Pushing filament with a gear was never a good design; it was just something that hobbyists could build easily. Injection molding machines use a screw drive to push molten plastic. This new machine uses a scaled-down version of the injector screw of an injection molding machine.
Design and manufacture of injection molding screws is complicated but well understood.[1] They're not easy to machine, but that's what CNC mills and four-slide machines are for. It's like making a drill bit, something that's done industrially millions of times a day but is very tough for hobbyists.
I'd wanted to use a laser for heating in a 3D printer, but not to heat the plastic being deposited. The fundamental problem with filament-type 3D printers is that you're welding a hot thing to a cold thing. That never works very well. That's what causes cold solder joints and weak welds in metal. In 3D printing, it's why the bonds between layers are much weaker than the plastic's inherent tensile strength. I wanted to have a small laser (about 10W) aimed ahead of where the filament is being deposited, to heat up the material already placed. That way, both sides are hot as the weld occurs, which should yield a much better weld. This is what the heated build plate and heated build chamber designs are trying to accomplish, but they're not putting the heat the one place it's really needed.
As lasers in the 10W range become more affordable, this looks like a better idea. A few years ago, it was an insanely expensive idea, but now, you can get 10W for about $300.
Heated build plates and enclosures are primarily to prevent the part from warping and pulling itself off the build plate.
The problem with trying to bond two hot layers together is that they distort. Typically the previous layer has not solidified enough and so most printer employ a cooling fan to lower the temperature of the previous layer, not raise it.
It's very hard to find the right balance of extrusion temperature, fan speed, feed rate, and a myriad of other factors that affect print quality.
Honestly your comment makes me think you've never actually used an FDM printer.
Actually the fan is to cool the current layer, immediately after depositing, to solidify it as quickly as possible. That it also cools the previous layer is just an accident.
I think Animats is right. It would be better to locally heat the previous layer right before depositing, and then cool the joint right after.
Unfortunately that requires highly precise heating and (even harder) cooling, so is probably still a few years away.
Localized cooling isn't hard. Laser cutters do it; they have a tiny nozzle spraying compressed air (in some machines, an inert gas) on the cutting site.
For temperature feedback, a non-contact IR sensor would work. Operate the IR sensor only during the off time of the laser (presumably pulse-width modulated) and you can read the temperature while heating.
The screw mechanism they describe is quite a bit different than an injection molding screw. The screw in an injection molder is like an auger, designed to push plastic pellets. A hobbed bolt is not an option because it can't pick up the pellets. Here they are grabbing the filament from the outside with threads, kind of like rifling.
Reading the paper, it seems like they managed to double the force compared to a traditional pinchwheel. However, there are now extruders that used geared pinchwheels on both sides of the filament, doubling the force without requiring threading the filament.
That's what the article says, but there must be some trick to make that work. Trying to drive a filament just by threading it and running it through a rotating nut will mostly twist the filament or tear off the threads. Maybe an inside conical thread, like the negative of a wood screw...
It's actually just a standard 4-40 thread cut into the filament. They use rollers to prevent rotation of the thread when cutting the threads into the filament, and then when they actually use the nut it's got enough thread engagement that it's better than a hobbed gear. If you think about it, it's almost like a hobbed gear that attacks from all sides instead of one or two.
Feel free to ask me questions on it, I work in the industry and have spoken with one of the authors of the paper, it's a shame it's behind a paywall.
I wouldn't compare it to the screw drive they use on injection molding. Those are for pushing pellets and melted plastic, not for pushing a rod (or strand) of continuous plastic.
This design uses threads on the outside of the filament [1] to grip it with a nut.
The main difference is that the design and manufacture of this system is extremely simple, contrary to what you've said. They use off the shelf 4-40 [2] taps and nuts to create a triangular thread. I would even go as far as to say this is _easier_ to manufacture than hobbed gears.
It seems that improving the annealing process of printed parts from current 3D printers could, at best, reach the maximum crystallization of pure stereoregular PLA polymers.
Due to the two enantiomers of lactic acid, there are four different PLA polymers with different stereochemical structures: PLLA, PDLA, racemic PLA (PDLLA), and meso-PLA. It is only the the PLLA and PDLA stereoregular polymers that show crystalline characteristics. However, research indicates that mixing the PLLA and PDLA polymers allows the formation of a stereocomplex that has a much greater degree of crystallization. Note that this is a physical mixture of the two stereoregular polymers rather than a polymerization of the two lactic acid enantiomers, so the PLLA/PDLA stereocomplex (greater crystallization than either of the PLLA or PDLA polymers) is very different than the PDLLA polymer (which is amorphous, meaning non-crystalline).
Perhaps what we need is a 3D printer that uses one PLLA filament and one PDLA filament. The two filaments would be melted and combined in the nozzle of the printer. As the mixture of the two polymers annealled, allowing the stereocomplex structure to form, the resulting material might have mechanical properties that are superior to either of the parent filaments.
I wanted to use the same design to deposit copper powder to printed circuit board. The laser would draw a temporal heat mask and nozzle would blow copper powder over the workpiece.
But really, if you want PC boards, use a board fab house. There are so many good services, they're cheap, and you get drilling and plated through holes.
I have 3 printers of various quality and design, and by and large the issue with printing I find most difficult to optimize is a constant wall quality. As you mention, extruding hot plastic is a solved problem, but doing so a layer at a time is complicated by travel time of the print head, previous layers, and ambient air temperature.
Without that it is very hard to get consistent strength and durability at any speed.
The problem with combining a laser with a print head to perform some kind of heat-ahead function is you cannot predict the vector direction the head will be traveling with current hobbyist firmware, like Marlin, without modifying it sufficiently enough to buffer the GCODE so it moves the focal point of the laser to the next vector direction while still extruding in the current vector.
You'd be effectively operating two independent print heads at the same time, otherwise if you're just focusing the laser at the extrusion nozzle, you're just heating already heated plastic, or worse, burning the current layer of plastic behind the traveling printhead.
The problem with combining a laser with a print head to perform some kind of heat-ahead function is you cannot predict the vector direction the head will be traveling with current hobbyist firmware.
That's not a particularly difficult problem, although doing it on a machine with Arduino-sized RAM might be difficult. Large machine tools which interpret Gcode do considerable look-ahead, because they have considerable inertia and have to slow down well ahead of stops and turns.
A four-slide or mills for making screws? Not sure what you mean... a Swiss Lathe or milling attachment?
Most I've seen are one large piece machined on lathes with a tool attachment, then ground. And a some that are made of smaller 2"/51mm blocks that you slide together over a shaft for different configurations or to replace worn sections.
"Finally, they devised a high-speed gantry mechanism — an H-shaped frame powered by two motors, connected to a motion stage that holds the printhead."
It seems a CoreXY or H-bot design, I always believed that the issue with this design (like any cartesian design) is with the limits of acceleration and "drag" of (relatively) massive parts, particularly because the X and Y axis have not the same weight/mass when changing orientation of the movement (the Z is less of a problem, at least with "large" surface prints, the issue here is only around precision), and that was the reason why Delta's are usually faster.
Maybe a Delta with that head/printing gear can move even faster ...
> It seems a CoreXY or H-bot design, I always believed that the issue with this design (like any cartesian design) is with the limits of acceleration and "drag" of (relatively) massive parts
there are enormous multi-ton machines which can move faster than tiny little delta printers.
delta printers are good at maximizing speed when you have constraints like price, noise, weight, "fits in my home", "uses single phase residential power" and "cannot literally tear my limbs off". don't confuse the constraints of consumer printers with the constraints of industrial machinery, which is what it looks like these folks are trying to design.
Here's Naomi Wu's delta 3D printer mod. She mounted one on a backpack carrier frame, powered it with batteries, and walked around Shentzen while 3D printing.[1] This one isn't "fiddly" at all.
In a CoreXY printer you can calibrate the X and Y axes more or less independently. In a delta you have to take several points to get exact values for arm lengths and therefore a solution to the mechanism.
Very cool, but big caveats as usual from a uni press piece.
However, they ran up against a small glitch in their speedier design: The extruded plastic is fed through the nozzle at such high forces and temperatures that a printed layer can still be slightly molten by the time the printer is extruding a second layer.
“We found that when you finish one layer and go back to begin the next layer, the previous layer is still a little too hot. So we have to cool the part actively as it prints, to retain the shape of the part so it doesn’t get distorted or soften,” Hart says.
That’s a design challenge that the researchers are currently taking on, in combination with the mathematics by which the path of the printhead can be optimized. They will also explore new materials to feed through the printer.
That’s a problem with most 3D printers I think. I usually get around it by printing multiple copies of the object at the same time. By the time the printer gets around to the first copy it’s cool enough not to cause issues. Obviously it’s a much harder problem to solve the faster you print.
Doesn't work as well when you just want to pop out a quick small prototype so you can test it, iterate, and do it again, but definitely that works for some applications. And as they mention, path optimization can also help here. As a contrived example, if you're building an upside down extruded "T" shape, maybe you could work on building up the stem as high as possible as quickly as possible, and spending the "cooldown" time working on extending and building up the base. I'm not aware of any 3D printers that move beyond the simple "build from the bottom up" approach[1], but it seems possible to take into account the bulk/size/shape of the printhead and do some advanced things -- if the payoff is there.
[1] I also know approximately nothing about this space, I'm not speaking from an area of expertise, just novice theory. Maybe this is already being done or maybe it's simply impossible because the printhead is just far too bulky.
There are some experimental 3D printers that deviate from the “build from the bottom up approach.” Here is a pretty cool 5 axis 3D printer: https://m.youtube.com/watch?v=w8Fl8L4yk8M
Most slicers have a "slowdown if layer time less than x" setting too...in Slic3r it is in the filament>cooling tab. That is typically a better solution...but in this case I suspect this machine will have issues running slow..so the best bet is to only print large items.
> So we have to cool the part actively as it prints, to retain the shape of the part so it doesn’t get distorted or soften
This is a (solved-ish) problem even on current-generation commercial printers. It's called a "part-cooling fan". It sits next to the hot-end and blows air on the plastic as it comes out of the nozzle.
You get a worse bond if it's too cold. as the plastic cools it warps and can crack along the interface between layers. Plastics like abs usually require a heated print bed as well.
I think the ideal solution would allow the entire piece to stay at a elevated temp somewhere between ambient and melting, but blow cool air over the freshly extruded plastic to quickly bring the temp down from the melting point
It probably depends on what material you’re printing with. Keeping ABS from warping in a cold environment would be difficult or even impossible. At the very least it would put new limitations on what shapes are possible to print and constrain your options when it comes to infill.
The main benefit to the speed will be realized when they scale the build area of the machine. As I suspect they realized it is difficult to get past the physics of the plastic cooling without causing other issues (delamination, warping, etc). The gain will be largely noticed when they are printing with this extruder using a build area larger than 1m cubed. The issue on large printers is being able to get the newly deposited material down while the previous layer has not cooled too much (delamination), although, this can be can be partially controlled with things like heated build chambers and such...they are patented.
This idea was tossed around in #reprap for a few years (I know I have heard the idea thrown out there) and I am glad to see someone working on it. Although, I disagree that a screw drive is needed. I use hobbed gearwheel drive on my extruders and rarely is the issue the hobb slipping. Typically, it is the motor stalling instead. This isn't so much a product of the drive wheel/extruder as it is the hotend not being able to keep up with the commanded extrude volume...since this hotend doesn't have that issue (or so it seems) I suspect the conventional extruder could be used with just a modification to hotend/heater design.
Once the commercial 3d printing market moves up to things like servos, encoder feedback, etc. I suspect the speeds of things will start ramping up greatly. The issue isn't so much the tech but the cost...people are cheap.
edit: After reading the linked arxiv page. I see they are using closed loop servo motors/drivers and a RAMBO controller. They could go faster if they used a 32 bit controller like a smoothieboard or duet instead of the 8bit "arduino based" controllers. {Not saying RAMBO is a bad board, they are known to be very reliable for OEMs}
Not to mention that Smoothieware has the control modules for lasers and cnc built in as well as the 3d printer so it requires probably no firmware modification to get this running...just config and maybe a custom bin. (Disclaimer: I work directly with the Smoothie project so I may be biased)
> Although, I disagree that a screw drive is needed. I use hobbed gearwheel drive on my extruders and rarely is the issue the hobb slipping. Typically, it is the motor stalling instead.
Although that may be the problem on your printer, it's simple to fix because one can always add a larger motor. The problem that they solved is that eventually, a stronger motor won't cut it because the hobbed gear will slip, and that's what this paper is all about. Optimizing the motor is done by just upscaling it, but optimizing the drive is what made them to move away from the existing hobbed gear setup.
> They could go faster if they used a 32 bit controller like a smoothieboard or duet instead of the 8bit "arduino based" controllers. {Not saying RAMBO is a bad board, they are known to be very reliable for OEMs}
I would actually predict a minimal increase in speed. 32-bit controllers don't make for inherently fast printers, they allow for it if utilized properly. What you'll find though is that Marlin and Smoothie both use the GRBL planner under the hood, and generate very similar output. The only major difference I noticed in the output between the two is that Smoothie generates smoother (hence the name) step waveforms.
I guess you can also generate steps at a higher frequency with smoothie, but you should be careful equating higher step frequencies with higher speeds, when most of the time people are running with microstepping settings. Higher frequency allows you to get more dynamic range between speed and precision, but that's not really what the paper was about.
I’d like to see some close ups of the prints. In the video you can see the print bed vibrating like crazy, and each layer can’t be anywhere near solid if there’s a tenth of the time to cool, even with a fan.
I wonder if you could have a something that squirts a fine jet of liquid nitrogen, just behind the print head? Expensive though I suppose.
>I wonder if you could have a something that squirts a fine jet of liquid nitrogen, just behind the print head? Expensive though I suppose.
Maybe - just a semi-random idea, mind you - a Ranque-Hirsch (or Vortex) tube (and a source of compressed air) would be enough and possiby be more practical:
Also though depends on the need right? If I'm making a 3-d print of something that I just want to see a very rough mockup in space, I might trade accuracy for speed if it goes from an hour to < 10 min.
>Also though depends on the need right? If I'm making a 3-d print of something that I just want to see a very rough mockup in space, I might trade accuracy for speed if it goes from an hour to < 10 min.
Yep, coincidentally I was just reading today about a "crazy" experiment increasing "abnormally" the nozzle hole size on a common printer, JFYI:
I’ve been printing with a 1.2mm nozzle on my printer for about a year now. I love it. The fat traces are strong as hell and I can print at 0.8mm layer height.
However I have to print at 10mm/s and I actually use 0.4mm layers so I’m not sure it’s faster in my case. I’m still limited by my ability to melt plastic.
But I’m building a robot [1] that is 200 hours of print time currently, and I’d love to make it faster.
Yeah...overall your hotend is still limited volumetrically...so overall for volume extruded you are still printing at the same rate.
I feel big nozzles printing wide traces have other benefits like overall strength and such (less welds is my guess) and thicker walls per number of perimeters (again...less welds).
In my shop I keep a .7mm authentic jhead for when I want to use a larger nozzle.
In the mid 1980 there was a similar craze related to dot-matrix printers and their cps (chars per second, or how fast they were). Then laser printers became cheap and everyone forgot about the dot-matrix. I guess this is the same thing. Fast forward 15 years.
Probably the next revolution in 3D printing speed will be something that prints an entire object at once, maybe via some sort of science fiction style hologram illuminating a resin. There are technologies that kinda resemble this on the cutting edge (like continuous liquid interface production), but we are still really far away from that ideal. For the foreseeable future we are probably doomed to incremental improvements.
It's probably possible to create a "matter hologram": we already know we can put molecules as large as C60 in quantum superposition. If we could create a coherent source of such molecules and pass them through a beamsplitter and suitable modulator (to shift their phase), combining the two beams, we could create an interference pattern in free space which would reify the object, like the replicator in star-trek.
The nicest thing about our mythical hologram printer is that supports would be unnecessary for stuff like overhangs and you could maybe even print complex internal structure, depending on just how capable your 3D hologram was. One can dream...
There are printers around that don't have any of these issues already - e.g. the resin based machines. So 3D printing is not in any way tied to extruding molten plastic.
"The team identified three factors limiting a printer’s speed: how fast a printer can move its printhead, how much force a printhead can apply to a material to push it through the nozzle, and how quickly the printhead can transfer heat to melt a material and make it flow."
None of these are the actual limiting factor on print speed. Many 3D printers are capable of traveling at 300 mm/s, and extruding at more than 100 mm/s, yet in practice the fastest parts of a print are usually done at 60 - 80 mm/s. This is because the faster you try to go, the worse the print quality is. The outer shells of a print, where quality is most important, are usually done at 30 mm/s.
Quality 3D printing is all about maintaining a delicate thermal balance. You want the plastic to cool to below it's melting point (if it has one) as soon as it leaves the nozzle so that it hardens and maintains it's shape. However you also want to keep it warmer than it's glass transition temperature in order to combat the effects of thermal contraction.
As you try to go faster the soft plastic will get dragged around harder and will have less time to cool. The layer below will be softer when the layer above is being deposited on top of it. This causes sharp corners to get rounded off and overhangs to curl upwards.
The article does not have any close up shots of the objects printed on this machine, but I expect that they are pretty rough. As QAPereo pointed out, the article mentions that they ran into this problem. This does not mean that this is a bad idea, though. Products like E3D Volcano and the Lulzbot Moarstruder have proven that there is a niche in the 3D printing marked for fast sloppy printing.
Another thing that they don't talk about is how this extruder deals with retraction. One nice thing about gear driven extruders is they give you very precise control over how much plastic is being put out. You can also quickly retract the filament a bit to drop the chamber pressure and minimize plastic leaking from the nozzle as it travels from one part of the print to another. I'd like to know how well the auger is able to do this.
I think you bring up a good point about the lack of precision at high speed. I think the next big thing in this industry is going to be diving into the factors behind the 'sloppyness' you mention.
> I'd like to know how well the auger is able to do this.
I can answer this one! it's actually more like a nut on a bolt, instead of an auger, meaning it can reverse just fine.
this is absolutely brilliant. i have no idea how nobody thought of this before. it was so obvious, just melt the plastic faster. so blindingly obvious. i have personally spent many hours trying to design a faster printer. its encouraging to know that there was a very elegant solution right under my nose the whole time. maybe there are a few more?
If it incorporates a gantry design and a laser powerful enough to melt plastic, how hard would it be to make it a multipurpose 3D printer/2D (+depth) laser cutter? Getting two useful tools for the price (and space) of one would make this an easy sell to me, at least.
what are the problems with printers today? for hobbyist extrusion printers?
- they are slow (fixed). but even 10x leaves something to be desired.
- they cant print all shapes elegantly. for many shapes you need scaffolding. its a nuisance.
- they are course. print layers, even at the thinnest settings, are still rather thick. also, there is a lot of inconsistency, stringing and surface imperfections.
find a way to solve these problems in a relatively cheap printer and we might actually see the revolution that was talked about when 3d printing first arrived.
Two others: They are big, and they are expensive. That makes owning a 3D printer something that still doesn't make a lot of economic sense for me. I'm not certain it ever will.
I'm waiting for the equivalent of a Kinko's for 3D printing. That will make 3D printing much more accessible to me, and I assume it would also get me access to 3D printers that can do much higher quality work than the at-home models are capable of.
> I'm waiting for the equivalent of a Kinko's for 3D printing
A lot of universities offer this to their students already. Commercially, off the top of my head, Staples and the UPS store have some 3D printing services available at a limited number of stores but I don't think it has been terribly successful. Problem is, 3D printing is IMO not much more than a novelty unless you are designing your own parts, but then iterating on a custom design is painful and expensive with a commercial service.
I use my 3D printer a ton, both for work and for hobby stuff, and it usually takes a few tries to get a design right. That's hard to do sitting in a store or by mail and typical markup for those services is like 10-20x actual cost. It quickly gets expensive enough that buying your own printer starts to make sense, especially when quality printers can be had for <$500 nowadays and quite serviceable printers for <$200.
There's really not a huge difference in quality between the higher end printers and the good examples of the affordable ones (at least, comparing FDM printers to their kin, SLA/DLP is a different beast entirely). Most of the difference is in convenience and secondary features like dual extrusion or self-leveling.
They're not that expensive. I bought one for £150 earlier this year, just to see what all the fuss was about. It is a fantastic machine and I use it for something new almost every week.
It took quite a lot of assembly, and there is an ongoing maintenance burden (there's nearly always something that isn't quite working as well as it could) but I couldn't imagine living without one.
I'd like to upgrade to a higher-quality machine but I'm putting it off as long as I can tolerate because they just keep getting better and cheaper the longer I wait.
But if you're at all interested, I highly recommend getting a cheap one, learning how to use it, and learning some basic CAD. It's a great hobby all of its own, and is a great addition to almost any other hobbies you might have.
They might end up with a higher resolution process that is only a a few times faster. If they reduce the diameter of the filament by 30%, the layers would take longer to lay down (more passes) and have less heat to shed. But building an object would take twice as long.
If they doubled the resolution their system would still build objects over twice as fast. Twice the resolution in half the time could probably work as a sales pitch.
It looks like they are using either a CoreXY or H-bot based design.
Nice thing about those designs, is that you don't need to accelerate the motors/wiring back and forth. Giving better acceleration rate per unit of motor.
The downside with those systems, is that they tend not to be stiff (due to large amounts of belting), or react to external forces well.
That (H-bot/Core-XY) design would be poor for a convention CNC router, that has to react the forces associated with conventional machining.
But for this application, it doesn't matter. The forces acting on an extruder are basically non-existent, with the exception of (the mass of the extruder * its designed acceleration rate).
For going that fast real time, it might have some trick servos, or a hybrid motor (Search for Technic ClearPath Motors).
If you like that video, then you might like this one.
This kind of 3d printer generally uses stepper motors, not servos.
From what I can see, this printer appears to use stepper motors too. I'd believe it's real-time.
The X axis is held in place vertically by the 2 shiny metal rods you can see, and is dragged back and forth by the black rubber belt, which is controlled by a stepper motor.
> The key to the team’s nimble design lies in the printer’s compact printhead, which incorporates two new, speed-enhancing components: a screw mechanism that feeds polymer material through a nozzle at high force; and a laser, built into the printhead, that rapidly heats and melts the material, enabling it to flow faster through the nozzle.
That's the exactly what is making this new 3D printer 10x faster
> a screw mechanism that feeds polymer material through a nozzle at high force
Practically all existing 3D printers have this.
The laser does seem novel - most use an electrically heated head nozzle to melt the plastic like a hot glue gun - but I remain skeptical when they list screw feed as an innovative ceature.
No, usually they have a rotary feeder, i.e. "pinch-wheel", it seems like what they propose is a screw feeder, likely a screw with the axis roughly in the same direction as the filament, with "higher grip" on the filament, but probably it depends also on the specific filament material:
>"In most desktop 3-D printers, plastic is fed through a nozzle via a “pinch-wheel” mechanism, in which two small wheels within the printhead rotate and push the plastic, or filament, forward. This works well at relatively slow speeds, but if more force were applied to speed up the process, at a certain point the wheels would lose their grip on the material — a “mechanical disadvantage,” as Hart puts it, that limits how fast the printhead can push material through.
Hart and Go chose to do away with the pinchwheel design, replacing it with a screw mechanism that turns within the printhead. The team fed a textured plastic filament onto the screw, and as the screw turned, it gripped onto the filament’s textured surface and was able to feed the filament through the nozzle at higher forces and speeds.
"
Design and manufacture of injection molding screws is complicated but well understood.[1] They're not easy to machine, but that's what CNC mills and four-slide machines are for. It's like making a drill bit, something that's done industrially millions of times a day but is very tough for hobbyists.
I'd wanted to use a laser for heating in a 3D printer, but not to heat the plastic being deposited. The fundamental problem with filament-type 3D printers is that you're welding a hot thing to a cold thing. That never works very well. That's what causes cold solder joints and weak welds in metal. In 3D printing, it's why the bonds between layers are much weaker than the plastic's inherent tensile strength. I wanted to have a small laser (about 10W) aimed ahead of where the filament is being deposited, to heat up the material already placed. That way, both sides are hot as the weld occurs, which should yield a much better weld. This is what the heated build plate and heated build chamber designs are trying to accomplish, but they're not putting the heat the one place it's really needed.
As lasers in the 10W range become more affordable, this looks like a better idea. A few years ago, it was an insanely expensive idea, but now, you can get 10W for about $300.
[1] https://www.plasticsportalasia.net/wa/plasticsAP~zh_CN/funct...