"Unnamed" plotter XY mechanism revisited

rich is credited with bringing this old plotter-style mechanism to the attention of the 3d printing community, with the "SLI3DR" design.



i've been thinking about this mechanism, and i noticed something fascinating: the x arrangement has a 2:1 gearing effect due to the return of the belt from the carriage. what is intriguing is that the y line on the original mechanism does not.

i'd therefore like to propose a mechanism that takes advantage of this 2:1 gearing effect on both X and Y, using a "cross" like that used on the ultimaker and the quadrap, but without the double-ended drive arrangement of those two designs: to use the "unnamed" X arrangement instead. the picture below, for clarity i have started the Y belt (in yellow) but have left out the motor and bearings.

also, i propose using a GT2 timing belt instead of wire.



the X motor is in the top right corner (green), with a GT2-20 or GT2-16 tooth. 625 bearings are marked in red. 608 bearings with a 3d-printed teethed outer rim, or better a 40-tooth GT2 idler, is marked in BLUE. the carriage therefore has FOUR such idlers (y-axis large idlers marked in CYAN). the linear rails / bearings are also left out for clarity.

now, as this is a symmetrical design, the weight distribution for X and Y is perfectly balanced. each axis has the additional 2:1 gearing down ratio due to the return-belt effect around the carriage idlers. the motors can be placed inboard a little bit (so that they do not stick out beyond the edge of the frame). also, i feel that due to the reduced angle on the motor gear, a GT2-16-tooth would be fine, resulting in even higher gearing compared to standard designs that have to use GT2-20-teeth due to the 180 degree belt looping that's typically used in RepRaps and CoreXY designs.

also, i feel that the carriage need only have single support-rods and single linear bearings / v-rollers onto the frame. even if there is a slight amount of wobble in the rods, taking them off perpendicular to the frame, i don't believe it significantly affects the accuracy. imagine that the X carriage rod is rotating back and forth by +/- 2 degrees, and the carriage is in the centre and is held static. will the green motor gear rotate? no it will not. why? because the belt travel cancels out. if the carriage is full to the left and the same thing happens, again there is no rotation of the green motor gear. yes the belt will stretch but no there will be no inaccuracies introduced.

which is extremely cool.

also what is convenient is that both belts may be fixed and tensioned at one end with mechanisms identical to those used on the mendel90's x belt (a tensioner screw). this removes all the complexity of having motors on rails or tensioners on carriages as in many CoreXY designs.

additionally, all idler bearings with the exception of the large carriage ones are belt-to-smooth, reducing the need for large numbers of (unusual and harder-to-source) GT2 teethed idlers.

overall i think this is pretty exciting, as in combination with reduced inertia by following the design of the quadrap (10mm square carbon-fiber carriage rods with double-v delrin rollers), i believe this design would result in significant accuracy improvements (2.5 times better gearing, less inertia) as well as speed improvements (reduced carriage weight).

i'd be interested to hear people's thoughts as there are a huge number of design advantages here.

[update] - one down-side i can think of, corexy belts are approx 4x the size of the printer's frame. 2x width + 2x depth. this design: belts are approx 5x the size of the printer's frame: 3x width, 2x depth. that would involve ordering 4m of GT2 timing belt for a 200x200mm printer.

Edited 2 time(s). Last edit at 05/11/2016 05:46AM by lkcl.

---

-- sandwich200: compact portable folding corexy printer [reprap.org]

... look at the images here - I've built several of this type for plotting or laser-cutting with the motors fixed on one side:
[forums.reprap.org]

---

Viktor
--------
Aufruf zum Projekt "Müll-freie Meere" - [reprap.org] -- Deutsche Facebook-Gruppe - [www.facebook.com]

Call for the project "garbage-free seas" - [reprap.org]

Quote
VDX
... look at the images here - I've built several of this type for plotting or laser-cutting with the motors fixed on one side:
[forums.reprap.org]

cool. those images definitely look like it's the "unnamed" arrangement. nice to see it was patented... and has expired.

note the key difference here: i plan to use GT2 timing belts, *not* wire and *not* a capstan effect. so unless a worm-drive is used, the motors would need to be upright, just like in CoreXY. hence why i was looking for a suitable place to put them. in the corner as in the diagram above allows the belt to be pressured against the GT2 gear's teeth, without needing too many extra idler bearings. if you put the motor(s) in the middle of the belt, you'd need one more extra bearing, plus the motors might get in the way of the carriage travel. in the corners they're out the way.

Quote
lkcl
... nice to see it was patented... and has expired.

... I've only patented it then to "hold my back free", so no other could make claims against my developments ... did this several times with other projects/systems too

---

Viktor
--------
Aufruf zum Projekt "Müll-freie Meere" - [reprap.org] -- Deutsche Facebook-Gruppe - [www.facebook.com]

Call for the project "garbage-free seas" - [reprap.org]

Quote
VDX

Quote
lkcl
... nice to see it was patented... and has expired.

... I've only patented it then to "hold my back free", so no other could make claims against my developments ... did this several times with other projects/systems too

good maaaan

ok so don't laugh, i made a mock-up in under 2 hours using paper, glue, cotton and tiny little 3d-printed circles to represent "bearings":




i've also done a video walk-through, with the kind assistance of my 7-year-old daughter, going over this in detail (so as to make sure apart from anything that i haven't made any glaring errors), as well as outlining some of the significant benefits. the main ones are: stability of the carriage, reduced dependence on perpendicular-ness of both the x and y linear assemblies, reduced *possibility* of the x and y linear assemblies going off-perpendicular, and that 2:1 pulley-gearing effect which i don't believe anyone else has actually noticed is there.

[www.youtube.com]

i have however made a mistake in the mock-up, i used 4mm circles instead of 8mm to represent the 16mm (605) bearings.... on a 1-in-2 scale model. whoops. anyway, a quick analysis, going over the positions of the bearings (bearing in mind that the X and Y belts are offset just like they should be in CoreXY), they will actually miss. whew.

actual bearing positions i will alter slightly from what's there when i get to the CAD model, but with this exercise done i am confident that a 3.0 metre belt can be bought, cut in half, and utilised for a 200x200mm printbed.... just like "normal" in any other CoreXY 200x200mm design. this answers a question that was seen on another forum, someone asked "if i use belts, will they need to be longer than for CoreXY" and the answer is NO.... as long as you keep the outer dimensions to below 312 x 360mm. it's pretty tight, but doable. with a bit more thought - perhaps for a version 3.0 of the portable 3D printer - i might even be able to reduce those outer dimensions by somewhere between 20 and 40mm. which would be pretty damn cool.

Edited 1 time(s). Last edit at 05/19/2016 12:56PM by lkcl.

correction, for people who may find this thread in the archives in the future: i believed that this design would be less affected by rotation of the carriage: the moving x-rod and moving y-rod not being perpendicular to the static rails that they're on. that is NOT correct. the only situation where that is true is when the printhead is directly in the centre, such that the belts which go to the centre are all exactly of equal length.


the forces on the carriage are i feel pretty much balanced. the only thing is, the location where the pulleys pull is not dead-centre with the position of the printhead, which *might* cause some moment of inertia and thus rotation. this could be fixed by putting in a counterweight to ensure that the centre of gravity of the entire carriage is directly at the cross-hairs of the X and Y pulley systems.

Edited 1 time(s). Last edit at 05/21/2016 07:10AM by lkcl.

It seems an interesting proposal. Belts can work in parallel planes, no belt crossings. You just need a fair amount of idlers :-)

Quote
misan
It seems an interesting proposal. Belts can work in parallel planes, no belt crossings. You just need a fair amount of idlers :-)

yeahh, turns out it's only a couple more than a corexy if you do the trick of putting motors "inboard", you know the one? where instead of having the 3d printer be 20mm wider on both sides because the belt goes direct to the GT2 gear, you bring the motors in by 40mm or so and put 2 90 degree turns. couple of people have done that.

so if you do that, it's 4 extra idlers on a corexy when you already have.... mmmm....10. so that would be 14 on a corexy that uses the "inboard" trick - and here it's... 18? ok yeah that's quite a lot, isn't it but hey, you get an extra 2:1 gearing effect from the 180 degree turn on the pulleys at the carriage, which means extra resolution on the movement of the printhead, so i'd say on a cost-benefit analysis it's a win.

got it! i have a satisfactory arrangement which fits the (extreme) space requirements as a drop-in replacement on the sandwich200v2:



details are on the forum including a 3D CAD walkthrough video [forums.reprap.org]

Hmm, I like that the X and Y axis have equal moving mass, and the simplicity of a dedicated motor per axis is reassuring. That cross arrangement always looks like it's begging for a quad hotend setup

Quote
JamesK
Hmm, I like that the X and Y axis have equal moving mass, and the simplicity of a dedicated motor per axis is reassuring. That cross arrangement always looks like it's begging for a quad hotend setup

ha, yeah it would be quite a span though. you can see i pushed the square rods to the ends of their holders. a quad hotend setup you'd no longer need to do that, because you'd need the range of travel on both sides (equally). that has the nice advantage of being able to re-centre the square rods.

actually perhaps you could instead go for double-rods (bear in mind that introduces another 6 double-v rollers) but the stability in doing so would be enormous. i'm currently having to use 9x5x3 bearings (!!!) and am getting some 11mm v-rollers made up which take 2 9x5x3 bearings each, so for an extra 2 sets of rods that's *12* extra bearings. they're extremely light-weight.

anyway by going for double square rods per axis you could move the 4 hotends closer together. almost certainly if i was going to try that i would try square solid 8mm carbon fibre rods, to reduce weight, it's a huge difference. i do however have to verify that they'll take the load. i've got aluminium ones at the moment.

I missed the fact that you were planning on using square section tubes. Is the choice over round rods primarily for weight saving or is there something else? It doesn't look like you need the non-rotation around the long axis (there must be a better way of saying that - axial non-symmetry?) for anything. V rollers are cool, but the simplicity, durability and cost effectiveness of smooth rods and ball bearings are hard to beat.

Quote
JamesK
I missed the fact that you were planning on using square section tubes.

nono, not tubes: solid bar. big difference. i got 8mm square-section solid bar, extruded. ordered it from the aluminium-shop.nl (think that's the right domain), they're about 12km from den haag in naldwijk so delivery was next-day.

Quote

Is the choice over round rods primarily for weight saving or is there something else? It doesn't look like you need the non-rotation around the long axis (there must be a better way of saying that - axial non-symmetry?) for anything.

i know what you mean. axial rotation is not prevented by square rods. if there is any it would actually damage the v-rollers as the axial rotation would torque the v-rollers sideways.

Quote

V rollers are cool, but the simplicity, durability and cost effectiveness of smooth rods and ball bearings are hard to beat.

yeahhh i know - i'll have to take that risk and see for myself.

yes it's two reasons. firstly, an aluminium solid square bar @ 8mm x 8mm is lighter - 52g - than an 8mm OD precision machined chrome-plated solid steel rod (118g), and a carbon fibre solid square bar @ 8mmx8mm is *even less* - something mad like under 12 grams for a 300mm length.

and, also, the 12 9x5x3 bearings are significantly lighter than either 2 LM8LUU (or four LM8UU) linear bearings. so yes, i wanted to experiment and see if the trade-off of massively-reduced weight of using these different materials is worth it.

secondly... what was secondly... oh yes: space. the *compact* bit of the "compact" design requirements that i've inherited from the failed vertical-rod version of the sandwich200v2 are just... ridiculously tight. i simply can't get 2 LM8LUU bearings into the available space: the carriage *has* to be below the length which would accommodate two LM8LUU bearings mounted at right-angles to each other.

but there is an additional issue (which someone with not so ridiculous space requirements could easily overcome, i'll explain it below), which is that you really do need the 2 rods to be as close to each other as possible. why? because any "pull" by the pulley system could result in a twist on the carriage if the rods are too far apart.

so to accommodate the rods being closer together, you can't stack LM8LUU bearings one on top of the other because they're 15mm diameter - that strikes me as too far apart. so you put the rods only 1mm apart, now you have to move the bearings - which are 45mm long already - another 10-15mm further along the rods. now the carriage is a whopping 55 to 60mm on a side *minimum*!

in the idea that you had, to do quad hotends that's not a problem, you're not worried about saving space already, that's for sure but, perhaps if you wanted to straddle 4 hotends one in each quadrant you could use 4 LM8UU bearings, 2 each side of the "cross" point, and the rods could be separated by 1mm no problem.

i mean it's kinda annoying: even if i were to stack 2 LM8LUU bearings one on top of the other separated by 15mm and actually touching at right-angles in the carriage, i *still* wouldn't be able to fit the damn carriage in because there is a 40mm height restriction in order to avoid hitting the outer box when the printer's stowed.

you guys who do desk or workshop static 3D printers that spend their time in one place are damn lucky

anyway i _would_ consider using rails for this because the linear carriages are so much smaller... but the fact that the rails would also be moving, it would be necessary to add some support for the rails (some 2040 extrusion perhaps) and that is outside of the weight boundary that i have (arbitrarily) set. however other people may not have such a limit and so could easily consider that using linear carriages on 2040 or other extrusion would be absolutely fine. some of the stuff from the openbuilds store would be perfect.

---

-- sandwich200: compact portable folding corexy printer [reprap.org]

Ah, I see where you're coming from. Yes, the bearings are a bit on the clunky side when you're really trying to squeeze the geometry. Nice challenge you've set yourself there - look forward to seeing how it goes!

Quote
lkcl

Quote

Is the choice over round rods primarily for weight saving or is there something else? It doesn't look like you need the non-rotation around the long axis (there must be a better way of saying that - axial non-symmetry?) for anything.

i know what you mean. axial rotation is not prevented by square rods. if there is any it would actually damage the v-rollers as the axial rotation would torque the v-rollers sideways.

sorry JamesK, forgot to say: what _does_ prevent axial rotation is the *other* rod... there's a word to describe this... each rod mutually prevents axial rotation about the other. something like that.

in other words, the only way that you can get axial rotation about the Y rod is if the X rod's ends are somehow not properly attached, and vice-versa.

the mistake that i did make - well a silly compromise which i am extremely relieved to be able to say i could take it back out again - was to *remove* a single y-rod rail-support roller. the theory went that the three on the X rod would act as one of the stabilising points, and the 2 remaining ones on the *Y* rod would be the other end of the stabilising effect. of course, if the carriage bounced up and down (i.e. the rods started to vibrate) then that would result in some seriously inaccurate bouncing-about of the printhead.

luckily after speaking to mutley3d, he has some tiny 9x5x3mm bearings, so i will be using those, and *that* means i can put in 3 rollers per rod again. whew. what that means is that the Y rod's triple-sandwich of rollers helps *stabilise* rather than *rely on* the X-rod positioning, and of course vice-versa.

Hi

Any progress on the build?
I am curious on how well this mechanism works in real-life and if there are any unforeseen gotchas...

I really like the gearing ratio the pulley system provides. I assume that in addition to the better resolution, the torque requirements on the stepper will be halved as well. Is this correct?
I am designing my own small printer based on the 28BYJ-48 12V "Toy" stepper motors (more as a fun challenge) and these mechanical advantages make this mechanism quite attractive for such a stepper.

Quote
jpconstantineau
Hi

Any progress on the build?

i've been very busy with a crowd-funding campaign for the _real_ reason why i started this 3D printer in the first place: [crowdsupply.com]

Quote

I am curious on how well this mechanism works in real-life and if there are any unforeseen gotchas...

I really like the gearing ratio the pulley system provides. I assume that in addition to the better resolution, the torque requirements on the stepper will be halved as well. Is this correct?

yes. and accuracy. however using 400steps/rev motors is generally recommended for any new printer these days.

Quote

I am designing my own small printer based on the 28BYJ-48 12V "Toy" stepper motors (more as a fun challenge) and these mechanical advantages make this mechanism quite attractive for such a stepper.

i'll let you know how i get on... just very very busy at the moment.

Hi,
I just put together a small prototype to check on the feasibility of using this mechanism for my small printer (See picture).
I took 1/4" rods, pressed some simple bushings in a few 3D printed parts, clamped the assembly to the bench and tried to move the carriage around.

Considering that this isn't a perfect setup, it did highlight a potential flaw in the mechanism you propose. When considering the force balance between the carriage being pulled and the friction at the outside rails, one can create a moment significant enough for twisting to occur; causing binding.
If the carriage is centered, the friction forces on the outside rails will create equal moments and the carriage will move. However, if the carriage is close to one of the edges, the moment are no longer balanced; causing a twist and potentially making the whole carriage bind.

If you look at the original plotter design, the one Rich implemented, you will note that the carriage sits on a set of parallel center rails, a bit like the h-bot/coreXY designs. Being on parallel center rails, the moments around the carriage are negligible enough to create significant twisting (and balanced if pulled from the center). The center rails themselves are "edge-pulled" (blue path). This ensures that the friction is balanced without any significant leverage to create any moments. I did try the Y-belt arrangement from the original SLI3DR design (blue belt) and it did move the center rail, regardless of where the middle carriage was located. With the "center-pull" (red path), I couldn't get the carriage to move reliably.

If you look at most "+" designs (Ultimaker, Tantillus,Quadrap,MPCNC), the center rails are all being "edge-pulled, both for X and Y.
If you look at most "H" designs (H-bot, CoreXY, Original Unnamed), the middle carriage is pulled in the direction of the rails it sits on, and from very close to these rails. Movement in the other direction is where there are significant differences (Y or blue path).

My little experiment has made me reconsider this "unnamed +" arrangement. I am evaluating whether to keep the "+" design or switch to a "H" design.
I do like the space saved that the "+" arrangement offers.

As I indicated, my prototype isn't perfect, the bushings do slide nicely on the rods on their own, however, when setup on 3 "parallel" rods, any misalignment increases the friction significantly.
Too bad I can't test the CoreXY arrangement right away. I'll need to dismantle what I have and print a few new parts to give it a go.

What was your original reason to switch from Core XY?

ipconstantineau, very very brief reply because i am currently extremely busy with this [crowdsupply.com] which is the crowd funded eco-laptop.

if you add in even some string wrapped around rods glued to the top of the prototype you've created i think you'll find it's less of an issue than it seems, the pull of the belt is equal tension.

ideally you would have two rods per axis creating a bridge on each that stopped the twisting, and use linear rails to stop that even more. the printhead would be bracketed by a square of 4 rods, one on every side.

i am using aluminium 10mm solid square rod, it is 1/3 the weight of steel round bar.

i am going to use 10mm solid carbon fibre, it is only 12g (!!)

plastic 8mm bearings are also less weight.

i am going to be doing spring-loading of the v-rollers onto the square rod because that avoids the accuracy issues associated with 3d printing and wear of v-rollers

also if the carriage holder is made rigidly the twisting is less because the cross-bars support each other to prevent each other from twisting. any twist in the X-ends must also result in a twist in the Y-ends if the carriage is rigid and there is no "play" in the carriage linear bearings / v-rollers.

sorry so brief, can't explain more fully, limited time now.

Can I get the CAD model (in Step or iges) please? I would like to dissect the design for possible use with a large printer.

i'm really really sorry, Trexation - i'm a software engineer so use the tools that i'm familiar with (programming languages). the files are written in the python programming language (which can then be revision-controlled and properly managed), there's a library called pyopenscad which auto-generates SCAD files for use in [openscad.org], that then can be used to compile the CAD objects to STL, and from there it goes to standard 3D printing software.

i don't have any kind of proprietary software that could give you what you're asking for, and it would be difficult to re-integrate any improvements that you made. i'm also having to focus on the crowd funding campaign for the eco-laptop that the 3D printer was designed to help me to develop, so have very very little time.

so sorry!

if you're a python programmer (or would like to learn) you can follow the instructions here:
[www.youmagine.com]

adapt them to use the git clone URL for the 3d printer instead.. .ah! found the brief instructions i'd set up:
[hands.com]

follow those okay?

sorry, so little time now, am tracking 17 different forums for the crowd funding campaign, and increasing.

Ok, I totally understand; I hope your crowd funding campaign goes well.

thanks man. talk to jpconstantineau, he may be able to send you the design files of the prototypes he created.