what is practical/possible with regards to size?

I saw a build on here that had a build size of 300mm, at 3m tall.. but what if someone wanted a 600mm build size? Is that possible for someone who doesn't have access to special machinery? Or is it just no practical.?

I read a few posts from 2013 that said it wasn't possible, but that was a long time ago...

Quote
Qdeathstar
I saw a build on here that had a build size of 300mm, at 3m tall.. but what if someone wanted a 600mm build size? Is that possible for someone who doesn't have access to special machinery? Or is it just no practical.?

I read a few posts from 2013 that said it wasn't possible, but that was a long time ago...

Of course it is possible, the thing that is holding back is the diagonal rods.

For a 600mm bed they would have to be very long, ~800mm, and to find something that doesn't have much weight but is very stiff at those lengths.....

Titanium maybe?

Quote
Tha_Reaper
Titanium maybe?

Perhaps, but it would cost you, I don't think tubes would cut it, might even work with carbon fiber, just not tubes.

You could also go with existing material, carbon fiber tubes, but print very slow to keep the accuracy.

I dont think stiffness of the diagonal rods need be a problem, because stiffness of a tube increases very rapidly with diameter. So you can use larger diameter carbon fibre rods. Also increase the separation between them to increase the stiffness of the assembly as a whole.

Getting a flat bed is more difficult as it gets larger, so you probably need an aluminium tooling plate.

The other issue is that the height may become unmanageable. My 300mm printable diameter build uses 355mm arms, so I think you will need arms around 710mm long or a little longer. So the printer will be at least 800mm high plus your build height.

---

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].

You would need wider towers too. Something like 9045 extrusions.
The stepper would better be 0.9° at 1/32 or else you loose a lot accuracy with such long rods. Planetary geared steppers would work too.
A flat and level bed will be your biggest problem.
-Olaf

Diagonal rods are not a problem. Just get thin-walled but big diameter carbon fiber tubes. Notice that tubes are mostly loaded in their long axe and they will be strong enough there.

I do not see how big diagonal rod length decreases accuracy if you use the same pulley. So not opinion here. But the claim sounds suspicious.

Getting big flat bed will be a problem especially if you do not want to go with thick glass.

Getting stiff frame will be a problem.

And if you want to print quickly then getting stiff belts will be a problem too.

Firmware may have problem with overflowing when doing cartesian to delta transformations. This will result in complete malfunctioning or (if you are lucky) only decreased precision.

And the most important at the end: How do you want your ABS to stick to the bed. The parts which have big bottom will just curve and you get a destroyed print. Maybe PLA will not have this problem. There is a chance. It is not sure with such a big bed. But I do not know.

Edited 1 time(s). Last edit at 11/08/2015 01:19PM by hercek.

so, then, the things you need would require things to be custom made? like custom carbon fiber rods, I imagine that would be extremely expensive.

Quote
Qdeathstar
so, then, the things you need would require things to be custom made? like custom carbon fiber rods, I imagine that would be extremely expensive.

Seems my assumption on the rods is erroneous, the consensus is larger diameter tube is enough.

The flatness of the bed and ridgity of the frame is Important.

The bed/heating of is probably where the main part of the budget should be.

But I'm wondering a bit about issues of a flat bed, I mean many use a mirror to print on (not super expensive), l have also read about people just using cardboard as bed andand then print with raft.

Someone even suggested using some kind of insulation board (fairly cheap) that you just dig your nozzle in some mm and then print using a raft.

But sure if you want a cast aluminum tooling plate at that size you will probably have shell out a bit of money.

Quote
Qdeathstar
so, then, the things you need would require things to be custom made? like custom carbon fiber rods, I imagine that would be extremely expensive.

No, you can buy 16x14mm x 1m carbon tubes (outerDiameter=16mm, innerDiamter=14mm, length=1m). This should be plenty enough. Nothing custom need. 14x12 mm x 1 m will be probably enough and better (lighter). This is a standard material you can get in almost any RC model shop.

There have been a lot of Building Big threads here, not so many for Deltas.

My 2 cents worth: building 3x as big is more than 6x as difficult. Not to discourage you at all... Just make sure the extra effort to engineer and build it will be worth your application.

You have to think outside the box, none of the components that work for a Kossel may be appropriate, but you can still get off-the-shelf components.
The basic design needs to be completely rigid and enclosed - even if you are printing PLA at that size. Insulated if the chamber is to be heated.
Bearings must be better (linear) and fully supported
Belts must be wider (like 9mm) since they will be very long
Motors may need to be higher torque, perhaps double precision, perhaps nema 23 (however don't plan on more than .01mm per step resolution, or the motors will be running too fast.)
Motor drivers must be tune-able (or you will have resonance issues) - you will still have some resonance in the long belts.
The controller must be 32 bit like a Smoothie.
Yes, you would need a cast aluminum tooling plate (a mirror on top works fine to print on). I found one as a remnant.
Plan on a large bore hot-end of course. This printer would not be appropriate for printing 1 inch chess men or little frogs.
Make sure the extruder gear is well finished - not sharp edged. Prints that take dozens of hours must not be stopped by accumulating swarf in the gear.
From experience - a magnetic only connection setup for the effector is not strong enough for very long arms.

You're going to have to deal with ringing as well as motor resonance.

There are additional issues with any slicer when preparing huge models, and dealing with the longer bowden tube and pressure latencies. I've been working on a post-processor to deal with this for a while now.

For the arms - they sell carbon tube like this on Ebay as "Quadcoptor arms". Not very expensive.
I used 12mm OD tube for arms of about 400mm length, very rigid. As mentioned, 14mm would work fine.

---

My printer: Raptosaur - Large Format Delta - [www.paulwanamaker.wordpress.com]
Can you answer questions about Calibration, Printing issues, Mechanics? Write it up and improve the Wiki!

out of curiousity, why couldn't you just use multiple mk3 heat beds with a piece of glass on top?

what is the hardest aspect of going big?

The heated bed must be absolutely dead flat, combined with auto tower calibration. When you start printing an object that is say 350mm or more across then every part of that needs to be within about .04mm to stick properly (I aim for .02mm), and to give an even first layer. If it starts squishing up in one part or not quite sticking to another then the print is toast. The larger structure also will respond to heat, so the auto-calibration is absolutely needed.

So I don't think any material other than cast aluminum tooling plate will stay flat enough, regular aluminum plate will warp too much over that distance over time and with heat. Perhaps heaters could be bonded directly to a thicker sheet of flat glass, I dunno. I have some mirror glass cut to size, and removable so I can easily clean it and apply hairspray on occasion.

The hardest aspect for me was sourcing the materials, big requires more pro materials. Any part you customize will take more time to design, build, and debug. Every part of mine was customized, but I also work out of a maker space, and so had access to lots of tooling and 3D printers, and some help machining. But it still took 6 months to design, 6 months to build, and longer to tweak (still doing that, mostly post-processing software now tho).

Make sure your components are known good for the task, like a Smoothieboard, so you don't have to upgrade the controller first thing, an excellent high performance hot-end and extruder.

Equally difficult is the fact that you are pushing every boundary - extruder performance, hot end performance, controller performance, firmware, slicer, build accuracy, and printing physics. You will have to learn more than you wanted to know about it. That's how this all advances.

I'd advise making sure you have created a very good 3D model before you start (and build a smaller printer first), and have a group of folks check your design and offer suggestions.

I wish you the best.

Edited 1 time(s). Last edit at 11/08/2015 11:05PM by Paul Wanamaker.

---

My printer: Raptosaur - Large Format Delta - [www.paulwanamaker.wordpress.com]
Can you answer questions about Calibration, Printing issues, Mechanics? Write it up and improve the Wiki!

Quote
hercek
I do not see how big diagonal rod length decreases accuracy if you use the same pulley. So not opinion here. But the claim sounds suspicious.

The steps/mm might be the same for long or short rods, but the "delta translation" will be different.
The longer the rods are, the slower the stepper moves when the effector is close to the tower.
Slower steps means less steps/ accuracy compared to the same effector move in the center of the bed.

Try to print a small upright cylinder but place it close to one tower. Observe the difference between stepper moves of this particular tower and the other two.
-Olaf

Quote
o_lampe

Quote
hercek
I do not see how big diagonal rod length decreases accuracy if you use the same pulley. So not opinion here. But the claim sounds suspicious.

The steps/mm might be the same for long or short rods, but the "delta translation" will be different.
The longer the rods are, the slower the stepper moves when the effector is close to the tower.
Slower steps means less steps/ accuracy compared to the same effector move in the center of the bed.

I'm not convinced that is true. The number of steps that a stepper motor has to move for a given XY movement depends on the angle of the rod to the horizontal, but not its length. So as long as your scaled-up printer doesn't use shallower rod angles at the edges of the bed, diagonal rod length shouldn't affect accuracy at the edges of the bed.

I've been trying hard to think of reasons why you can't scale up a delta printer arbitrarily while maintaining the same resolution. The only ones I have thought of so far are:

1. Getting the bed flat enough. Can be solved by using an aluminium tool plate.

2. Ensuring that the rods remain rigid. Can be solved by using larger diameter rods, at some cost in extra mass.

3. Ensuring that the spider as a whole remains rigid. Can be solved by scaling up the diagonal rod spacing.

4. Ensuring that the frame remains rigid. Can be solved by using thicker extrusions for the towers and/or bracing the corners.

A risk is that one of the resonant frequencies of the spider assembly or the frame will be low enough to be excited by the stepper motor vibrations. It helps that the mass of the effector does not need to scale with size, because the effector can be a lightweight piece of plastic or PCB so that most of the mass is in the hot end and the fans.

This is all theory, because the largest delta I have built has a 300mm print diameter.

---

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].

Interesting reading as this is what I'm planning to build and stumbled across this thread on google,

looking for a 600x600 build volume.

As already stated the build platform would need to be strong, i was planning 6mm or 8mm tooling plate rough cost would be around £70

Arms i was already planning on carbon tube

I was thinking of using 20mm stainless steel smooth rods doubling as the frame this should eliminate the risk of bending and then using more 6mm and 4mm aluminium plate for top and bottom

Smoothie board for electronics

using the deltacalc page gives me the following specs, one question i have is the dead area below the arms. Im curious to know if the belts must extend below this? or could i use a jack shaft to reduce the main belt length and drive with a pair of shorter belts from the stepper?

Edited 2 time(s). Last edit at 11/09/2015 02:13PM by GR4EME.

Another thing to mention, at least from my own design.

The rods should not be part of the support structure. Design the case in 3 parts. Top/Bottom/Support structure.

Personally I have been thinking of a larger Delta for some time. I currently run 330x360 delta. My goal is 500x600 or there abouts.

I would like to use cast iron top and bottom plates. Black Cast pipe in 3/4inch x 6 to support the plates and create an even separation. Then use all thread rod to compress the top and bottom together ontop of the pipes. This should give me a very heavy yet very ridgit base to build on.

The next thing I would like to use is some 12mm hardened steel rods that are threaded on both ends and drill through the top and bottom plate. This will put the rods under tension, which reduces resonances as well as deflection. The carriages will have to be machined aluminum holding lm12uu bearings that are compacted with axle grease. All spherical ball ends on the design.

Double 6mm gt2 pulleys and belts in parallel in each axis would reduce some stretch. Tension mechanism should be on the bottom of the printer, with nema 23 motors ontop. All powered with CNC style stepper controllers and a 32bit board ontop.

The end effector should be machined aluminum with the hot end directly mounted on it. No heatsink. if properly positioned the bowden tube should be no longer than 60-70cm.

One more thing is that the build plate would have to be glass with 2-3mm aluminum heat spreader and either 4 PCB beds or a single silicone bed. 1200w server spec psu with a bunch of boost converters converting 12v power to 30v for the beds and electronics. (iffy if I can get away with this arrangement) The bed control maybe a DC-DC SSR of good known quality.

All said and done nearly 2k worth of parts and many gray hairs is what I envision, however the outcome would be a fast larger scale 3d printer that can have enough surface area to 3d print a full size human bust. I honestly feel that a delta 3d printer may not be the best platform to go big. I think CORE XY is a better platform overall for such a project.

---

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

Modicum V1 sold on e-bay user jaguarking11

@GR4EME
If your motors are on the top, then the idler pulleys can be mounted just below the lowest point the cars will go.
I would avoid any extra belts and connections, as that will inevitably add lash and complexity.

Quote
Paul Wanamaker
@GR4EME
If your motors are on the top, then the idler pulleys can be mounted just below the lowest point the cars will go.
I would avoid any extra belts and connections, as that will inevitably add lash and complexity.

Thats another option, that thought had crossed my mind but i can't think what stopped me and got me onto the more complex jack shaft idea lol

Quote
jaguarking11
... Tension mechanism should be on the bottom of the printer, with nema 23 motors ontop. All powered with CNC style stepper controllers and a 32bit board ontop.

Check the maximum step rate and minimum step pulse width before you choose external stepper controllers, especially if you intend to use 0.9deg/step motors. CNC machines don't move nearly as fast as delta printers.

A Duet run from a 24V or 30V supply could drive some types of Nema 23 motors directly, for example these [www.omc-stepperonline.com].

---

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].

Regarding stepper resolution vs. rod-length:

I made a little drawing to explain my thoughts:
Two triangles with same angles but different "rod-length".
Both make a move of 10mm horizontally.

I find the result pretty obvious, right?
-Olaf

o_lampe, your picture is not valid since there is a significant diagonal rod length difference in the smaller version of the sketch. It depends on DPI but on my monitor the lower rod is long 104 mm and hibher one 99 mm. The rods should be the same.

Anyway your point seems to be that you want to show us that for a small change of x or y (i.e. the planar dimension at the bed) there needs to be a bigger change in tower position when the printer is small. Therefore the smaller printer does need less resolution at the belt movement.

In other words, you want to show us that a partial derivative of tower position function based on x or y dimension is different for a small printer compared to a big one. So, lets look whether it is true:

a(x) := √(r² - x² - y²)   // function defining tower position based on planar (x,y) position
∂a(x) / ∂x  =  - x / √(r² - x² -y²)   // partial derivative with respect to x

Now notice that when you scale everything two times then the a(x) is also nicely scaled two times (as it is expected):

√((2 r)² - (2 x)² - (2 y)²) = √(4 (r² - x² - y²)) = 2 √(r² - x² - y²) = 2 a(x)

But the partial derivative is not scaled:

- 2 x / √((2 r)² - (2 x)² - (2 y)²) = - 2 x / √(4 (r² - x² - y²)) = -2 x / (2 √(r² - x² - y²)) = -x / √(r² - x² - y²) = ∂a(x) / ∂x

Therefore for small movement in the x or y direction on the proportionally same place in the XY plane we need the same movement on the tower regardless on the printer size.

I would be glad if somebody can find an error in the above reasoning. Because this reasoning was the source of my claim that a need for a smaller (micro)step size in a bigger printer is suspicious.

My bad, I actually misread the length of the small rod.

All taken back
-Olaf

What angles are we looking at on the joints when printing with a large delta?