Vertical X-Rods vs Horizontal X-Rods in CoreXY

ok, soo, i've built my first corexy printer (sandwich200) and love it. i'd like however to shrink the footprint, and i've come up with a design concept that is a 280x280mm footprint using the standard 214x214mm printbed PCBs... but it relies on "stacking" the components, so you would have (vertically) belt YRod belt, and the XRods would likewise go one *underneath* the YRods, so it would be lower XRod, YRod, upperXRod.

now, the only reason that i am considering this is because after flexing the XRods on the sandwich200 (which is a fusebox derivative) i noticed a *massive* amount of play (shear) was possible on the XRods when there are no belts. now that i've tensioned up the belts, i'm having difficulty shearing the XRods sideways.

and that's what i'd like to check.

QUESTION: is it DEFINITIVELY the case that the BELTS are the reason why CoreXY designs are accurate? let's assume competently-arranged and sourced (i.e. ideal) linear bearings etc. clarification of question (asking in a different way): if i only had *one* X-Rod (let's say it was square instead of round), would the CoreXY belts still give accuracy?

if "yes" then i can go ahead and design a 280x280mm footprint corexy printer, which would be awesome.

Belts on a CoreXY
are NOT structural tensioning devices!

Even tho belts have kevlar cores -- they are not structural tension devices

If the belts make your COREXY more rigid the structure itself is bad
and one should look for ways to tighten structure.

The belts are to provide the motion in X & Y axis

If you want to optimize print quality while reducing printer bulk, start by making a rigid frame, then use linear guides instead of end supported round rails.

I went from this X axis:




To this:



by using a linear guide instead of the 1/2" rails I started with

Edited 1 time(s). Last edit at 04/15/2016 03:12PM by the_digital_dentist.

---

Ultra MegaMax Dominator 3D printer: [drmrehorst.blogspot.com]

Quote
the_digital_dentist
by using a linear guide instead of the 1/2" rails I started with

i was aiming to get down to only 36mm (or so) on each edge of a 214mm printbed, for a total width of 280mm. i've not managed that - i have however got it down to 305mm total width. i honestly don't feel that linear rails would help achieve that kind of design goal: they're off-the-shelf parts that i'd not be able to alter.

so for example i've got the upper belt bearings sitting above an LM6LUU, and the lower belt bearings below. LM6LUU's are only 12mm high so i feel comfortable having the belt bearing centre lines separated by 23mm. i can't use LM8LUU's because i don't think the extra separation distance for this design concept would be a good idea.

if i were to try to use linear rails, the width of the linear rails themselves would result in increased outer dimensions. right now, the belts run *literally* at the outer extent of the design's dimensions (or to within +/- 0.5mm).

unless.... unless... are you talking about y-raiils left and right that are *underneath* the x assembly?

Quote
cozmicray
Belts on a CoreXY
are NOT structural tensioning devices!

you may have misunderstood.

ok, let's say that the x-assembly is allowed to rotate freely at each end that it is attached to each left and right y-rod, so it can be come a Z rather than an H (let's ignore the fact that if the H became a Z the x-assembly would fall off the y-rails because it would be too short - let's ignore that for now, assume that the x-assembly gets longer if each end is permitted to freely travel *independently* along the respective y-rods).

now, let's tension both belts - any tension will do. then, go "pinggg" on the long left and right belt lengths (the ones running down the outside between bearings). what happens? regardless of how sloppy the x-assembly is off of perpendicular, you *will* get *EXACTLY* the same note out of both belts, because (assuming they have the same characteristics) their tensions will equalise out, as the x-assembly is not impeding that in any way. the x-assembly will be MASSIVELY off perpendicular, but that's precisely WHY the belts would have equal tension.

now.

let's now bring the x-assembly back to a "normal" x-assembly, where it uses linear bearings that try to keep it perpendicular to the y-rods at all times.

*now* let's tension one of the belts far less than the other. what happens?

you get one belt going "ping" and the other going "pungggggg", don't you?

why is this?

it's because the x-assembly, by virtue of being a rigid (or more likely SEMI rigid) body that is designed to stay perpendicular to the y-rods, is resisting.

but, the way that corexy works, when you turn the belts, the forces are all supposed to equalise (unlike in an H-Bot design) such that even if the x-assembly is fairly loose, the forces equalise and don't pull the x-assembly out of kilter.

my question is, basically, is this REALLY really true - such that if the x-assembly is vertical and resistance to going off of perpendicular is really quite weak (e.g. because the attachments to the Y rods are made of very thin plastic), will you really REALLY get the x-assembly still operating at perpendicular, when (of course) the belts are correctly tensioned and aligned?

i did NOT say "are the belts structural tensioning devices".

Well I don't have a belt tensioning tuner to get the right note out of a timing belt.
But you can use your perfect pitch to screw up the belt tensioning.

A GOOD DESIGN and Assembly

The Core XY motion software is built for orthogonal X Y Axis
So one should design / build for this, and make it so X and Y are rigidly held that way.
Having a " x-assembly is vertical and resistance to going off of perpendicular
is really quite weak (e.g. because the attachments to the Y rods are made of very thin plastic)"
is just asking for 3D printing frustration and making the printer closet king.

A small metal T bracket would probably rigidly keep the X axis at 90 deg to Y axis.

It is not a good idea to think your going to 300mm of printing travel out of a 300mm guide rail
The trolley on a 12mm MRN guide is 38mm long -- so a 300mm rail may provide 260mm of printing travel

It is my contention (I may have to prototype or simulate)
Core XY is a pull- pull system ( using a timing belt in compression is useless)
If belts are attached at a center point where there is no twisting torque
the pull- pull nature of the system would not turn it into the Z you talk about.

Designing and building an accurate 3D printer with a 300 x 300mm
that fits in a back pocket is indeed a challenge!!
But when you slide your printer across a table and it gets racked out of square
will we be getting questions "my square prints are a rombus --- what is wrong?"

Quote
lkcl
i was aiming to get down to only 36mm (or so) on each edge of a 214mm printbed, for a total width of 280mm. i've not managed that - i have however got it down to 305mm total width. i honestly don't feel that linear rails would help achieve that kind of design goal: they're off-the-shelf parts that i'd not be able to alter.

I don't understand - what does vertical vs horizontal arrangement of a pair of guide rails have to do with the length of the rails?

Linear guides (and round guide rails) can be cut to any length you want. A cutoff wheel on a grinder will do the job.

Is 280mm the target width of the whole printer or the target length of the rail(s)? The travel you get is the length of the rail(s) minus the width of whatever is moving on them. If the extruder carriage is 100mm wide, and you need 218 mm of travel, your rail(s) need to be 318 mm long.

If your target is getting a 214 mm wide bed into a 280 mm wide printer, the rails aren't the issue, the width of the extruder carriage is. Use a bowden extruder and mount only the hot end on the carriage to minimize its width.

---

Ultra MegaMax Dominator 3D printer: [drmrehorst.blogspot.com]

Quote
cozmicray
It is not a good idea to think your going to 300mm of printing travel out of a 300mm guide rail

i made, earlier, the following statement:

Quote
lkcl
i was aiming to get down to only 36mm (or so) on each edge of a 214mm printbed, for a total width of 280mm.

so if you can read "214mm printbed" and "total width of 280mm" and conclude that "300mm of travel is necessary" and "the rods are 300mm" in a way that then allows you to infer that i must be a complete idiot, then i'm sorry but i'm going to choose to no longer respond to what you write. i may read it - because i might learn something, but i'm not going to respond further to what you write. what you're doing is to be deliberately sarcastic and write in a way that taunts the reader by deliberately misinterpreting what they write in a negative way, creating opportunities to criticise and hint disparagingly at their efforts at the same time. *mixed in* with this is some good advice, which is why i'm happy to *read* what you write, but if i were to respond it would only give you further opportunity to criticise. you get *ONE* chance - this message *is* your one chance - to say "eek, sorry, i didn't realise i was causing offense, i'm very sorry, i'll work harder in the future not to do that again", and we can carry on a reasonable and rational conversation. if however you choose not to respond with an apology or an acknowledgement, choosing instead to respond with either sarcasm, dismissal or further criticism, you will get no further response. are we clear?

Quote
the_digital_dentist

Quote
lkcl
i was aiming to get down to only 36mm (or so) on each edge of a 214mm printbed, for a total width of 280mm. i've not managed that - i have however got it down to 305mm total width. i honestly don't feel that linear rails would help achieve that kind of design goal: they're off-the-shelf parts that i'd not be able to alter.

I don't understand - what does vertical vs horizontal arrangement of a pair of guide rails have to do with the length of the rails?

partly it does, partly, it doesn't the primary reason why i'm turning the sandwich200v2 round to a vertical arrangement is so that i can remove the hotend (which will be on a flex3drive) by way of some clips or wing-nuts. what i found from the sandwich200v1 is that the bowden tube sticks up a *long* way. i "solved" that by swinging the top up and over through 270 degrees so the bowden pokes through from the other side, the top corexy part being *behind* the z-assembly when stowed.

instead, in the v2, i want the top corexy assembly to swing *down* through 90 degrees so it sits in *front* of the z-assembly when stowed. thus i have to remove the extruder because it sticks up about 120 to 150mm if you take the curve on the flex3drive into account.

to explain as best i can - it's complicated - about the vertical vs horizontal rails, the FuseBox has horizontal rails that straddle a hole where a bowden E3Dv6 drops in. the carriage is 45mm x 70 (or so), and has a mount for a 10mm fan at the end... it's now up to 80mm. the idlers overhang a bit so that means you can't hit the y-end with the fan, it hits the idlers about 10-20mm before the end of the y-rods. the y-rods have 10mm support at each end, but because of the space used up at either end they have to overhang the edges of the extrusion so the outer-edge to outer-edge of front and back is 330mm (inner edge to inner edge is 300mm)...

... there are basically dozens of places like this where an extra 10-20mm is added, both in the x and y directions in the FuseBox. don't get me wrong: i love that it's a $300 corexy design, i wouldn't have picked it as a starting point if i didn't love it.

so learning from that, i've turned the x-bars to vertical, so that saves about 30mm straight away, because the hotend doesn't sit *between* two bars, it rides along-side two bars. the fan will go *on* the E3Dv6 hotend (i'd like to use the clip-mounted 30mm one that's supplied with the E3Dv6 kit). the y-rods i'm placing *directly over* the extrusion, whereas alex mounted them a full 40mm further inwards.

so in a very very indirect way the arrangement of the rods is, through a complex interdependent chain of positioning decisions, affecting how long the rods *can* be if you decide to do a 214x214 printbed and to minimise width and depth.... that then means that the rod lengths are thus "open to debate". picture at the end, it's a work-in-progress, those are v1 idlers and motor holders, on a v2 carriage and v2 x-ends. current outer dimensions: about 305mm x 370mm appx.

Quote

Linear guides (and round guide rails) can be cut to any length you want. A cutoff wheel on a grinder will do the job.

Quote

Is 280mm the target width of the whole printer

it was... but it's too tight. i've upped that to about 305.

Quote

The travel you get is the length of the rail(s) minus the width of whatever is moving on them. If the extruder carriage is 100mm wide, and you need 218 mm of travel, your rail(s) need to be 318 mm long.

plus there's the corner-supports in the way it turns out. i learned from the sandwich200v1 that it's absolutely essential to have rock-solid corners on the top corexy frame. edvardas mentioned kindly the idea of drilling the extrusion in one corner and dropping an allen-key bolt down through it, into the self-tapping hole of a perpendicular extrusion: i love that... but i don't want people to have to do drilling in order to make one of these printers. and you need the extrusion machined dead-flat, too.

so in the end i went, "y'know what? what's 20mm here or there, why not just use the standard size of 300mm and go with that".

Quote

If your target is getting a 214 mm wide bed into a 280 mm wide printer, the rails aren't the issue, the width of the extruder carriage is. Use a bowden extruder and mount only the hot end on the carriage to minimize its width.

yeah i'm going with a flex3drive, the sandwich200v1 is a bowden. first time i used one - don't like them, won't be doing it again. the flex3drive by jason is like seriously, seriously well-engineered and thought out. i love it. it's the best of both worlds of what bowden is trying to achieve, combined with the advantages of direct-drive. 105GBP including shipping and you can easily do 350mm/s print-speeds, with *good* quality as well.

Well as someone said here someplace
Sorry to pee in your cheerios

I am just an OLD boomer
if you Millennials can't take a poke for a design/ build miscue
and listen a bit, you may learn something.

You post to get feedback and questions answered

How many are giving you answers and feedback?

nuff said

I just noticed in a previous post that you're using 6 mm bearings and guide rails. I don't think that's a good idea at the length you're going to be using. 8 mm rails are questionable over that span. I would never consider 6mm rails. I know you have a goal for the printer's size, but don't you also have one for print quality? Will a few extra mm in printer size really matter so much?

Off the shelf bed heaters are typically 200 mm square. Do you need the extra 14 mm in the bed?

---

Ultra MegaMax Dominator 3D printer: [drmrehorst.blogspot.com]

Quote
the_digital_dentist
I just noticed in a previous post that you're using 6 mm bearings and guide rails.

yeahh i upped them to 8mm.

Quote

Will a few extra mm in printer size really matter so much?

no not compared to print quality.

Quote


Off the shelf bed heaters are typically 200 mm square. Do you need the extra 14 mm in the bed?

the standard reprap printbed, as well as the Alu MK3 (which is what i'm using) is 214mm. 7mm surround so that there's room for the screw-holes in the corners. the print *area* is 200x200. the actual physical dimensions are 214x214.

I looked at the specs on that heater- on the good side it has holes to allow 3 point leveling. On the bad side it looks like it only dissipates about 100 W. The description says it's flat enough to print on directly, without adding a glass plate, but unless it's quite thick and preferably cast aluminum, I have doubts about that. If you put glass on top of it, it may never reach 105C to print ABS (but maybe you don't intend to print ABS, so it won't matter). Other people must be using these heaters. Do they get hot enough and in a reasonable amount of time? Are they actually flat enough to print on? Do they stay flat when heated?

This heater has about 0.25 W/cm^2 power density. I run 0.5 W/cm^2 in my printer and it get to 105C in about 5 minutes (heating up a massive aluminum plate, 317 x 305 x 6.35 mm).

---

Ultra MegaMax Dominator 3D printer: [drmrehorst.blogspot.com]

Quote
the_digital_dentist
I looked at the specs on that heater- on the good side it has holes to allow 3 point leveling. On the bad side it looks like it only dissipates about 100 W. The description says it's flat enough to print on directly, without adding a glass plate, but unless it's quite thick and preferably cast aluminum, I have doubts about that. If you put glass on top of it, it may never reach 105C to print ABS (but maybe you don't intend to print ABS, so it won't matter). Other people must be using these heaters. Do they get hot enough and in a reasonable amount of time? Are they actually flat enough to print on? Do they stay flat when heated?

This heater has about 0.25 W/cm^2 power density. I run 0.5 W/cm^2 in my printer and it get to 105C in about 5 minutes (heating up a massive aluminum plate, 317 x 305 x 6.35 mm).

i've got glass on top - quality's good. it does take a while to get up to 60C. i'm reaaasonably happy with it at the moment. i'll see what happens with ABS one day. if i don't like it, i'll rip it out. it's only $EUR 15 or so.

ok, so just an update, for anyone considering vertical rods (as opposed to vertical rails), the answer is, after a rather expensive development cycle: DON'T [forums.reprap.org]