Delta printer build vs "buy"

So i've decided to try my hand at building a delta printer. For a lot of reasons not really worth going into at the moment I want to design my own but my experience with 3d printing in general is limited to the last month and change that I've spent with my ff creator pro. Part of me thinks, screw it, I'm a reasonably intelligent guy with an engineering background and I have the interwebz to refer to if I get stuck, design my own and make it how I want right from the start. The other part of me is a bit more reserved and thinks that I might be better off downloading parts from an established design and simplifying my first build to replicating what others have already done and then work on designing my own from there. So there is my debate. Buy/print the parts to build a clone of what's already out there and just copy/paste everything I need to make it work, or design my own and get it done the way I want the first time while accepting the higher probability of failure? Those of you that have built a delta, how difficult is it really? Was yours a "kit" or did you design it yourself? Based on the experience you gathered from your build what would you do differently if you were to do it again?

For the hell of it here are some of the goals for the one I plan to design myself:

12" triangular heated build platform (no apparent reason to limit myself to a cylindrical build area)
12" MINIMUM build height
lead screw motion
80/20 extrusions (much easier for me to get than the openbeam stuff)
active print cooling (switched)
enclosed build area with active cooling (switched)
power supply and other electronics housed in base for a cleaner finished look and easier transport
provisions for 2nd extruder (will not be part of the initial build)

one thing I'd add to a self design is a flexible drive shaft to the print head. That way you can keep the motor on the frame but have a direct drive feed on the head that's really lightweight. Dual extruder maybe.

I saw this done at the tct show.
It was one of these guys who'd built it: http://www.wedo3dprinting.com/

Lets you have use 1.75 filament. Gives you the control of a direct drive with the lighweight design of a bowden setup.

I'm sure if you email them they'll tell you where to get the driveshaft.

Quote:

---

Originally Posted by curious aardvark

one thing I'd add to a self design is a flexible drive shaft to the print head. That way you can keep the motor on the frame but have a direct drive feed on the head that's really lightweight. Dual extruder maybe.

I saw this done at the tct show.
It was one of these guys who'd built it: http://www.wedo3dprinting.com/

Lets you have use 1.75 filament. Gives you the control of a direct drive with the lighweight design of a bowden setup.

I'm sure if you email them they'll tell you where to get the driveshaft.

---

Didn't see anything on their site about that but it's an interesting idea for sure. Definitely worth exploring further, thanks for the idea.

Not on their site - but one of the guys in the company had built a printer with a flexible drive shaft and they had it with them at the tct show in october.

The SeeMeCNC RoStock kit is very nice. I bought the Assembled Delta Orion and I love it.

Well my design is well underway. I'm on effectively my 4th or 5th iteration. All my earlier attempts would have worked but were, frankly, ugly as hell. This one seems to be going well and I'm pleased, if not necessarily ecstatic about the aesthetic quality of it. A quick teaser rendering of the chassis is attached, obviously I have a lot of work left to do.

I've done a bit of searching but haven't found much useful info, can anyone tell me or point me in the direction of how to calculate the length of the arms and/or the angle the arms should be at relative to vertical when all three carriages are at the same Z? I assume that an absolute minimum arm length would be whatever is necessary to reach from a given carriage's pivot points to the associated effector pivot points with the arm horizontal to the build plate and the print nozzle at the extreme edge of the build platform. That would result in an angle of about 45° when the carriages are all at the same position. Is there any reason to make the arms longer than that other than to improve movement speed? Things I've found so far:

1) longer arms = more error
2) longer arms = less print height for a given machine height
3) accuracy improves as the arms approach horizontal because the carriage moves farther for less effector movement
4) shorter arms = reduced movement speed as more carriage movement is required for a given amount of effector movement

Longer arms does not mean more errors, it's the fact of control on tolerance and the use of uneven length of arms. Any well picked arms will work just fine.

Speaking of error caused by the arms, it's probably a fraction compared to the end results when you print. Shrinking is the killer and the error can be as big as 10 times your accuracy of positionning.

Here's my tip, if you design a delta, make sure it's overconstrained when it comes to the carriage running up and down on the tower. As well as the arms attaching to the hotend device.

Quote:

---

Originally Posted by richardphat

Longer arms does not mean more errors, it's the fact of control on tolerance and the use of uneven length of arms. Any well picked arms will work just fine.

Speaking of error caused by the arms, it's probably a fraction compared to the end results when you print. Shrinking is the killer and the error can be as big as 10 times your accuracy of positionning.

Here's my tip, if you design a delta, make sure it's overconstrained when it comes to the carriage running up and down on the tower. As well as the arms attaching to the hotend device.

---

Maybe error was not the correct term, mathematically longer arms would result in less resolution because for a given amount of vertical movement, there would be more horizontal movement with longer arms vs. shorter. Basically the math tells me that the closer to horizontal the arms are, the more vertical motion is required to obtain the desired horizontal motion. More vertical motion = more rotation of the stepper motor which means more steps are being used which means higher resolution because I don't have to rely on microstepping which isn't necessarily precise. As for controlling the length of the arms relative to each other, that shouldn't be too much of an issue.

As for your tip on over constraining, I'm not sure what you mean. To me, over constrained means failure. Telling two things they are the same length and then telling one that it's 1 inch while the other is 2. That would be an example of an over constrained condition at least as far as I understand the term. As far as ensuring "pure" vertical movement and solid fixing of the arms to the carriages and effector, that's being dealt with. Rather than the usual slide bearings that I see a lot of in homebrew printers I will be going industrial on mine. I can't afford preloaded ball bearings so I'm going with high precision plain bearings. Based on what I've seen of them in use in my shop they are as precise, suffer very little wear even in extremely abrasive environments, have minimally more resistance to sliding, and carry the same or greater axial and cantilever loads as ball bearings do. Not to mention I can get 4 of them for the cost of a single ball bearing. There will be 4 in each carriage riding on 2 rods. For the moment I've spec'd out 3/8" rods as a compromise between rigidity and excessive size but I have the option to increase that and/or go with open bearings and a fully supported rod if I decide it's needed.

Movement will be controlled by acme lead screws with doubled up nuts in an adjustable housing so I can set preload on them and eliminate backlash. I will be using heim joints to connect on both ends but I'll be using aluminum rod ends so they will be stiffer than the usual plastic bits. I thought about going industrial with the joints as well but even the smallest ones are rated for such extreme loads that I'm concerned they will introduce excessive resistance to motion on the system. As for the rods themselves, I haven't decided on a material or source yet. I would love to go with carbon fiber for the low weight, extreme rigidity, and cool factor, but I have to see if I can find stock material in the length and diameter that I need. If not I'll go with aluminum.

Oohh ,if you talk about the resolution, again, should you be worrying that much? I don't think so, because you're likely going to use NEMA 17 stepper motors(1.8deg) and you can do microsteps. Even that, depending on if you use belt or string then it may or may not matters. But keep in mind that your nozzle will likely cripple all movement and you will have srhinking that outcomes when speaking of error/tolerance.

When I was mentionning overconstrained, I did not mean on a 3D cad perspective, but what you mentionned about the number of rods for each carriage. Because having myself with 2 Delta self assembled, I see flaw where the hot-end mechanism wobble.

I designed my laser burner printer based on 3/8th rod I found from inkjet printers, the linear bearing cost for this typical dimension cost way too much and I have to scrap the whole design. I decided to downgrade to 8mm shaft and use LM8UU bearing.
The cost for the linear bearing drop from 15 to 2-3$/unit.

I like your design so far. We built a couple of Kossel Deltas, we printed all of our own plastic parts and sourced out the hardware to various vendors. Both have heated build platforms and both print very well. We have pony printed PLA on these machines because they are open framed and we already have six different closed frame printers for ABS.

Of course, the downside on the Kossel is the Power Supplies do not fit under the machine. We also have our Ramps Boards mounted on the outside of the frames so we can have active cooling although I have seen some Deltas with the boards mounted under the Build Platform

Quote:

---

Originally Posted by richardphat

Oohh ,if you talk about the resolution, again, should you be worrying that much? I don't think so, because you're likely going to use NEMA 17 stepper motors(1.8deg) and you can do microsteps. Even that, depending on if you use belt or string then it may or may not matters. But keep in mind that your nozzle will likely cripple all movement and you will have srhinking that outcomes when speaking of error/tolerance.

When I was mentionning overconstrained, I did not mean on a 3D cad perspective, but what you mentionned about the number of rods for each carriage. Because having myself with 2 Delta self assembled, I see flaw where the hot-end mechanism wobble.

I designed my laser burner printer based on 3/8th rod I found from inkjet printers, the linear bearing cost for this typical dimension cost way too much and I have to scrap the whole design. I decided to downgrade to 8mm shaft and use LM8UU bearing.
The cost for the linear bearing drop from 15 to 2-3$/unit.

---

Microstepping imo is not a reliable method of positioning. Yes it works to a degree but the greatest precision comes from using the steps of the motor directly. I will be using 17's which gives me 200 steps per revolution. Couples with .1 pitch lead screws gives me 2000 steps per inch or a potential vertical resolution of .0005" without microstepping. I don't care to do the math to figure out how the vertical resolution relates to horizontal at the moment, it's been a rough day and I'm a little burnt. I realize that I will have material shrinkage in my prints but that can be compensated for fairly easily IF the machine is built properly and can print repeatably. For instance, a 1% shrink which is typical for ABS means all I have to do is scale my parts to 101%. I don't need half thou accuracy in my prints but I use my printer(s) for prototyping purposes first and trinkets second so accuracy is important to me.
As far as the linear motion itself, acme screws with doubled nuts as I mentioned, industrial rods and bearings, and so on. It's not going to be a cheap build by any stretch of the imagination. I'm guessing somewhere in the area of 1k US by the time I'm done just to buy the parts I need (not including printed parts that I will make on my existing machine). So on par or above the cost of many fully assembled machines but with vastly superior components. Maybe that will translate into a better printer, maybe not. Time will tell I guess.

Quote:

---

Originally Posted by JohnA136

I like your design so far. We built a couple of Kossel Deltas, we printed all of our own plastic parts and sourced out the hardware to various vendors. Both have heated build platforms and both print very well. We have pony printed PLA on these machines because they are open framed and we already have six different closed frame printers for ABS.

Of course, the downside on the Kossel is the Power Supplies do not fit under the machine. We also have our Ramps Boards mounted on the outside of the frames so we can have active cooling although I have seen some Deltas with the boards mounted under the Build Platform

---

Part of my design criteria is to fit the power supply under the machine. It means a taller base and a taller machine but a cleaner finished product and one that is easier to situate on a desk and/or transport. The board I haven't decided on a mounting solution for yet. It may live under the machine as well with a fan ducted to outside the frame but I'm still working on the basic mechanical design so I haven't gotten to the electronics yet. The frame is being designed with the intent of closing it off so I can print with whatever I want. I haven't decided on hinged or fully removable panels yet though.

Progress continues to be made. I was on the verge of saying my basic mechanical design was complete moments ago. Unfortunately I've decided to make some concessions to off the shelf parts since the project cost really is getting up there and I don't want to be dealing with custom made industrial parts ($$$$$$$). So based on those concessions and design choices I've already made I spent some time checking my available build volume and discovered that it's not what I was hoping for. It's fairly close but despite starting with a layout sketch that gave me a 12" build circle, the combination of motion control components and effector has left me with just under 10.5". Build height is a respectable 14" so I'm not terribly upset about that though I would have liked to get a bit more given the height of the machine. I'm won't exactly be starting over but I will definitely be increasing the size of my base sketch to boost the overall machine size and get me the 12" circle that I wanted as a minimum. Putting together this model has definitely been an interesting process and I can certainly see why many start with an established design and/or kit before trying to build one of these on my own. For sure if I didn't have the ability to put it together in CAD first I would not have gotten this far without getting horribly frustrated and giving up. Anyway, here's what I've got so far. The finished version will be nearly identical other than a larger footprint. I may also move the arms closer together on each carriage as that directly affects the size of the effector plate which in turn helps to determine how much of the footprint is space that I can't print on. Also, the effector itself as shown is nothing more than a place holder for now. Working towards maximum build height I am positioning the arm pivots down by the nozzle rather than hanging the hot end as most designs do. This effectively increases total build height by the height of the hot end so it's a significant difference.
Attachment 3805

Having built now several Kossels, both mini and full size and currently in the middle of a supersize.. the one thing I would honestly suggest is a round build plate. Think about it this way..

Alot of the printing you do.. think about it relying on the size of the largest square that will fit on your build plate. Now - the question I have (and it is a question...correct me pls someone.. ) is would you not be able to fit a bigger square on a round build plate than a triangular one?

The round plate does not add any width to the machine, so I am just trying to figure out why you want to go triangular? I mean sure it will work but if you use the Delta calculator, it seems like you might actually be cutting yourself short on available build area by going triangular.

Quote:

---

Originally Posted by Geoff

Having built now several Kossels, both mini and full size and currently in the middle of a supersize.. the one thing I would honestly suggest is a round build plate. Think about it this way..

Alot of the printing you do.. think about it relying on the size of the largest square that will fit on your build plate. Now - the question I have (and it is a question...correct me pls someone.. ) is would you not be able to fit a bigger square on a round build plate than a triangular one?

The round plate does not add any width to the machine, so I am just trying to figure out why you want to go triangular? I mean sure it will work but if you use the Delta calculator, it seems like you might actually be cutting yourself short on available build area by going triangular.

---

It's debatable,some people take in consideration the type of printer before throwing the file. There was one guy who knew I was dealing with a rostock max with 11" diameter bed. The guy went straight and shoot 9x9" square parts, and did not consider the diagonal size when he asked me to print it. (Diagonal about 12,7" )

It is misleading when you use a round plate.

Quote:

---

Originally Posted by richardphat

It's debatable,some people take in consideration the type of printer before throwing the file. There was one guy who knew I was dealing with a rostock max with 11" diameter bed. The guy went straight and shoot 9x9" square parts, and did not consider the diagonal size when he asked me to print it. (Diagonal about 12,7" )

It is misleading when you use a round plate.

---

Sorry, how is it misleading? would it not depend on the shape of the object you are printing?

I suppose it all depends on what you print mainly.. For me it's alot of busts which tend to be an odd shape.. not round, not square... So to have a build area of 180x180 instead of 220x150 rectangular like say my flashforge tends to be more efficient as I can actually print larger busts.. and I am more talking the higher layers.. where the head of the bust is much larger than the base of the model and you are trying to print at max scale for your printer.

But each to his own I guess, we all have our preferences, I am not saying one is wrong and one is right, I have 4 different beds on my machines now, a 225x150, a 200x200, a 160x160 and a 180x180, and am now building a Kossel that uses a 500x500 build plate.

The way I figure it, for a given size of delta there is a maximum size circular build plate that will fit which is the circle that fits within the confines of the three sides of the machine. This assumes that you do not want the build plate to extend beyond the edges of the machine itself (I don't). So with the machine dimensions as the constraint, the circular build area is defined. BUT, once you put that circle within the confines of the machine you have small areas between the circle and the vertical frame members that are essentially wasted space because there's nothing there for the machine to print on. By filling those areas in you increase the build size (marginally to be sure) without increasing the machine size.

You are correct in that a round plate could be used that extends outside the bounds of the machine base without increasing the overall width but unless that plate is positioned off center it will also extend off the front (rear?) of the machine which would increase it's depth. You are also correct that a round area will fit a larger square than a triangular area IF the round area is defined by the three points of the triangle rather than the three sides of it.

Good points overall, it seems I could have been clearer in describing some of my design goals, especially as they pertain to the aesthetic/functional compromises that I am making. I want this machine to work well but imo, it must look as good as it works or there's no reason for me to put the effort into designing it myself.

Bumping this with a bit of an update. My mechanical design is, as far as I can tell, finished. The frame, carriages, and effector are all fully defined and ready to go. My original plan to use spherical joints isn't going to work as the joints simply do not have enough freedom of motion to give me the build area I want. In one plane they're fine but on the perpendicular plane their maximum angle is too limiting. My solution was inspired by the yoke joints on the rostock max. My issue with that design is that it relies on a precise fit between the metal pins and the plastic holes. Too tight and excessive friction results, too lose and things can wiggle. Either way you have a wear point which will introduce progressively more error as the machine ages. I've gone with ball bearings, 2 on each end of each arm. Since they can only move on a single plane, adding a second perpendicular to the first gives me all the freedom of motion I need, with precise control, and minimal friction.

The effector is multi-piece and has provisions for a cooling fan and duct (not yet designed). Vertical motion is handled with acme screws and constrained by two case hardened parallel shafts per carriage. The carriages themselves each contain 4 linear plain bearings with one side being a pair of precision fit parts and the other side being compensated to allow for slight misalignment of the shafts. The double stack of bearings ensures that the carriages can handle hugely more cantilever load than should ever be applied to them. Each carriage also has an adjustable preload flange so I can tune out backlash.

Lastly, the extruder is nothing extraordinary other than being designed to integrate into a matching horizontal extrusion. I'm debating right now between leaving it as an additional element of the frame or replacing one of the top rails with it. Unfortunately, to do that I will either have to space the top rails a bit further apart to make room for the tension arm, or redesign the extruder. The first option isn't horrible, the second one I'd like to avoid. The extruder itself mimics most other designs except for the internal 90° turn that the filament will make prior to reaching the hobbed bolt. While this likely will make my design useless for stiffer materials like carbon fiber fill, it eliminates the need for a loop of material to extend out to the side and/or above the machine in order to make the turn from the side mounted spool, down to the print head. An internally routed ptfe liner should help to reduce any possible scraping or snagging while the filament makes it's turn.

Overall I'm fairly pleased with the theory of my design. I have no idea if some of my ideas will work in practice or not. The good news is that the questionable items are largely printed parts so if I have to iterate it won't be terribly expensive to do so.

http://i135.photobucket.com/albums/q...psfa256298.jpg
http://i135.photobucket.com/albums/q...ps2e90a16f.jpg
http://i135.photobucket.com/albums/q...ps2f9c8b36.jpg
http://i135.photobucket.com/albums/q...ps1b69c384.jpg

Now for some questions to all of you. With the mechanics basically done it's time to turn to the electronics. From what I've seen, the ramps/mega combo seems to be everyone's go to option for controls. It's cheap, and parts are everywhere, both of which are excellent qualities. The question is how well suited is that setup for a delta machine? I definitely want an LCD, SD reader, and onboard controls for untethered printing. Rumba seems to be another option that is similar to ramps. Assuming cost isn't a factor but support is, what other, better options do I have or is the ramps/mega setup my best bet?
I'm basing a lot of my work on the assumption that I can download an existing delta compatible firmware and simply load it into whatever controller I end up with rather than having to write my own. Is this a valid assumption? If I have to write my own software this whole project is dead in the water. It's not that I can't learn it and make it happen, but I don't care to put that kind of time into it only to end up with unproven software running an unproven machine. One or the other I'm ok with but if I don't have a known good in the equation, troubleshooting is going to be a problem.

Quote:

---

Originally Posted by Geoff

am now building a Kossel that uses a 500x500 build plate.

---

Not sure how I missed this previously. What are the overall dimensions of your new build that it will fit a 500x500 plate? That's an impressive size.

If you are experienced and like to fix, read manuals etc, go with and buy a kit, but u will need much time to know exactly how to assemble it and calibrating it!