RepRapMicron
This page is a development stub. Please enhance this page by adding information, cad files, nice big images, and well structured data!
//identifies pages under construction. remove when the page is relatively stable.
Release status: Early development
| Description | Micron-scale 3D Printer
|
| License | |
| Author | |
| Contributors | |
| Based-on | |
| Categories | |
| CAD Models | |
| External Link |
Contents
Summary
The µRepRap is intended to be a RepRap capable of micron (1/1000th of a millimetre, or 1µm) and sub-micron fabrication. This degree of accuracy was initially made possible by 3D printed microscope platforms designed by The OpenFlexure Project. Variations more suitable for prototyping and fabrication have since been developed, but kudos to them. Expectations for the initial prototype are to demonstrate repeatable positioning to better than 3µm on a work area 10mm across (accuracy is highest at the centre) and to produce a probe tip in lieu of a print head that is suitable for manufacture in a simple home workshop.
Overview
Microelectronics and Micro-electromechanical Systems (MEMS) are essential components of most of the electronic wizardry we use in our everyday lives, whether we realise it or not. In the way that RepRap brought Open 3D fabrication to the masses, the aim of the µRepRap Project is to bring users the same capabilities on a much smaller scale and allow those components to evolve in the same way.
Background
Precision manufacturing began somewhere in the 1700's, and the first micron scale electronic devices were fabricated from silicon in the 1960’s. The Intel 4004 CPU, for example, was made with a precision of 10μm. Ever since then the technology focused largely on silicon, with fabrication systems becoming ever more complex, esoteric, and costly. The techniques used are difficult for the average hobbyist to manage, and in many cases are downright dangerous.
In biology and medicine, equipment to measure and manipulate objects on the micron scale are relatively common - though these devices tend to be large, specialized, and expensive. Recently though, microscope platforms capable of sub-micron resolution were developed by The OpenFlexure Project, and these have created an interesting opportunity for developing micron scale fabrication.
If micron scale manufacturing can be achieved by RepRap-like technology, it is likely that these fields will be advanced in the same way that manufacturing was by the RepRap. The biological sciences will gain from inexpensive, rapidly-evolving equipment. The microelectronics field will regain the potential for independent communities develop on the micron scale, and break away from its fixation on silicon as its main platform. As with 3D printing, there will certainly be new developments in fields that do not currently even exist.
There is also the possibility that this technology could replicate in the future, and do so at a yet smaller scale.
Requirements
Micron scale 3D printing has many of the same requirements that The RepRap Project developed when initially printing on the macro scale:
- A 3-axis positioning system
- CAD/CAM software
- Axis zeroing sensors
- A deposition system
- Building material
In addition there is the practical aspect that humans are unable to directly manipulate micron scale assemblies and sub-assemblies. Novel systems are therefore needed to:
- Detach printed items
- Transport items
- Rotate and position items
- Conduct micron scale maintenance tasks
- Operate and interface with items
While conventional optics are readily available to allow humans to initiate and inspect the fabrication processes, it is likely that some customisation of the optics will be desirable.
Adaption Of Existing RepRap Technology
The 3-axis positioning systems developed for the RepRap are largely applicable to operation on the micron scale. They are also readily available and understood by many potential collaborators. Likewise the CAD/CAM systems developed can largely describe the volumetric and control aspects on the micron scale. Early RepRap designs catered for many initial design issues experienced, such as backlash and the management of delays in the extrusion system, and these will likely have parallels.
One example would be the positioning system. Current 3D printers use microswitches, optoelectronics, and hall-effect sensors to detect the zero position of an axis. Others simply slam the axis into a physical stop. One possible solution is a light gate closing off a light source, the light minima indicating a known position. Actual probe height above the work area needs to be determined, and this initially is likely to be a manual process. A piece of burnished 30μm aluminium foil laminated to a glass working surface using UV resin has proven effective as a touch plate for rough alignment, and as a bonus provides sensitive materials with protection from illumination sources under the glass.
The control of the deposition process and the choice of building material will need to be reconsidered due to the practical issues of creating fine extrusion orifices and moving phase-changing materials through them. Photosensitive resins as used in resin printers do scale however, and similar materials are already used widely in the microelectronics industry. Their wide availability to the 3D printing community makes them worthy of consideration. If the work area is illuminated by UV for photosetting from underneath, the reservoir of resin can be kept in a shielded portion of the work area above. The probe can shelter in this area when exposure is in progress, and reload itself with resin at the same time.
An opaque resin could be sparingly used as support material. Under an intense light pulse such as a laser, the darker resin in the finished part will be able to absorb enough light energy to vapourise. A similar strategy could be used to weld manipulated parts, though it may be simpler to strategically apply liquid resin before manipulating them.
It is likely that a number of substances with desirable physical properties will be experimented with. Conductive and electrically active materials are an obvious step. A magnetic material would allow a means of activating assemblies by means of an external magnetic field. Droplets of catalyst could be used to solidify a substrate or render it soluble etc. The actual cost of the consumables is relatively unimportant due to the small quantities used. Accessibility is of greater concern.
Novel Requirements
The simplest form of deposition system is an old-fashioned dip pen. It requires no more than being dipped in an inkwell, and then to be touched to a surface. If the ink can be persuaded to change phase by thermal cycling, photosetting, or application of electricity etc., the print head itself need not have any complex or moving parts.
A sufficiently sharp tip, in the sub-micron range, can be easily made on the workbench from fine wire. 22 gauge (0.12mm) titanium or nichrome wire work well. Place a large electrode in the bottom of a container of 5% sodium chloride solution and connect this to the negative side of 3 AA cells in series. Suspend a length of wire vertically in the salt solution with about 8mm immersed, and apply +4.5V across it. Electrochemical erosion occurs, and when the end of the wire falls of, cut the current. The process takes a few minutes and can be automated or done manually.
Combined, these items allow the formation of a test system for deposition, operated by conventional CAD/CAM systems attached to an OpenFlexure stage.
Assembly Manipulation
Seeing the object being fabricated during development is crucial. As micron scale optics tend to be very 2D and have a limited depth of focus, gauging the height of things is particularly difficult. There are some tricks that can be applied such as creating shadows, and illuminating the object with different coloured LEDs from several angles. These techniques are also useful for observing completed objects. They will be useful for flat printed parts that are subsequently manually folded up into 3D structures.
Once an object has been fabricated, one way to detach and manoeuvre it would be to simply use the “ink” to glue the probe tip to the assembly. While adequate for initial experimentation, eventual re-use of the probe is desirable and a release mechanism such as heating the probe tip could be implemented.
To rotate parts does not necessarily require a rotating manipulator. Parts could be made to rotate around built-in pivots when moved or operated with the probe. To move in the vertical plane, an assembly could contain joints that allow it to erect itself at the desired angle by manipulation of hold points with the probe. Once the assembly is in the required orientation the probe can be glued to the angled assembly. By use of multiple probes, each attached to an OpenFlexure stage, and the ability to apply glue, assemblies can be combined arbitrarily to produce macro-scale items and either positioned with the probe or more conventional manipulation. Another possibility is electroadhesion or plain old magnetism, but this only works with some materials.
The maintenance tasks are currently unknown. Likely more convenient tools – grippers, rotating devices, probe recovery systems, ink well fillers – will need to be manufactured. The early stage of development will likely have a high attrition rate.
Future Advances
As with the original RepRap, it is hoped that once the first μRepRap is operational, it will be able to replicate. Unlike the original, the replicant will not be identical to its parent but will be a much smaller functional equivalent. Control of the replicant will be difficult, but not impractical. Options include perpendicular magnetic fields, inducing vibrations to operate ratchet mechanisms or tuned structures, or at its most basic cranking the mechanism with the tip of a probe until something better can be worked out. It should be possible to run more than one printing mechanism simultaneously.
It might also be possible to fabricate a μRepRap that is scaled down further. Assume that a RepRap has a precision of approximately 0.2mm on a 200x200mm work area, and a μRepRap has potentially a 1 micron resolution on a 10x10mm work area. We see roughly two to three magnitudes in reduction in scale. If flexure systems also scale, then it might be possible to use a μRepRap to create a smaller printer with a resolution on the nanometre scale. Unlike current attempts to operate with nanometre precision with macro scale hardware, errors in the device due to thermal expansion and other material-related noise would be much reduced. It would at least be interesting to see how far this limit could be pushed.
Practical Progress
Initially a modified OpenFlexure Delta Stage was used. After sufficient experience was gained with the larger working volume, the Delta Stage was abandoned in favour of the OpenFlexure Block Stage.
With GRBL CNC firmware, the Block Stage not only allowed the use of end stops, but the "Z Touch" feature used to probe workpiece height allowed the conductive probe to be used as a rudimentary scanner on conductive objects.
The OpenFlexure designs proved excellent for digital microscopy, and their one piece design made assembly simple for the novice. This did not suit the requirements of prototyping, however, as it was not possible to replace parts or observe the movement of components inside the integral hull. Prototype testing was also more rigorous than regular use, and breakages occurred. Reconfiguring or repairing the design was not possible without reprinting the entire mechanism, and so it was necessary to adopt a more modular approach.
With the experience gained operating in the micron environment, the next prototype was designed from scratch without using any files or unique features of the OpenFlexure design, allowing it to be released under the GPL V3. While not an ideal licence, it is suitable for the ethos of the RepRap Project, which has always been more concerned with enabling development than controlling it.
Maus Prototype
The Micron Accurate Universal System- "Maus" was developed as a toolkit for design testing and development, rather than as a specific configuration. It could even be implemented in delta configurations, but this section covers the initial use of the modules to create a testbed for the prototyping system.
The Maus toolkit consists of several sections:
- "Metriccano" modular mount system
- Axis driver
- Flexure stage mounts and linkages
- Probe system
- End stops (not yet implemented)
- Software suite
"Metriccano" Modular Mounting System
A common mounting system was designed using M3 fasteners on a 10mm grid. As these measurements had already been used to develop probe hardware to repair a Block Stage, it seemed a sensible standard. This allowed a system of strips, struts, plates, and spacers to be used to connect the modules of a stage assembly together. Due to its resemblance to the 20th Century modelling kit "Meccano" and being in Metric Units rather than Colonial Units, this was christened "Metriccano" and an OpenSCAD library of parametric parts created. Once developed, a Metriccano design can be joined, condensed, and compressed in OpenSCAD to produce a module that requires no fasteners and yet is still capable of being used in a reconfigurable manner.
Standardised Axis Driver
The axis driver design was standardised, each axis using the same flexure-based module with its own NEMA17 stepper motor. This was driven by the same GRBL firmware and RAMPS hardware as earlier prototypes. Being identical axes the work area became a 4mm x 4mm x 4mm cube, though it is intended that an 8mm cube can be traversed without permanent damage to the mechanism. This is achieved with a dual parallelogram arrangement of flexures, driven by a lever with approximately a 3:1 advantage. The OpenSCAD model is parametric. A 3D printed concentric flexure was used to attach a M3 x 50mm drive screw to the NEMA17 by means of a tapered hexagonal well in its tip, and an M3 nut screwed down onto the head (diameter slightly reduced by filing) of the M3 screw. An M3 drive nut is inserted into a similar well on the lever, and retained against backlash by several silicone "Loom Bands" on the overhead bearing block. The bearing block holds an M3 nut drilled out to 3mm, which provides a loose bearing surface to protect the plastic block. A locknut and washers on the drive screw press tight against this bearing, providing enough downward pressure to keep the drive screw in its coupling.
After problems with the driver frame flexing, it was realised that mounting driver modules by the motor was a bad idea, and the mounting points were moved closer to the lever's pivot.
Motion Stages
The initial design made from Maus components used an XY Table with two degrees of freedom. The joints were flexures, 3D printed in flexure pairs. Two pairs formed the flexure assembly for the top of the table, and two identical pairs were inverted and aligned at the base of the table. The "legs" of the table were a vertical 100mm Metriccano strip, fastened each corner (note: allow clearance for the screw ends when table is in motion!). Some space is required under this assembly to allow the flexure joints freedom of movement.
To connect the XY Table to the X & Y axis drivers a simple 3D printed flexible linkage of 40mm x 5mm x 0.8mm was used. The link was oriented with the thinner dimension in the Z direction, allowing the X & Y axis drivers to move the XY Table with minimal effect on each other's motion.
A second design using pairs of vertically mounted flexures for the XY axis and using the same driver mechanisms is being tested. This version has an integral frame, dispensing with the need to screw the modules down to a piece of plywood. Providing stability by attaching the base to a piece of 10mm pitch perforated board is still recommended however. Development files are on github.
Probe
The construction of the probe tip remains the same as earlier prototypes, though the holding system was changed significantly to provide a reliable means of electrically connecting the probe. The probe was mounted on a separate Z axis, removing the potential for movement in the XY plane directly being affected by the positioning of the probe. This required a redesign of the probe holder. One convenient feature of the previous "Titch" prototype was that the static probe allowed the mounting of mechanically complex manual probe levelling apparatus. It is likely that with Maus the whole Z assembly will need to raised and lowered for coarse probe alignment.
Print Bed
The print bed is a glass slide with an aluminium foil square at one end, held in place with cured UV resin. This foil section serves several purposes:
- An electrical contact that can be used to calibrate the probe height using CNC "Z Touch" techniques.
- A reservoir that can hold a very thin smear of UV resin (an actual drop of resin presents issues with surface tension).
- Protects the resin reservoir from a UV source underneath the slide.
- Provides a location reference point to make finding the output easier.
In Maus V0.01 the slide is held in place with jaws. In V0.02 a magnetic mount system was implemented by embedding magnets in the stage. Placing magnets on top of the slide holds it firmly to the magnets beneath. This turned out to be very convenient as it allows off-axis placement of the slide. Jaws can also be fitted to V0.02 if desired, without colliding with the embedded magnets.
End Stops
As the interior was now visible, the need for developing end stop hardware was given lower priority, as the axes could be repositioned manually.
Software Suite
Two significant changes were made to the software used. Firstly instead of operating in microns, the GRBL board was reconfigured to work in millimetres and all toolpath development done on a scale of one millimetre to one micron. This was done because functions such as arc approximation and cornering are sensitive to floating point and rounding errors, which were producing artefacts. Also the control console, cncjs, would not zoom in far enough to see previews of GCODE operations in millimetres.
Secondly the means of converting SVG paths into GCODE was handed over to PrusaSlicer, also moving the project towards the goal of true 3D printing. This was primarily done as jscut would always close a toolpath, and this is a fundamental feature of the software's polygon engine. PrusaSlicer requires some configuration and output filters but will reproduce the intended path. A printer configuration was devised based on the Marlin driver, and Marlin-specific functions (such as GCODE 'M' commands, and movement on the A axis) removed with an output filter.
A "dipify" python script has been written that will convert GCODE movement from the slicer into a series of waypoints. The concept is that these waypoints will be where drops of UV resin are deposited. The code allows for the probe to return to dip into the reservoir for resin replenishment after a certain number of droplets have been deposited.
The resulting GCODE could be fed to the GRBL controller using cncjs as per the previous prototype.
GRBL continued to be used rather than Marlin as there was no pressing need to change, though it should be noted that Marlin does support the GCODE G38.2 command used for the Z Touch probe.
Notes On Microscopes
Three microscopes are in use on the prototype: An inexpensive, unbranded, 2MP USB microscope to observe initial Z probe height. A 4MP Digitech USB microscope used for observing XY probe positioning. A Biolux NV turret microscope for examining the results of a run, which is fitted with a plug-in low resolution USB optic module that provides pictures for experimental results. Some photos of output were made using a Konus Crystal-45 trinocular microscope and a generic 64MP camera.
To maintain even lighting throughout the day and night it was found necessary to cover the stage with a sufficiently large cardboard box when observing. This also reduced dust and debris.
USB Microscopes
Magnification values given my USB microscope manufacturers are essentially meaningless. There is no optical image to compare with the original, and comparing original image size with the image displayed on a monitor is dependent on monitor dimensions and screen resolution. It is recommended that the maximum number of true pixels is used as a guide for performance when obtaining USB microscopes.
The USB microscopes provide illumination and are held in an ad hoc frame. This frame will be developed and detailed later, as stable mounting is essential. The current version involves too many unsupported ends and too much gaffer tape.
Turret Microscopes
To look at what you've made you will need a real turret microscope. Your eye will do better than any USB plugin on a real microscope - the author uses the optical USB plugin to post photos and keep records.
On the turret microscope a 4x objective and a 5x eyepiece is frequently used and a 50 micron line is quite visible. For fine detail a 10x objective and a 16x eyepiece is used. This is only 160x but if your naked eyes can resolve a 0.5mm object, at this magnification you will be able to resolve 3 micron objects. Do not use the largest magnification objective on a turret microscope. It usually requires contact with the slide and will destroy your sample when you focus on it. If you need more magnification use a "Barlow Lens" adapter.
Binocular Microscopes
These allow observation of the object in 3D, and some have an additional camera port. Zoom capability is highly desirable. The maximum magnification is generally low - 45x to 90x - but the ability to spatially resolve components and output during prototyping is valuable. You won't need one, but you will want one.
Laser Projection Microscopes
If you suspend a drop of water on a point or slide, and shine a reasonably coherent laser through it, the laser will project an image of things in the drop onto a screen (or the ceiling). This is a bit fiddly, but an inexpensive and a useful trick for STEM education.
Cats
Cat hair varies between species, but Ragdoll hair has a diameter of around 30μm. If you have a cat, it is strongly advised that you keep them off the workbench or you too will be observing the diameter of cat hairs.
Collaboration
The project will be conducted as Open Source under the terms of the GPL 3 or later licence where possible, and documentation distributed under the terms of the GFDL. Note that the OpenFelxure Project uses the CERN Open Hardware Licence v1.2, which will be respected. Progress will be blogged on the reprap.org blog, and the primary repository for technical details and conclusions will be on the reprap.org wiki. Software and configuration files are on GitHub. Participation is encouraged. There is likely to be some discussion on the Facebook RepRap page - note that due to spam you must answer the group question before posting on Facebook!
As a side note, the original Z80 processor was manufactured using an 8 micron process. Construction of assemblies on this scale seems an achievable goal.
Conclusion
Hopefully this will bring the RepRap project to smaller and smaller things. VikOlliver (talk) 22:44, 6 March 2024 (EST)
