Vertical X-axis standard !And Contest W/ Prize Updates

The spacing would need to be exactly correct to get even distribution, which it never will be. If the carriage is slightly too short all the weight will be on the top bar and it it slightly too high all the weight will be on the bottom bar.

Other than shorter printing time I can't see the attraction. It seems a less suitable platform for mounting extruders, etc.

---

[www.hydraraptor.blogspot.com]

Actually the printing time is the main reason for me since that is the primary design goal of my test tube mendel. That and elimination of the fragile Z smooth rod bushing tower of the Prusa design. I have broken more than one of those and had to glue it back together. Others have gone the vertical route for their own reasons. There were enough projects adopting that configuration that I thought we ought to get together and hash out a standard for interoperability. I guess time will tell whether it's a good idea overall or not.

As for load balancing, I can see your point for an ideal system with absolutely rigid rods, but I wonder if that will be true IRL. Seems like with a little flex of the rods, mounts etc, you could still get back to a fairly even balance of loads on each rod. It might not be ideal from that point of view, but perhaps the other advantages will "balance out" the unbalanced load issue

Yes if they flex they will balance the load but the whole point is that we don't want them to flex. I must admit I don't know how big an issue it is IRL, mechanical engineering is not my strong point really, I only play at it.

I do feel that the current evolution has gone too far towards less printing time and has compromised the functionality and reliability. I am evolving the design in the opposite direction.

---

[www.hydraraptor.blogspot.com]

I agree on the whole, but I don't know that you can't have both. It seems like there is still room for reduced print designs that don't compromise on stiffness/strength. It may be that this particular design branch does need thicker smooth rod, but that may still be an overall win if the print times are significantly reduced (depending on the value of your time, $, and complexity multipliers) , and some of the torsional axes may be even stiffer than a Prusa while at the same time using less plastic.

Quote

Yes if they flex they will balance the load but the whole point is that we don't want them to flex.

Now, mechanics loaded with weight, accelerations or other forces always flexes. If you want it or not. What you probably mean here is, we want to keep flexing a small as possible.

- One thing is, as soon as both rods are not properly aligned, friction will increase. So you can assume mostly aligned rods.

- Another thing is, the weight of the extruder is constant, so it always makes the mechanics flex exactly the same amount. The static weight of the extruder doesn't matter much.

- There's substantial acceleration. _This_ is what needs to be compensated. For example, it matters where the toothed belt is mounted, especially how close this is to the center of the extruder's gravity, and where the motor puts the counter-force into the frame.

- Flexing not only varies by force, but even more by distance. Accordingly, the bigger the plastic parts are, the more they flex. So, smaller parts do not only print faster, they also flex less - as a rule of thumb.

- It's not so important how much a single part flexes, it matters how much the bed flexes compared to the extruder tip.

With the last two, I see good chances for enhancements by a vertical X axis. The rods for the X axis are much closer to the rods of the Z axis, so I expect less bending.


BTW., I hope you don't recognize this long list as a blow in your face, I just tried to outline a few basic rules.


BTW2: I like the idea of doing a small scale contest.

---

Generation 7 Electronics	Teacup Firmware	RepRap DIY

Quote

Another thing is, the weight of the extruder is constant, so it always makes the mechanics flex exactly the same amount.

It flexes more when the extruder is in the middle, so the nozzle gets closer to the bed in the centre. That is the main deficiency of the current X axis design in my opinion. The rods are not thick enough to support an extruder that weighs about 0.5Kg. I think the y-axis has the opposite effect where it flexes more when the bed it at the edge, so the nozzle to bed distance is a saddle curve even when the bed is perfectly flat and level.

Quote

Flexing not only varies by force, but even more by distance. Accordingly, the bigger the plastic parts are, the more they flex. So, smaller parts do not only print faster, they also flex less - as a rule of thumb.

On the original Mendel I don't have a problem with the plastic parts flexing, just the rods and the studding. On the Prusa I think the only plastic part that has a problem with flexing is the y- motor bracket, possibly the x-ends to a lesser extent. On my redesign I have got rid of the studding, upped the rods to 10mm and made the plastic parts strong shapes, so I don't anticipate them flexing significantly.

Quote

BTW., I hope you don't recognize this long list as a blow in your face, I just tried to outline a few basic rules.

No not at all, not anything I wasn't aware of already.

---

[www.hydraraptor.blogspot.com]

And it's on like donkey kong! There are now 4 contestants so the contest carries. Deadline to enter is Oct 12.

[reprap.org]

I dont think its all that easy and clear as it was presented, i think truth is somewhere down the middle, hence i write this to try to get on bottom of this stuff.

I apparently disturbed a guy with a post latelly, and i want to be a little clear, so this time i should add this. DISCLAIMER: The following (and any other post i write) is obviously my opinion, pov, etc, only! I say it now so i dont just have to say it each sentence. I know as a matter of fact that i am stupid (hence, my nick). Stuff i dont know could fill up 10 universes like the one we live in. So i could be all bad, wrong, etc, whatever. Nobody should use my head. Its mine and only mine to use. Everyone should use theirs. But enough with this and here we go.

There is a static deformation under load. It is usually calculated midway between supports (as max point). Big concern for all structural stuff, for example in arhitecture. Its not important by itself, as it only shows how the castle card stays in place when no wind blows, nothing more. So it is a mere starting point, because the static deformation is well ... static.

Many other things will happen which are not included in static values. If for example, we want to see how a beam like that in a building, would behave in an earthquake, static deformation wont help. We would have to make it dynamical deformation, which is more complex issue as it tries to estimate the momentum value at which it will turn into a plastic deformation, e.g. rupture or a deformation that wont revert to its initial state (energy ends in these cases, compared to elastic deformation in which energy is just reflected). In extremis (idealized situations), there are 2 momentum forumlas for impacts, for elastic impact and static impact, while both include mass, the difference is that one has velocity square (=m*v^2) and the other has velocity double (=2m*v). These are idealized situation (extremes), because afaik most of the real world situations will have 2 parts of each formulas, more one or the other, more or less closer to one type, but basically things will fit in between. As far as we are concerned for our printer we will have to deal with elastic formula, which is the one with square value. Basically bars will deform elastically, after which will regain the former shape. All beams have more or less static deformation. Minimal values, but will have some, even with no load, under their own weight. What we care is the change in it under dynamic forces, and in ideal elastic conditions (e.g. parts will have infinite life).

The fact that we are gonna use elastic formula gives the following: the more speed the carriage will have before it stops, than the dynamic deformation will increase with the square of it. In reverse, the slimmer the bars are, the lesser speeds and accelerations we can use in our machine.

Further on, I consider the dynamical part as incredibly tricky. Basically would be what happens when the carriage has a certain speed and then stops suddenly. Momentum is transmitted in the rod, trying to deform it.

To be correct, when the moment goes into the rods, it moment splits, and the vectors (direction) of how it splits is hard to estimate. In fact, i believe both Z rods and X rods will get deformed, each more or less. I think its complex to see exactly what the new vectors are. First of all, its like being over a pool table and trying to model how the pool balls will go and at which speed/direction each. Or better yet, when a car at a certain speed hits the breaks, the front suspension will lower itself to lowest position (our planet = infinite mass). These are extremes but our case is not like that. Second in our case, even with such a model, it would require us to know exact mass of every component, basically the overall machine, mostly because bars are supposed to make a solid body with the overall frame. Which again is not quite so, and almost impossible to estimate, since for a start, everybody is using different type of parts, uses more or less nuts, and even the thread rods will have different metal compositions and be significantly different in weight. Not to mention that the parts could be ABS or PLA, and both could of been printed at different infill solidities, which both can be treated as making the joints to be more or less elastic types, and finally the frame being a solid body to itself isnt quite so. Only to compute the equivalent masses of a complex assembly that uses somewhat elastic parts and joints it would take a very hardcore study. On the other hand if we consider everything in ideal conditions, and entire momentum only acting on X bar, we would probably end up with a gross overestimate with little relevance

So its basically rule of thumb + safetly margin time. Like i remember using approx 2-3uF for each amp in a linear psu (although its a lame example as i believe this could be fairly easy calculated in joules and maybe few other ways etc), but i also remember that for a cnc cartesian system, like common sense size and speeds, to consider the static deformation to be 10x times lower than the smallest significant distance (i believe 0.1mm for us).

Dunno how this thumb rule relates to our situation. There are red flags about how relevant this rule of thumb is (in both senses). First of all, for that typical machine, the Z is an axis to itself, and it can jog it during the process, so the minimum Z height is "more" relevant for them than it is for us. For us Z is just a layer change, so maybe we have slightly more room for errors in it. If we add the warping effect, then it would probably be of no relevance at all. Also, compared to a machine made from thick / proper sized aluminium, our 8mm rod skeleton is lots of times less rigid, which can be considered both good and bad from different perspectives. As long as deformation goes, who knows, maybe its all good because the horizontal momentum vector (perpendicular on Z) will end up bigger and the vertical one who makes the vertical deformation on X bar will be less.


So here we go. Static deformation is easy enough to compute. Static deformation under load, for a (single!) rod of 55 cm length, and a carriage of 2kg (my carriage is 1,2kg, i weighted some time ago, but also thinking of those trying with 2 extruders; btw, that 0.5kg carriage is a bowden one or what kind?):

fi8mm.....................fi10mm.....................fi12mm
0.0336.....................0.0137.....................0.0067 (mm)

Notably here the values are for regular steel. Our smooth rods are actually induction - hardened so are ofc different, but i have no information /coef for them. So real values might be lower than that although general idea should stand the same.

Classical setup comments:
So, with (many!) reserves regarding the accuracy or even the validity of this estimation of static deformation x10 being smaller than smallest distance, we can see that if there would be a single rod, it would have to be at least 12mm to fit the rule. And if we exclude the linear bearings / bushings, e.g. like the classical mendel with its small bearings, then its clear why the 8mm rods are horizontal, in fact it probably couldnt of been vertical because with 360+180 constraints, the deformation would of been 100% on the 360 bar. And looking at values, even with the classical setup and perfect distribution on 2 X-bars, it still does not fit the rule. But well, we overestimated some, so maybe its allright, but just at the limit of what it should be. But i am inclined to say that 8mm setup clearly shouldnt be any longer than it is, even with 2 horizontal bars.

New setups comments:
- If we use the classic bearings: bars should be at least 12mm
- If we use the pla bushings: supposing these should have increased wear in their top points, untill the maximum contact would be reached, so at a certain point an equal weight distribution on both bars can be assumed
- If we use linear bearings, as things cant get perfect and these wont subdue easily, i would like to assume and inequal weight distrubution, plus a premature wear on the bearings

Now, the funny stuff, prices:
- supplier price of 1m of hardened and chrome plated ( ! std WV) guiding bar: fi8mm=11.98 eur/m; and fi12=12.83 eur/m
(note on bar prices: the bars not chrome plated are at half price, but well, those are crap)
- supplier price of linear bearings: LME 8 UU = 5,50 Euro/buc and LME 12 UU - 5,60 Euro/buc

About fi 10, is not a good option because the linear bearings are open type, and not like LME xx UU. Well that is my supplier at least. For some reason they dont make LME 10 UU, but instead LME 10A UU, which is not common and more expensive than 12, as it has metal seals instead of resin. Different one is like KH 1026 PP (INA) (6.80 eur !) and its a different model, again metal seals and you can see the balls from the outside. I think this one is designed to be encased in a custom case which will have a hole to one side, made for lubrication with a special tool that i dont know its english name. Could make a case from plastic like that, but i doubt adding lubricant in abs or pla case, for the metal bearing inside, would work on long term, as it usually its contradictory, if its ok for one side (either metal or plastic), will probably attack the other one. Bottom line, about fi10 and fi12, the fi10 bearing is an issue which usually can make the fi10 setup more expensive than fi12.

So, the question is why do we use fi8 in the first place - or at all?
12 mm bars would add ... less than 3 eur to overall machine?
and 12mm linear bearings would add ... 10 cents each?
its like ... are we serious about our machine or what?

There is no question in my mind that a proper machine will use fi12. But well, as i said before, thats me.
Only imaginable downsize is that for each X and Y stage, using 12mm linear bearings would add a few grams .... but well, we should take that out from somewhere else. Even in the classical horizontal setup, i would rather use fi12 instead of 8, adds alot to rigidity and i would consider it a case of huge benefits for some insignificant price.

For the closing, as i said in the opening, thats my opinion only ... blah..blah..blah.

Edited 1 time(s). Last edit at 09/14/2011 07:51PM by NoobMan.

Just posted and seen the size of it. So, if somebody skipped the wall of text before (which is .... i suspect... everybody) ... here is the conclusion:

- ! using fi12mm rods and bearings (instead of fi8 ones), at supplier prices, will add ... less than 4 eur to entire machine, for a huge gain in material size and much more other benefits.

Could work with 8mm threaded rods structure vertexes etc, but if it would be up to me i would of chosen 12mm there also. In cnc world prices like for any ball screw, or trapezoidal screw, or linear slide, etc are just .... uh, no comparison or further comment possible... 4 eur is really dirt in comparison.

And a secondary thing on the bigger picture, the deformation is not the same. Its directly proportional to the mass of the moving assembly, but its squared value to the speed of the carriage has before it suddenly stops / breaks / decelerates to a halt (i guess teacup ramping likely smoothens it). Also it splits between Z and X rods, so estimating how much only X gets is also a vector (has direction, pivots, gravity centers, angrenating point, and Z part which oposes it depends on the total machine mass to which its supposed to be considered a body width, but this can be an issue because the structure is more elastic in joints instead of rigid, so it doesnt quite make a proper body with it, etc).

Basically splitting deformation between Z and X vectors is frankly way beyond me. For example if geometric center has all mass, the X deform will be min and Z max. If mass would be 10m in the air with an ideal perfekt rigid link to carriage, it can pivot as a lever and the X bars would probably break as toothpicks from a 10mm/s sudden stop (just more probably wont even start moving in the first place in such conditions). Otherwise would be simple to say that kinetic conserves, the deformation momentum amount is know (m*v^2), if we know how to split it in two, then exact deformation can be figured on both directions.

But in the general dynamical picture, the most important factor which increases the deformation is speed (squared).

Rule of thumb for an acceptable dynamic deformation on an axis, the max static deformation should be ~10 times lower than accuracy planned. Sry i just dont recall the source of this, and no clue how relevant is for us coz we are so totally different than a rigid and proper cnc structure, so this should be considered with a good amount of skepticism.

Again, just my opinion, could be very bad ... bla-bla-bla etc.

Edited 5 time(s). Last edit at 09/14/2011 08:34PM by NoobMan.

You're right about the prices. Zapp Automation [www.zappautomation.co.uk] sell SFC8/10/12/13/16 Precision round rail are all £1.20 per 100mm, LM8UU and LM10UU bearings are £3.20 each, LM12UU are £2.50 and LM16UU are £2.60. So if you buy from them, 12mm is actually cheaper!

Perhaps the best way forward is to allow people the choice of which size to use, by building parametric axis-ends and carriages. It's what I'm aiming to do.

droftarts Wrote:
-------------------------------------------------------
> You're right about the prices. Zapp Automation
> [www.zappautomation.co.uk] sell
> SFC8/10/12/13/16 Precision round rail are all
> £1.20 per 100mm, LM8UU and LM10UU bearings are
> £3.20 each, LM12UU are £2.50 and LM16UU are
> £2.60. So if you buy from them, 12mm is actually
> cheaper!

£1.20 per meter ...huh!
From the price those rails are clearly not chrome plated. By all means spend like a little extra and buy chrome plated. This way at least ... wont rust!!!. Otherwise you will go on vacation and after 1 month when you come back you will have small dots of rust on the rails. Unless you live in a really dry place. Also, nobody says stuff like this, but the ones that arent chrome plated are crappy, not to mention that they are usually much worse tolerances, some probably dont even make it into the worst tolerance class.
sry didnt seen its price per ... 100mm ... lol and checked the sheet and are chrome plated and good durity ~60+hrc
where i get from its always 1m, i couldnt buy less even if i would want to


> Perhaps the best way forward is to allow people
> the choice of which size to use, by building
> parametric axis-ends and carriages. It's what I'm
> aiming to do.

If ppls would be advised of all aspects, ofc. But in most cases probably just wont be like that. Look at what i said about 8mm vertical, classical setup (100% on one rod only), i had 0.0336 mm static deformation. Static as in "just sits there". Well asuming that 10x rule of thumb is adequate and not too far off, that turns into 0.336 mm deformation in dynamic conditions. HUGEEEE, 0.336 mm its like a layer height. Which means when you put the skirt, or first layer, if the tip is like only 0.3mm above bed, along a line it will travel ok, but each time it takes a corner it might hit the bed. You know, it might not be so bad since it will keep punching the plastic part, maybe it will end up with less warping effect on large parts .... hahahaha. Not to mention that unlike a car's springs which come with "telescopes" - when our "spring" releases there is nothing to slow it down, so that means vibrations. And 0.336 mm deform turned into vibration cant be all that healthy. So, if we make parametric a system which is calculated for fi12mm, its nice, but somebody will want at a point to make it 8mm, and can ran into problems easily, so a downgrade warning ought to be somewhere, which complicates things a little.

Those values were only for a single rod context. For the setup we use now, 2*8mm bars splitting the load, its a static deform of 0.0168mm only, which does fit allright-ish on the overall, especially considering each parameter had a little extra. Again i do know those values are probably more or less off from the real thing, but in lack of any better i would orient myself from what i can, and work with what i have.

So the parametric stuff: its very nice to be able to adapt a piece to needs of more ppls. Great, really. But in the lack of such, if there would be a "standard", i would opt for fi12mm stuff instead of 8 ... any time, every time or circumstance. Again my opinion.

Edited 6 time(s). Last edit at 09/14/2011 10:10PM by NoobMan.

Just an observation, in favor of vertical axis : [youtu.be] (hit pause)
The vertical configuration of x-axis seems to be quite common in the industry (due to probably some good reasons)

Someone may have said that somewhere, but that hit me when looking at the video, and thinking back at all the 2D printer interior I remember. Also with the belt between the rods (which is certainly good for transmission as said above).

(can't wait to have more time and finish the v0.7 of the Vert-X^^)

The choice of orientation of X axis rods in commercial machinery has more to do with compactness of form and ease of assembly/access to the work head than any performance characteristic. I do think the vertical X axis provides some simplification opportunities for Mendel machines, but there isn't any inherent performance advantage that couldn't be obtained with a horizontal orientation. There is room for improvement.

Hello everyone. In our design we took care of the weight distribution. Brian is telling the truth. Details are printed quickly. Extruder disassembled very easily and it is very compact.

[www.reprap.org] This.

Looks nice! Can you comment on time savings on printed parts for your x axis parts? Looks like the extruder will be significantly faster to print than a prusa carriage and wades extruder.

Also how do the rod spacings compare to the standard that we're trying to get started. Any interest in the contest

I don't remember exactly. X axis motor holder about 60 min( 60 mm/s speed ), X axis bearing holder about 40 min, extruder about 40 min and x carriage about 50 min. I can write more accurately tomorrow. About rod spacings tomorrow too. Your comments?)

Probably saving a couple hours print time by my rough estimate.

In our version we have 30mm between the z rods also.
see www.skb-kiparis.ru.

I apologize for the delay in response. Too busy now.

I understand about being busy. It looked like the Z motor mounts were pretty standard, so I pretty much expected the z rods to be 30 mm. What is the spacing between the X smooth rods? And good luck with the new store. It's exciting to see so many electronics designs and new bot designs running around these days.

Trying to do any calculation on dynamic deformation is a waste of time at this point, as it will entirely depend on belt attachment points, mass distribution, vibratory modes, and end conditions of the smooth rods. Did I mention it is also super complicated?

Really, the only difference between vertical and horizontal mounting is cleanliness/simplicity of design (which is semi opinion based.) Also, because the horizontal rods can be mounted very closely to the Z axis, the deformation of the X/Z mounting is minimized. Weight will be pretty evenly distributed between the two rods regardless of orientation.

I've made a number of printers with 8mm and 12mm rods in vertical and horizontal orientations, and hands down the 12mm rods with vertical alignment are just better to work with (for me)

Anyway, the orientation of the axis is far less important to printer performance than the rigidity of the frame and having high enough belt tension.

Andrew Diehl Wrote:
-------------------------------------------------------
> [...]Weight will be pretty
> evenly distributed between the two rods regardless
> of orientation.

If we look at "official" stuff, the constraints on the 2 rods use small ball bearings (radial), e.g. 3+3 bearings for one rod to get a 360deg constraint and 2+2 bearings on the other rod to get a 180 constraint. The latter needs to have the bearings perpendicular to the plane where rods are, e.g. to allow changes in the distance between the two smooth rods - misalignement. The only way to do that with a vertical setup is to put these latter 2 bearings to be horizontal to ground. Which means in lack of the 360 deg constraint, the assembly will fall directly to ground. Nothing to stop it falling except the 360 deg constraint.

Even in the case of linear bearings, thinking that weight will distribute perfectly is a utopia, the distances will never be perfect like that. In case of the horizontal setup however, they dont have to be perfect, it doesnt matter since the normal to ground is perpendicular to the rods plane, so it will distribute evenly regardless of anything else, as long as the rods plane is fairly horizontal to ground.

Edited 2 time(s). Last edit at 10/04/2011 12:13PM by NoobMan.

I agree with Andrew, it is far more complicated than most people are assuming, and therefore there's not much point in arguing the physics, it will come down to real world results in the end. For instance, a horizontal x axis does not automatically distribute the weight evenly. On my prusa the center of mass of the extruder/hotend is roughly 1 cm from the center of the bar closest to the belt. So static weight distribution would be about 80% on that rod and 20% on the opposite rod. Now start torquing on that system back and forth with a belt attached on the other side of the bar, and tell me where all the forces end up, or even just tell me the max forces. Next calculate what the deformations will be and then a simple geometrical transformation will get you the movement of the tip of the print head. Is it within a tolerance of 10% of the width of the extruded plastic? Is it dwarfed by the flexing of the frame?

Therefore, you should all be designing vertical X axis parts for the contest so you can try it out to see if it's better or worse than what now exists.

bryanandaimee Wrote:
-------------------------------------------------------
> [...]Therefore, you should all be designing vertical X
> axis parts for the contest so you can try it out
> to see if it's better or worse than what now
> exists.

I never intended to defuse ppls from it, although i can see now maybe it could be interpreted like that - i was just trying to have the correct perspective. I can join in and see what i can think of getting done, if i can make up for it this way. Do late starters get any cookies?

Hmm, cookies, I've been wondering what to add to the prize pool. You may be on to something. An no offense taken, I enjoy a lively discussion as much as the next guy.

I'll join then on the last 100m, and i just read the stuff so i think i should say ahead that i dunno if i can meet the deadline, as i am usually very slow at doing things. Doesnt matter for me as i would do it primarily as experiment for myself, but just saying it now for the fairness of things. Also i will most probably go for 12mm stuff instead, havent noticed anything about that yet.

Actually, Oct 12 is only the deadline to enter. Nov 12 is the end of the contest, so there is still quite a bit of time.

You just need to post a link to your project page by Oct 12 to the contest page. It doesn't have to have anything on it by then.

Contest Page

Yup i read that and I was refering to 12 nov - i can get *that* slow
For the time being I made a page and will update it there - will see how it goes.

Right now it looks like i will go for a little oversized machine, and about the x axes i will theoretically use 4 rods in a square pattern 7cm each side, in order to get a second x axis with its own carriage. The second axis arrangement will get to be symmetrical to the other one, mirrored both ways. This way the rods can be vertical, and also could be used in a horizontal (classical) arrangement.

The second axis could be used for experimental toolheads and such, to be toyed with, without having to interfere with first toolhead. It could let ppls have more fun with the printer.

The horizontal distance between 8mm rods on my mendel is 50, but i increased for the square config at 70 horizontally aswell as vertically. Not sure of it yet, because at 70 the rods are pretty far away from Z drive and it means lots of extra plastic so it might not be so good, maybe i will revert to a smaller distance even, but dont know it yet.

About t5 belts, probably will have 10mm. And certainly some mod like the "doubler accuracy mod", or at least have the flexibility to be arranged like so.

This is what i came up with, so i hope its allright so far.

Any feedback on the general picture of what i intend to do?

Edited 2 time(s). Last edit at 10/05/2011 06:31PM by NoobMan.