What's Wrong With This Picture...

OR "Yet Another Gantry Design Question"

So, I'm starting a printer build and am scratching my head over all the X/Y designs, especially the ones with the long, complex belt routing and the moving mass. What's wrong with simple and basic cartesian X/Y?



I certainly can't be the first to try this but the design doesn't seem to show up very often.

• The outside dimensions are 500mm square. The usable build area will be less of course.
• The rods are all Arris 3K rolled carbon fiber, 10x500mm (25g each), stock from Amazon.
• The bearings are all Igus Iglide press-fit sleeve bearings. Since the force is applied evenly at the ends of the cross pieces, there's no need for floaters or a 2-1 design. That may change as the design progresses of course.
• The cross assemblies will be driven on both ends by the same dual-shaft stepper, 1 for each axis of course, fixed to the frame. 4 belts, each 1m in circumference driving each end of each axis. I suppose I could do it with 2 3m belts, each in a H design, and even have the belts cross the build area inside the cross piece tubes. But, why?
• The tool platform in the picture is embarrassing but it's the best I could do with what I currently had. WAY too much plastic.
• Even with parts printed on a decidedly imprecise MP III Select Mini, and a metric tape measure I got in 1976, the tolerances are good enough that with only minor fiddling, the tool platform glides with only a flick of the finger and has zero rotational slop. With the cross piece ends fixed as though they were attached to the belts, there's no slop in any direction. I'm thinking of maybe using the Igus drylin N series rails for the perimeter to reduce the dependence on 3D printed parts for the corner braces.
• The moving mass (as pictured except the corner mounts and perimeter rails) is only 210 grams.
• The center of gravity for the hotend (bowden setup) will be right in the center of the tool platform.
• There's plenty of room for improvement in all aspects of the design.

What am I missing?

1) Extra weight due to having to haul the X or Y stepper with the moving platform. With COREXY both X and Y stepper are fixed.
2) Equal tension as you move in any direction which should help keep both X and Y axis in alignment as they move.

That said there is nothing inherently wrong with a decently designed X/Y setup as long as you are not going for the latest in cool or trying to make a speed demon.

Quote
WZ9V
1) Extra weight due to having to haul the X or Y stepper with the moving platform. With COREXY both X and Y stepper are fixed.

That's just it...the steppers for both axes are fixed to the printer frame. They don't move at all. The moving mass will be what I showed above plus hotend and whatever portion of the bowden tube and cables that will be moving.

Quote
WZ9V
2) Equal tension as you move in any direction which should help keep both X and Y axis in alignment as they move.

I can't tell if you mean that as a downside or upside.

Quote
WZ9V
That said there is nothing inherently wrong with a decently designed X/Y setup as long as you are not going for the latest in cool or trying to make a speed demon.

I am going for speed, hence the low moving mass, close tolerance bearings, short-ish belts and force applied to both ends of the cross members.

Quote

The cross assemblies will be driven on both ends by the same dual-shaft stepper, 1 for each axis of course, fixed to the frame. 4 belts, each 1m in circumference driving each end of each axis.

Can you draw the belt loops and stepper positions in the above picture for clearing up your intentions? In my mind there would be 3 belts for each axis....

Edited 1 time(s). Last edit at 08/08/2017 02:20AM by o_lampe.

It looks interesting, and if you can make it work, it will be great. What jumps out at me is using CF tubes and bushings in lieu of linear motion components. The tubes aren't made with the sort of surface finish/tolerance of a guide rail for linear motion- motion could be rough. CF is abrasive stuff, and plastic bushings sliding on CF probably won't last long. Also, plastic bushings are made to slide on steel. Friction sliding on CF/epoxy will probably be much higher. Then there's the diameter. The CF tubes have some nominal, poorly controlled diameter. Chances of getting bushings made to fit a very specific size steel rail to fit on a CF tube without any slop or being too tight seem unlikely. Finally, is the CF going to be rigid or behave like a bunch of springs?

You are going to learn a lot developing this mechanism.

---

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

Igus is offering CF tubes for use with their Drylin bushings
[www.igus.de]
Not a cheap option though, but definitly one with very little moving mass

---

[www.bonkers.de]
[merlin-hotend.de]
[www.hackerspace-ffm.de]

Quote
the_digital_dentist
It looks interesting, and if you can make it work, it will be great. What jumps out at me is using CF tubes and bushings in lieu of linear motion components. The tubes aren't made with the sort of surface finish/tolerance of a guide rail for linear motion- motion could be rough. CF is abrasive stuff, and plastic bushings sliding on CF probably won't last long. Also, plastic bushings are made to slide on steel. Friction sliding on CF/epoxy will probably be much higher. Then there's the diameter. The CF tubes have some nominal, poorly controlled diameter. Chances of getting bushings made to fit a very specific size steel rail to fit on a CF tube without any slop or being too tight seem unlikely. Finally, is the CF going to be rigid or behave like a bunch of springs?

All good points. The Arris CF tubes have a high-gloss finish and Igus has bearings that are meant for CF and other composites. I've also tried different CF manufacturers and I've settled on Arris because of tolerance and accuracy. I bought a dozen Alston tubes a few months back (for other purposes) and returned all of them because no 2 were alike and the surface finish and diameters varied over the length of the tubes. I've got 30 of the Arris tubes now and only 1 was out of spec enough that I wouldn't be able to use it in the printer. The bearings themselves are meant to be press-fit and compressed into the bore, nominally 12mm for a 10mm OD rail. When the bore is correctly sized, the CF tubes slide like they were in a good LM10UU with slightly less slop and much less noise and weight. Because my current printer is not all that accurate, I wound up having to slightly oversize the bores and shim the bearings with kapton tape around the outside before pressing them into the bores. That worked great actually. Having said that though, I will probably replace the perimeter tubes with linear motion rails and lightweight sliders. That'll knock out half the bearings even lessen the moving mass.

The CF tubes are pretty rigid. Not as rigid as steel but about equal with aluminum. I experimented with filling the tubes with an epoxy/balsa mixture and that made the tubes even more rigid at the expense of 10g per rod. I don't think I'll need to do that though. The static load thankfully isn't all that much and it's spread across 4 tubes. It takes way more force than even a triple hotend could apply to deflect them at all. The dynamic load is tricky (as always) but given the low mass, the 2 cross tubes for each axis are driven from both ends, the tubes are about 60mm apart and clamped, and they are aligned with the plane of motion, I don't think those forces will be causing a flex issue.

Quote
the_digital_dentist
You are going to learn a lot developing this mechanism.

You got THAT right! That's half the fun though and this wouldn't be the first project I've gotten half way through and thought 'Well, that was stupid".

I think you've got a project! Start a new thread and keep updating it as you go...

---

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

This is commonly known as an "Ultimaker Style" Cartesian. Named for the company that first commercialized it. The general concept is open source, and pretty solid, but people want to be unique and make something new rather than use what's been proven to work.
For what it's worth, CoreXY has advantages over Ultimaker, but is indeed more complicated.
The Tantilus is the most well-developed Reprap I've found to use the Ultimaker style gantry.

You're planning to use carbon rods for the stepper shaft extensions too?

BTW, before you start a build blog, this baby needs a name
As mentioned before the gantry style isn't new, but using carbon rods and suited plastic bearings is.
All you need is an extruder that can deal with the potentially high speed.

Edited 1 time(s). Last edit at 08/09/2017 03:09AM by o_lampe.

Quote
Feign
This is commonly known as an "Ultimaker Style" Cartesian. Named for the company that first commercialized it. The general concept is open source, and pretty solid, but people want to be unique and make something new rather than use what's been proven to work.

Ultimaker may have commercialized it for 3D printing but the simple cartesian x/y design has been around for decades. I saw it first back in the 70's. <gasp>

Quote
o_lampe
You're planning to use carbon rods for the stepper shaft extensions too?

Nope. Simple 5mm stainless steel. Rotational torque isn't one of CF's strengths and it's not like I need to save the mass since the shafts aren't getting lugged around. At 500mm in length, they'll have a little inertia to overcome to get them spinning but I can't imagine it'll be significant.

Quote
o_lampe
BTW, before you start a build blog, this baby needs a name
As mentioned before the gantry style isn't new, but using carbon rods and suited plastic bearings is.

Well, the printer itself will be named "HyperCube OverKill"

For the gantry name, maybe "Carbon Cartesian" or maybe "Cartesian Renaissance" since the actual X/Y motion has been around forever.

Quote
o_lampe
All you need is an extruder that can deal with the potentially high speed.

Yeah. No kidding.

Quote
Srek
Igus is offering CF tubes for use with their Drylin bushings
[www.igus.de]
Not a cheap option though, but definitly one with very little moving mass

Yeah even though I'm not going cheap, I think they're a little outside my price range.
The more I think about the perimeter though the more I'm convincing myself that the Drylin N series will be a better choice.

Do eeeeet, build thread!!!!

ikcl is building something similar. See: [forums.reprap.org]

---

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

Oh yeah. I really need to go through all the forum sections.
Lotsa belts though.

Quote
the_digital_dentist
ikcl is building something similar. See: [forums.reprap.org]

Quote
gtj
Oh yeah. I really need to go through all the forum sections.
Lotsa belts though.

Ah, I didn't read the whole thread. He changed the design. Cool.

Quote
Srek
Igus is offering CF tubes for use with their Drylin bushings
[www.igus.de]
Not a cheap option though, but definitly one with very little moving mass

the technical specifications tab just below says that even those are 0.1mm tolerance. i really really wanted to use carbon fibre in the design i'm doing, and investigated solid 8mm square bar, which is a common standard, it's even used for large kites (wind-surfing for example). i designed and made some 7mm polycarbonate V-rollers and put 6mm x 3mm x 2.5mm ID absolutely tiny bearings in them. but i made the mistake of trying to position the V-rollers in static positions.

so the design mod i planned was to make the middle V-roller spring-loaded, so that tolerancees on the V-rollers and the bars wouldn't matter. the top 2 V-rollers would rest on the top and the 3rd would be the spring-loaded one.

however as i was still quite concerned about rod tolerances changing along the length of any one rod i decided instead to go with twin 6mm rods per axis and LM6UU bearings (total QTY 8). strictly speaking it might even be possible to just use 4 LM6UU bearings but i quite like using 8, it's possible to fit fans (and cables) down the gaps



so that's twin 25mm fans above, through the circular holes. the square ones, i use one to get the E3Dv6 cooling fan duct through it, and the other one i use for cables. the carriage is i think 80mm wide or so, but if you look closely you see the corners are truncated.

btw don't make the mistake i did, make sure that the carriage can be placed without disassembly onto the bearings, don't get fancy, use cableties or some screw-based plastic attachment underneath ffor clamping onto the bearings.

i initially made the mistake of designing the bearings to be push-fit into the carriage, asseembled everything, locked the rods down into the x/y-ends (16 35mm M3 screws in total, takes a hell of a long time....) and then realised FRICK!! i have to redesign the carriage to make the fan holes bigger...

... but i couldn't get the damn carriage off without either disassembling the *ENTIRE* top... or *breaking* the damn carriage plastic, destroying it in the process.

so... do consider that ok?

Quote
the_digital_dentist
ikcl is building something similar. See: [forums.reprap.org]

what is it with the "i"?? why do people always ... it's the frickin variable-width fonts, isn't it, that's why people think it's an "i" not an "L", aiyaaa. it's L K C L - my initials. luke kenneth casson leighton. not to be confused with the 25-year-old drug dealer from southampton who was arrested and jailed a couple years ago, or the member of the royal family, both with the same first and last name as me, nor with the moron who keeps giving out my email address to people and even used it to defraud paypal about 4 years ago... L, dd! it's an L!

Quote
gtj
*] The center of gravity for the hotend (bowden setup) will be right in the center of the tool platform.

What am I missing?

please don't make this a bowden. look up the flexx3drive instead: mutley3d does a very small extruder originally designed for a kossel mini which will have a good enough weight for your needs. please don't use the zesty nimble, either.

so the flex3drive has a 40:1 wormdrive (the zesty has 30:1 gears, which are smaller so more force is exerted on the plastic, and also have backlash), and a properly-spec'd automotive driveshaft where mutley3d have done the proper engineering analysis and years of testing to make sure it's up to the job.

why would you *NOT* want to use a bowden tube for a high-speed printer?

well, it's down to latency due to the PTFE tube. that's almost a METRE you're pushing a filament down, and it doesn't properly sit absolutely flush with the walls: it buckles. this is why you have to have such a stupid amount of "retract" on bowden tubes (4 to 5mm???? insane). so you're doing 350mm / sec or higher travel speeds, and you want the filament, which had to be over-pushed by a whopping 5 MILLIMETRES, to suddenly just... stop oozing out? naah not a chance. if you're only doing even 150mm / sec that's enough time for oozing or delays not to be really noticeable, particularly if you keep the acceleration low enough.


but high speed in turn means high acceleration, because if you don't dial up the acceleration the printhead *never reaches* high speed, particularly on smaller parts (theere are online calculators you can find which will demonstrate this for you - anyone know a link?)

now, if 3d printer firmware happened to have some sort of PID control on the filament, along-side some sort of measuring system on how much *ACTUAL* filament was coming out the hot-end, *THEN* we might be able to get somewhere with bowden tubes operating at high-speed.

so the advantage we have with the flex3drive is that the motor is static, yet the force iis being applied *at the extruder* just like a drect-drive... but without the weight *and* you have the higher torque and accuracy from a 40:1 wormdrive.

rhen there are all the problems associated with bowden. broken filament. wasting a METRE of filament every time there's a problem.. or you just get to the end of a spool. or want to change colours. chewing filament through the hobbed gear you can't trust it. jamming means total disassembly.

i ran a bowden tube once, just to see what it was like. three weeks of absolute hell where literally every single problem i'd ever heard anyone else have, all occurred regularly and repeatedly. yes it was a 3mm bowden. yes i was trying 250mm / sec. yes it was a complete f*******g nightmare and i was both relieved and delighted to have found the flex3drive. bought one immediately and am confused as to why it's not more well-known.

please. do not use a bowden tube with a high-speed design ok?

Edited 1 time(s). Last edit at 08/13/2017 03:19PM by lkcl.

Quote
lkcl

however as i was still quite concerned about rod tolerances changing along the length of any one rod i decided instead to go with twin 6mm rods per axis and LM6UU bearings (total QTY 8). strictly speaking it might even be possible to just use 4 LM6UU bearings but i quite like using 8, it's possible to fit fans (and cables) down the gaps

so that's twin 25mm fans above, through the circular holes. the square ones, i use one to get the E3Dv6 cooling fan duct through it, and the other one i use for cables. the carriage is i think 80mm wide or so, but if you look closely you see the corners are truncated.

btw don't make the mistake i did, make sure that the carriage can be placed without disassembly onto the bearings, don't get fancy, use cableties or some screw-based plastic attachment underneath ffor clamping onto the bearings.

i initially made the mistake of designing the bearings to be push-fit into the carriage, asseembled everything, locked the rods down into the x/y-ends (16 35mm M3 screws in total, takes a hell of a long time....) and then realised FRICK!! i have to redesign the carriage to make the fan holes bigger...

... but i couldn't get the damn carriage off without either disassembling the *ENTIRE* top... or *breaking* the damn carriage plastic, destroying it in the process.

so... do consider that ok?

Over the weekend I did a little test. I took that 1 CF tube that was out of spec and taped it into the groove of a piece of extrusion that I verified was perfectly straight. Then I vacuum sealed the entire thing and baked it at about 120c for a few hours. I then let it cool slowly in the oven without disturbing it. When I took it out, it was near perfect. It was straight and the diameter varied no more than 10 microns across the entire 500mm length.

The carriage and end sliders are all just proof of concepts really. The concept being that, with the tools I had, I could create parts that were close enough in tolerance that the entire assembly was square and parallel enough that the carriage and sliders wouldn't bind. Mission accomplished. In reality, the carriage itself will be a flat surface with Igus low profile pillow bearings similar to the following...



The printed pillow blocks are just placeholders of the same size as the Igus ones. The hotend assembly will be its own bracket and will secure to the base with locating pins and screws.

Quote
lkcl

Quote
gtj
*] The center of gravity for the hotend (bowden setup) will be right in the center of the tool platform.

What am I missing?

please don't make this a bowden. look up the flexx3drive instead: mutley3d does a very small extruder originally designed for a kossel mini which will have a good enough weight for your needs. please don't use the zesty nimble, either.

I was actually thinking about some sort of remote or split drive. I had a crazy idea of some really light piezo driven thing at the hot-end for accuracy while a stepper mounted to the frame handled the heavy lifting. I'll check out the flexx3drive.

Quote
gtj
Over the weekend I did a little test. I took that 1 CF tube that was out of spec and taped it into the groove of a piece of extrusion that I verified was perfectly straight. Then I vacuum sealed the entire thing and baked it at about 120c for a few hours. I then let it cool slowly in the oven without disturbing it. When I took it out, it was near perfect. It was straight and the diameter varied no more than 10 microns across the entire 500mm length.

cooool that's pretty amazing. i wanna know where you got those rods

ok next major concern: 500mm is one hell of a long way. normally it's recommended to use 8020 extrusion for such massive distances. 8mm solid steel rods can barely manage 300mm without bending significantly: 8mm rods are therefore NOT recommended for use in 300x300mm printers. and that's *steel* rods.

so the next question would be: how much bend do you get in the middle of those rods if you place say.... 250 grams of weight in the middle, compared to a straight rod with no weight?

the answer to that is going to be absolutely critical as to whether you'll be able to use the carbon fibre rods at all at this sort of span (which is very ambitious, btw. 300mm using 8mm rods you would almost certainly get away with it).

remember that for supporting weight the deflection is a square law (roughly) with length. so if you had only a 250mm span you would be able to support FOUR times as much weight for the same amount of deflection at the middle. also, the middle is your worst-case.

Quote

The carriage and end sliders are all just proof of concepts really. The concept being that, with the tools I had, I could create parts that were close enough in tolerance that the entire assembly was square and parallel enough that the carriage and sliders wouldn't bind. Mission accomplished. In reality, the carriage itself will be a flat surface with Igus low profile pillow bearings similar to the following...


The printed pillow blocks are just placeholders of the same size as the Igus ones. The hotend assembly will be its own bracket and will secure to the base with locating pins and screws.

my point is: if you want to get that carriage off (the one described as a flat surface with low profile pillow bearings) do make absolutely absolutely sure that you can get the damn thing off.... *without* total disassembly of the x-y gantry. i.e. that the carriage itself is some form of multi-part assembly which can be bolted, snapped or cable-tied to the igus bearings *WITHOUT* needing to take the rods out the bearings, which will involve being able to take the rods out of the x-y gantry, which will involve disassembly of the x-y ends of some kind (which, in my case involved 16 35mm screws), by which time you've spent about an HOUR on a task that should have taken 3-5 minutes. total pain in the ass

anyway i mention that because it's extra weight.

Quote
gtj

I was actually thinking about some sort of remote or split drive.

yep. flex3drive. automotive-grade flexible driveshaft (750 to 950mm long), 40:1 wormdrive in the extruder.

Quote

I had a crazy idea of some really light piezo driven thing at the hot-end for accuracy while a stepper mounted to the frame handled the heavy lifting. I'll check out the flexx3drive.

*thumbs-up*. i've never heard of piezo-driven hotends: i've heard of sensors in the hot-end to measure actual filament extruded (proprietary....) but nothing about piezo-driven hotends. if you find something i'd be interested to know.

With 250g suspended in the middle of a 500mm span, I could hear the rods laughing at me. The rods I reinforced with an epoxy/balsa slurry were hysterical.

My caliper displays to only 10 microns and since measuring sag with contact-type tools is kinda risky, I took 10 measurements unloaded then 10 loaded. In both unloaded and loaded I got a swing of about 35 microns and the average loaded was 10 microns more sag than the average unloaded. I then repeated the whole process and this time loaded was 10 microns less sag than unloaded. Given methods and tool precision, I don't see that as statistically significant.

The CF rods are Arris 3K Rolled from Amazon.
[smile.amazon.com]

Also, I've made up my mind to go with the Drylin N17 linear slides for the perimeter since the rails obviously don't mode and the weight of the slider would balance with the weight I could remove from my printed ones. That leaves only the 4 rods that make up the cross. Making sure that the force is equally distributed across all 4 will be interesting.

The final pillow blocks would screw to the carriage plate with M3 screws into M3 press-in nuts on the other side of the plate.

I think the piezo thing would have to be some kind of ratchet and pawl thing. It wouldn't be able to pull the filament all the way from the spool (or push it back), the fixed stepper would handle the bulk of the force needed to pull the filament up from the spool and over the gravity hump. The piezo thing would just be for accurate metering.

The Flex3Drive 4th gen XY version looks cool if we can lop off the bearings.

Quote
gtj
The Flex3Drive 4th gen XY version looks cool if we can lop off the bearings.

remember there's several different designs, for different types of printers, but you could always design your own: the nice thing however about what mutley3d supplies is, they're all done in polycarb. also if you ask nicely he does actually do custom-modifications... i'm just amazed by his generosity and expertise: it's... rather unfortunate that some people take serious advantage of that. oops.

Quote
gtj
With 250g suspended in the middle of a 500mm span, I could hear the rods laughing at me. The rods I reinforced with an epoxy/balsa slurry were hysterical.

that's an absolutely fascinating addition. balsa wood as a light-weight filler i take it? that's.... totally ingenious. i love it.

Quote

My caliper displays to only 10 microns and since measuring sag with contact-type tools is kinda risky, I took 10 measurements unloaded then 10 loaded. In both unloaded and loaded I got a swing of about 35 microns and the average loaded was 10 microns more sag than the average unloaded. I then repeated the whole process and this time loaded was 10 microns less sag than unloaded. Given methods and tool precision, I don't see that as statistically significant.

daiyhm

Quote

The CF rods are Arris 3K Rolled from Amazon.
[smile.amazon.com]

really appreciated. made a note.

Quote

Also, I've made up my mind to go with the Drylin N17 linear slides for the perimeter since the rails obviously don't move

yehyeh. you're not trying to save frame weight.

Quote

and the weight of the slider would balance with the weight I could remove from my printed ones. That leaves only the 4 rods that make up the cross. Making sure that the force is equally distributed across all 4 will be interesting.

that's why, after a lot of arseing about, i'm going with that ultimaker-style. it's actually what's used in industry for professional equipment: it's just... not very common in the 3D printing community.

Quote

The final pillow blocks would screw to the carriage plate with M3 screws into M3 press-in nuts on the other side of the plate.

sounds very sensible to me...

Quote

I think the piezo thing would have to be some kind of ratchet and pawl thing. It wouldn't be able to pull the filament all the way from the spool (or push it back), the fixed stepper would handle the bulk of the force needed to pull the filament up from the spool and over the gravity hump. The piezo thing would just be for accurate metering.

i did some cautious research, apparently piezo is being researched for nanotech / bio-engineering but nothing stood out in the mainstream 3d printing world. some inkjet stuff as well. don't want to discourage you but it does sound... either highly experimental/risky or insanely expensive or both

Edited 1 time(s). Last edit at 08/16/2017 03:37PM by lkcl.

Quote
lkcl

Quote
gtj
I think the piezo thing would have to be some kind of ratchet and pawl thing. It wouldn't be able to pull the filament all the way from the spool (or push it back), the fixed stepper would handle the bulk of the force needed to pull the filament up from the spool and over the gravity hump. The piezo thing would just be for accurate metering.

i did some cautious research, apparently piezo is being researched for nanotech / bio-engineering but nothing stood out in the mainstream 3d printing world. some inkjet stuff as well. don't want to discourage you but it does sound... either highly experimental/risky or insanely expensive or both

Oh yeah, this would be a side experiment completely separate from the build. Maybe I'll just patent the idea and hand it off to a patent troll.