Making a huge 3D Printer!
Posted by PulsedMedia
|
Making a huge 3D Printer! January 05, 2014 09:57PM |
We are about to build a huge 3D Printer - A LOT of variations has gone through my mind - and i'm finally setting into a construct.
After seeing maker slides, i'm pretty sure we'll do the basic structure using them:
X: 1800mm maker slides, travel approximately 1600mm
Y: 1200mm maker slide, sliding on top of X, travel approximately 1000mm
Z: 600mm maker slide, mounted vertically on the Y, travel approximately 500mm
WHY?!
Once properly calibrated i want to put the plate full of stuff, let it print over the weekend (or the week) and then just go pick the finished pieces up, use the whole filament reel on a single print job ;D (One of the issues to be resolved!)
Ofc, eventually i want to experiment with HUGE prints which is the ultimate goal here, think about car side skirts, front grille, headlight mounting assemblies, rear spoilers and other aero stuff, fender flares and such! With polycarbonate we could experiment with fully 3d printed intake manifolds too!
I have not heard of this size of plastic printing, so it would probably be world first? How cool would that be!
Further, eventually when we have printed the mfg batch of stuff we need for our business (we need 150+ mini-itx trays, dozens of fan ducts, psu mounts, and hundreds of odd bits and pieces), we would like to run customer jobs, and we could output a huge amount of them at one go. Get the machine to pay for itself. Not sure is there sufficient market here in Finland tho, but for the large pieces we could probably find quite a bit of use after custom designing some parts, plus i think some businesses will be interested in getting custom parts for their machinery etc. since we can make them at practical sizes required.
ABS is such a good material in automotive use, we could do very complex shapes at near the weight to strength ratio of fiberglass! But because we can do so complex shapes, we can reinforce the part design leading to lighter parts for motorsports!
Also the first touch to 3d printing was Prusa V1, parts gathered in late 2012. This became a VERY expensive machine, which completely sucks. The mechanics are not sound at all, from end stop play to bed play, Z backlash, XY backlash etc.
and this machine was supposed to be as good as a prusa gets at the time. But, it's a learning experience!
All of the issues with Prusa V1 i want to solve in this machine, calibration which stays dead on etc. Minimal backlash, SOLID axes which don't vibrate like hell and give away etc. etc. Then again we truly punished that machine, we have printed at speeds up to around 300mm/s! With accelerations of 9000mm/s etc. The structure just don't seem to be able to handle all that vibration etc. It works always for a while after making everything just so, but that only lasts for mere hours before something goes wrong again, always a different part tho.
Plus i like hacking things together the way which other's don't think to make, and pushing boundaries into the territory where everyone thinks it will not work etc etc etc.
When?
This is a long term project, seeking completion sometime around summer, some of the parts have been acquired already tho.
While 3d printing is important for our business - we need to make stuff now, not months from now, so we will be working on this when we got time from the more important projects (projects which gets the bills paid now)
Plus our budget for this machine is constrained, we are a small business and we've just made huge investments in our business waiting for them to pay off (3d printing is a key here as well!).
Open source?
Kinda. This is a one off machine, but i will not be doing build instructions, bom etc. since being a one off iterative process of trial & error and therefore not keeping count of the smaller things. This is as much a fun research project, as much as a tool for our needs.
I will probably document quite a bit of the process here, especially when getting to do huge prints, the issues and hurdles.
*IF* this is successfull, and it turns out our machine is working 24/7 we might build another few, iterating on the previous ones, at that point we might make a full BOM and full design documentation.
We will likely build one platform as a CNC router tho after this one!
How?
All of that is going to be mounted on a very thick piece of wood, meant originally as a higher end desk for professional use, so it's ~50mm thick and much more massive than say MDF.
Then we'll make legs for it to be standalone and make them very strong as well, and add some weight to it, might as well go and make it a cool looking, happen to got a few ruined engine blocks, pistons, con rods etc. (motorsports, motorsports...) around for the counter weight portion
Heated bed would need to be custom built, i'm still not certain how exactly the heating will be done, a huge loop of nichrome wire or a bunch of resistors. Using MK2B heated beds would be prohibitively expensive since close to 40 would be required.
So maybe a bunch of heater cartridges mounted on thick piece of aluminium or sheet metal as a heat spreader, and bottom insulated with fiberglass. IDEAS welcome!
Bed itself would be glass, and due to the size i'm leaning towards getting one custom cut from 4mm thick piece.
Bed mount would be using studs, one every 400mm or so, the studs itself will be either thin ABS prints for flexibility or pieces of rubber hose, so adjustability is gained.
Steppers i'm leaning towards to 6.5Kg-cm Nema 17s since these are relatively cheap and can use anything to drive them. One option would be to use Nema 23s but these will probably cost more for any gain on torque.
for extruder(s) maybe Nema 14s since these don't need as much power and the weight savings would be a big thing!
We would mount pieces of acrylic all around the backside and half way the top to partially en case the build area to keep the heat in.
All printed parts would be ABS (given i get the old version Prusa to work long enough to make any proper work - which is part of the motivation for this build since it's mechanically ... POS)
However, i've not yet decided on how to drive the Axes: Belts, Rack & Pinion, Leadscrew, Fishing line are still all options.
Also i'm not dead set on the steppers either - this size of build might require some heaftier steppers for sensible travel speed (Gecko, SLAm, Hacking better drivers on RAMPS?)
Also i'm not set on the extruder yet - kinda would like to build this myself, but then again E3D has gained such a praise and my design would in the end to be similar all metal construction, just ghetto style. I'm kinda leaning towards getting a Kraken - but how to drive 4 extruders?
Electronics wise should i stick with RAMPS at all? One of the guys in our team likes to play around with electronics, so in that regard we can even do a complex setup from readily available pieces, say, hack in big stepper drivers on the usual ramps or what not.
Remote monitoring would be via a webcam, and we would have a rasperrybi, probably octoprint etc. so we can remotely check how it's going and stop if it screwed up.
Other ideas i've toyed with:
X axes Ghetto style: Use thick L-bar mild steel, or rectangular bar, since using very thick one they are bound to be somewhat straight, then use very precise laser distance gauge to smooth it out. In the end only thing that matters is extruder level (well X+Y axis) against the bed, so it being bend is not that big of a problem. As long as i don't weld it, it's going to be relatively straight. Then by hand even it out and make smoother
Sleds would contain 6-8 bearings depending on the type of bar i would use, 2 on each side 2-4 on top, and since we can't have it lifting, use cheap couple 2$ a piece neodymium magnets embedded in the sled as close to the rail as possible to keep it positioned, a few kg of force should be enough which can be easily created with a few round magnets running 1-3mm off the rail, behind 1-2mm of ABS plastic (total distance 2-5mm) depending on what design i set into. Using magnets idea came from Berrybot!
BUT makerslides are sooo cheap, this ghetto X axis would cost something in the vicinity of 40€ total + some manual work. Makerslides would cost something like 100€ (inc. shipping to finland) and be pretty much bolt -on and done!
The benefit of ghetto style would be weight and therefore overall rigidity of the build platform. and the cool factor: it would look so steampunk! ;D
Printed rack & pinion: Precision might be an issue here, since gearing tends to come out slightly loose with 3d prints i've done with the prusa v1 and what i've seen due to layer alignment issues. Doing long enough gearing it would probably take the forces tho, and using ABS naturally. Apply some grease etc. This would require several plates worth of racks printed and i would probably want to run double steppers with this. Probably 4 plates of slow precision prints for racks and then figuring a solid precise attachment for the racks, probably slots on bottom side, plastic welding/super epoxy(ie. jb weld) and then mounting all of these on the wood platform. Steppers on the sled would add weight to Y-sled (2-4 steppers) which causes speed issues.
Rack & pinion setup however is very precise and would be low backlash as well, especially if make it herringbone type of gearing
Fishing line drive line: The premium high strength one made by honeywell for which i forgot name. This would not stretch etc, and would be very accurate according to some posts. Difficulty is in making the pulleys. If strenght is not enough one could quite trivially add a long pulley for multiple lines running parallel, which will imply an extra bearing and what nots..
This would be relatively easy to make, relatively cheap, but i don't see this used too often - what are the drawbacks? The fishing line burrowing into printed pulleys?
DCs as steppers, using Gecko? One could then use RC Hobby motors, some really strong ones here in small package! Even with brushed one can achieve 300W+ of power on a motor fitting in palm of your hand. Brushless has very weak starting torque (needs minimal movement for torque, and often it would start by going to wrong direction before correcting itself. Sensored brushless motors are even harder to control and generally weaker?!) and harder to control, but going that direction we could drive thousands of kW through motors fitting on my palm (Been there, done that!). Problem with these is that high RPM is required for high power -> Higher rating the higher RPM usually implied. We are talking RPMs in the range of 50k now! This would require some mighty pulleys!
One could easily use larger ones, with shaft of 10mm etc. but anything with sensible price on the larger size will be very expensive for the same power output. A 15-20€ Chinese brushless RC car motor can pull of 5-7kW for limited time periods!
Of course, we don't need that kind of powers, perhaps 100W or so? Better to have a big margin!
All of this would probably also necessitate closed loop system.
Closed loop system: This would be a nice to have thing with something this large, but this is quite experimental right?
Besides, doing closed loop system is not a fix for weak construct! Let's build such a strong construct missed steps won't be an issue! ;D
XY positioning accuracy: This is OBVIOUSLY a huge issue, if the head is first doing at X position 1670 something then moves to 75mm, you want it to be just right, and not almost.
Besides even the small printers have a lot of positional accuracy play with belts and what not, so a GT2 belt will probably be very weak here even if i would be having a serpentive system with multiple tensioners along the line, and mount the belts against pulleys with 250degree+ contact surface area. (Y-shaped tensioner + stepper assembly)
15mm wide belts are available- but round teeths etc.
We want to get the accuracy to level of "good enough", that means if we make a 3" OD piping, it needs to be within 5%. Most important thing is repeatibility: Even if the accuracy turns out to be rather weak, it needs to make the same part with same mistakes day in and day out without recalibrations, and minimal maintenance. The maintenance needs to be really the occasional spray of oil into bearings and occasionally clean up of the extruder.
If it winds up having positional accuracy of only 0.25mm or so, for many of the parts we do, that's just fine as long as it's repeatable! I would prefer a position accuracy of around 0.05mm, Z we can make as accurate as we can
Filament lasting long enough:
Obviously going to be huge issue when we want to print large parts with final weigh of multiple kilos!
5lbs spools are the largest i've seen, so i've been toying with an idea, when everything else is done and working, setup a Filastruder, and on top of which put a huge bucket/pallet feeding it, fill that bucket with 5kg of pellets, and let it make filament so that it just gets to cool before being fed into the printer
Ofc, have a blender handy to recycle those failed prints and make fun stuff like rainbow colored prints
Question is that going to work reliably enough and fast enough. This is somethign i don't want to make ourselves because of the amount of trial and error involved in making a filament maker.
If it's not fast enough, maybe have two of those, and between prints change from which one the filament is coming from so you have a back buffer of few kilos to use.
At the end of day, one long term possibility is to make our own filament extruder for 5-7mm size filament for faster making of it. It starts to make sense when you are using kilos of the stuff at a go, and thus hundreds of meters of filament being fed, and longer the filament is, more possibilities for tangling.
Also, we could ask a chinese mfg to make us 10kg spools or something crazy like that.
But at the end of day we need to be able to make our own too, to recycle the failed prints - which there will be plenty of from experimenting with extra large prints!
There will also be a bunch of mostly used spools of filament which might need to be ground up and made part of longer piece of filament.
For some parts the color etc. simply doesn't matter at all, ie. car body parts will always be sanded and painted before mounting on the car, so recycled filament of differing colors do not matter here.
Plus for those we need to make our own filament to introduce fiberglass, carbon fiber and kevlar strands to it for added strength (and finding a reliable way to extrude that filament ... Maybe with a 1mm nozzle?)
So many possibilities, so little time! So got to set into a design and would like the input of community on what actually works and what does not.
After seeing maker slides, i'm pretty sure we'll do the basic structure using them:
X: 1800mm maker slides, travel approximately 1600mm
Y: 1200mm maker slide, sliding on top of X, travel approximately 1000mm
Z: 600mm maker slide, mounted vertically on the Y, travel approximately 500mm
WHY?!
Once properly calibrated i want to put the plate full of stuff, let it print over the weekend (or the week) and then just go pick the finished pieces up, use the whole filament reel on a single print job ;D (One of the issues to be resolved!)
Ofc, eventually i want to experiment with HUGE prints which is the ultimate goal here, think about car side skirts, front grille, headlight mounting assemblies, rear spoilers and other aero stuff, fender flares and such! With polycarbonate we could experiment with fully 3d printed intake manifolds too!
I have not heard of this size of plastic printing, so it would probably be world first? How cool would that be!
Further, eventually when we have printed the mfg batch of stuff we need for our business (we need 150+ mini-itx trays, dozens of fan ducts, psu mounts, and hundreds of odd bits and pieces), we would like to run customer jobs, and we could output a huge amount of them at one go. Get the machine to pay for itself. Not sure is there sufficient market here in Finland tho, but for the large pieces we could probably find quite a bit of use after custom designing some parts, plus i think some businesses will be interested in getting custom parts for their machinery etc. since we can make them at practical sizes required.
ABS is such a good material in automotive use, we could do very complex shapes at near the weight to strength ratio of fiberglass! But because we can do so complex shapes, we can reinforce the part design leading to lighter parts for motorsports!
Also the first touch to 3d printing was Prusa V1, parts gathered in late 2012. This became a VERY expensive machine, which completely sucks. The mechanics are not sound at all, from end stop play to bed play, Z backlash, XY backlash etc.
and this machine was supposed to be as good as a prusa gets at the time. But, it's a learning experience!
All of the issues with Prusa V1 i want to solve in this machine, calibration which stays dead on etc. Minimal backlash, SOLID axes which don't vibrate like hell and give away etc. etc. Then again we truly punished that machine, we have printed at speeds up to around 300mm/s! With accelerations of 9000mm/s etc. The structure just don't seem to be able to handle all that vibration etc. It works always for a while after making everything just so, but that only lasts for mere hours before something goes wrong again, always a different part tho.
Plus i like hacking things together the way which other's don't think to make, and pushing boundaries into the territory where everyone thinks it will not work etc etc etc.
When?
This is a long term project, seeking completion sometime around summer, some of the parts have been acquired already tho.
While 3d printing is important for our business - we need to make stuff now, not months from now, so we will be working on this when we got time from the more important projects (projects which gets the bills paid now)
Plus our budget for this machine is constrained, we are a small business and we've just made huge investments in our business waiting for them to pay off (3d printing is a key here as well!).
Open source?
Kinda. This is a one off machine, but i will not be doing build instructions, bom etc. since being a one off iterative process of trial & error and therefore not keeping count of the smaller things. This is as much a fun research project, as much as a tool for our needs.
I will probably document quite a bit of the process here, especially when getting to do huge prints, the issues and hurdles.
*IF* this is successfull, and it turns out our machine is working 24/7 we might build another few, iterating on the previous ones, at that point we might make a full BOM and full design documentation.
We will likely build one platform as a CNC router tho after this one!
How?
All of that is going to be mounted on a very thick piece of wood, meant originally as a higher end desk for professional use, so it's ~50mm thick and much more massive than say MDF.
Then we'll make legs for it to be standalone and make them very strong as well, and add some weight to it, might as well go and make it a cool looking, happen to got a few ruined engine blocks, pistons, con rods etc. (motorsports, motorsports...) around for the counter weight portion
Heated bed would need to be custom built, i'm still not certain how exactly the heating will be done, a huge loop of nichrome wire or a bunch of resistors. Using MK2B heated beds would be prohibitively expensive since close to 40 would be required.
So maybe a bunch of heater cartridges mounted on thick piece of aluminium or sheet metal as a heat spreader, and bottom insulated with fiberglass. IDEAS welcome!
Bed itself would be glass, and due to the size i'm leaning towards getting one custom cut from 4mm thick piece.
Bed mount would be using studs, one every 400mm or so, the studs itself will be either thin ABS prints for flexibility or pieces of rubber hose, so adjustability is gained.
Steppers i'm leaning towards to 6.5Kg-cm Nema 17s since these are relatively cheap and can use anything to drive them. One option would be to use Nema 23s but these will probably cost more for any gain on torque.
for extruder(s) maybe Nema 14s since these don't need as much power and the weight savings would be a big thing!
We would mount pieces of acrylic all around the backside and half way the top to partially en case the build area to keep the heat in.
All printed parts would be ABS (given i get the old version Prusa to work long enough to make any proper work - which is part of the motivation for this build since it's mechanically ... POS)
However, i've not yet decided on how to drive the Axes: Belts, Rack & Pinion, Leadscrew, Fishing line are still all options.
Also i'm not dead set on the steppers either - this size of build might require some heaftier steppers for sensible travel speed (Gecko, SLAm, Hacking better drivers on RAMPS?)
Also i'm not set on the extruder yet - kinda would like to build this myself, but then again E3D has gained such a praise and my design would in the end to be similar all metal construction, just ghetto style. I'm kinda leaning towards getting a Kraken - but how to drive 4 extruders?
Electronics wise should i stick with RAMPS at all? One of the guys in our team likes to play around with electronics, so in that regard we can even do a complex setup from readily available pieces, say, hack in big stepper drivers on the usual ramps or what not.
Remote monitoring would be via a webcam, and we would have a rasperrybi, probably octoprint etc. so we can remotely check how it's going and stop if it screwed up.
Other ideas i've toyed with:
X axes Ghetto style: Use thick L-bar mild steel, or rectangular bar, since using very thick one they are bound to be somewhat straight, then use very precise laser distance gauge to smooth it out. In the end only thing that matters is extruder level (well X+Y axis) against the bed, so it being bend is not that big of a problem. As long as i don't weld it, it's going to be relatively straight. Then by hand even it out and make smoother
Sleds would contain 6-8 bearings depending on the type of bar i would use, 2 on each side 2-4 on top, and since we can't have it lifting, use cheap couple 2$ a piece neodymium magnets embedded in the sled as close to the rail as possible to keep it positioned, a few kg of force should be enough which can be easily created with a few round magnets running 1-3mm off the rail, behind 1-2mm of ABS plastic (total distance 2-5mm) depending on what design i set into. Using magnets idea came from Berrybot!
BUT makerslides are sooo cheap, this ghetto X axis would cost something in the vicinity of 40€ total + some manual work. Makerslides would cost something like 100€ (inc. shipping to finland) and be pretty much bolt -on and done!
The benefit of ghetto style would be weight and therefore overall rigidity of the build platform. and the cool factor: it would look so steampunk! ;D
Printed rack & pinion: Precision might be an issue here, since gearing tends to come out slightly loose with 3d prints i've done with the prusa v1 and what i've seen due to layer alignment issues. Doing long enough gearing it would probably take the forces tho, and using ABS naturally. Apply some grease etc. This would require several plates worth of racks printed and i would probably want to run double steppers with this. Probably 4 plates of slow precision prints for racks and then figuring a solid precise attachment for the racks, probably slots on bottom side, plastic welding/super epoxy(ie. jb weld) and then mounting all of these on the wood platform. Steppers on the sled would add weight to Y-sled (2-4 steppers) which causes speed issues.
Rack & pinion setup however is very precise and would be low backlash as well, especially if make it herringbone type of gearing
Fishing line drive line: The premium high strength one made by honeywell for which i forgot name. This would not stretch etc, and would be very accurate according to some posts. Difficulty is in making the pulleys. If strenght is not enough one could quite trivially add a long pulley for multiple lines running parallel, which will imply an extra bearing and what nots..
This would be relatively easy to make, relatively cheap, but i don't see this used too often - what are the drawbacks? The fishing line burrowing into printed pulleys?
DCs as steppers, using Gecko? One could then use RC Hobby motors, some really strong ones here in small package! Even with brushed one can achieve 300W+ of power on a motor fitting in palm of your hand. Brushless has very weak starting torque (needs minimal movement for torque, and often it would start by going to wrong direction before correcting itself. Sensored brushless motors are even harder to control and generally weaker?!) and harder to control, but going that direction we could drive thousands of kW through motors fitting on my palm (Been there, done that!). Problem with these is that high RPM is required for high power -> Higher rating the higher RPM usually implied. We are talking RPMs in the range of 50k now! This would require some mighty pulleys!
One could easily use larger ones, with shaft of 10mm etc. but anything with sensible price on the larger size will be very expensive for the same power output. A 15-20€ Chinese brushless RC car motor can pull of 5-7kW for limited time periods!
Of course, we don't need that kind of powers, perhaps 100W or so? Better to have a big margin!
All of this would probably also necessitate closed loop system.
Closed loop system: This would be a nice to have thing with something this large, but this is quite experimental right?
Besides, doing closed loop system is not a fix for weak construct! Let's build such a strong construct missed steps won't be an issue! ;D
XY positioning accuracy: This is OBVIOUSLY a huge issue, if the head is first doing at X position 1670 something then moves to 75mm, you want it to be just right, and not almost.
Besides even the small printers have a lot of positional accuracy play with belts and what not, so a GT2 belt will probably be very weak here even if i would be having a serpentive system with multiple tensioners along the line, and mount the belts against pulleys with 250degree+ contact surface area. (Y-shaped tensioner + stepper assembly)
15mm wide belts are available- but round teeths etc.
We want to get the accuracy to level of "good enough", that means if we make a 3" OD piping, it needs to be within 5%. Most important thing is repeatibility: Even if the accuracy turns out to be rather weak, it needs to make the same part with same mistakes day in and day out without recalibrations, and minimal maintenance. The maintenance needs to be really the occasional spray of oil into bearings and occasionally clean up of the extruder.
If it winds up having positional accuracy of only 0.25mm or so, for many of the parts we do, that's just fine as long as it's repeatable! I would prefer a position accuracy of around 0.05mm, Z we can make as accurate as we can
Filament lasting long enough:
Obviously going to be huge issue when we want to print large parts with final weigh of multiple kilos!
5lbs spools are the largest i've seen, so i've been toying with an idea, when everything else is done and working, setup a Filastruder, and on top of which put a huge bucket/pallet feeding it, fill that bucket with 5kg of pellets, and let it make filament so that it just gets to cool before being fed into the printer
Ofc, have a blender handy to recycle those failed prints
and make fun stuff like rainbow colored prints Question is that going to work reliably enough and fast enough. This is somethign i don't want to make ourselves because of the amount of trial and error involved in making a filament maker.
If it's not fast enough, maybe have two of those, and between prints change from which one the filament is coming from so you have a back buffer of few kilos to use.
At the end of day, one long term possibility is to make our own filament extruder for 5-7mm size filament for faster making of it. It starts to make sense when you are using kilos of the stuff at a go, and thus hundreds of meters of filament being fed, and longer the filament is, more possibilities for tangling.
Also, we could ask a chinese mfg to make us 10kg spools or something crazy like that.
But at the end of day we need to be able to make our own too, to recycle the failed prints - which there will be plenty of from experimenting with extra large prints!
There will also be a bunch of mostly used spools of filament which might need to be ground up and made part of longer piece of filament.
For some parts the color etc. simply doesn't matter at all, ie. car body parts will always be sanded and painted before mounting on the car, so recycled filament of differing colors do not matter here.
Plus for those we need to make our own filament to introduce fiberglass, carbon fiber and kevlar strands to it for added strength (and finding a reliable way to extrude that filament ... Maybe with a 1mm nozzle?)
So many possibilities, so little time! So got to set into a design and would like the input of community on what actually works and what does not.
|
Re: Making a huge 3D Printer! January 05, 2014 10:24PM |
Parts acquired are:
Wood Platform
608ZZ ABEC 7 high quality bearings big bunch (50 pack)
608ZZ ABEC 9 High quality *actual* skate board bearings for the most precise bits (40 pack)
If going with GT2 belts those already got
If going with RAMPS, those already got with standard step sticks. RAMPS 1.4, Arduino Mega 2560
RasperryBi
Power supply (750W)
2F Capacitors for stabilizing power supply if need be
Relays for controlling the heated bed power outside of RAMPS
Nema17s for extruder(s) if not going with Nema14
Micro drill bits to make our own extruder nozzles from 0.15mm to 1mm
Bunch of 40W heater cartridges (10x)
Fans for extruder cooling
Magnets if that is the design we will go for (even with maker slides we could utilize magnets for stabilization, extruder tightening etc.!), Neodymium discs 10mm x 3mm
3mm thick pieces of aluminium to use as heat spreader.
All the M8 threaded rod we can imagine to use (Uhm, that would be none if i have something to say about it!)
Fiberglass insulation, aluminium tape etc. for the heat insulation bits
So basicly, now just have to decide how the linear motion is done, which extruder(s). I'd like to have the Kraken, but i'm wary of the sensibility of the water cooling bit (wasted energy and all that) and how to drive it. Also i'm not sure should i just do my own extruders.
My basic idea was to take 3mm ID stainless steel pipe, with 5-6mm OD, do outside threading, then i can just put nuts in place for the mounting bit, for the cooling portion use washer - fender washer combos, for the heat chamber weld a extension nut of larger diameter on the right size nut, well the ending shut, grind it to a pointy edge, drill on side hole for heater cartridges (might need to weld in nuts of just below heater cartridge OD on the side and drill them out) on each side for more even heating (and maybe drop the voltage a bit of them), and the tiny hole for thermistor.
On the pointy edge, just grind it enough to make just enough flat surface to tap in starting marking for the drill bit and drill the desired nozzle size into it.
Between heat sink portion and heated chamber grind the piping down a bit to make it thinner to slow the heat transfer process, and put some fiberglass heat insulation in that gap.
Tighten it all with counter nuts, or maybe even spot weld the heatsink portion tight, and the counter nut for the nozzle size.
Quite simple design infact, needs a bit of tinkering while needing, but very simple construct for someone like me who has the tools and odds and bits at hand.
Heated chamber i have an alternative design in mind for very quick nozzle swaps without taking the heat chamber away.
Cost of making hot end like that less than 10$, and for the first one something like 2hrs, further iterations and versions 10-15mins each, so in just 4 hrs i could make a dozen of these.
Stainless steel pipe with 3mm ID is most likely going to be rather smooth and even, if that is a concern could probably somehow polish it with a dremel and custom made bit but washing off that wax is going to probably be a slight pita.
I could make a bunch of these if people want to experiment with a ghetto style hot end!
Wood Platform
608ZZ ABEC 7 high quality bearings big bunch (50 pack)
608ZZ ABEC 9 High quality *actual* skate board bearings for the most precise bits (40 pack)
If going with GT2 belts those already got
If going with RAMPS, those already got with standard step sticks. RAMPS 1.4, Arduino Mega 2560
RasperryBi
Power supply (750W)
2F Capacitors for stabilizing power supply if need be
Relays for controlling the heated bed power outside of RAMPS
Nema17s for extruder(s) if not going with Nema14
Micro drill bits to make our own extruder nozzles from 0.15mm to 1mm
Bunch of 40W heater cartridges (10x)
Fans for extruder cooling
Magnets if that is the design we will go for (even with maker slides we could utilize magnets for stabilization, extruder tightening etc.!), Neodymium discs 10mm x 3mm
3mm thick pieces of aluminium to use as heat spreader.
All the M8 threaded rod we can imagine to use (Uhm, that would be none if i have something to say about it!)
Fiberglass insulation, aluminium tape etc. for the heat insulation bits
So basicly, now just have to decide how the linear motion is done, which extruder(s). I'd like to have the Kraken, but i'm wary of the sensibility of the water cooling bit (wasted energy and all that) and how to drive it. Also i'm not sure should i just do my own extruders.
My basic idea was to take 3mm ID stainless steel pipe, with 5-6mm OD, do outside threading, then i can just put nuts in place for the mounting bit, for the cooling portion use washer - fender washer combos, for the heat chamber weld a extension nut of larger diameter on the right size nut, well the ending shut, grind it to a pointy edge, drill on side hole for heater cartridges (might need to weld in nuts of just below heater cartridge OD on the side and drill them out) on each side for more even heating (and maybe drop the voltage a bit of them), and the tiny hole for thermistor.
On the pointy edge, just grind it enough to make just enough flat surface to tap in starting marking for the drill bit and drill the desired nozzle size into it.
Between heat sink portion and heated chamber grind the piping down a bit to make it thinner to slow the heat transfer process, and put some fiberglass heat insulation in that gap.
Tighten it all with counter nuts, or maybe even spot weld the heatsink portion tight, and the counter nut for the nozzle size.
Quite simple design infact, needs a bit of tinkering while needing, but very simple construct for someone like me who has the tools and odds and bits at hand.
Heated chamber i have an alternative design in mind for very quick nozzle swaps without taking the heat chamber away.
Cost of making hot end like that less than 10$, and for the first one something like 2hrs, further iterations and versions 10-15mins each, so in just 4 hrs i could make a dozen of these.
Stainless steel pipe with 3mm ID is most likely going to be rather smooth and even, if that is a concern could probably somehow polish it with a dremel and custom made bit but washing off that wax is going to probably be a slight pita.
I could make a bunch of these if people want to experiment with a ghetto style hot end!
|
Re: Making a huge 3D Printer! January 11, 2014 03:54AM |
Some progress!
Spent a while thinking how do i mount the Y makerslides so that i can fine tune it, then i recalled how panhard rod chassis mount is being adjusted in Nascar Simple really. Just need to get fine pitch M8 bolts for each corner to do the final tuning - the question remains: is it fine enough or will it be too coarse. Ultimately i would like to make adjustment as precise as my measuring tools can do.
Openscad isn't too good in making visual presentations of this type
The y mounting is designed to be made from 3mm thick flat iron bars or L-profile, by welding. It will bend somewhat when making those welds but it doesn't matter because i can just use the adjustment to make it level.
I didn't worry about making the design too exact - at the end of the day the details will depend on what kind of flat iron i do happen to have at hand, it's just important to get down the basic design.
Next up is designing how the X axle is moved on top of the Y axle, leaning towards basic sleds and magnetic tightening on the wheels.
Also you can see basic view of the spool holder.
Spent a while thinking how do i mount the Y makerslides so that i can fine tune it, then i recalled how panhard rod chassis mount is being adjusted in Nascar
Simple really. Just need to get fine pitch M8 bolts for each corner to do the final tuning - the question remains: is it fine enough or will it be too coarse. Ultimately i would like to make adjustment as precise as my measuring tools can do.Openscad isn't too good in making visual presentations of this type
The y mounting is designed to be made from 3mm thick flat iron bars or L-profile, by welding. It will bend somewhat when making those welds but it doesn't matter because i can just use the adjustment to make it level.
I didn't worry about making the design too exact - at the end of the day the details will depend on what kind of flat iron i do happen to have at hand, it's just important to get down the basic design.
Next up is designing how the X axle is moved on top of the Y axle, leaning towards basic sleds and magnetic tightening on the wheels.
Also you can see basic view of the spool holder.
|
Re: Making a huge 3D Printer! January 11, 2014 08:12PM |
I have no idea if anybody has tried this. But here's an idea: use square box stainless steel or aluminium extrusions and fill them with cement for rigidity.
Bearings with delrin or machined teflon rollers could be used as guides if you use soft extrusion like aluminium.
Heck, if you make it really big on the order of meters you could even use concrete reinforcing steel! That way it'd holds its strength in case the cement inside cracks.
Something like this: [secure.flickr.com]
Some pictures illustrating what I had in mind: [forums.reprap.org]
As for a heated bed. Lots and lots of resistors on the bottom of a sheet of metal could work. If you go this route make sure to use plenty so the heat is spread evenly and use a material that conducts heat well. I wouldn't use glass, as it gets to hot where the resistors are and cracks ( mine did
).
Making an enclosure and using an oven heating element with fans to circulate the hot air might be an idea. But just the hot bed in an enclosure might be enough heat. Just try with only a heatbed and take it from there I'd say..
Good luck!!
Bearings with delrin or machined teflon rollers could be used as guides if you use soft extrusion like aluminium.
Heck, if you make it really big on the order of meters you could even use concrete reinforcing steel! That way it'd holds its strength in case the cement inside cracks.
Something like this: [secure.flickr.com]
Some pictures illustrating what I had in mind: [forums.reprap.org]
As for a heated bed. Lots and lots of resistors on the bottom of a sheet of metal could work. If you go this route make sure to use plenty so the heat is spread evenly and use a material that conducts heat well. I wouldn't use glass, as it gets to hot where the resistors are and cracks ( mine did
).Making an enclosure and using an oven heating element with fans to circulate the hot air might be an idea. But just the hot bed in an enclosure might be enough heat. Just try with only a heatbed and take it from there I'd say..
Good luck!!
|
Re: Making a huge 3D Printer! January 19, 2014 03:10AM |
Quote
cnc dick
It's nice to see somebody else attempt this With very long running times reliability is more important than speed take for example if the printed parts take one hundred hours and it fails at 90 hours just lost 90 hours. Versus if it was a little slower will say 150 hours and print was successful obviously you'd be much better off. To achieve that kind of reliability you are going to have to use all industrial quality machine parts that are meant to be reliable for millions of cycles in production which makes it is expensive. Supported linear rails not the round one's ballscrews and machine operating system from a CNC machine such as Mach 3. You're probably going to have to design your own hot end and extruder very important to be reliable
This is an important question as well, that's why we are going to start small and keep going larger and larger until it works out
Reproducibility is the most important aspect for me, it needs to work with the same reliability, good or bad, for every single print there, not a single part must loose it's calibration by itself.
Speed is two edged sword: WIth high speed, say that 100hr print taking 40hrs, you'd be loosing 36hrs of print time if it fails, but then again.... This is such a large that some speed needs to be built in, therefore, i'm still gauging options on the steppers. The strongest Nema 17s probably is not enough, and there really isn't that many options for Nema 23. Gecko, SLAm are options to research for.
I'm not going to be using industrial quality parts, ie the most expensive ones, rather, i'm trying to find design solutions where the tolerances don't matter as much, and they are accounted for. I have no need for a machine which costs 10k + , and can only produce a single 200€ part per week. I need to produce in volumes, which is part of the reason i'm building this, with this, i can print the parts we need in our datacenter in mass quantities at one go, and complete parts instead of requiring assembly afterwards. After that demand has been fullfilled we will move on to printing car grille's, side skirts, fender flares etc. to keep it operating 24/7.
The makerslides will have in between supports tho, these are yet to be designed as i'm fluctuating between multiple deisgn options. Further, this will be an iterative process, will design it as good as i can with the parts available at sensible prices, and then upgrade one bit at a time once weak spots have been identified.
Quote
bobc
I like the idea of Makerslide, I have an Ordbot which uses it. The makerslide is quite stiff and the trucks roll really easily.
However, I bought a 1 meter length of makerslide for a custom machine, but over 1m the makerslide has a distinct bend. Not sure whether this is a one off fault, or a general feature of extrusions, but it's not really good enough for precision work.
On this I would agree with dick, if you are going "big" the robustness and precision of CNC equipment is desirable, if expensive. I guess the alternative is a somewhat flexible frame which requires a lot of adjustment to get it true.
The biggest printer I could find in this sort of class has a 600x600x600 print volume, although people have discussed building big printers before, I'm not sure much came of them.
Hmm, according to the data sheet they should be straight! If they are not -> ask your money back
For our solution, we can have midway supports which are heavy duty steel and built-in adjustment, to force them true, if it's just some millimeters, hoping the aluminium doesn't crack in the process (or maybe heating it all upto 500c first might help)
that is my original plan, not to care that much about the trueness of the parts rather than the ability to force them true
Further, my plan has always been to make this very stiff heavy duty machine, not a flimsy prusa mendel.
Some people have made huge printers, i think in Denmark some architects made a huge printer to be able to print out whole rooms at a go, some people are working on printing with cement big scale etc.
The biggest hurdle will be on how to make the parts stick! If we are printing a 1meter+ long object, the warp forces will be immense if there isn't built-in stress reliefs in the object itself. OR the opposite might happen, the runs are so long that the other end cools off completely during each layer, and there's no build up of warp forces.
Quote
TheTechnicalNoob
I have no idea if anybody has tried this. But here's an idea: use square box stainless steel or aluminium extrusions and fill them with cement for rigidity.
Bearings with delrin or machined teflon rollers could be used as guides if you use soft extrusion like aluminium.
Heck, if you make it really big on the order of meters you could even use concrete reinforcing steel! That way it'd holds its strength in case the cement inside cracks.
Something like this: [secure.flickr.com]
Some pictures illustrating what I had in mind: [forums.reprap.org]
As for a heated bed. Lots and lots of resistors on the bottom of a sheet of metal could work. If you go this route make sure to use plenty so the heat is spread evenly and use a material that conducts heat well. I wouldn't use glass, as it gets to hot where the resistors are and cracks ( mine did
Making an enclosure and using an oven heating element with fans to circulate the hot air might be an idea. But just the hot bed in an enclosure might be enough heat. Just try with only a heatbed and take it from there I'd say..
Good luck!!
I'm opposed to the idea of using counter weighs - i think the rigidity should come from the structure, not by adding extra weight to it. THO, weight is excellent on eliminating vibrations, so i've been considering using bitume on the underside of the main makerslides. That's made for vibration elimination
But only enough the stuff, and sutff made for vibration elimination (bitume is used in cars for that). Concrete is stiff and brittle, might not be so good in eliminating vibration.
Bare sand might be better.
Delrin/Acetal/POM: I've already ordered a roll of this filament, so i might just print the rollers
Heating the bed:
I was thinking using several strong resistors embedded into heatsinks and fans, fully enclosed bottom area, and the fans would distribute the heat, more heatsinks attached to bottom of the heat spreader aluminium sheets.
On top all of that would be a 4mm glass.
The fans would distribute the heat quite efficiently i believe, i would have them circulate the underside air. Hot spots will still form, but one can control them to a degree with the heat spreader heat sink placements.
We will probably need something along the lines of 500W just to keep it all warm, 1kW to heat it all up in a sensible time frame. Heated chamber will ofc lower the requirement, but we need to calculate for the worst case scenarios.
|
Re: Making a huge 3D Printer! January 19, 2014 03:14AM |
Quote
garyhlucas
The stiffness of a long narrow part comes from the material on the outside. Steel has a tensile strength of 50,000 psi or greater. Concrete 5,000 psi in compression, a couple of hundred psi at most in tension. So your frame will be much heavier but not stiffer. Doubling the circumference of a part about quadruples it's stiffness, and only doubles its weight. Making a part twice as long reduces it's stiffness by about 4 times. Hence making something twice as long requires making the circumference about double to handle the same load at the same deflection.
Interesting!
I don't know much about maths regarding to strengths, but i do know some. Didn't know about those ratios!
Mild steel is fortunately very cheap when bought from the right place, and you don't care that much about the exact dimensions and quality
Makes building frames rather cheap when you can buy steel at 0.5€/kg, at what ever shape, strength and lengths happens to be in those piles Very often there is even 6meter long pieces, and all sorts of choices.
|
Re: Making a huge 3D Printer! January 19, 2014 03:24AM |
Quote
cnc dick
Yes pretty much so I don't think 20 x 20 mm is strong enough for even a small machine. And other things that are going to be mounted to the structure are also going to get heavier. For anything horizontal I think you should use rectangle with the largest side taking gravity load
This makes me think that the narrow beam, which i incorrectly named in the designs X axis (i intend to have the long axis X because it's easier to handle wide machine than long one, but at the end of the day it doesn't matter how they are named), which mounts hotend etc. I should perhaps do a triangle support for it as well.
In that case, M10 or M12 threaded rod is probably sufficient strength, 1M piece from either side will make quite a bit of angle, to support the middle position of the axis.
Weight is becoming an issue on this long axis tho, will need some strong steppers to be able to drive it travel speeds of 2000mm/s, which requires quite a strong drive system as well.
I haven't still decided on the drive system i want to use
Rack & Pinion would be great, but would necessitate yet more steppers on the sled.Lead screws are VERY expensive. Normal threaded rods don't have the accuracy and suffers from backlash. Any thread or screw method will have slow speeds.
Perhaps, one of the methods i saw on thingiverse, which uses 4x608 to make a smooth rod as an lead screw, forgot the name of the system
I could scale that up to use say 20x2mm stainless pipe as the lead, and there would be no backlash issue, only getting sufficient grip & pressure for the bearings and angling them right.
|
Re: Making a huge 3D Printer! January 20, 2014 03:56PM |
The relationship between added size and maintenance of accuracy seems to be exponential in terms of cost. This is why you do not see a plethora of school bus sized printers. Not that it can't be done, but it gets way, way more complicated and expensive to go big. Do not assume anything is ever straight, or square. Getting it that way is difficult. Keeping it that way, once you heat everything up is more difficult. At issue with larger printers is first desired output. What are you making? Decorative arts require far less precision than things like car parts, robot parts etc. The more distance you have to travel in each direction, the more deviation you will encounter. This is, after stiffness, the key problem. You will not solve this problem on a big printer without a great deal of resources. I would state that once you get over a cubic meter, you are at the price point of a big commercial machine, even a few cubic feet, like the machine I am building requires a major investment in terms of producing all of the non standard parts. I am building everything from scratch, and attempting to pre solve as many potential problems as possible. From a time standpoint, it is in a sense like going to the moon. The machine you are describing could take over a month to do a full size print. So it is the cost (in time) of failure that dictates the requirements. Sure nobody dies, but if you are say 3 weeks into a print and everything goes south...Now, I would encourage anybody who feels they can to push the size envelope, but with the following caveat...The process of making a complex thing that works and is reliable is an unforgiving one, one which seeks what flaws it can find, one which will make you earn every success. I have made things my entire life. At times it is a war. A war you can only win by overcoming every obstacle and challenge that which you are building throws at you. Not to try is to fail, but trying is just the first step to success. The road to glory is oft paved with time, money, and band-aids....
|
Re: Making a huge 3D Printer! January 21, 2014 12:14PM |
I'm thinking a CoreXY type setup for the X and Y axis drive to save weight on the sled.
The sled itself will not be very heavy, not sure how much makerslide weighs in at but will know then.
I'm saying NO to inch wide belts, right now i'm leaning towards finding the strongest fishing line at sensible price, then weaving it together to form a even stronger one for the XY. Some fishing lines are very durable - forgot the name, but the final research into that is to be done later on.
As for the stiffness, i will be using heavy gauge steel... LOTS of it, and incorporate adjustment pieces to force the makerslides straight enough, i've learned that aluminium extrusion rarely really are straight even tho claimed to be. At the end of the day, only straightness that matters is bed level vs. carriage at every position, so the ultimate fine tuning is probably easier to do on bed - which needs to be done at temperature. yikes, i can see couple very sweaty days leveling that off!
I still need to figure out the proper precision measurement tools as well, something which can measure distance to the thousandth on these scales - so somekind of good laser based tool is required.
I have also considered software solutions: Bed autoleveling and closed loop system, crude closed loop (every Nth layer home X and Y) etc. Needs more research into that.
If i make the whole sled with Z axis lightweight enough, i doubt the cost creeps insanely high, then i can probably get away with strong Nema 23s. Need to calculate some forces! Droop is going to be an issue however, and the fact that makerslide will not be straight enough. Droop cannot be fixed with leveling from the ends, might need an software solution there to stay in cost effective range.
If the sled stays below let's say 3kg, using CoreXY i have 2 motors strength to move it to begin with. Wonder if i can wire into single channel multiple stepper drivers to just put more cheap steppers, ie. one in each corner while driving single belt/fishing line.
What i'd REALLY like is to come with a way to put the power remotely to run the extruder and Z axis. Cables will probably flex too much for accuracy, hydraulics get messy really fast, pneumatic might be the solution, but i have no clue how to make a stepper using pneumatics - i'm sure someone has come up with a solution to do that, but i've got no idea how to do that. Fortunately, i've got someone to ask about that and if possible, help with the design and choosing parts.
There won't be much weight remainaining on the axis if that can be solved!
Makerslide weighs in at 0.98kg/metre, something like 1½ metres of that is required -> ~1.5kg on the sled, and if extruder + Z can be operated with remote power, that still leaves 1.5kg for all the mounts, extruder, hotend etc. Mounts will probably be 3d printed, if ABS is not strong enough then from POM or PC.
Ultimately, this machine is not a must-have, so i'm working on this as a hobby, while setting up many small printers to do our production runs initially with, still targeting summer for building this!
Cost wise i'd like to stay below 2000€ total. Huge chunk of that can be directed towards motors
My belief is that very precise calibration will be easier as well because of the long distances, if steps are off 0.1% it's easier to measure it across 150cm than it is across 20cm
The main hurdle will be leveling everything and making sure it's square etc.
The sled itself will not be very heavy, not sure how much makerslide weighs in at but will know then.
I'm saying NO to inch wide belts, right now i'm leaning towards finding the strongest fishing line at sensible price, then weaving it together to form a even stronger one for the XY. Some fishing lines are very durable - forgot the name, but the final research into that is to be done later on.
As for the stiffness, i will be using heavy gauge steel... LOTS of it, and incorporate adjustment pieces to force the makerslides straight enough, i've learned that aluminium extrusion rarely really are straight even tho claimed to be. At the end of the day, only straightness that matters is bed level vs. carriage at every position, so the ultimate fine tuning is probably easier to do on bed - which needs to be done at temperature. yikes, i can see couple very sweaty days leveling that off!
I still need to figure out the proper precision measurement tools as well, something which can measure distance to the thousandth on these scales - so somekind of good laser based tool is required.
I have also considered software solutions: Bed autoleveling and closed loop system, crude closed loop (every Nth layer home X and Y) etc. Needs more research into that.
If i make the whole sled with Z axis lightweight enough, i doubt the cost creeps insanely high, then i can probably get away with strong Nema 23s. Need to calculate some forces!
Droop is going to be an issue however, and the fact that makerslide will not be straight enough. Droop cannot be fixed with leveling from the ends, might need an software solution there to stay in cost effective range.If the sled stays below let's say 3kg, using CoreXY i have 2 motors strength to move it to begin with. Wonder if i can wire into single channel multiple stepper drivers to just put more cheap steppers, ie. one in each corner while driving single belt/fishing line.
What i'd REALLY like is to come with a way to put the power remotely to run the extruder and Z axis. Cables will probably flex too much for accuracy, hydraulics get messy really fast, pneumatic might be the solution, but i have no clue how to make a stepper using pneumatics - i'm sure someone has come up with a solution to do that, but i've got no idea how to do that. Fortunately, i've got someone to ask about that and if possible, help with the design and choosing parts.
There won't be much weight remainaining on the axis if that can be solved!
Makerslide weighs in at 0.98kg/metre, something like 1½ metres of that is required -> ~1.5kg on the sled, and if extruder + Z can be operated with remote power, that still leaves 1.5kg for all the mounts, extruder, hotend etc. Mounts will probably be 3d printed, if ABS is not strong enough then from POM or PC.
Ultimately, this machine is not a must-have, so i'm working on this as a hobby, while setting up many small printers to do our production runs initially with, still targeting summer for building this!
Cost wise i'd like to stay below 2000€ total. Huge chunk of that can be directed towards motors
My belief is that very precise calibration will be easier as well because of the long distances, if steps are off 0.1% it's easier to measure it across 150cm than it is across 20cm
The main hurdle will be leveling everything and making sure it's square etc.
|
Re: Making a huge 3D Printer! January 21, 2014 12:20PM |
Pneumatic stepper motors: [www.bibus.hu]
Am i reading this right? Aren't those total crap?
Weighs too much as well.
Seems to be rather new thing, i found a paper from 2007 describing it as new.
Oh well!
Any ideas for remote power, which doesn't flex and doesn't weight too much?
My bellief has always been that all cable based remote power flexes, and therefore will be way too inaccurate for this use.
Am i reading this right? Aren't those total crap?
Weighs too much as well.
Seems to be rather new thing, i found a paper from 2007 describing it as new.
Oh well!
Any ideas for remote power, which doesn't flex and doesn't weight too much?
My bellief has always been that all cable based remote power flexes, and therefore will be way too inaccurate for this use.
|
Re: Making a huge 3D Printer! January 21, 2014 12:41PM |
If you had a MASSIVE SLS machine (which typically (I say typically as if more than a handful exist) cost millions of dollars) you could fabricate reasonably thin rods with a stressed apposed strut internal structure, from titanium, heat treated and precision ground. Then half of your problems go away. You still have to deal with step error, though...As to wiring, keep an eye on my thread. I am doing a variety of things to minimize moving weight and wiring, like you I am not producing plans at the moment, but my progress should give you some ideas on limiting not only moving mass, but wires as well. Wireless solutions would be far too error prone due to RF issues, at any price.
Then half of your problems go away. You still have to deal with step error, though...As to wiring, keep an eye on my thread. I am doing a variety of things to minimize moving weight and wiring, like you I am not producing plans at the moment, but my progress should give you some ideas on limiting not only moving mass, but wires as well. Wireless solutions would be far too error prone due to RF issues, at any price.
Sorry, only registered users may post in this forum.