New Delta Printer - Look Ma, no rails!

Here's a first look at a delta printer I designed that instead of using linear rails and fixed length arms, it suspends the carriage from 6 fishing lines. Each pair of fishing lines wraps up directly on the same motor shaft. It also uses 3 spring-loaded posts to press down on the carriage to keep the strings in tension. Before putting on the extruder, I tested the mechanism with a single post pressing down in the center, and it worked incredibly well - you could push the carriage way out to the side with no wobbling. Unfortunately, with my fat extruder in the center, I had to use 3 outboard posts which don't work nearly as well. This is really crying out for a Bowden extruder so that I can put the post back in the center, but in my impatience, I slapped on the extruder I had.

I have no idea if this is a good idea or not, or if the fishing line will break after an hour, but it does print and it can be incredibly cheap to build.

BTW, I have seen Brandon Heller's SkyDelta prototype from 2013 which is very similar, but I haven't seen if he has gotten it to print. Does anyone know if he's completed it?

Youtube video




Edited 1 time(s). Last edit at 06/25/2015 12:49AM by LoboCNC.

Nice experimental design!

However, your video does not show how the wood poles are spring loaded and I wonder how you manage the Z-height of the effector, is that by simply compressing the springs that load the wood poles?

I guess you have encountered the same problem as Brandon Heller's SkyDelta, which is how to keep the effector plane parallel to the heatbed, something not exactly easy when using fishing lines instead of rigid arms.

Great effort Jeff....I'd like to see how a Bowden tube setup works and more details on your design....thanks!

Quote
AndrewBCN
Nice experimental design!

However, your video does not show how the wood poles are spring loaded and I wonder how you manage the Z-height of the effector, is that by simply compressing the springs that load the wood poles?

Thanks! Here are a couple of photos of the target design using a single tension post -- without the extruder. (My temporary 3 post setup is too embarrassing to post photos of.) It is a little ungainly, but as you can see, it works pretty well because it automatically adjusts its angle to apply a biasing force in the right direction (which isn't simply always downward). And the platform stays rigid through a wide range of motion. However, it is quite critical that the contact point on the carriage be in the center and down low as possible. Hence, I'm now working on a Bowden extruder that sits underneath the carriage and that feeds the filament in from the side. (We'll see how well that works!)

You'll also notice the set of guitar tuners on the top plate. There's not a lot of room on the motor shaft to wind up the fishing line, so getting all the string lengths just right was a total pain. With a $5 set of guitar tuners, though, I am able to wrap up arbitrarily long lengths of slack at the fixed ends very precisely. The downside is that it is very tempting for people to twiddle your tuners, putting the printer out of whack.




Edited 2 time(s). Last edit at 06/25/2015 01:20PM by LoboCNC.

Very impressive and 100% creative, original solution! Hats off to you, this should win a prize!

That is all sorts of crazy. I would never have guessed it would work. Congratulations!

Quote
JamesK
That is all sorts of crazy. I would never have guessed it would work. Congratulations!

Thanks! But the jury is still out as to whether it will really work as a viable printer, let alone as to whether it is a good idea. To figure that out, I need to design a right-angle Bowden extruder because a normal Bowden extruder would get in the way of the tension post. Has anyone tried building a hot-end with the nozzle at right angles to the filament feed? Seems like it ought to work as long as the filament has melted completely before taking the corner.

Does the hot end really need to be centered

The E3D Cyclops has the filament feed offset at right angles to the print nozzle. So yes that approach does work. Also the CEL-Robox printer feeds the filament 90° to the nozzle. Lay the Cyclops on its side where the Bowden tubes feed horizonal to the bed; drill and tap the nozzle 90° to the stock location and set it all on a lower stage effector plate. Should work well.

Edited 1 time(s). Last edit at 06/25/2015 11:21PM by simspeed.

I wonder what the life-cycle of the line will be at the effector wear points..?

Quote
simspeed
I wonder what the life-cycle of the line will be at the effector wear points..?

That's the $64,000 question. Right now, I've got the 0.010" dia fishing line running over 0.035" dia polished wire eyelets. I expect there will be wear from just the rubbing, which is load dependent, and wear from the flexing and unflexing, which is more bend radius dependent. I may try putting some PTFE sleeving over the wire which would both increase the diameter some and reduce the rubbing wear, although I'm afraid the line may cut into the PFTE. Once I get a proper extruder, I should be able to do some reasonable life testing. (Fortunately, the braided fishing line shows visible fraying before it actually breaks!)

Quote
thetazzbot
Does the hot end really need to be centered

In theory, no. But inaccuracies in the kinematics can cause the carriage to rotate slightly, and the effects will be minimized if the extruder is centered. The bigger problem, though, is that the tension post sweeps out a pretty large area above the carriage, so having the filament coming in anywhere from the top will probably interfere.

I'd be more inclined to use a base pivoting micro pulley machined out of Delrin or even aluminum. They make them pretty small.

[static.rcgroups.net]

I'd think maybe a little heavier line too.

Quote
simspeed
The E3D Cyclops has the filament feed offset at right angles to the print nozzle. So yes that approach does work. Also the CEL-Robox printer feeds the filament 90° to the nozzle. Lay the Cyclops on its side where the Bowden tubes feed horizonal to the bed; drill and tap the nozzle 90° to the stock location and set it all on a lower stage effector plate. Should work well.

Thanks for the references. Good to know that I should be able to make it work.

Quote
simspeed
I'd be more inclined to use a base pivoting micro pulley machined out of Delrin or even aluminum. They make them pretty small.

[static.rcgroups.net]

I'd think maybe a little heavier line too.

Wow, those are quite cute. The one thing, though, that I don't have a good feel for how much a non-zero diameter pulley will distort the "idealized" kinematics. Could be insignificant, or it might require some level of correction in the kinematic equations. I doubt that my Printrboard has enough spare cycles perform much extra math, but going to an ARM processor would certainly make this feasible. Also, part of my motivation for doing this is to see just how little "stuff" is required to make a printer. However, if you wanted to scale this concept up to a very large printer, real pulleys and heavier fishing line would certainly be the ticket.

[www.modelbox.info]

Here's an even smaller pulley that's structured to work with your existing pins. Your point about the zero diameter pulley seems logical of course, but looking at the physical layout the pull is always consistent between the two transition points. Even if you had a large diameter pulley at each point the pull center line stays the same if the pivot stays the same at each end. Yes...?

Yes, I like these little brass pulleys. With the eyelet free to pivot, it would remove most of any distortion from an ideal path. There will still be some deviation in the overall string length as the whole pulley pivots side-to-side, but probably minimal. (I should actually do the math, but it'll make my head hurt.) If fishing line wear becomes an issue at all, I'll switch to these.

One other question for everyone - any ideas on homing? Right now there is nothing really to mount a limit switch on.

I'd think about using a Hall sensor mounted to the top plate and a magnet or magnetic tape on the center pole for TDC location.

Definitely the Innovation of the month! And a nominee for the anual award
I like the simplicity of the concept, especially the low count of "special parts". ( The RepRap-blues )
I hope you/we can make up a version where these parts are printable.

I´m not sure, if I fully understood the principle, but if you´d print a tall part, the fishing line would wind up on the stepper shaft completely?
I thought about the growing "shaft diameter" if the fishing line builds up layer after layer on the shaft.
Would that cause a nonlinearity in z-movement?

At the same time the springload would raise and cause more and more friction to the center rods ball. This may cause binding.
Olaf

Quote
o_lampe
I hope you/we can make up a version where these parts are printable.

I originally planned on 3D printing top and bottom plates as 3 pie-shaped pieces that screw together, but realistically, laser cut acrylic is way less work. Laser cutting services for acrylic are pretty readily available these days, although not as fun as printing all your own parts.

Quote
o_lampe
I´m not sure, if I fully understood the principle, but if you´d print a tall part, the fishing line would wind up on the stepper shaft completely?
I thought about the growing "shaft diameter" if the fishing line builds up layer after layer on the shaft.
Would that cause a nonlinearity in z-movement?
Olaf

One big downside to this printer design over a standard delta is the limited range of motion in Z. In a standard delta, you can pretty much make the printer as tall as you like without major accuracy issues. With this design, as Z grows taller, the lines become much more vertical when at the bottom of the build volume, and your X-Y accuracy decreases. And as you point out, of course, you run out of room to wrap up string in a single layer. Overall, this printer is 11.5" in diameter and I am shooting for at least a 6" dia. build area and 5" to 6" build height.

Loss of resolution as the Z lengthens is a deal breaker I'm guessing. Of course you can move the bed through the Z range and keep the print head in the sweet spot, but that adds costs and complexity to what now is a nice and simple printer. Still Jeff...I'd be interested in seeing what can be accomplished with your idea as it stands. Keep up the good work!

Quote
simspeed
Loss of resolution as the Z lengthens is a deal breaker I'm guessing.

The loss of resolution is really not as dire as it seems. With the way I've got it strung up now, I'm starting with about 5x the resolution of a standard delta configuration. The "sweet spot" is actually reasonably large, but it's just not unlimited in the Z direction like with a standard delta. The real limitations with respect to this being a practical printing solution have more to do with the fishing line stiffness and the the fishing line wear. Not to mention the overall fussiness of having strings running everywhere.

Very, very nice design. Simple and innovative. Respect!

---

Blogs:
Meine 3D Druck Abenteuer
[3dptb.blogspot.de]
FLSUN Delta Drucker für Deutschland
[flsun-deutschland.blogspot.com]
Books on 3D patents:
[goo.gl] (english)
[www.amazon.de] (deutsch)

I've managed to fashion a Bowden extruder for this printer, which in turn allowed me to use the single tension post in the center of the flying carriage. Overall, it works much better. There are some stringing and blobbing issues to sort out with the extruder (this is my first time working with a Bowden setup), but otherwise the print quality is not too bad. You'll also notice in the video that I got the right-angle hot end working without any major grief.

Edit: I almost forgot - so far 3+ hours of printing and no worn (or broken) fishing lines. It's a start.

youtube video



Edited 1 time(s). Last edit at 07/03/2015 01:30AM by LoboCNC.

Nice print!

About the homing issue:
Can you add black tape marks on the fishing line, so when they run through an optical endstop, will trigger the home position?
Not very accurate, but a starting point
-Olaf

Excellent quality I'd say... Nice work indeed.

Wow LoboCNC, nice work! This is just fantastic!

About SkyDelta: Brandon Heller is currenty investing all his time in a startup and doesn't know if/when he will finish the SkyDelta.

How do you deal with line building up on your spools?

I'm currently doing a master's thesis on another cable driven parallell RepRap very similar to yours, check it out:
Blog (occational status reports, some pics): [vitana.se]
Repo (both scad-files and unfinished Marlin-modification): [github.com]
Development thread: [forums.reprap.org]
I have currently printed my prototype, and its soon ready for its first moves.

I'm very excited to read about your project, we probably have a lot of interesting engineering problems in common!

---

torbjornludvigsen.com

Quote
tobben
How do you deal with line building up on your spools?

It's pretty important that you only wrap a single layer of string on the motor shaft - otherwise, the diameter is too unpredictable. I've got the in-feed height of both the right and left strings for each pair located in the center of the shaft and as the shaft turns, one string winds up the shaft and the other winds down the shaft. There's a slight bias pulling the strings back towards the center so each string ends up with a tight, single layer wrap.

In my implementation, the amount of string you can wrap up is the limitation on the range of motion because standard NEMA 17 shafts are not long enough to hold a lot of fishing line. My current thinking is to make 8mm dia sleeves to go over the standard 5mm shafts. These will both take up more line per turn and I can also make them longer.

I see, predictable wrapping is obviously very important. I'm quite impressed that you manage to make the strings wrap onto a nice single layer.

Have you thought about winding the strings around threaded rods or larger spools? Mounting the steppers on top of threaded rods, and letting them slide up and down in a stepper-shaped tube could make the strings' winding-radius and winding-point constant. Sketched what I mean and attached picture of sketch...

This would increase build complexity and part count, but would scale better than simply larger spools (like my design will use).

---

torbjornludvigsen.com

@Tobben - The way I see it, there are two ways to go with a cable driven printer - cheap or really big. If you go big, you can afford to put some extra money and complexity into the cable handling, but then the whole thing can be scaled very large. It looks like this is the direction you are heading with your project. I suspect that making a really big printer is what would interest most people with this type of design.

For me, I started this as an exercise in exactly how cheap could you make something that actually prints (without cheating and using surplus components). By winding the line up directly onto the motor shafts and using bent wire eyelets instead of pulleys, the cost approaches the bare minimum of motors, extruder and control electronics - plus some for the structure which could be made more cheaply out of plywood and threaded rod. I don't actually see this approach really going anywhere outside of maybe being an interesting build-your-own printer variant. It's way too fussy to be of general interest.