Making a huge 3D Printer!

I'm with you CNCD, machines care nothing of cost, I think what the OP may not be considering is that a single belt (line) slip will destroy an entire print...FDM is not a fully scalable technology as far as singular design is concerned. Not that you can't have a printer that makes massive high quality prints, but it wont look, or work like any of the established designs. It will need a whole new design paradigm, new software, new materials, etc, etc. The ultimate question is what constitutes a Working printer? I would, speaking for myself, submit that a working printer is one that once setup can consistently produce clean, tight, properly registered (if multi material/color) prints that have accurate consistent detail and do not fail under desired use. So I personally feel it is important to consider these things, as we attempt to build larger printers. I would further reiterate, that the bigger the printer, the more critical these provisions are, due to printing times.

There's one huge issue you didn't seem to address: weight. A build platform of nearly (over?) a meter squared is going to be massive, probably at least 20 times heavier than your typical reprap. How do you intend to get print speeds anywhere near what a small printer can do using the same nema 17 motors? You're probably looking at reliable axis acceleration orders of magnitude below what a prusa can manage, so even a prusa sized print will take far longer on the large machine vs the prusa. Being able to print larger things would be nice, of course, but your stated goal of filling the build platform with small objects and just letting it run over a weekend is much less efficient than just using several smaller printers, and more prone to failure as well.

Edit: I see you're actually going with a fixed bed, but my point remains. Moving the entire axis gantry style is no small matter, as any supports sturdy enough not to flex at that length will be fairly massive as well. I'm not saying it can't be done, but you're going to need much larger motors to have any hope at all of achieving acceptable print speeds.

Edited 1 time(s). Last edit at 01/23/2014 01:38PM by Rezer.

I have a large repstrap,





There is no way in hell your design is going to work.

If you are using any kind of rollers your boned,

If you don't have your Y-Axis spread out enough from allowing your Z-Axis from wobbling around your boned,

I'm using NEMA 23's. If you attempt to use NEMA 17's your boned.

And to top it all off, your foot print is larger than mine, and mine is made from a professionally (expensive) re-enforced frame and rail system that I still had to upgrade to get stiff enough to 3D Print.

The only way anyone will get a 3D Printer that large working well is if they drop some serious cash on proper linear slides from an automation company.

I don't want to be a Debbie Downer, but I don't want to see anyone waste their money,

It will be a learning experience for you, but seriously, don't put too much money into it thinking it will work.

If it was possible for someone to build a volume that big on a low budget, it would already have been done.

Edited 2 time(s). Last edit at 01/23/2014 03:30PM by ShadowRam.

To the issue of weight, I expect to end up at around 300LB, for 3.25 CUFT. And that is for a minimal sized machine. I'm with you S-Ram, many of us , myself included come here to learn about these machines. And I can say, for someone starting out in this pursuit, the less money they have to spend on it, the more critical it is they do things the right way to be successful. If you are going to go big, sooner or later you realize you will have to spend big, if not on parts, then on labor. If you have the talent but not the tools, maker clubs are becoming available. If you just want a huge printer with little cash, and little work, well, we'd all like to see how that works out. You can get into printing with a few hundred bucks or so, but small, one color, PLA, and don't expect miracles. I'll have a year in my printer at least before I even begin calibration. That's kind of what it takes...

I agree with Maxx I haven't seen too many large-scale printers that turned out what I would say would be okay. It's not easy you know you're going to need a enclosure to keep temperature up in print area so everything has to be designed around that. That means motors and everything inside is going to have to deal with temperature some of it needs to stay warm and some needs to be cooled. And reliability is very important because of long print times. Pulsedmedia if funds are low I think I would try something about half the size of what you're trying to build or whatever size you can afford to and use good industrial components on. High acceleration and deceleration is really what's needed for printing. And I might add safety so it does not catch fire

Edited 2 time(s). Last edit at 01/23/2014 08:53PM by cnc dick.

GT2 is too weak for my use, plus getting proper lengths is going to be an issue.
We have a local shop who specializes in belt drives, i could go there to shop around, but the cost to quality ratio is going to be way off compared to weaved thick fishing line.

As for the temperature limitations of fishing line: Wikipedia says 80 to 100c, if i keep the enclosure at 60c, there shouldn't be any issues what so ever

For now, far bigger issue is getting the correct stepper motors and drivers. These are expensive! :O
For i need both speed and accuracy, it gets very very expensive. I'm satisfied with sacrificing some positional accuracy tho.

Quote
Rezer
There's one huge issue you didn't seem to address: weight. A build platform of nearly (over?) a meter squared is going to be massive, probably at least 20 times heavier than your typical reprap. How do you intend to get print speeds anywhere near what a small printer can do using the same nema 17 motors? You're probably looking at reliable axis acceleration orders of magnitude below what a prusa can manage, so even a prusa sized print will take far longer on the large machine vs the prusa. Being able to print larger things would be nice, of course, but your stated goal of filling the build platform with small objects and just letting it run over a weekend is much less efficient than just using several smaller printers, and more prone to failure as well.

Edit: I see you're actually going with a fixed bed, but my point remains. Moving the entire axis gantry style is no small matter, as any supports sturdy enough not to flex at that length will be fairly massive as well. I'm not saying it can't be done, but you're going to need much larger motors to have any hope at all of achieving acceptable print speeds.

Indeed, i intend to have a fixed bed, so issue resolved. Current designs BIGGEST issue is the moving bed.

The long axis is supported in many places, with built in adjusters -> no flex issue, the shorter axis will house the Z as well with hotend. This will be shorter, therefore less droop.
Makerslide is quite big. and still low weight: [www.makerslide.com]
The short axis will weight less than a kilogram for the axis itself! the plastic parts will be very low weight to fix it and adjust it into position.
The Z axis and the 2 steppers there will be the heavy weight.

I'm thinking of battling droop, if it becomes an issue, by putting the short (Y) axis on tension 0.5degrees upwards or so, and at the end of the day what matters is that it's level against the bed, so in theory i can put in middle of the bed 0.1-0.5mm higher studs which will force it upwards. Right now the larger issue is droop on the bed to be honest.

Quote
ShadowRam
I have a large repstrap,

20131107_171821.jpg

There is no way in hell your design is going to work.

If you are using any kind of rollers your boned,

If you don't have your Y-Axis spread out enough from allowing your Z-Axis from wobbling around your boned,

I'm using NEMA 23's. If you attempt to use NEMA 17's your boned.

And to top it all off, your foot print is larger than mine, and mine is made from a professionally (expensive) re-enforced frame and rail system that I still had to upgrade to get stiff enough to 3D Print.

The only way anyone will get a 3D Printer that large working well is if they drop some serious cash on proper linear slides from an automation company.

I don't want to be a Debbie Downer, but I don't want to see anyone waste their money,

It will be a learning experience for you, but seriously, don't put too much money into it thinking it will work.

If it was possible for someone to build a volume that big on a low budget, it would already have been done.

There will *always* be nay-sayers. When the first mendel was being developed, i'm sure there was plenty of nay-sayers. Look at what happened.
Infact, more there is nay sayers, the more i see i'm doing something worthwhile.

The worst thing that can happen with this is we deem it doesn't work and just downscale it until it works, but being stubborn as i am, i will probably battle to find a way of "good enough" quality, if it cannot do the fine intricate designs we need, fine, we'll just do car parts, sideskirts, front masks, air ducts etc. a millimeter here or there, even a few doesn't matter s* on those parts, they get anyways acetone steamed, then attached, sanded, filler, sanded, filler, sanded, filler, sanded, filler, sanded, primed, sanded, filler, sanded, primed, sanded, paint, sanded, paint, clear, polished
or alternatively they goto motorsports and no one gives a f* if they are a bit bent or a bit misaligned, they get ruined anyways by the season's end. and misalignment here can be centimeters and the guys will just force it into right shape.

Y-axis spread enough: 64.4mm is the makerslide wheel spacing, the axis will be mounted probably on a about 80mm sled. 80mm ought to be sufficient, if not, it just means it needs to be either tensioned, or i will need to add a guide linear axle to make a triangle. No biggie to be honest. That third doesn't even need to be precision machined, as long as it will keep the hot end within couple degrees it's gonna be just fine.

I actually hate it when people say "you need precision that and you need precision this" - most of the time there will be an engineering solution to be found to live with the tolerances without going wildly expensive on the mfg process. Rather, i'd look for what is "good enough" for the job, ie. performs to the target performance, with desired reliable, with close enough repeatability.
Sure the makerslides aren't going to be that straight, i've actually heard there is quite a bit of bend, that's why the makerslide i cannot easily adjust mid length is the shorter one, and the long ones will house multiple adjustment positions to tune it in to be true. I don't give s* if it's 0.01mm out of alignment - i just don't. As long as it's every time out of alignment the same amount AND the out of alignment is much less than the desired resolution.
If the makerslide is 10cm higher on the other end than the other on the long axis (X) i will simply adjust it, not a big deal, if the Y (short) axis is drooping, i'll add tension or make a triangle to make it not droop.
If the v-wheels are proving to give backlash and become loose too fast, i will simply add a tensioning system and choose wheel material better suited for the task - again not a big deal.

If the steppers i choose are skipping steps, i'll make a gearbox, or lower microstepping, or i'll get stronger steppers.

These are called engineering challenges. They are meant to be solved
I much prefer to design myself out of a challenge than go overboard with expenses.
There is no doubt in my mind this machine could be built with limitless budget - the trick is doing it within a constrained budget.

They way biggest question now is, what kind of steppers do i need to drive 4-5kg at speeds of 1000mm/s+ with accelerations of 5000mm/s2++ - and i have no idea of the maths i'd need to be doing to calculate the forces required
Being CoreXY i already have the benefit of having 2 steppers driving the axis.
or should i just go with multimotor route, add more basic cheap stepper drivers and stepper motors. one motor in each corner.

Does anyone know can one turn standard brushless motors (with gearing obviously) into stepper drivers?
Fully closed loop system could do that probably, with a lot of custom coding tho.

Single slip of drive system:
I was considering that initially when doing huge prints we'd modify firmware to do homing every 2nd layer or so, that way it gets realigned and not the whole print is ruined IF the lost steps/slip is not too great, most of the things i do we can live with 1-2 layers being a bit misaligned when doing very large prints. Or if it's ultra large print, we could put a built-in timer, rehome every 10-15minutes or so.

I also want to look into what it would require to make a closed loop system - that would solve that issue Who cares about occasional slip if the position is ALWAYS checked

Quote
cnc dick
I agree with Maxx I haven't seen too many large-scale printers that turned out what I would say would be okay. It's not easy you know you're going to need a enclosure to keep temperature up in print area so everything has to be designed around that. That means motors and everything inside is going to have to deal with temperature some of it needs to stay warm and some needs to be cooled. And reliability is very important because of long print times. Pulsedmedia if funds are low I think I would try something about half the size of what you're trying to build or whatever size you can afford to and use good industrial components on. High acceleration and deceleration is really what's needed for printing. And I might add safety so it does not catch fire

half of the point of this whole hobby project is to find out if we can do it on a budget. You'd be surprised how many things can be done with quite a little of money when some thought is put into it.
Besides, if we manage to do it on a sane budget, let's say 3000€ and it works, and it works brilliantly, and turns out to be OK labor wise to build more of them, and we know the demand is there ... You see where i'm going with this?

That's actually how many neat projects gets started - as budget conscious things. Original Mendel was never ever planned to be like Stratasys machines, it was meant to be on a budget. And look where it has lead things. People are now buidling machines which exceed stratasys quality a lot, with a fraction of the cost.

If to make this printer to work needs 100 000€ - it's not anymore something i'm interested in at all, not the tiniest bit!
It would require that printer to operate 5 years in a row, without malfunction, 24/7 for me to ever recoup the initial investment.
But done with 3-5k € or even less, i only need to make a few dozen prints to get my money out of it and best of all: Worst case scenario is that i repurpose the parts for a normal size printer and can say "Oh well, i tried!"

Right now tho i'm working to get many small printers to work autonomously to make some production parts, and then i need to work on some gear for our DC, and build up 3 cars before i can get back to this project So July-August looks now a bit optimistical to have the frame built, it's going to be more like fall. Next 2 months is going to be hectic with the season starting in April and exhibitions coming up, then a bit of a breather before the racing season starts. Then it's going to be a hectic race between events and keeping car operational and functional, which with day-to-day work doesn't leave much time for hobby projects until about end of July Hopefully by June tho i have fully designed this in openscad

Any ideas for the stepper motors btw?
Or should i pursue the multiple cheap motors multiple cheap drivers venue?
I'd really like to get cracking on the drive system and get the motor mounts designed, as those might affect how the sleds need to be designed.

Speaking from experience as someone who has designed and is building a machine of the same size I will tell you a few things.

Extruder:
Your extruder design is god awful. It will fail badly. Designing an all metal hot end isn't to tough and there are ALOT of examples out there that you can draw from. The biggest thing I will tell you is that if that picture is a section view of the extruder, the chamber before the tip is way WAAAAAAAY to big. Ghetto or not it will also do very, very bad things if you use the wrong kind of washers and get to much heat build up. They could cause their coating to smoke which could give off some really bad gasses.

Frame:
Extruded aluminum will work for a frame but expect problems with flexing. An exterior frame will help but the idea of filling it with concrete will only cause further issues with bowing and flex due to the added weight. If you did this, id only fill the base parts of the frame with concrete to help add weight to compensate for momentum at the top.

Stay away from steel where possible. Steel is strong but its heavy. This can cause issues.

Motors:
DO NOT SKIMP ON MOTORS! For the love of god drop more money into them. They will make or break you.

Filament:
There are larger spools of filament out here, but a 30lb roll directly from china will run you $500 starting.

If you get raw PLA or ABS pellets, expect to have issues. They ship these things in rail cars that get rained and snowed on which in turn causes the dreaded issues with filament. Research how filament is made and the processes used prior to going into an extruder.

Research:
Look up every damn thing you can find on big machines. They are out there, just hard to find. Sometimes purposely so.

Look at mechanical movements and issues with the designs of the small machines and the issues they have had with the frames. That will give you ideas of what to look out for because when you go big, those issues do one of two things. They go away or they get exponentially worse.

Cost:
Whatever your budget it, expect it to double or triple (trust me). You will do something that you think will work and it will fail. Hard. This will cause costs to go up as you back track to do repairs and reproach the problem.

If you have some questions, PM me. I will answer what I can. Just realize, there is a reason that I am being vague.

This looks really awesome, glad you're documenting everything!

some inspiration ...

a huge printer....dont know how it works though

3der

Extruder -> You are actually talking about the hotend
Haven't built yet, and have to still do testing.
Melt chamber size is infact adjustable on the "ghetto" design, i can just loosen up the actual hotend, screw it upwards, retighten, shorten the nozzle and screw it back in. Done, smaller melt chamber
One of the things i need to test out is outside shape of the nozzle, in theory a bit sharper than usually seen might be better ... But, at the end of the day it's all about flow dynamics.
Coating heat up gasses : I'm not too worried about that one, this printer will be built into a 900m3 space, where welding is often done and other metal fabrication - the tiny amount of gasses in a few washers ONCE is not going to be issue unless i'm intentionally breathing them in
One thing: I can easily experiment what's a good amount of washers, cooling power etc. I don't believe for a sec there is going to be a issue with the cooling beyond the initial attempts to find out proper spacing etc.

I have a J-Head on my "bench" printer (awful awful prusa v1, just crap to work with), and it extrudes so that the extruded plastic curls to the "left", most visible symptom being that when doing first layer when the head is moving to roughly opposite direction than the curl, it comes up quite nice, but when it comes back, it slightly "lifts", creating a reeeeeaally awfull first layer.

Next up is testing Makerfarm metal hot end, and i guess for testing out i'll shell out for E3D all metal hotend as well, to see how they function.
But with my own hot end what i REAAAALLY want is the ease of changing nozzles, and with that comes, the ease of cleaning it out

Larger melt zone: While obviously retraction is going to suffer, on my testing the original j-head for example has waaayy too small one, often the plastic doesn't get heated up enough even when printing PLA at 230C :O then at times it gets heated up too much. depends on what is being printed. I hate that per print experimentation, so in perfect world the filament comes into melt chamber at say 80C for ABS, and the melt chamber has sufficient surface area to transfer enough heat to heat it quickly enough to printing temps. It's always going to be an compromise, and single hot end / melt chamber size will not work for all situations -> or that's the conclusion i've drawn. Unless you go really really slow.

Actual extruder:
I will be using the normal Greg's wade.

Frame:
At no point did i say i was going to fill it up with concrete. I am not. I *hate* that idea.
The long X axis makerslides: They will have supports every 30cm or so, no issues on flexing or manufacturing intolerances, as i can just screw it to alignment.
Y axis: I will mount it straight, and i decided that instead of extruder going up and down, i will be making the bed as the Z axis to lower weight on this, i will see what the droop is, and if there is one, i will probably print mounts with 1degree or less angle upwards, and mount it under tension to see if that works. If that doesn't work, i need to make the same amount of droop into the print bed.
Z axis: I have done now a mental decision to make the print bed, despite i already designed mostly complete bed mount, heating etc. to save weight on Y axis, despite i had an idea of a possible way to add to CoreXY a Z-axis to it by changing length of the belt.
It will be mounted on a threaded rod on each corner, and a smooth rod, i will probably use M8 size since that i got plenty around, and same goes for 8mm linear axles (i got like 60x 0.5m pieces of those lol), and i have the corner pieces already designed in my head, and rough ballpark idea how to operate all 4 with single stepper.

Steel: Actually, the weight is only an issue if it's on moving part
Steel is my friend here, with that i got plenty of experience working with I will use steel as the base support structure everywhere i can. Big and heavy duty steel.

Only place i'm going to add weight is the base, if it ends up vibrating far too much.


Motors:
Research still on going, i did not intend to skimp here, but that doesn't mean i'm going to buy 500$ a pop Nema 34 steppers and 300$ a pop stepper drivers neither!
But before i know what i need, i need to do some maths on the forces involved with the targeted accelerations, weight etc.
At the end of the day, you'll never know if the strongest available Nema 17s on sensible budget are sufficient, 6.5kg-cm vs. 4.4kg-cm usually used with a reprap.
One method of getting lots of power on the cheap is somehow figuring out how to make a completely closed loop system, requires a bit of custom coding and quite a bit of research what's on the market since standard sensorless brushless motors are a no go (they attempt often to start to the wrong direction), which is such a shame.
Finally, if the motors i get end up being too weak, it simply means i can't run it fast until i get some new ones.

One thing i never seem to find about steppers is: How fast can they go? What kind of power do they have at, say 500rpm? 1000rpm?
What happens to their torque when going fast?
That is one of the reasons why i want to look into a closed loop system, then i can go in and choose a simple brushed DC motor which pushes out 200W @ 15k RPM and gear it down by 1:10 to get to the RPM range i want and torque. That ought to give about 12.5Kg-cm PER MOTOR. Put 4 of those to drive XY, you get at minimum 25Kg-cm per axis, total 50Kg-cm.
Vs. traditional reprap total of 8.8Kg-cm.

No specs on RobotDigg high torque Nema17 @ 24VDC driver as they suggested

stepper driver higher voltage: [www.circuitspecialists.com]
Rather cheap high torque Nema23s: [www.circuitspecialists.com]
Or Nema34 with 63Kg-cm: [www.circuitspecialists.com]

Tho, stronger the single motor is -> The more expensive it is to get fixed on the frame and the drive system.

Not that expensive at the end of the day if i have to choose 2xNema 34 and 2x expensive stepper drivers. ~150$ for the steppers, and ~200$ for the drivers is still just 350$, and in grand total and compared to the plastic going through this machine, it's not that much. Just about the same as ~90lb of pellets, which i will go through in matter of couple first months doing car parts.
Hell, now i'm tempted to get those Nema34 beasts!
it's just a bit over 14x increase compared to standard Reprap and i will be running 4 fold the weight, but also 1/4th microstepping, which is unknown how it affects torque @ speed, so the end result in speed is what ... ? Is it able to go 3.5 times faster as the maths suggests without considering microstepping?

Filament:
Looked already into extrusion / making own filament. This is kinda important.
Basicly, one needs to bake the pellets at around 95C for couple hours to get the moisture out.
Some materials you can get as cheap as 2.5$/lb.
Plus of course if we have capabilities of making our own filament, it means we can recycle failed prints, and where surface quality is not important, pretty much throw any plastic in

I want to eventually be printing car body parts, which means we need our own filament to begin with, infuse it with some carbon fiber or glassfiber etc. as for the resin material (plastic) it doesn't matter that much what it is - as long as it's not biodegradeable and bulk of it has working temperature of close to 100C at the very least (summer sunshine ..), even ABS is not the best material for the job. Someone suggested polypropylene and HDPE but i have yet to check up on the details of those two. POM is very cheap as well, and used in heavy duty automotive applications already: Bushings for motorsports.
Ultimately they are large and no need for sub couple millimeter details, hence for those parts i will be using 1-1.5mm nozzle, which pretty much guarantees the need for 5mm filament as well.
Running that large nozzle gives interesting options as well, carbon fiber "dust" is very cheap, i could try running through few millimeters long strands of carbon fiber, hell even aluminium or stainless steel dust or strands for added rigidity and strength. Those experiments would be so interesting and fun to do!

Big machines in general:
I've looked into a few, interesting issues and dilemnas!
But we must remember, in grand scale of things, my machine is not *that* big, the weights or not *that* high, hence the forces are not *that* high.
We must remember that my whole Y axis will weight less than 2kilos, which is not *that* far beyond the 0.6-0.7kg on a standard reprap. The bulk of added weight on that is the makerslide and the mounting pieces that go with it. The extruder and hot end will be standard. and i will be driving it with 2 steppers to begin with (still need to do maths to verify my initial assumptions about forces with CoreXY), and i have easy means to add 2 more.
That would be *4* steppers driving the X and Y axis, that's on best case scenario quadruple the power, and worst case scenario double - if using the average Nema17. Further i ca

Further, i intend to reduce microstepping, and might potentially use gearing. Not sure of the figures yet, need to do maths. But in any case, holding torque will increase by dropping microstepping to 1/8, 1/4... I've even been considering full steps or half steps. Know too little about step motors to make a decision yet
Ultimately, the usual reprap has what 0.012mm "accuracy", but precision is out of whack, hence they never ever achieve anywhere that kind of positional precision. I think 0.024 (1/8 microstepping) even is more than needed, so 1/4 with 0.048 would be a good option, for increasing holding torque. going half step is cutting a bit too close if i decide to print something with say, M3 nut sockets, or long smooth curves. but when using a huge nozzle for large parts, full step is well enough

It's definitely going to take a lot of experimentation etc. but also trying to do it on a budget is important -> someone needs to, eventually.
I'm targeting now 2-3k €, that sounds insanely low, but that already buys huge amount of materials, if i choose the right parts. Use as many as commonly used parts, avoid all specialty parts like plague (specialty==expensive), i believe it can be doable, if the motors can be solved.

thanks for the link about BigRep.
Cool machine!

Quote
PulsedMedia
Melt chamber size is infact adjustable on the "ghetto" design, i can just loosen up the actual hotend, screw it upwards, retighten, shorten the nozzle and screw it back in. Done, smaller melt chamber

Larger melt zone: While obviously retraction is going to suffer, on my testing the original j-head for example has waaayy too small one, often the plastic doesn't get heated up enough even when printing PLA at 230C :O then at times it gets heated up too much. depends on what is being printed.

Got a pic of your hot end with the adjustable melt chamber?

Melt chamber size of the J-Head:
Other than extruding a low volume vs. a high volume, I don't understand how the melt chamber can be too small, and too big for the same material.
How noticeable is the difference, does every one experience this?

When printing small things @ low speed, then it heats up too much, since the high temp is to compensate small melt chamber. Manual adjustments is a bad idea as eventually i want to be running these machines autonomously, i got a ton of things to print, just to begin with when design is finalized, i have about 3500hr printing queue :O Just for the ITX Blades + chassis parts.
IMO, better to have large melt chamber and just use bigger retraction

No i don't have a pic, as mentioned, i am yet to build one, which is a bit stupid as i could use it right about now, i got Nylon and POM (Acetal) coming in.
The design has a bit changed, from 5mm SS pipe to M5/M6 threaded rod. Got to build one soon, as i want to start testing with 0.2mm nozzle as well

Hi. At work there was a guy who was designing a 400x400 mm reprap. But desisted because the massive amount of energy needed to heat the heatbed.

You should do that math first to see what pops out.

400 on a side is not that big. That is unless of course you intend on using your ramps board as the power supply. I'm 18 x 26 in or so, and taking power responsibilities completely off the control board. You're looking at six or seven hundred bucks to do this properly. Long story short, more than a square foot, mains power is the only way to go. The larger you are looking to build , costs will, let me clarify that, will go up exponentially given the reasonable assumption of the following fact: The filament is in control here, kind of like the high seas in a small boat - the filament does not care the level of resources you have to invest, because filament is not a capitalist, it's a piece of string made out of plastic. If you lack the proper heat, the proper degree of rigidity, of repeatability, the plastic will not make plastic parts. It will make something akin to pasta, re imagined by Andy Warhol. Now I have been working on a reasonably large machine for about a year. Considering its size and projected capabilities, the materials cost, with little regard thereto will come in at about the price of a Makerbot x2 at retail. However, you have to figure in over 1000 hours of fabrication labor. And that puts you at about the price of a a Zcorp 650 with materials,software, and service for the first year. In the end, I'd say that a large machine can be built at a reasonable cost, but in reality only if you have access to a machine shop, and a whole lot of time...

Quote
PulsedMedia
When printing small things @ low speed, then it heats up too much, since the high temp is to compensate small melt chamber. Manual adjustments is a bad idea as eventually i want to be running these machines autonomously, i got a ton of things to print, just to begin with when design is finalized, i have about 3500hr printing queue :O Just for the ITX Blades + chassis parts.
IMO, better to have large melt chamber and just use bigger retraction

Seems like a 3500 hour print queue could be better handled by two 1725, or four 875 hour queues.


There's a few links to large machines on: [reprap.org]

I need the size to make the parts to begin with. Then again, it's not one of the reasons i want to build a megasized one.
Fully autonomous with many smaller printers will handle the huge queue.

as for heating the heat bed:
My intent was always to go away from PLA, biodegradeable in automative OR datacenter applications not so good... especially with the low temps
and hence, make the printer enclosed, as it will also reduce warping on ABS parts.

I've also had great success with no heated bed printing as of late, with PVA glue, my own mix of "abs juice" (contains also PLA and Nylon) on a *very* clean plate of glass.
What i mean with clean is not just washed, not just acetone rubbed... Infact, acetone alone doesn't come anywhere near the cleanliness i want, but something every household has does: Dish washing soap, which is then finished off by drying using black + white oldschool newspaper.
The newspaper ink also partially transfers to the glass, filling imperfections - my guess is that those particles have also some say on the initial sticking to the bed - at this point i can print ABS without problems to bare glass.

To get even more sticking very strong PVA mix seems to do quite nicely for PLA and Nylon.

Costs does not need to increase exponentially. Makergeeks have apparently had quite a bit of success with 16" by 16" and 32" by 32" printers.
What we need is a design out of the situation.
Thing is, as research goes on by, technology gets better, more people use, there is only one direction for price to go: Down. Same holds true for repraps, my first reprap cost me a total of probably 2500€, today i can build the same machine at 300-350€ and actually get it to work swiftly. Just couple short years ago unassembled arduino & ramps costed MORE than assembled ones with extras today. Actual linear axles were harder to come by and costed much more, now i purchased a lot of 50x500mm 8mm linear axles for like 150$. etc etc.

i looked into big stepper drivers and way bigger Nema steppers - they aren't *that* expensive in grand scheme of things - power always costs.
But i'd much much prefer to take a standard DC motor and adding the new "make stepper from anything" sensor kit just released, forgot the mfg, maybe texas instruments or some other, a friend who is automation engineer told me about it. Much much higher precision, just need the right encoders.
and when we are at a point we can use standard DC motor - perhaps we could even use a brushed one? with sensorless brushed the main issue is initiating movement from stand still which can try to begin by going to wrong direction - since with that sensor we are essentially doing a closed loop system it doesn't even matter.

And i know from experience that we can push 7kW out of a 15-20$ chinese brushless RC motor .... We want something a bit more long lasting, so "just" 700W peak per motor with a gear down box. Gear down boxes are probably 30$ a piece.
Tho we will also need ESCs for those brushless motors, tho, from my RC days i know i can get 100A+ 40V+ capable chinese ESCs for fairly cheap as well, just something like 25$ or so what they cost years ago, and then we just need to input the right kind of signal to the controller.
That does get to cost about 90$ per motor - but only need 2 of those: 180$

Tinkering and making it all happen is about 1 week's worth of work, let's say something like 35hrs.

Now we have practically limitless speed and power for the X & Y axis. Slamming the whole Y axis at speeds of 2 000mm/s+ won't become an issue. Strength of drivelines becomes a issue tho.
Even if there is 5x GT2 timing belts side by side they will fail at these power levels.

Closed loop is kind of a must for this system, and i've also been looking into some methods for distance sensing, but either they are über expensive, or do not scale

On other news, i've been running successful production batches for smaller things on small reprap lately. I'm burning through a couple of kilos a week now on just 1 printer Time to get a filastruder finally.
Running reliably ABS and Nylon right now, nylon is annoying to work with, need to bake it prior to each print etc. ABS has become very pleasant to work with, but nozzle pressures are way too high. PLA makes better quality but my current hot end tends to get clogged after a few hours of PLA printing. Better get a E3D hotend or make my own soon.

Time for this project has been extremely limited.
Next sea freight leaves in a few weeks, i better get the makerslides on this shipment so i have them in september.

Let's hope i get enough time sometime soon to start researching again, fall is going to be very busy tho, winter as well, but somewhere along the line i'll probably find the couple of weeks of time to design and assemble this beast

I've been considering the power requirements for heated bed.

A standard prusa with 200mm * 200mm has surface area of 400cm2. Poorly insulated, open air contraption and requires roughly 120W or 0.3W per cm2
A printer with 1 000mm * 1 000mm heated bed surface area is 10 000cm2 and at same relation requires 3kW. However, well insulated bottom on the heated bed and a chamber will lessen this requirement significantly, it could get down as low as 1.5kW or maybe even lower.

Segmenting the heater so we only need to heat up the portions actually being used makes sense, tho manual thing. That electrical consumption is not a power, if a print lasts for 2 weeks, 1.5kW is worth 75€ in a place which needs to be heated up in any case!

Next up is speed.
I think we can drive that up by going higher voltage on the steppers and dropping microstepping altogether and get the precision by gearing.
Precision around 0.05mm is sufficient.

If stepper has 180 steps, it's 2 degrees. If 2 degrees needs to be 0.05mm it means the diameter of pulley should be 9mm which is a little bit tiny to do printed.
So perhaps gearing, 27mm gear on stepper, pulley axle 34mm gear and 18mm pulley, which becomes 0.0636mm roughly per step.
This can be worked on later, probably i will move towards larger gears for strength and less relative play.

The usual reprap has 1/16th microstepping, so moving totally away from microstepping we are increasing the torque 16 times!
Then becomes the question how fast can the steppers turn without microstepping - or does it change at all in actual nozzle speeds.

To further increase power per axle we can go for more steppers per axle
2 steppers = double the power.

This is if no one has come up with a cost effective system to use standard DC motors or brushless motors as a a stepper. I would much prefer standard DC motors if they can be controlled because of the simple fact they can be driven much faster most likely.

Other way to drive up the speed is to make a 1mm nozzle hotend. That's double the line width, double the speed while losing from quality. But when printing very large parts they will be post processed in any case. We will be experimenting on laminating glass fiber or carbon fiber directly onto the printed parts, brushing with acetone and filaments.

Now need to estimate time for a big print, and i'm wondering how i should do those maths. Should i look up print times on a smaller reprap per cm3, or perhaps slice something humongous and see the end result?

The priority of this machine has shifted now, for other projects it's likely we'll need this machine this winter and for work related projects.
It is possible we will once more change the design almost completely to work with off the shelf local materials.

Also we will feed this machine directly out of a filament extruder most likely - unless it causes extra issues. Just need to work out how to buffer the filament most conveniently to avoid tangles.

I loaded up a basic piece on slic3r, made settings changes etc. for a 34 178cm3 build volume object, loaded up in pronterface. For speeds i used what one of my prusa v1s can do with basic steppers, tho that printer has been getting slower and slower over time needing to turn down the speeds.
I used 50% honeycomb for that part.
Slicing was surprisingly fast and it turned out to be 79meg gcode file.

but loading it into pronterface took much longer and the preview doesn't reveal much
Pronterface gave these values:
2494052.61906 mm of filament used in this print
The print goes:
- from -1509.98 mm to 2784.70 mm in X and is 4294.69 mm wide
- from -1334.67 mm to 2960.03 mm in Y and is 4294.70 mm deep
- from 0.00 mm to 174.50 mm in Z and is 174.50 mm high
Estimated duration: 3 days, 10:57:53

These values are obviously screwed the part is supposed to be 1125mm by 174 by 174
So pronterface cannot handle those values

But let's assume the filament and print time is correct.
Previous experience tells that actual print time is 25 to 50% longer than pronterface estimated duration, so this one would take about (at 35% longer duration) 4days and 16hrs.

1kg spool of ABS has about 130meters of filament - and that estimate says about 2500 meters or 19kg. I don't think that's correct
The normal size part weighs like 40grams, so i think the actual weight is more like 1.9kg rather than 19kg.
Since i scaled it up 700%, 40gr * 7 * 7 = 1.96kg. (is that correct?)

So that's an another hurdle - should i scale everything down 1/10th on slicing and pronterface?

I would try another host before making sketchy changes like that. I'm a long time Repetier Host user so I can't give you much advice there.
As a fellow larger scale builder in progress, I think you're spending too much time rebutting people's "nay-saying," so you call it, and not enough listening to what they actually have to say. A lot of your "nay-sayers" are the ones who are truly developing the RepRap world with ingenious designs. I think they know what they're talking about, and not those old people with tons of experience, they are those with fresh, inventive ideas. I started out with the goal of less than half your build area, at 1000X500X500mm. I've shrunk to about 650X310X[unknown yet]mm and I'm still finding hurdles to get over after five months of research. If I had waited more patiently and listened better to what people had told me, I would likely be finished by know. I sadly have wasted half of my spending on parts that aren't suitable in the slightest. These people really know what they're talking about and when I really started tapping in to their wisdom things really started moving much faster. I've only got my frame right now, but it is rock solid. Did I follow everything they said? Nope. The things they had to say really helped me develop my own ideas. They're not nay-sayers, they make no benefit in reducing your dreams, but they do want to help you get to your dreams better.

It will work, but be aware that it won't work easy or near what you may expect. Building large scale is an entirely different game. You can't just add more stuff to it, if scaling were that easy, anybody with a budget would do it. The larger you go the more inefficient your money will be.
"strong steppers to be able to drive it travel speeds of 2000mm/s" If that is still your goal [I haven't read everything] you will be extremely disappointed. The Ultimaker 2 is made from a whole team of geniuses and pushed to its sizes's speed limits, and it doesn't reach anywhere near that. With all the weight and suspension spanning your planned dimensions I'd be so bold as to say you'd never reach half of that and print reasonably. Your travel does affect print quality, especially when a lot of weight is in play, I can tell you that from personal experience. Look at the Objet1000, it's large scale and still very accurate. Built with top industrial parts, but it isn't fast. Size and speed are near direct trade offs when maintaining accuracy/quality. Don't even dream of doing that with just NEMA 23's, let alone 17's if that's still your plan. That RAMPs is also incapable of such things. The game your playing cannot be played with the standard pieces. The parts, both mechanical and electric, that everyone uses on standard prints aren't going to work up to those.

"Nay-saying" aside, I honestly do hope you achieve what you want. I really do. You'll be pulling so many world records that you'll have plenty of wanting investors knocking on your door and maybe even Stratasys will try and drive you into the ground with patent violations I'm sure they can make up. Just don't waste all that money like I did. Wait until you know to a depth the science and engineering behind what you're trying to achieve. In my mind, you've already wasted a lot, you've already failed and you'll have to start over from scratch. I hope I'm wrong.

---

Realizer- One who realizes dreams by making them a reality either by possibility or by completion. Also creating or renewing hopes of dreams.
"keep in mind, even the best printer can not print with the best filament if the user is the problem." -Ohmarinus

Actually i'm quite familiar with mechanics - just not exactly this kind of mechanics, but the physics underneath are very familiar.

The main objective is to keep it upgradeable.

RAMPS is perfectly capable with some hacking, it does give the right signals, so it's just matter of choosing the right stepper drivers.

Speed: 2000mm/s is not any more a goal, steppers are not suitable high speed motors.
For 2000mm/s at 0.0636mm steps, 180 steps per revolution is 11.448mm, which means 2000mm/s requires 174.70 steps per second, or about 10 482RPM. Stepper motors i think can't come near that - that's not what steppers are designed for and that kind of figures are not even revealed.
That would need standard DC motors OR significantly decreasing precision.

As for the weight, X axle will have to lug around WAY much more weight than in a small printer, Y not that much different.
Some older figures and maths are above.

My main goal is now just to build it after i decide stepper drivers and stepper motors and just see what happens. Those very same parts can always be recycled

So for these BIG -- 24 day prints
do you have a 1000m spool of filament to use?
a way to splice roll together?

Perhaps a 1 ton bag of pellets into pellet extruder?

Will bed and extruder be steam heated by small reactor?

How are you going to get a FLAT bed?

There was a giant printer (48" x48" x 48") at MakerFaire NYC last year --- never saw it work.

Have Phun

We are building big printers. One thing that comes to mind is something a friend in the inkjet pri t business told me: "you got any idea how many things can go wrong on a 72 hour print?"

I use lasered steel frames, CNC bent and heavy industrial components.

Sometimes you got to take a giant hammer to the problem and that costs either cash or time if I got the shop to build.

We too moved all the power consumption off the board for the heated bed and extruders using SSR. Power supplies if they say 30 amps and you need 28 amps, go with bigger or a second power supply. 30 amps 12v are around 25 bucks so 2 of them are cheaper than a big ATX not to mention the heat is dissipated better.

Stepper drivers got huge custom heat sinks. So many things to do to make a 72-96 hour print successful. You almost really need to build a custom arduino board and ramps with big heavy tracings, big mofsets, and big terminal blocks for heat dissipation.

If you really want big and reliable, ditch the arduino, go with 1hp servos, industrial controllers, and big steel frame. Course you are in the 20k dollar range for parts already, but it's the same with any CNC. Bigger machine the costs and difficulty go up exponentially. No shortcuts

Plan is to make an filament extruder which makes the filament just prior to using and buffers a bit - this way we can also ensure the filament is 100% dry and not moist without special measures

We've been right now been looking into motor solutions. Could easily get huge power if somehow one could control a standard BLDC with a separate speed controller and separate positional controller ...
With a normal BLDC motor it's easy to get 3kW+ power per motor - speed and direction is controlled with PWM output if i recall right.

Power supplies: always use the 80% rule, if you got continuous load you can only use 80% of the rated capacity. Same for power circuits. Nothing special.
As for power supply, i got dozen or so 69A 12V power supplies which can be put in series for higher voltage

Even surface:
Base plate a *thick* wooden plate meant for professional use office desk, 30mm+ thickness - on top of that a 4mm glass sheet. Nothing special
It's even *enough*, if we use 1mm nozzle, first layer can be about 0.8mm thick, and also it means we can have 0.8mm variance in the leveling.

RAMPS & Arduino: People are already successfully running multi day prints without anything special.
For safety, i was thinking adding something which logs the last line printed, or maybe have a webcam recording constantly the process and a display showing line of the GCODE so we can after the fact check the pictures to find out when it stopped and restart from that position.
Further, i think i will postprocess the GCODE to add homing instruction every 1mm or something like that - if any missed step issues arise.

Biggest prints will be about 5-7kg final weight.


I don't think CNC or mill level of sturdiness is needed, we are not doing hard work against hard objects, we are only moving around a few kg axle which is moving around a less than 1kg axle.
Not pushing a milling tool against metal slowly but with sufficient pressure, and not having potentially 20kg+ item moving around on the level.


In my mind it's all about achieving acceptable level of quality, precision and accuracy - not about getting an absolute always within 0.01mm tolerance.
We are going to be printing car panels and other parts ffs, a bumper, if it's out of alignment even by 20mm from end to end is still better than 99% of the other aftermarket products! FRP panels usually have very very bad fitment and more often than not can even be 200mm out of alignment, and it's not even unusual for a product to be this shitty quality!

Now i need to do maths on power requirement to accelerate 3kg to 2 000mm/s within 0.2s, which is about 10m/s^2 acceleration - rather slow infact - about the same as free fall.
It's been a long time since my physics classes so anyone of you know how to do rest of the maths?

As for the heaviest axle, this isn't too badly priced: [www.ebay.com]

1.8degree = 200 steps per revolution.
with 9mm pulley 0.1413mm per step

and in any case, it would need to be geared up for the required speeds and do microstepping.
But i think 300 kg-cm might be enough torque to do that

Now as i read that driver doesn't even do full steps - it doesn't actually supply enough power neither
but with microstepping to 800 steps per revolution and 18mm pulley we get 0.07065mm per step.

That's both sufficient speed AND sufficient precision for the targets.

But does it have enough power ...