the uprights prior to assembly). M3 allen-key style bolts might also work very well, but would require confirmation of the clearance height for the heads, inside the cubes.
To-order or cutting aluminium extrusion
There are a couple of options here. Ths simplest, if you have a Ltd Company (or access to one) is to purchase cut-to-order Misumi1515 extrusion. Other sources of compatible 15x15mm extrusion such as Makerbeam should also do cut-to-order. If you are going to cut them to length yourself, you can either use a metal hacksaw and a metal file (if you have time) - remember to overcut then file down the last bit to get a clean, flat edge.
ONLY if you are competent with them you can CONSIDER a 10,000 RPM disc cutter with 125mm x 1mm metal-cutting disc. But - BE WARNED - Disc cutting tools are extremely dangerous: despite weighing 2 to 3kg if you get them even slightly off-centre when half-way through the metal they can jump and kick out of your hand, send the metal flying off at high speeds, and, because they get smashed between two pieces of metal , this can cause the disc to disintegrate and send bits flying off at extremely high speeds (well over 200mph). Basically, please think twice before considering it, but if you're going to use one, get the appropriate safety equipment: goggles, leather gloves, face-mask, ear-plugs, leather protective apron and/or trousers, and steel-capped safety shoes. If you've ever worked in the building trade you should have all this equipment to hand. But, seriously: please only use disc cutters as a last resort.
A better, safer alternative is to find a metal-working shop or an ironmonger's or hardware store that can do metal cutting. If they can do it, ask them to over-cut by 1 to 0.5mm, then "grind" down to the lengths required: this will give better accuracy and cleaner ends, which is important.
Development
Discussion on the reprap forum is here: http://forums.reprap.org/read.php?397,639675,648097 and there are some additional references on the fusebox forum here: http://forums.reprap.org/read.php?397,557542,649164
Version 1.0
Initial variant. Uses a non-intersecting CoreXY design to eliminate belt crossing issues. 3mm Hardboard for the outer box keeps each box half rigid (no diagonal bracing needed). Folded steel corner braces from a hardware store keep the CoreXY assembly rigid.
Issues
- Z-axis issues - pressing down hard on left side of the printbed causes right side to go up.
- Untidy wiring
- Huge number of nuts and bolts for the box (almost 100 to hold the panels in place)
- Wing nuts stick out and often catch on nearby parts, making the printer harder to assemble and disassemble than it should be
Todo
- Convert to 24V in order to increase speed without back EMF[2]
- Replace wing nuts with rotating, self-locking plastic parts (or spring-loaded "guitar-case" clips)
- Add 2.8mm rail-runners to all parts (to fit into extrusion grooves), locking PLA parts in place.
- Extend plastic parts where required so that no placement of any part requires measurements to be made during assembly.
- Analyse the practicality of putting the Y-rods parallel with the CoreXY belts (above, below or in between), going through the idlers and motor supports, in between bearings. The general idea is to reduce the width of the box by another 80mm as a result, which would translate into a potential reduction in stowed height of 30-40mm (170 or 180mm box height instead of the current 210mm).
- Install and test the E3Dv7 "Volcano" hotend kit with their current experimental 0.4mm nozzle. "Volcano" hotend kits provide a 20mm heat chamber instead of the more usual 10mm, thus greatly increasing the surface area of filament heating and thus allowing much faster prints.
Gallery
A photo gallery is maintained here: http://lkcl.net/3dprinters/portable.sandwich200/rev1/
Videos
<videoflash>3n4vfIlU1B8</videoflash> <videoflash>KcDm6pHVeBo</videoflash> <videoflash>c2Yowf7puvQ</videoflash> <videoflash>vMktV4MyZA4</videoflash> <videoflash>ttiOFbyHWZc</videoflash> <videoflash>enHXQ1z3BZQ</videoflash>
