Comprehensive search for full strength material printers

From RepRapWiki
Jump to: navigation, search

Hello, this is my first page on this wiki.

I have been interested in the last few months in printers capable of printing in highly useful full strength materials such as metals, ceramics, high performance plastic etc. to directly produce end products which can be used in applications that are more demanding than jars or doorknobs. That is, made to adequate tolerances etc. to be part of machines and tools, the main source of humanity's material wealth.

As far as I can tell there is no index anywhere of printing methods on the RepRap project, so this may morph into that.

The stated overall goal of the RepRap project is to develop von Neumann universal replicator like replicator that can produce directly end-products. So I figure narrowing the focus down to printers that can print eminently practical full strength parts is a good start. The Official RepRap machines, when using practical materials like nylon etc. may fall into this category. The printer should also print to a practical tolerance though.

Even wikipedia omits a lot of information from it like particular printers. So I am making this page, where links to such printers can be more easily collected in one place, and the possibilities and ideas can be either listed, or where it takes up too much room, linked to (as it matures the information can also be added to wikipedia).

Secondly, fundamental information about e.g. the build rate that can be expected for a given laser power and type and material should be made available in a way that is useful for developers and other interested parties, which apparently is not available right now on the wiki. I will try to contribute to these things in the coming months but if anyone can pitch in that will help.

This page is organized along the lines of the typical product development/design process: Level one is the initial thinking, looking at existing options and so forth. Level 2 involves trying to come up with more specific ideas. And it usually continues, zeroing in on particular implementations and getting them working to product release. You might end up backtracking to previous levels but that's the idea.

If you notice something that is not here, please add it right away.

Contents

Level 1

Comprehensiveness

Let's make an index of absolutely all the options, starting hierarchically from the top with all possible methods of heating, depositing materials etc. for a comprehensive search for better printers that can produce real world, useful goods from full strength materials.

Yes, biology might lead to molecular assemblers. But I'm not willing to wait that long, are you? There is still absolutely enormous potential for getting manufacturing capabilities into the hands of real people through truly versatile printers that can make it easy to make real-world goods.

A good metal printer for instance could make practically anything a lathe, milling machine, forging press, casting, cold rolling mill, wire mill, etc. within size limits can, but be far cheaper than buying all those machines.Given the home power supply,s limitations , and the variable usefull proportion of this power (eg laser in steel 98% usefull but in Aluminium 3% a factor of 32difference (given you need in metal 600 W to print in hours not weeks for Aluminium the home supply is too small)). Overcoming these obsticles mean it can be owned and controlled by real people.

A good multimaterial printer (maybe three or four materials on a single workpiece, several microns accuracy, good surface finish, no stray particles, water soluble supports, would suffice?) could print motors, PCBs, sensors, gears, journal bearings, structural parts, screwdrivers, wrenches, and combine them all into a single object in one print operation making robots, household appliances, light bulbs, stirling engines, welding guns, parts for industrial machinery, and who knows what else. Certainly not anything, but it would be incredibly useful. It could nearly replace the functionality of a large amount of not just production equipment like lathes, but assembly equipment too.

Sorting things in a comprehensive way will help with the attempt to avoid missing potential and promising methods. The methods may be combined if it were advantageous, with e.g. most of the heat deposited with one method and the rest deposited with a more precise method to achieve reasonable resolution.

Also remember there are undoubtedly many options that include stuff that most people had no idea existed, like the plasma detection to determine melt pool temperature in the temperature feedback section, which I came across in a patent. Patents could be mined and also brainstorming from people who know physics would help a lot to fill in this index of possibilities.

The design process can also be divided in a sort of hierarchical way, with the principles and ideas at the bottom, then through more specific ideas, to designs, to prototypes, and to fully working prototypes that meet the design goals at the top. So that is a handy way to organize a search like this I think. Imagine it like a tree. Most of the branches will end before they get to the fully working, adequately performing prototype stage. Let's map and grow that tree - to the top.

Precision/accuracy/tolerance requirements and material properties requirements

See [[1]] for information on tolerance and precision/accuracy. A list of what exactly is particularly desirable to print should be made so that printers being considered can be compared, putting their capabilities in context a bit.

Material properties are to some degree an issue. See library section, in particular the roadmap PDF. Very strong metal and ceramic parts can certainly be produced with additive methods and often are. In fact they can be quite a bit stronger than normal cast parts with the same material due to changes in crystalline structure etc. Additive manufacturing also gives some unusual control over properties, see [[2]].

However standardized, uniform properties require additional development of the processes and technology and even science to achieve. This is a concern for large scale manufacturing but it's a long way off. I think we can more or less ignore it for now. The main thing is to even get to the point where real people even have access to this type of manufacturing capability to start with. Then the necessary process monitoring and higher level of closed loop control needed to get standardized material strengths could be added.

Objects you'd particularly want to print and the precision and material requirement of them

See main page wanted objects.

With a metal printer

Here is a list of the sort of practical things that can lead to serious economic wealth generation which could be produced by printers.

  • Journal bearings. Ball and roller bearings would need a finishing step, as they need to be machined to a tolerance in the range of 0.5 microns roughly, too high for any printers on the horizon for now.
    • Journal bearings include the hydrodynamic bearings used in engine crankshafts etc. though they need a high surface polish.
  • Pistons and cylinder sleeves for engines, compressors. Maybe piston rings too?
  • Other expander mechanisms like small turbines (multi micron accuracy can work), worm compressors, roots blowers. Can be used for heat engines like rankine engines (steam), Brayton engines and Stirling engines. Also hydraulic pistons and motors.
  • Custom fiddly bits like screws and bolts and fittings, valve seats etc. in sizes that are a bother to obtain. Indeed, if the entire print volume can be filled with bolts or screws then that is not so uneconomical as it adds up to a large number of parts.
  • Flexures, machined springs.
  • Wrenches, and other tools. Parts for things like hammer drills.


With other full strength materials

Castable materials

Castable materials like polyethylene could be used in printers like DMLS printers perhaps. Thermosetting materials and epoxy do not fall in this category.

  • velcro
  • fabric could be produced theoretically, if fiber diameters are higher than the resolution of the printer. You would literally print each fiber in the garment. Thus clothes could be produced with synthetic fibers such as rayon, or polyester. Embedding elastomeric fibers or producing fiber blends would require a multimaterial printer. You could also just print a spool of material.

Multimaterial objects

  • Stepper motors, and brushless AC and DC motors of substantial power output. Need to use journal bearings, or add ball bearings yourself after printing.
  • Printed Circuit boards, conceivably they could be printed in complex 3 dimensional shapes, with the parts added as you go along with a pick and place machine, and soldering done with the laser or other heat source. High performance ceramic boards could also be produced.
    • Electrical components like small value inductors and capacitors as well as resistors could be feasible. Practicality depends on the printing speed for precise or detailed components though.
    • Electrodes for batteries often require or benefit from complex structures which could be printed. The active materials could even be printed right into the battery, and the casing around it.
  • Screwdrivers


Things that appear hard to make with existing printing methods

The limitations stem mainly from the resolution, the precision and accuracy, and materials limitations.

Remember that any printer in the forseable future will only occupy a corner of the manufacturing equipment ecology needed to make all things. However, in many cases it may be possible to print (or the parts for) a cleverly designed, low cost widget which can, in turn make particularly desirable things like a laser tube (or other resonator). If the design is open source, it becomes easy then, given a printer and an internet connection, to make the desired lasers or other product.

  • anything that falls into the category of chemical engineering basically
  • Parts that require particular microstructures which are below the resolution of the printer
  • Parts that require specific materials that cannot yet be printed in. Most full strength printers use melting during the processing. So if the material cannot be cast or injection molded it may be not possible to print with it using existing techniques. Includes semiconductors and other things that require manipulating atoms or molecules at higher resolutions than the printer can manage.
    • In some types of plastics which cannot be melted.
    • Rubber and some other elastomers. May be meltable elastomers.
    • leather, wood, cotton, etc. in many cases there are meltable, usually synthetic materials that can be used in most circumstances.
    • Paints or other specially formulated materials in some cases, although there are often satisfactory substitutes for paints e.g. a themoplastic with pigment mixed in applied to the surface and heated to bond to the underlying surface can make good paint substitute.
    • Motor brushes made from graphite? Theoretically graphite may be melted but might not be graphite upon freezing. Same for other materials with specific microstructures or atomic or molecular orientations that need special processing. Polarized polymers like mylar or Spectra.
    • A laser tube may be hard to make, although a CO2 tube is conceivable. The shorter the wavelength the more precisely and accurately it must be made with regards to the alignment and shaping of the mirrors at the end of the cavities, and the length of the cavity and purities of the materials involved. A Yb-fiber laser might be unproduceable for the fprmer reasons, and processing Yb doped glass might be hard as well. Plus the pump laser is usually a diode laser, so semiconductors.
  • In many cases end products like a shoe could not be made in exactly the same form as we know them, but perhaps a satisfactory shoe could be made with processable materials.

Heating methods to melt material or otherwise weld or join

Light

(do calculations do find out what intensity will give what material build rate for what types of materials, then find out what the intensities the various sources can produce and if it is adequate. There may be something which can produce intense enough light which is likely to be cheaper/easier to make than a laser.)

  • laser
  • Thermally radiant: sun, halogen light, plasma, gas and/or other extremely hot objects maybe magnetically confined, like a piece of tungsten, molten, suspended in a vacuum and heated inductively... but the tungsten would evaporate though. Hm. Maybe there are materials less prone to evaporation.
    • melt probe, basically a tungsten halogen light bulb with a method to circulate the halogen to cool the glass, then highly transparent glass similar to fiber optic glass (maybe not needed), then the tungsten right up close to the surface of the glass. Could be used for bulk melting of material. Under a vacuum a melt probe could just be a piece of tungsten sans glass.
  • synchrotron radiation?

Main issue is brightness and cost per watt of radiation, remember cost is up in the air and probably a lot lower than current market price if you can print the object.

Modulation or direction of light

  • galvo mirrors
  • essentially a super bright projector i.e. a raster image bright enought to melt the material being printed, may have e.g. only one line of vertical resolution or similar
  • mechanically moving the light source or a fiber optic head which emits the light (often used with fiber lasers on DMLS machines)
  • move the workpiece instead of the light source
  • materials which change refractive index when exposed to electric feild
  • combo, could use motion for gross motion and galvo etc for fine control
  • modulation of laser per se depends entirely on the type of laser usually and needs to built into it's design though an element could be added in between laser and workpiece to selectively reflect or absorb some light

Radio waves or microwaves

  • probably not because the resolution would be low, although it could be used in combination with radiant heating maybe

Conductive

  • A hot melt probe?
  • a wire comes down, gets extremely hot by some means and promptly touches the surface, but this is sort of a convective method since it would melt.
  • the bulk of the powder bed is often heated in laser sintering


Electrical direct ohmic

  • have wire much like wire welding but with very small wire so can get decent resolution,

Electrical inductive

  • magnetic induction (seems like it would be hard to get the precision, but maybe with multiple coils and areas where the fields cancel out?)
    • Could be used with wire to form a large melt pool, then followed with subtractive machining.

Particle beam

  • electron beam commonly used, currently known how to make low cost per watt of heating power to the melt pool:
    • magnetic fields produced around the melt pool might move the particle around a bit? Not much if any since it is low current
    • Might not be practical to print anything nonconductive due to the accumulation of charge that would occur, redistributing powder and causing deflection of the electron beam.
    • Requires high vacuum, but making good metal parts sort of requires vacuuum anyway maybe not as high. Oil diffusion pump is a low cost way to get high vacuum though so probably not a major problem.
    • stray electrons could be an issue? Make everything inside print chamber out of metal and ground it.
    • relatively easy to aim and modulate
  • ion beam, probably no good because would deposit ions in the material

Friction

  • see ultrasonic,
  • Friction stir welding? Other rotating rod?

ultrasonics

  • ultrasonic welding, 2 pieces touch and friction welds them together, vibrating rod or wire touching surface? There is that one that uses a sonotrode and foil layers followed by milling (add link to it in the right section)
  • maybe just heating due to the expansion and compression of the material.

Convective (hot fluids)

  • hot gas blown on the surface?
  • hot plasma blown on surface
  • hot liquid metal drizzled or sprayed on as droplets (there is a method of surface coating that uses sprayed droplets but apparently always leaves a porous bulk material)


chemical

  • Small hydrogen flame or other flame, like welding
  • decompostion of a chemical gas


Transferring the material from the main reservoir or print head to the workpiece, by physical state

Gas

-vapor deposition (seems like it would be hard to get high res or accuracy, even with heating - or cooling maybe (by exposing to a cold object nearby so heat lost by radiation or convection or even conduction) some parts of the workpiece to deliberately increase the deposition rate while keeping it low in others, also seems limited to certain materials)


Liquid

  • droplets
    • heated droplets,
    • electrospray, would be powder by the time it gets to the surface probably

Solid

Powder

  • distribute in thin layer over the surface, then sinter
    • textured roller (common)
    • electrospray with molten metal
    • powder gun with electrostatics, nozzle can be round or wide and flat, beam of charged powder particles could be steered maybe, (one problem maybe the different charge/mass ratio of the different particles, could be reduced in various ways like using sorted powder particles), ultrasonic atomization type thing to get the powder to jump off the surface of the reservoir bed if it is needed to overcome the particles sticking together, or maybe just charge it strongly enough it jumps off due to mutual repulsion (much like with electrospray).
    • use a spinning impeller to simply throw the powder around in various ways, down a tube and onto the workpiece, or dispersed randomly over the surface, thrown at a plate so it spreads out into a sheet-like spray, then move the spray back and forth across the surface.
  • fire directly into melt pool with powder gun of some sort (accuracy could be a problem) there is the LENS system, the company that makes it claims they can focus a particle stream entrained in gas (helium probably) to a 10 micron spot which is pretty good.

General points relating to powder: Powder distribution could be messed up by nearby charges or magnetic feilds or gravity, vibration, etc. the uniformity of the layer and the thickness errors would add up without feedback and the ability to manipulate the powder bed after application or otherwise correct for nonuniformities in the layer thicknesses (i.e. z axis errors).

Manipulating powder after being deposited on surface:

  • Intense pulse from laser or electron beam could cause rapid local heating at a spot, causing expansion, causing some powder to jump away from that area. Powder could be "herded" in that way maybe.
  • magnetic feild from electron beam
  • saw a paper once about moving objects around on a flat surface with a phased array of transducers, peizoceramic elements could be used and ultrasonic for finer control but will always be limited control presumably as stray vibrations will mess up powder distribution elsewhere. But it only has to be good in the are being sintered...
  • ultrasonic or sonic lubrication could help to level the powder bed after application.
  • electrical charges and electrostatics could be used in various ways, apply some charge in an area with an electron beam or use a charged probe to "herd" the powder, maybe combined with ultrasonic lubrication


Removing powder

  • The surface of the workpiece could be charged enough to get the powder to jump off by mutual repulsion. For nonconductive surfaces spraying it with an electron beam might be used to give the charge. But the residual charge may need to be removed to get a nice even powder layer layed down afterwards.
  • Brushes etc. might be practical.
  • Have the workpiece upside down or turn it upside down, then vibrate with ultrasonics etc. to remove any sticking powder.
  • blow it or vacuum it off with gas
  • attract neutral powder away from surface with electrostatics
  • centrifugal force, probably not practical
  • light pressure (photon pressure) unlikely to be practical but maybe
  • other vibration
    • thermal expansion getting it to jump off, as mentioned above, just herd it over the edge of workpiece or to a disposal area


very small Non spherical particles like powder

-nasa preferred small rods apparently for some reason instead of powder in an ebm for their use

Wire or rod

  • see existing additive methods section for several methods that use metal wire
  • could use short rods, a device that orients the rods then feeds them down to the surface to the melt pool. Much like a laser welding system with metal wire except ceramics might be be easy to make into wire. The rods could be handled like a powder until they reach the print head, whereupon they are oriented and fed to the surface. Or it could be a fiber like glass fiber, as long as it's thin enough I guess it could be sufficiently flexible and.
    • feeding from a spool of very fine wire breakages are inevitable so system needs to be able to recover often and reliably from them.

Plasma

  • sputtering, heat some areas to prevent deposition while allowing in others (probably no good accuracy)


Feedback methods for the workpiece shape, powder layer thickness etc. in the print volume

In some cases they are specific to or dependent on specific printer approaches. Where this is the case, add some details in brackets after.

Shape (of workpiece, powder bed etc.)

  • Scanning electon microscope, also other types of electron microscope (tunnelling, non scanning lensing type (like the first electron microscope? not common?) (under vacuum)
  • cameras, normal cameras, pinhole cameras, light feild cameras, microlense array camera (microlense for each pixel eliminates the focus problem)
    • stereo cameras
    • the thing with the laser line generator and only a single camera to produce a 3d image (needs an xyz table)
  • use a laser to scan the surface and a photodetectore of some type so it's like a scanning electron microscope but with light (good for laser sintering maybe) (main point is photosensor cheaper than camera)
  • a single LED or laser, measuring resonant frequency to nearby objects by being part of a resonant circuit that resonates with the frequency corresponding to the optical path, then apply fourrier analysis not very accurate though. Have more than one emitter/reciever for 3d and maybe can interpolate or similar for higher res. Maybe too difficult computationally for any image area of reasonable size.
  • radar presumed useless
  • touch probe like AFM or with a much larger probe (too slow and disrupts the powder layer? could be used after powder is removed or along with powder bed manipulation methods)
  • using resonant frequency measurement plus fourrier analysis and other computation to determine the shape of a workpiece surface probably only works practically with conductive workpiece, maybe not though, could work with sonics or ultrasonics, maybe multiple emitters/recievers makes computation easier. much the same as the led resonant circuit one and similar problems. Like an ultrasound microscope?

temperature of melt pool feedback

  • pyroelectric sensors, slow though? Maybe there are good ones available.
  • electrical characteristics of a photodetector give information about the color of light striking it?
  • Use refraction of diffraction to divide the wavelengths and them multiple photosensors, examples are a color camera, spectroscopes to get emission spectrum info
  • saw a patent that used the plasma emitted by the melt pool heated by a laser, sensed by the current between 2 nearby electrodes to estimate melt pool temp (under vacuum I think) , mine patents more.
  • ultrasonics to determine changes in propagation speed in the hot melt pool
  • amount of laser light reflected from melt pool change with temp?
  • modulate laser beam and listen to the sound produced
  • what else can do with modulating the laser or electron beam? Modulate the beam while doing something else? Don't think of it as a steady state system.
  • comprehensive information gathering with a complete model of the printer system and as much feedback as possible from the real printer for various parameters might reveal ways to infer the temperature of the melt pool etc.(see general) ( also dead-reckoning using computer simulation if there are in-between areas where sensing leaves blinds spots etc.)
  • light reflected back from melt pool, and the shape of the light feild may reveal info about smoothness of melt pool etc. maybe smoother when molten.
  • overall conductivity of path, how changes with time etc. (for electron beam heating based printer especially since already have vacuum and ebeam)
  • modulate electron beam and watch the response, noises produced, net current flow, backscatter using the sensors for the SEM, etc. (for electron beam heating especially)

General approaches to improve a printing process

  • Remember the higher level possibility of combining all the characteristics of the system that can be manipulated - everything down to the current/voltage ratio of the electron beam in an electron beam printer and the ambient temperature of the print chamber for example - and all the things that can be sensed with whatever hardware is already being used or low cost additional hardware (for example, a microphone to detect vibrations that are being produced), then create a system that allows manipulation and sensing (for example, modulate the electron beam current and see what sort of vibrations are produced or form a graph of the voltage vs. current relationship.

This information might be used to reveal information about the temperature of the melt pool in an electron beam printer. Or even totally passive sensing to be combined with computation to infer what you want to know. In many cases there may be manipulatable values and easily sensible parameters or other information that we forget we can use.

Humans of course tend to be pretty good at doing this when using a machine manually because we pick up on clues. In this case we broaden our own sensing capabilities to include things like the voltage vs. current relationship of the electron beam, and do something similar. Knowing in detail how the printer works - ideally having a nearly complete computer simulation - would help a lot to interpret the signals available.

Another example of information from somewhere you might not expect, information from the SEM shape feedback mechanism might conceivably, for instance, give information about the melt pool temperature through the amount of electrons reflected from the melt pool, or the shape or smoothness of the pool surface or similar.

A genetic algorithm might help if it had a good computer simulation to work with. Digital prototyping tools in e.g. autocad may reduce the need for custom software for the simulation?


  • Full strength printers seem almost invariably to use a single point to add material to i.e. a single melt pool. Having a large array of lasers or ebeams or whatever and therefore many melt pools might be better in a number of ways, allowing faster build times with finer detail and also eliminating the x and y axis motion requirements if galvo mirrors or similar were used. Might reduce the per watt cost of laser light too. If a large number of lasers were used the whole thing would be close enough to the print bed that you could eliminate some of the optical path components, but then would need to be able to lift it vertically as printing progressed which would be a source of innacuracy due to play in the mechanism but not much. An array that was high above the printer could use electrorefractive stuff instead of galvo mirrors and maybe be accurate enough without z axis motion.
As mentioned in the roadmap PDF (see library section), moving from single point methods to multipoint like a row or array of points may inherently increase speed without compromising on precision.  The Arcam printers have recently been endowed with multiple electron beams for this reason - same power as a single larger gun so similar material addition rate, but the process can get more precision.  Just basic engineering really, work with what you got at all levels.

In some cases the resolution vs. the size of the print bed may not be satisfactory because the angular resolution with which say a laser or ebeam can be directed is too low. In that case, maybe using multiple printheads which are closer to the workpiece may be better.


  • On thing that is interesting in looking at the existing methods is the sheer number of different ways to do it. There are just so many different possibilities, including so many unexplored surely some can be made to work or harnessed in a RepRap eventually.


Idea for metal deposition head: have heard electron beam welding can be done under a low vacuum with the electron gun having a plasma window, it is a high vacuum inside the gun. Could maybe be used to make a process similar to EBF3 but without the expense of a large high vacuum chamber, with all parts in it made of specialized materials for low outgassing (including the stepper motors, electronics, sensors etc. Dunno about you but I bet versions of those things rated for use in a high vacuum are pretty expensive).

Electron beam welding but with large number of very small heads, wire could be made on the fly hierarchically from larger wire to reduce breaks in the wire, could rotate and move so the entire printe surfeace is covered by a large number of heada so they do not all have to work, various numbers and sizes of heads, various ways of moving the heads and also coudl be ultrasonic welding to weld the wire to the piece .

In many cases ultrasonic lubrication may be useful to keep the powder moving through the reservoir or delivery channels or whatever. Also could be entrained in a fluid.

  • A lot of problems with controlling the temperature of the melt pool, if it is too high boiling or evaporation of the material results when under vacuum. Under argon atmosphere is much less of a problem.
  • Speed is increased while adding material to the interior of the object by sacrificing accuracy etc. in many cases there are similar ways to improve performance greatly compared to e.g. what you would be stuck with if you used only a powder layer 4 microns deep and a melt pool 5 microns wide, which would give very fine detail but be very slow. But you only need detail at the surface of the object. If you print up a sort of wall where the outline of the object is that were say 40 microns high, the basin that is formed between the walls can be filled with powder and sintered relatively fast with a large melt pool. Can similarly be applied to wire welding methods i.e. use very fine wire for the outline of the part for high precision but low deposition rates, then use large wire for the bulk.
  • Other ways of increasing real world performance and reducing demands on e.g. laser modulation speed, by similarly recognizing the limitations on the sort of object we actually tend to want to print.

Laser cost:

One of the main problems with using the electron beam is that it would leave nonconductive objects charged, complicating powder deposition and beam aiming. Multimaterial printing therefore would probably be better done with light as the heat source, either laser or with a magnetically confined very hot plasma or other extremely bright light source. Lasers are proven to work already. The main problem is the cost per watt when purchased.

If the printer is sufficiently capable it is not unreasonable to expect it can be used to make e.g. a CO2 laser tube or a machine which can make a tube. It may be better to make an array of laser tubes each with their own galvo mirrors, to acheive the same net power output but without the difficulty of handling a lot of high intensity light with a single mirror/lense, like an array of 10 by 10 40 W tubes, each with their own galvo mirrors. You may have to calibrate them so they work together effectively, with adjacent printed areas being aligned. Since you could print most or all of the galvo mirrors the cost is probably not a major issue, though the first prototype might be expensive. You could simply use 2 tubes and galvo systems for the prototype, and scale up the size of the array later.

Another issue for the laser is the modulation of the laser, and the power supply, which is not all that cheap or easy. There will be a ton of waste heat too. Ultimately maybe the problem needs to be handed off to some photonics people and maybe they can find a way to produce a suitable system which can provide the light we need with a complete system that is low cost (after factoring in the ability to print it or a widget that can make it ourselves, so not talking about the market price cost).

Notes on printers which use E-beams to melt material

Would it be possible to periodically remove some of the powder, so that the level of the powder bed remains slightly below that of the top surface being printed rather than above it? That could provide the benefits of the surrounding powder bed acting as a support structure, but also give the more desirable shiny finish rather than matte as particles from the surrounding bed would not stick to the outer surface as it were sintered.

Mutual repulsion of powder particles could maybe be used to get them to spray off the powder bed surface and transfer to another area.

Notes on printers that use light for melting materials

Need a cheaper laser, that is the main thing, but direct laser sintering similarly suffers from matte finish due to the surrounding powder bed. A support material may be needed to avoid it, or like the metal printer reducing the level of the powder bed?

List of existing working printers that can print in or have the potential to print in full strength materials

There are a great number of these additive manufacturing processes, it turns out. No surprise given the numerous ways to do it. However many of them are only near net shape methods, but I'm including them too since they are interesting and could be combined with another method, and are examples of processes that could be adapted to printers.

(Try to make it the official manufacturer page if there is one)

  • The radiative selective inhibition sintering one, can't find it again now. Used inkjet to deposit fluid in areas where you don't want the powder bed bonded, then a nichrome heater element to heat the bed. Fluid evaporates, keeping some areas cool and unbonded. Existing embodiment can do solid high density copper alloy objects but temperatures are too low for steel.
  • Direct metal laser sintering, DMLS. The original process that was dubbed DMLS was developed some time ago and there are many variations now each with different properites. Some only achieve a low accuracy of +/- 50 microns or 95% dense parts, while some can reportedly achieve several micron level precision and higher density. Some use an XYZ head with laser light brought to it through a fiber optic cable, and others use only galvo mirrors that scan across the powder bed. Supposedly it is already used for more than one material on the same workpiece with limitations, using some approaches [[3]].
    • Co2 laser low precision:

http://www.3axis.us/direct_metal_laser_slintering_dmls.asp

    • fiber laser DMLS:

http://www.morristech.com/dmls.asw (25 micron accuracy reported)

  • Interesting table that gives an idea of the variety of processes out there:

http://www.additive3d.com/tl_tab2.htm

Additive processes that might be adaptable to printing

By "printing" I mean the production of finished parts made to a fair precision. Some of these processes may be considered printing process but they cannot do detail well so I put them here instead.

http://www.freeformfabrication.com/

  • microcasting:

http://npl-web.stanford.edu/user/files/papers/sff1994a.pdf http://npl-web.stanford.edu/user/files/papers/sff1994a.pdf


prototypical freeform fabrication methods

  • I don't remember exactly which these were but at least one of them was a freeform fabrication followed by milling method:

http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.11.7390&rep=rep1&type=pdf. http://utwired.utexas.edu/lff/symposium/proceedingsArchive/pubs/Manuscripts/1998/1998-55-Song.pdf. http://www.aimme.es/archivosbd/observatorio_oportunidades/design_development_layer_based_additive.pdf http://utwired.utexas.edu/lff/symposium/proceedingsArchive/pubs/Manuscripts/2002/2002-54-Mizukami.pdf. http://utwired.utexas.edu/lff/symposium/proceedingsArchive/pubs/Manuscripts/1999/1999-092-Song.pdf. pretty good welding followed by millin gdiff machines

Currently, layered manufacturing have had liquid-based, solid-based, and powder-based systems. Among them, Selective Laser Sintering (SLS), Selective Laser Melting (SLM), Electron Beam Melting (EBM), ProMetal 3D Printing (3DP), Selective Laser Cladding (SLC), Lasform, Laser Engineering Net Shaping (LENS), Precision Optical Manufacturing (POM), Laser Additive Manufacturing Process (LAMP), Wire-Arc Spray (WAS), and Multiphase Jet Solidification (MJS) can be used for fabricating components made of metals (i.e. satisfy Requirement (1)) from http://edge.rit.edu/content/P10551/public/SFF/SFF%202004%20Proceedings/SFF%20Papers%202004/29-Chen.pdf. Page 1

  • Multiple Material

Selective Laser Sintering (M2SLS) http://repositories.lib.utexas.edu/handle/2152/674 http://repositories.lib.utexas.edu/bitstream/handle/2152/674/jepsonlr029.pdf?sequence=2 on multimaterial sinteringt only functional gradients thgouh


prototypical printing approaches

  • http://inderscience.metapress.com/app/home/contribution.asp?referrer=parent&backto=issue,8,10;journal,17,24;linkingpublicationresults,1:113605,1 ubtractive manufacturing (CNC machining) has high quality of geometric and material properties but is slow, costly and infeasible in some cases; additive manufacturing Rapid Prototyping (RP) is just the opposite. Total automation and hence speed is achieved in RP by compromising on quality. Hybrid Layered Manufacturing (HLM) developed at IIT Bombay combines the best features of both these approaches. It uses arc welding for building near-net shapes which are finish machined to final dimensions. High speed of HLM surpasses all other processes for tool making by eliminating NC programming and rough machining. The techno-economic viability of HLM process has been proved through a real life case study. Time and cost of tool making using HLM promises to be substantially lower than that of CNC machining and other RP methods. Interestingly, the material cost in HLM was also found to be lower. Synchronisation of this two-step process offers a new accelerated way of building metal tools and dies. HLM can also be used as a cheaper retrofitment to any three or five axis CNC milling machine or machining centre.


  • http://www.sciencedirect.com/science/article/pii/S0924013608000976 very interesting, plasma followed by milling, ****no distortion*** the support material could be deposited on various areas to prevent vapor deposition , maybe could also selectively heat various areas too , Actualyl that could be really good if precise heating could be done and precise feedback . New idea for printer: have vapor deposition to the part and a head that goes around and mills areas to completion, then deposits material (maybe thermally insulating) on the surface to prevent further vapor deposition (maybe have another material that is thermally conductive for use as support material ) maybe a separate head for deposition May not be full strength parts though. "plasma deposition" usually means plasma spray deposition.


High level notes on Multimaterial printing

Multimaterial printing indicates printing in multiple materials on the same workpiece.

  • Multimaterial printing is already done in limited forms. Some DMLS printers can already print in multiple different metals in a limited way by melting powder onto the workpiece, which is then consolidated with the laser, rather than distributing a thin layer of powder (presumably the powders would contaminate each other or the differences in density between powder and solid metal is an issue).

The capacity to print a different support material in some types of printer is also arguably multimaterial printer.

Potential ideas:

  • Lay powder down, print layer, then remove *all* the excess powder, lay a powder for a different material down, print, remove all excess, repeat. Needs supports since no surrounding powder bed.
  • If could direct powder very carefully and precisely into the melt pool? Some stray powder probably inevitable though.
  • More than one printer simply side by side, the workpiece is moved from one to another.


Level 2 in the design process hierarchy: Specific potential printers and points to consider for them

Single material part output (may include support material)

  • in some cases the resolution vs. the size of the print bed may not be satisfactory because the angular resolution with which a laser or ebeam can be directed is too low. In that case, maybe using multiple printheads which are closer to the workpiece may be better.
  • Electron beam welding can be done under a low vacuum with the electron gun having a plasma window, it is a high vacuum inside the gun. Could maybe be used to make a process similar to EBF3 but without the expense of a large high vacuum chamber, with all parts in it made of specialized materials for low outgassing (including the stepper motors, electronics, sensors etc. Dunno about you but I bet versions of those things rated for use in a high vacuum are pretty expensive). As mentioned on the MetalicaRap page, not useful for ebeam printers due to beam spread.
    • Electron beam welding but with large number of very small heads, wire could be made on the fly hierarchicaly from larger wire to reduce breaks in the wire and store the wire supply off the print head, could rotate and move along the x axis say so that each area on the workpiece is passed over by a large number of heads, so they do not all have to work all the time. Various numbers and sizes of heads, various ways of moving the heads.
  • Selective inhibition sintering machine but with a brighter light source so it can sinter a wider range of materials practically. (question: the solid material must be more dense than the powder so presumably the powder layer top slowly get higher than the workpiece and may topple onto the workpiece? Never seen mention of this beign corrected for.) How good is the accuracy though?
  • A printer much like a the MetalicaRap but smaller (maybe 5x) and cheaper. Like with a print volume the size of a mason jar or so?
The smaller size would make prototyping much cheaper so it might actually happen.   It would also be intrinsically more accurate due to the accuracy of the current metallica being limited by the electron beam deflection system, allowing the printer to be simpler, and eliminating perhaps the subtractive machining.  Secondly exactly what you want to print should be nailed down a bit better; I suspect that most of the value is in smaller, more precisely made parts not large parts. 

Specifically, I suggest using electrospray with molten metal (the metal can be wire that enters the electrospray unit before being melted) for uniform powder dispersion over the build area. The vacuum chamber is a faraday cage (when the valve later mentioned is closed) so hopefully a good taylor cone can be obtained.

Feedback using stereo SEM approach using the same electron gun or a smaller lower power side one, there are examples of DIY Scanning electron microscopes. This allows software compensation of the layer thickness variations

2 chambers connected by an all-metal ball valve: one for building and one for waste powder. To transfer powder from one chamber to the other give the build chamber a strong electrostatic charge. Powder would stream from one chamber to the other by mutual repulsion much like a gas. Ultrasonic vibrations may be applied to shake any particles that are stuck to anything. This prevents much of the toxic metal powders from escaping.

Secondly, during the print process, I wonder if it would be possible to make the powder bed lower than the top surface of the object being printed. If this were done, the edges of the object would not be surrounded by a powder bed while being formed, which would be good because no or little powder would stick to the object then - you might get a smooth, powder free finish that requires no further processing. But to do this the edges of the object need to retain enough powder as it falls from above and settles on the surface of the workpiece - you have to be able to build a vertical cliff face so to speak even when to printing plane is not surrounded by powder. However I doubt that would be a problem. The capacity for shape feedback from the SEM allows flexibility with regards to inventing ways to ensure that the edge of the object builds at the same speed as the bulk of it.

The powder bed can still be used for supporting objects as the surface of the powder bed could be only say 50 microns lagging the uppermost surface of the part i.e. the printing plane. Of course the parts of the objects printed in the bed because they need it as support would get powder stuck to them as usual but that would only be a very small area of the total part.

This could be accomplished by periodically removing small amounts of powder from the powder bed using the electrostatic method just mentioned. If the bed needs to be re-leveled after this electrostatic process it could be fluidized with ultrasonic lubrication briefly. The surface of the bed can be imaged with the stereo SEM too as needed to ensure flatness. These might cause problems though with areas of the parts that are not connected to each other yet shifting relative to each other, I don't know. It may be possible to fluidize only the top surface of the bed or hopefully it would not need it. Powder excess needs to be removed somehow though as the solid metal is more dense than the powder so the powder bed will otherwise become taller than the object with the sides of the powder bed eventually perhaps caving. In DMLS a scraper is used to level the bed.

The metallica says it can get to 20 microns across the pritn volume with the intitial Ebeam printing, divide by 5, 4 microns is plenty accurate for a lot of things. Bolts are like 50 microns tolerance or so. A typical CNC lathe gets in the range of 20 I think.

Something like this could concievably be able to make (if it could be worked out), directly with little or no finishing, just to give an idea, small piston heads and cylinder sleeves for engines or compressors, gears, journal bearings, hinges, custom fittings, motor couplers, obscure types of screws or bolts, screwdrivers, wrenches, many of the parts for say an air tool like an air drill, valves and valve seats, springs, custom objects of all sorts including of small size and relatively high precision. In other words real parts for real machinery and tools and household items :).

Probably not easy to put side to side to have print large objects but something like this would be a good start.

Auxiliary ideas and potential methods

  • use a normal inkjet piezo pump thing to deposit the soluble but heat resistant support material in solvent carrier, solvent evaporates and leaves the material behind as a deposit. Metal can be deposited on top.
  • Put powder onto a rotating cylinder like a laser printer, then have it brought very close to workpiece surface and then repelled and attracted to piece surface or maybe touch the surface, to release from cylinder. Then melt layer to melt the powder and build up a layer, clean cylinder and repeat. Best method I can think of for depositing material very precisely, might be useful for something. A system that deposited material precisely then heated more or less indiscriminately might be easier to build at some point.


Multimaterial printers

  • Single head for cartesian robot with several different heads maybe combined into one head - subtractive (edm or milling most likely) and additive heads (FDM, inkjet, metal deposition, ceramic powder with laser melt pool heating, and a cleanup tool to remove any stray powder or chips from the milling process befor the next step, for example) on it, each suited to different materials and methods could form a functional and capable multimaterial printer. Head might get pretty heavy but that can be compensated for with a heftier structure etc.
  • one idea with laser heating:

An array of lasers, say 5 by 5 or 10 by ten or whatever the photonics people indicate is appropriate, with galvo mirror(s) on each one. The focus is directed to the workpiece at a spot several microns across or whatever is a good melt pool size for the laser power that turns out to be convenient (or more for a co2 laser). The galvo mirrors in the adjacent array have scan areas that overlap slightly. The array is calibrated so they work in concert effectively. Their closeness to the surface allows high resolution without the need for insanely accurate or galvo mirrors and also reduces the demands on the flatness and shape of the optics a bit. That provides the precisely directed heating needed.

The lasers would be designed so that they can be produced with a machine which itself can be mostly printed.

The atmosphere is argon, like in the LENS system.

Then the capability to put a powder layer down is provided, probably textured roller as it's proven. Brownian motion helps to provide a uniform powder layer as the powder falls through the argon. Then you need some way to remove the powder highly effectively, like blowing it away with highly pressurized argon, maybe ultrasound to loosen any stuck particles, then vacuuming up the cloud of dust to reduce the mess and prevent it from settling elsewhere on the workpiece (metal and other dusts harmful if inhaled etc.). Apply a different powder. Repeat. Support structures as needed.

Bulk melting could be done with some other light source maybe, which was less precise than lasers but way cheaper per watt.

    • It has to be able to print a vertical cliff face sort of thing somehow, might be a problem if powder falls off the rounded edge of the cliff face. Could be fired at the surface with enough force to allow slight sticking to the surface? Or maybe would stick naturally enough. Could be given a slight electric charge.
    • Powder particles might remain on the workpiece and cause abrasion or otherwise shorten the life of bearings that were printed.
    • Plastic next to metal plastic would burn at the melt pool temperature of the metal if you tried to add it after. As long as the high temperature material is printed first should be okay so maybe not a major problem. So during the print process the metal would protrude above the plastic layer at all times until the final layer of plastic.
    • one of the materials would need to be a support material, ideally water soluble or similar for easy removal. (ideas anyone? Has to be something that can be sintered, not corrosive or too toxic either)


Level 3: Explicit designs

The MetalicaRap. Electron beam heating. Tolerance goal is 20 microns on all dimensions and distances.

All the offioial RepRap machines are capable of printing in full stregth plastics. Tolerance varies greatly depending on the circumstance and material used. Have not been able to find the typical tolerance achieved.

Level 4: Open Full strength material prototypes

library

Documentation on printers that is particularly interesting http://wohlersassociates.com/roadmap2009.pdf roadmap for the future of additive manufacturing

http://www.astm.org/Standards/additive-manufacturing-technology-standards.html

http://www.deskeng.com/articles/aaayyx.htm roundup of additive manufacturing technologies as of 2010, they mention some documents that look even more interesting. I looked for the first Wohler report one but could not find it.

Documents that are from the peer reviewed literature are usually copyrighted by the publisher but can be shared to a level of 3 people passing them around after the initial purchase.

To do list for this and related pages

idea for metal or other printer: put powder onto a rotating cylinder like a laser printer, then have it brought very close to workpeice surface and then repelled and attracted to piece surface or maybe touch the surface, to release from cylinder. Then melt layer to melt the powder and build up a layer, clean cylinder and repeat. Best method I can think of for depositing material very precisely, might be useful for something. A system that deposited material precisely then heated more or less indiscriminately might be easier to build at some point.

Idea for multimaterial printer : Single head for cartesian robot with several differrent heads maybe combined into one head - sutractive (edm or milling most likely) and additive heads (FDM, inkjet, metal deposition, ceramic powder with laser melt pool heating, and a cleanup tool to remove any stray powder or chips from the milling process befor the next step, for example) on it, each suited to different materials and methods could form a functional and capable multimaterial printer. Head might get pretty heavy but that can be compensated for with a heftier structure etc.

General improv section: as mentionied in the roadmap PDF, moving from single point methods to multipoint like a row or array of points may inherently increase speed without compromising on precision. The Arcam printers have recently been endowed with multiple electron beams for this reason - same power as a single larger gun so similar material addition rate, but the process can get more precision. Just basic engineering really, work with what you got at all levels. Idea for metal deposition head: have heard electron beam welding can be done under a low vacuum with the electron gun having a plasma window, it is a high vacuum inside the gun. Could maybe be used to make a process similar to EBF3 but without the expense of a large high vacuum chamber, with all parts in it made of specialized materials for low outgassing (including the stepper motors, electronics, sensors etc. Dunno about you but I bet versions of those things rated for use in a high vacuum are pretty expensive).

electron beam welding but with large numbe of very small heads, wire could be made on the fly heriarchically from larger wire to reduce breaks in the wire, could rotate and move so the entire printe surfeace is covered by a large number of heada so they do not all have to work, various numbers and sizes of heads, various ways of moving the heads and also coudl be ultrasonic welding to weld the wire to the piece . poitt out there are just so many different possibilities surely some will work in face there are many commercial examples that do more or less 

-move on an xyz a small number of very small wire welding things add summary page on product ecology t put pruinters in peerspctive -almost anything can be scaled down and multiplied - other variations of net shape layer then milling or foil then milling , dip peic ein molten metal tehm mill it or wipe molten metal of maybe , wire welding ebeam resistive followed by milling, divide existing printers into existing sections add "projects in progress always break down the critical parameters involved e.g. number of heads and try variations on that themee , also put in the right place in teh tree the varisous varaitions on the idea but signal to noise ratio matters http://www.aero-mag.com/features/41/200910/59/ afdd to existingprinmter section, "ohmic wire weldi f direct manufacturing could eb combined with milling steps after each layer for accurate parts add the roadmap pdf also add

wire arc or resistive http://ntrs.nasa.gov/archive/nasa/casi.ntrs.nasa.gov/20080013538_2008013396.pdf (mostly about electron beam freefrom Several arc-welding processes, using either wire or powder feedstock or both, have also entered the arena, including Plasma Transferred Arc Solid Free Form Fabrication (PTA SFFF) and Shaped Metal Deposition (SMD) which uses either Metal Inert Gas (MIG) or Tungsten Inert Gas (TIG) welding techniques in a layer-additive fashion.[10 has other interesting stuff about ebeam Shaped metal deposition: http://www.rapolac.eu/smd_info.html http://www.amrc.co.uk/featuredstudy/shaped-metal-deposition/

pTA SFFF http://www.freepatentsonline.com/y2005/0268517.html http://www.dtic.mil/ndia/2006smallarms/christou.pdf

interesting on freeform fsabricatiom http://aries.ucsd.edu/LIB/REPORT/CONF/SOFE99/waganer/ mentions Laser or plasma arc forming and Spray Casting Google terms that coudl use more mining:

freeform fabrication

http://infohost.nmt.edu/~hirsch/homepage.html#fab robocasting, says metals and ceramics

microcasting: http://npl-web.stanford.edu/user/files/papers/sff1994a.pdf http://npl-web.stanford.edu/user/files/papers/sff1994a.pdf add stuff to wikipedia too when can


search the acronyms in the roadmap pdf

Keywords for high leevl concepts "direct digital manufacturing" "additive manufacturing" "freeform fabrication" "solid freeform fabrication" "layer by layer manufacturing"

ebm http://www.freeformfabrication.com/

idea: a two-fabber-system: one makes the right micro-shapes, the other one assembly them to the right macro-form (ultrasonic or resistance welding or welded together somehow?) except if you have a micofabber why not build the obkjects right on top of each other instead and just move the micro fabricator around the object

take thing at http://forums.reprap.org/read.php?1,92359,92359#msg-92359 and put it on the wiki somewhere

- put longer sections in their own pages with a link to main page and summary where it used to be link title

related pages on the reprap wiki

Ultrasonic consolidation Hybrid printer Concept multimaterial laser printer Metal deposition print head