Large envelope printer

Hey all,

I've almost finished a delta with a respectable print area (based on AndyCart's Cherry Pi III), but I'm already looking forward to my next project.

I was hoping to build something big enough that could potentially print an engine block. They aren't as big as you'd think. One of the largest in terms of length I'd require would be 70cm long for a large bore straight six. To accommodate a small-ish V8 I think it'd need to be 50cm wide. 50cm height would be more than enough too. I'd also like a design with dual extruders.

Does anyone know of an existing design with those parameters?

[www.re3d.org]

Quote
the_digital_dentist
[www.re3d.org]

I did find that, but I was hoping to find something a little less proprietary, and a little more DIY. I've read a few posts on forums and blogs with people aiming to complete one for around $3k...

What's proprietary about it? Look at the photos and build one.

Quote
the_digital_dentist
What's proprietary about it? Look at the photos and build one.

With that mentality I should be building space shuttles in my shed. Just because I can see one and know roughly how it works doesn't mean it's able to be built at home. I'm not a great inventor either... mostly I copy.

I was hoping to find something a little more open that may follow the reprap philosophy. I know there's little in the way that is produced in plastic when getting up to that size (weight and rigidity requirements of most components dictates metal), but I'm looking for a design where even a rough BoM is available.

I can theorize how things might need to change from the small-ish delta I've built but I'm no engineer. Hence why I'm asking.

3D printers aren't rocket science- you already know how they work, what sort of parts are needed, what they cost, and where to get them. All you need to do is scale up a smaller design or copy a larger design. The motors, belts, pulleys, extruders, bearings, electronics, etc. are mostly the same. Nothing is hidden. There are no secrets. And of course, there's plenty of help available in forums like this one.

Here's a little known fact: great inventors mostly copy other people's work and then expand on it a little. If you can copy someone else's work you have the potential to become a great inventor.

I strongly desagree. Engineering a machine needs some skills. The down scale does not matter, but upscale may not work properly. Most of the larger printers I saw may not work that good, because it's not engineered. You know at first sight no structural studdy was done. A 8mm guide itself flex more than 0.005mm at its centre while supported at only 300mm of distance. How can you pretend print 0.1mm layers with the same guide along 500 or 600 mm ? And it's the guide itself under no load at all. A 20x20 profile will flex too much along 500mm and a 20x40mm too. Calculate by yourself and make up your mind.

---

Collective intelligence emerges when a group of people work together effectively. Prusa i3 Folger (A lot of the parts are wrong, boring !)

I was thinking about this the other day but with a delta printer with a build volume of about 60cm/60cm/60cm the Height of the printer would be nearly 1.3M at the minimum. Probably around 1.5m in height. Then I sat down and thought about the stability of the frame.

As far as I can think through. I would need some large steel plates for both the top and the bottom. Around 1.5-2cm thick. Then some 40x40mm aluminum tubing to create the frame. Also 3 smaller steel plates to hold the moving mechanism.

All in all I think its doable. But the cost associated with building one would probably be in the 1.5K range if your frugal. With print volume in that range, you need 1mm nozzle to even have a chance at slicing and printing anything that large. Dual extruder would be an option. But honestly not necessary. Most slicers are not going to like slicing something that large in sub 500micron layers anyway. The other problem is feeding enough filament. 1kg spools are too small for something that big. 2KG spools may come close.

For now I would like to focus on a smaller desktop sized one. I have a mid range unit that does 330mmx330mmx560mm and it is running well.

---

My Personal Blog. Build blog.
[engineerd3d.ddns.net]

Modicum V1 sold on e-bay user jaguarking11

Quote
jaguarking11
I was thinking about this the other day but with a delta printer with a build volume of about 60cm/60cm/60cm the Height of the printer would be nearly 1.3M at the minimum. Probably around 1.5m in height. Then I sat down and thought about the stability of the frame.

As far as I can think through. I would need some large steel plates for both the top and the bottom. Around 1.5-2cm thick. Then some 40x40mm aluminum tubing to create the frame. Also 3 smaller steel plates to hold the moving mechanism.

1.5cm thick steel? That sounds like serious over-engineering to me. I can't believe that you need anything as heavy as that.

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

1.5cm thick steel? LOL!

I think over engineered is better than under engineered. 3/4 inch steel plate cut to my requirements would be ~200USD which in the grand scheme of things would be a spit in the bucket vs the liner motion parts. Resonance during printing needs to be dealt with, the best way to deal with it is to add weight and stiffness.

One more thing, the larger a printer gets the higher the error rate it will have. Therefore the precision needs to be closer to a CNC than a printer. Its a balancing act between precision and speed. Nema 23 motors with matching high amp drivers would probably become a necesity. Delta printers have an advantage in weight, but even then its probably prudent to add some torque to the drive system.

Edited 1 time(s). Last edit at 03/16/2015 02:30PM by jaguarking11.

---

My Personal Blog. Build blog.
[engineerd3d.ddns.net]

Modicum V1 sold on e-bay user jaguarking11

A steel plate of that thickness will be verry heavy. Beware it will not force your stiles to bend. What about thinner plates with welded straighteners ?

---

Collective intelligence emerges when a group of people work together effectively. Prusa i3 Folger (A lot of the parts are wrong, boring !)

Well I'd imagine that certain components are fine to be over-engineered (eg, stationary parts of the structure where the extra weight isn't going to be an issue), but others will need to be pretty spot-on (such as most moving parts).

I was considering this last night on my commute home. Here are some issues I think might be relevant with the scale up:

Timing belts - We're used to playing with small rubber timing belts but there might be a use case for a timing chain and/or the use of a tensioner mechanism. Light duty timing chain is actually quite cheap. [www.hobbyparts.com.au] but even having a tensioner (kind of a pulley forced onto the belt/chain with a compression spring) may alleviate any issues here.

Print beds - It's hard to get a small, perfectly flat surface and I think this might increase with difficulty in a scale-up. It would add complexity but I think the preference would be to have a large static bed where the print head/s calibrate to the bed (like a delta does), as opposed to the bed moving. If the bed remains still we can consider heavier options.

Calibration - As mentioned above I think this might be an issue.

Speed - I think that more contemporary designs might not scale up so well. Speed will suffer (which is already going to be terrible given large print sizes). We might actually be looking at models which take longer to print than they do to design...

Flex - Use of heavier components to remove flex over the longer distances is going to be an issue. Motors and a whole lot of other things might need to be used.

Weight reduction - Certain components may need to be made more modular so that they can be supported elsewhere. For example, if it was going to be Bowden fed (my preference to reduce weight at the print head), it would probably be wise to have a connection from the roll of filament to a static point that can be bolted to the chassis, before further connecting to the print head. This would help reduce strain.

There's probably a few uses where dual ballscrews connected by gears or chain would be best used to reduce flex or twist, too.

Anyone else have ideas?

I think I'd like to start putting together a rough design... I've not built a standard XY type before, only a delta, and that's barely functioning, so it'd be nice to have input.

Oh, and I'm thinking it needs to be at least dual extruder, if not an E3D Kraken...

As for heavy plates. For me I am comfortable with Delta printers. If properly designed you can use some very heavy plates and use them under compression. You need to design a frame that is stressed and not simply placed. Using the materials in the proper spots will yield a exceedingly strong frame. Once you have that. You can design a light weight moving mechanism.

---

My Personal Blog. Build blog.
[engineerd3d.ddns.net]

Modicum V1 sold on e-bay user jaguarking11

Why timing chains ? do you need to go on a single way with heavy load ? A 3d printer is the opposite. Timing chains are for conveyors or bikes, or any unprecise application. You can get no precision with them ! Many reasons to don't use that :
- no smooth motion
- large pitch
- weight = low acceleration
- noisy
- needs to be greased often
- needs to be tensionned and controled often
- to much tolerance between links = no precision

Autolevel procedure won't replace a dead flat surface. It's just to measure and compensate the whole bed level position. If your bed makes a cup, you will never print a flat part. Simple as that. As you guess, a dead flat surface is exponentially expensive with size.

Edited 1 time(s). Last edit at 03/17/2015 05:35AM by Zavashier.

---

Collective intelligence emerges when a group of people work together effectively. Prusa i3 Folger (A lot of the parts are wrong, boring !)

Scaling up printer designs isn't really that hard, it just requires a different approach. The real challenge is extruding plastic fast enough.

Hobbyists regularly build 8'x4' routers that can hold within 0.010" fairly easily.

I'd recommend an H-Bot using timing belts ( [www.youtube.com] ). I'm not a huge fan of CoreXY in a build that uses linear block and rail (which you should be using). Alignment isn't that hard if you have the correct tools and experience (indicators, possibly a granite square and straightedge). It's not a particularly great idea to rely entirely on software for calibration.

Heated bed can be cut from Mic6 or similar tooling plate.

Rigidity/Weight/Damping is a fairly complex trade-off when designing something of this size. Steel plate is probably massive overkill, but if you only use it on non-moving components that isn't really a problem. You can achieve lightweight rigidity using aluminium tubes (or just large tslot extrusion). Resonance can be a problem in a welded aluminium frame, but if you are bolting a bunch of stuff together each joint will dissipate energy.

I'm hesitant to recommend delta for large build areas because the ratio of build to overall volume isn't great, and having a roughly cylindrical build area isn't efficient for most prints.

I've played with 16mm rod/linear bearings, v-slot, and the Delta-Pi linear motion setup using 623zz bearings on square steel tube. Out of all those think the best option for your scale-up will be v-slot. Smooth motion, quiet and can be rigid using 20-60 profiles. What I've found is that you can easily machine your own carriage plates out of 1/8" aluminum using a printed drill template from a standard inkjet/laser printer. Once you place the template on the aluminum blank, use a centerpunch to punch the holes, then drill. By using metal carriage plates with the v-slot, I believe you could have a pretty rigid, large machine. As 691175002 stated, an Hbot or CoreXY would be a good design.

Edited 1 time(s). Last edit at 03/17/2015 11:40AM by runninfarmer.

I think I can get away with using smooth rods on a larger envelope as long as I can cut threads at the ends of the rods and torque them. If you stretch them 100-200 micron the tension they should reduce the flex dramatically. As long as the strech is within the spring stage of the rods and not on the plastic/deformation strength.

---

My Personal Blog. Build blog.
[engineerd3d.ddns.net]

Modicum V1 sold on e-bay user jaguarking11

I don't get it. I'm nuts for suggesting that scaling up a printer isn't rocket science, but people are talking about 1.5 cm steel plates and pretensioned guide rails?

It really isn't that difficult. If you're worried about guide rails flexing, use fully supported rails or linear guides. If you want to make a rigid frame, use stuff that you know will work. I'm not a mechanical engineer. I don't know how much 1.5" square 8020 flexes over what distance, all I know is that when I built my printer (http://mark.rehorst.com/MegaMax_3D_Printer/index.html) I saw laser cut plywood printers that flexed every time I pushed on them a little. It was obvious that that sort of cheesiness would not do for scaling up to a 300x300x300mm build envelope. Then I saw a CNC machine made of 8020 and pushed on it and it didn't move and that was all I needed to know. I saw 8mm rods that flexed even in a small envelope printer so I went to 1/2" which were available for free at the time. I started with bronze bushings and quickly replaced them with linear bearings. I eventually replaced the end-supported rails with fully supported linear guides. I used NEMA-23 motors from the start because I figured the extra moving mass in the Y axis would need more than the torque a NEMA-17 motor could deliver.

The hard thing is not scaling up a printer. The hard things is sourcing the parts without going broke. I relied on the hack rack at the local makerspace, scrap yards, ebay, and friend's junk collections. A few things were purchased new because they weren't available from those other sources.

I don't know why anyone would think they need 1.5cm steel plates to make a 3D printer, and I have no idea how much they might flex, but I'm sure it would be strong/stiff enough. Overbuilding isn't the same as engineering, but it can get you to the end result you want. If you want all new parts in your printer, then by all means engineer it properly because that's the way you'll end up at the lowest cost (for a printer made with all new parts). You can build the same performance printer for less money and a lot fewer calculations by scrounging parts and making do with parts that are probably bigger than needed but available cheaply.

Quote
Zavashier
Why timing chains ? do you need to go on a single way with heavy load ? A 3d printer is the opposite. Timing chains are for conveyors or bikes, or any unprecise application. You can get no precision with them ! Many reasons to don't use that :
- no smooth motion
- large pitch
- weight = low acceleration
- noisy
- needs to be greased often
- needs to be tensionned and controled often
- to much tolerance between links = no precision

Autolevel procedure won't replace a dead flat surface. It's just to measure and compensate the whole bed level position. If your bed makes a cup, you will never print a flat part. Simple as that. As you guess, a dead flat surface is exponentially expensive with size.

Yeah, you're right, not a great idea to use chains. It was an idea, but not a good one. I just am worried that the amount of weight that will need to be moved may take a toll pretty quickly on GT2 or GT4 belt. Maybe there's a reinforced automotive belt that could be used? These are made for high speed and repetition under pretty serious circumstances, so should last a lifetime in a printer? Furthermore they're not directional.

In regards to bed leveling I was actually referring to making sure the bed is parallel with the horizontal planes, not so much whether it's flat or not. I think a flat bed is a requirement regarding whatever platform is used.

For a bed, would glass (6mm) bolted onto a sheet of aluminium (maybe 8mm?) be suitable? It's going to take a lot to heat up, but given the thickness it should distribute heat quite well. It might even be possible to use strategically placed resistors to heat the bed rather than an expensive mat of sorts (wow, a cost benefit to scaling).

I think the biggest concern for me would be to keep rods parallel to each other and the print bed over large distances. How would this be achieved? I'm guessing the parallel to print bed part could be achieved by software, as long as the bed is flat, but keeping rods parallel to each other may be a challenge. I want the same precision in this large envelope as there would be in a small one... Still aiming for 70cm long, 50cm wide and 50cm high (30x20x20 for those who prefer imperial measurements). I'd imagine overall size without filament would be at least 100x75x75cm. HUGE.

@the_digital_dentist (Mark?), that's an awesome printer you've made. You certainly know what you're talking about.

I think I'm a bit limited because there aren't really junkyards where I am. I guess the alternative that's relatively open now is Aliexpress and other cheap Chinese sources. The savings in electronics alone are phenomenal, but a lot of other bits and pieces should be relatively cheaper too.

If you're talking about a printer in which the bed moves in the Y axis, your biggest problem is going to be controlling the motion of that large moving mass, unless you don't mind very low print speeds.

Fully supported rails or linear guides can easily fix the sag problems, as long as the frame of the machine is sufficiently sturdy.

Edited 1 time(s). Last edit at 03/17/2015 08:52PM by the_digital_dentist.

Quote
the_digital_dentist
If you're talking about a printer in which the bed moves in the Y axis, your biggest problem is going to be controlling the motion of that large moving mass, unless you don't might very low print speeds.

Fully supported rails or linear guides can easily fix the sag problems, as long as the frame of the machine is sufficiently sturdy.

What would you recommend? A moving print bed or a stationary one?

Coming from a delta my inclination is a stationary print bed. For a printer this size I have a feeling it'd be easier to solve problems in moving a print head gantry than dealing with a whole print bed moving. It might even be worth mounting most electronics on the print gantry... with really the only connection being power. But getting that gantry to move in parallel with the print bed... that's an issue.

For a large envelope Cartesian machine, a bed that moves in the Z axis is the way to go. The mass of the bed won't matter because you only move the bed when the extruder is not extruding.

Engine blocks are generally rectangular shaped objects. A cartesian type printer's bed is usable over almost it's entire surface. To print something 600 x 600mm, you need is a bed that's maybe 620 x 620mm. I think the bed for a delta that can print that large has to be a bit bigger.

Either type of machine will have it's own set of problems to solve. For example, a cartesian machine can be built into an easy to build rectangular frame (nothing but 90 degree angles). A delta is usually built into a triangular frame (though you could use a larger rectangular frame). Trying to accurately cut and join parts at 60 or 120 degrees can be tricky.

You can try to reinvent the wheel or look at other machines and see what others have done. Once you start comparing them, you'll develop a critical eye for what is good and what isn't. The final reference should be industrial machines. Those are usually designed by guys who went to school for years to learn how to do that sort of thing and the machines are built for customers who are willing to pay big $ for high quality output. Copy their work and modify to suit parts available on your budget. Hunt through web sites for photos of industrial printers and look at how they mount things and how big things are.

The single most fundamental contributor to print quality is controlling the motion of the extruder nozzle relative to the print bed. That requires a solid frame, smooth, low slop bearings, and motors/electronics that have sufficient torque to do their jobs well. Those are the sorts things that are mostly missing from hobbyist grade printers because they drive up the price of the machine. This was the idea I started with when I designed and built my machine. I scaled up a reprap machine (at the time they were made from threaded rods and none of them had the build envelope I wanted), and eliminated plastic parts from the frame structure, a divergence from the reprap philosophy. To me it seemed silly to use plastic pieces to join much stronger, stiffer metal pieces when the metal pieces could just be bolted directly to each other.

I've been rebuilding my machine for about a month now and hope to have it up and running in its new form in the next few weeks: [plus.google.com] and [plus.google.com]

Edited 1 time(s). Last edit at 03/17/2015 08:54PM by the_digital_dentist.

I agree with most of what has been said in the previous posts. Choose either a delta design so that you have a fixed bed, or a printer design such as CoreXY that moves the bed only in the Z direction. I suggest you ask on the Delta and CoreXY sub-forums for advice on building large printers of these types.

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

I am a newbie, and this may sound stupid, by why not use a bridge (like on a cnc) with a stationary base?

Moving mass is very high. Printers generally operate at higher speeds than CNC machines (there's no resistance to the motion of the head- we're not cutting anything). High speeds and high masses don't go together- Newton's laws at work.

Quote
pugzor
Hey all,

I've almost finished a delta with a respectable print area (based on AndyCart's Cherry Pi III), but I'm already looking forward to my next project.

I was hoping to build something big enough that could potentially print an engine block. They aren't as big as you'd think. One of the largest in terms of length I'd require would be 70cm long for a large bore straight six. To accommodate a small-ish V8 I think it'd need to be 50cm wide. 50cm height would be more than enough too. I'd also like a design with dual extruders.

Does anyone know of an existing design with those parameters?

Imho these are unrealistic specifications for current FDM printing technology. Have you calculated how long it would take to print such an object? What is the largest (in terms of volume) or longest (in terms of time) print you have ever executed successfully in your delta with a respectable print area?

There are good reasons why other technologies are used for rapid prototyping of large parts, you should keep that in mind.

And I don't agree at all with the phrase "scaling up a printer isn't rocket science". Scaling up a 3D printer design is only possible within a very small margin, beyond that you really have to redesign the whole thing following good engineering principles. And costs escalate rapidly.

One small thing to keep in mind. Don't expect to print with a normal nozzle. At this point 1.0 or larger nozzle is the only way to go. You probably need more thermal mass for such a large nozzle. In my experiments it's possible to print with a 1.5mm nozzle with 1.75mm filament. The output will be in the 1.8mm range. 500 micron or taller layers Need to be extruded.

---

My Personal Blog. Build blog.
[engineerd3d.ddns.net]

Modicum V1 sold on e-bay user jaguarking11