what's the current best engineering practice for stopping lead screw lateral travel?

hi folks,

i'm working on the sandwich200v2 compact portable printer, and will be using 3 belt-driven lead screws with a 900mm closed loop, 3d-printed GT2-6mm-40tooth gears and a GT2-16t on a NEMA17 motor. i've seen an application note about how to hold the lead screw top and bottom with 608 bearings that are pushed in alternate directions using grub screws: for a 300m rod it is recommended to use 2 such bearings top and bottom.

HOWEVER.... i have a set of 16 ABEC-5 skate bearings (608) that have done a few thousand miles over the past 10 years, and i don't know if they're different from the kinds of bearings normally used, but they have about 0.2 to 0.3mm of lateral travel on the central hub. assuming that even 0.005mm is an unacceptable amount of lateral travel in a 3d printer's z lead screw (such that the fact that i am using ABEC-5 skating bearings is irrelevant), my question is: what's the best current engineering practice used in 3d printing and industry generally to stop lateral travel (along the axis of the lead screw itself) in lead screws?

i am not so concerned about taking up slack in the x and y planes, as that's dealt with by both the belt as well as being partially irrelevant because of the way that the printbed is supported (2 rigid 12mm bars with LM12UU linear bearings). i would however *not* like to rely on gravity or any other method that would result in wear and tear of the frame, rod or bearings, so resting the bottom of the rod against the aluminium of the frame strikes me as not being particularly sensible, but i would be interested to hear from people as to whether that would in fact be a suitable method.

if anyone has specifically solved this issue or knows of a printer that has i'd be interested to know how.

---

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

It sounds like those bearings are worn out. I'd look at getting some new ones- they're very cheap.

If the screws were horizontal you'd have to do something about it to prevent backlash from the whole screw moving back and forth whenever it reversed direction. The Z scews are vertical, so I don't think it matters if you take extra pains to ensure the bearings and screws are forced to one position. Gravity will be pushing the bed, undercarriage, screws, and bearings to their lowest possible positions.

---

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

See how I modified my Prusa Z system. 608 bearings do the jobs. The Z axis moves only one way and slowly. Deep groove bearings can take axial load and they are far below their specs here.
These bearings are used because they are common and dirt cheap.

Geetech overhaul

doing some research, cross-reference to prior discussions, quick note:
[forums.reprap.org]

this post talks about "pre-loading" the bearings
[www.cnczone.com]

Quote
the_digital_dentist
It sounds like those bearings are worn out. I'd look at getting some new ones- they're very cheap.

i need an absolute minimum of 6, preferably 9 even 12, so it starts to become an expensive addition. might look at wedo3dprinting KFL08 self-aligning flange bearings. yeah thank you replying dd, i thought about replacing them, but have checked them carefully, and thought about it. when the lead screw's vertical with these bearings there's no x-y travel - the central hub doesn't rattle about, especially when i've tightened up the two grub screws to lock the rod in place. the z-travel is what was concerning me.

Quote

If the screws were horizontal you'd have to do something about it to prevent backlash from the whole screw moving back and forth whenever it reversed direction. The Z scews are vertical, so I don't think it matters if you take extra pains to ensure the bearings and screws are forced to one position. Gravity will be pushing the bed, undercarriage, screws, and bearings to their lowest possible positions.

mm nom nom mumble ok. i think i may just be overreacting from a past experience: i'm being extra-paranoid after just having received some slightly out-of-spec LM12UU rods (since kindly replaced), where the bed actually managed to stay jammed on the rods because of them. assuming clean travel.... ok, yes i agree with you. thanks!

Quote
MKSA
See how I modified my Prusa Z system. 608 bearings do the jobs. The Z axis moves only one way and slowly. Deep groove bearings can take axial load and they are far below their specs here.
These bearings are used because they are common and dirt cheap.

Geetech overhaul

okaay, so skate bearings clearly aren't "deep groove" ok found this post by you [forums.reprap.org] you made something called SC8UU_P.stl - i'm curious, there, there's no grub screws to push the bearings sideways so that they trap the rod. have you found there to be any issues without doing that? this is what i made (i forgot to switch on underside print-support.... oh well) but you can see one M3-plus-nut on the left and one lower down on the right.



p.s. i love the idea of successfully loctite-gluing an open belt to make a closed one....

Edited 1 time(s). Last edit at 05/01/2016 04:27AM by lkcl.

Quote
lkcl

Quote
MKSA
See how I modified my Prusa Z system. 608 bearings do the jobs. The Z axis moves only one way and slowly. Deep groove bearings can take axial load and they are far below their specs here.
These bearings are used because they are common and dirt cheap.

Geetech overhaul

okaay, so skate bearings clearly aren't "deep groove" ok found this post by you [forums.reprap.org] you made something called SC8UU_P.stl - i'm curious, there, there's no grub screws to push the bearings sideways so that they trap the rod. have you found there to be any issues without doing that? this is what i made (i forgot to switch on underside print-support.... oh well) but you can see one M3-plus-nut on the left and one lower down on the right.

IMG_0140.jpg

p.s. i love the idea of successfully loctite-gluing an open belt to make a closed one....

Not deep groove ? Anyway, they are not thin and considering their size they will have no problem taking the loads we are dealing with !
No grub screw ? Of course, you never use grub screws to assemble bearings ! Press fit or loctite ! Metal epoxy is also good here.
The part SC8UU_P I designed is to replace the SC8UU with the linear bearing inside held by two O rings to allow some preload and take care of play, misalignment of the two smooth rods.

All what I did works and I print at 120mm/s. 180mm/s for prototype with PLA. You can see the video.

I have to work out the hotend and a few other thing now like cooling ...

And the belt is holding ! I don't need to order one Note that you better not use an idler on the smooth side with this kind of spliced belt but it is a practice to be avoided anyway.

Quote
MKSA
Not deep groove ? Anyway, they are not thin and considering their size they will have no problem taking the loads we are dealing with !

yeah they've been carrying... mmm... well over 85kg of load for a few thousands of miles, now, and will have been doing a few thousand RPM at the time as well: i think they can cope with a couple hundred grams and a couple of revolutions per second for a bit longer

Quote

No grub screw ? Of course, you never use grub screws to assemble bearings ! Press fit or loctite ! Metal epoxy is also good here.

that goes directly against the advice of best engineering practice that i found online, which explains that the rods can slip in the central hub. imagine that there is a tiny amount of offset between the top and the bottom end-supports, such that the rod is *not* perpendicular to the bearings. or that there is a slight amount of bend in the rod (end-result the same: rod not perpendicular to bearings). what will happen to the rod? it will end up rotating like a screw, won't it? *especially* if it's a screw-thread (which a TR8 is), it will actually *unscrew* out of the bearings, won't it? that unscrewing will result in unpredictable changes in z-height. the way to stop that is: clamp the rod between a pair of bearings which are pushed from opposing sides, each with a grub screw.

the application note - wish i remembered where it was - then went on to explain that if you go over 300mm you should use *four* bearings per end, each offset alternately left-right-left-right. the housing should be altered (alternate ovals) to allow the bearings to do exactly that.

press-fit of the bearings i have no problem with. metal epoxy? sounds like a maintenance headache, and a recipe for having to throw away parts and/or clean up bearings during initial assembly as well as on a maintenance cycle.

as a one-off i'm sure it would be fine, but for a repeatable build (so that others can copy the design, and you can make more later), relying on the tolerances of rods fitting precisely under friction into the central hub of bearings so that they don't move about, when you *know* that even the slightest offset from perpendicular will result in the rods "unscrewing" as they turn - doesn't strike me as good engineering practice.

Quote
lkcl

Quote
MKSA
Not deep groove ? Anyway, they are not thin and considering their size they will have no problem taking the loads we are dealing with !

yeah they've been carrying... mmm... well over 85kg of load for a few thousands of miles, now, and will have been doing a few thousand RPM at the time as well: i think they can cope with a couple hundred grams and a couple of revolutions per second for a bit longer

Quote

No grub screw ? Of course, you never use grub screws to assemble bearings ! Press fit or loctite ! Metal epoxy is also good here.

that goes directly against the advice of best engineering practice that i found online, which explains that the rods can slip in the central hub. imagine that there is a tiny amount of offset between the top and the bottom end-supports, such that the rod is *not* perpendicular to the bearings. or that there is a slight amount of bend in the rod (end-result the same: rod not perpendicular to bearings). what will happen to the rod? it will end up rotating like a screw, won't it? *especially* if it's a screw-thread (which a TR8 is), it will actually *unscrew* out of the bearings, won't it? that unscrewing will result in unpredictable changes in z-height. the way to stop that is: clamp the rod between a pair of bearings which are pushed from opposing sides, each with a grub screw.

the application note - wish i remembered where it was - then went on to explain that if you go over 300mm you should use *four* bearings per end, each offset alternately left-right-left-right. the housing should be altered (alternate ovals) to allow the bearings to do exactly that.

press-fit of the bearings i have no problem with. metal epoxy? sounds like a maintenance headache, and a recipe for having to throw away parts and/or clean up bearings during initial assembly as well as on a maintenance cycle.

as a one-off i'm sure it would be fine, but for a repeatable build (so that others can copy the design, and you can make more later), relying on the tolerances of rods fitting precisely under friction into the central hub of bearings so that they don't move about, when you *know* that even the slightest offset from perpendicular will result in the rods "unscrewing" as they turn - doesn't strike me as good engineering practice.

Quote
lkcl

Quote
MKSA
Not deep groove ? Anyway, they are not thin and considering their size they will have no problem taking the loads we are dealing with !

yeah they've been carrying... mmm... well over 85kg of load for a few thousands of miles, now, and will have been doing a few thousand RPM at the time as well: i think they can cope with a couple hundred grams and a couple of revolutions per second for a bit longer

Quote

No grub screw ? Of course, you never use grub screws to assemble bearings ! Press fit or loctite ! Metal epoxy is also good here.

that goes directly against the advice of best engineering practice that i found online, which explains that the rods can slip in the central hub. imagine that there is a tiny amount of offset between the top and the bottom end-supports, such that the rod is *not* perpendicular to the bearings. or that there is a slight amount of bend in the rod (end-result the same: rod not perpendicular to bearings). what will happen to the rod? it will end up rotating like a screw, won't it? *especially* if it's a screw-thread (which a TR8 is), it will actually *unscrew* out of the bearings, won't it? that unscrewing will result in unpredictable changes in z-height. the way to stop that is: clamp the rod between a pair of bearings which are pushed from opposing sides, each with a grub screw.

the application note - wish i remembered where it was - then went on to explain that if you go over 300mm you should use *four* bearings per end, each offset alternately left-right-left-right. the housing should be altered (alternate ovals) to allow the bearings to do exactly that.

press-fit of the bearings i have no problem with. metal epoxy? sounds like a maintenance headache, and a recipe for having to throw away parts and/or clean up bearings during initial assembly as well as on a maintenance cycle.

as a one-off i'm sure it would be fine, but for a repeatable build (so that others can copy the design, and you can make more later), relying on the tolerances of rods fitting precisely under friction into the central hub of bearings so that they don't move about, when you *know* that even the slightest offset from perpendicular will result in the rods "unscrewing" as they turn - doesn't strike me as good engineering practice.

Most probably you misread the recommendation or misunderstood what I wrote !
Reread them.
The way you mount the bearing is a big NO ! If you made this based on your readings, there is definitively a problem.Try to tighten the screw and see how your bearing runs. You can squeeze it such that is binds !

An other thing, my explanation are for the Z axis, with the leadscrew hanging from the bearing, with its end free, just the nut below. If the screw slightly wobble (~ 1/10 ), no problem, it will flex to compensate plus the nut has play. The vertical rod is the reference.

If you don't want to use epoxy or loctite to secure the leadscrew in the bearing plus have the pulley attached, you will have to machine the leadscrew accordingly or make the ad hoc part..
I used epoxy and for the pulley, I used the grub screws. Good enough here.

Regarding maintenance, even seen how bearings are installed and replaced in real machines ?
Do you intend to replace these bearings often ???? Install proper ones and they will last forever. Anyway, should I replace them, there would be no problem either.
Do as you wish, I know that what I did is correct , good enough in view of the requirements, and it works.

BTW, 608 are DEEP GROOVE

Edited 3 time(s). Last edit at 05/01/2016 10:18AM by MKSA.

Quote
MKSA
Try to tighten the screw and see how your bearing runs. You can squeeze it such that is binds !

why would i want to do that? i've no intention of over-tightening the screw until it binds the bearing, and no need to do so.

i've already tested the arrangement and i am happy with a reasonable compromise point that is sufficient to bend the plastic [in the x-y plane only, which has already been established to have no detrimental affect on the z height and z travel accuracy], yet not sufficient to prevent the bearings to bind, as you say *may* occur [but does not]. in short: the bearings run fine, whilst at the same time achieving the additional critical goal of holding the rod firmly in place so that it will not travel.

bear in mind that this is a *portable* printer. when stowed the lead screws will be removed from their housings in the frame, with the bearing assembly still attached to the lead screw(s). having the bearing-end-assemblies remain firmly in place is therefore an important design criteria so that on next setup the housings will be in place, as is not using epoxy metal resins or other glues which would require additional special instructions and procedures to explain to non-technically-minded individuals who may wish to build one of these printers. "basic tools only" is part of the design requirements.

so, i appreciate your advice and i respect your design decisions - your advice worked for you, with the design requirements that you have in mind, and for the target audience that you are covering for the specific printer that you are designing.

So if you think that squeezing your leadscrew between two used skateboard bearings stacked together, by means of screws pushing them sideways is good engineering practice, no point continuing this discussion.

Edited 1 time(s). Last edit at 05/01/2016 12:34PM by MKSA.

Quote
MKSA
So if you think that squeezing your leadscrew between two used skateboard bearings stacked together, by means of screws pushing them sideways is good engineering practice, no point continuing this discussion.

so, after i said, "i respect your decision to go with what you feel is best for your own design", you responded with words to the effect of "you are incapable of making rational engineering assessments", which is incredibly insulting. i trust that you recognised that you are completely out of line, will apologise immediately, and will not attempt to directly or indirectly insult or denigrate anyone on this forum ever again.

The Z Axis leadscrew should just impart +Z and -Z on the gantry or bed.
Everything should be done to isolate X, Y movement to gantry or bed.
really should not rely on guide rails for this.
Leadscrew to bed float in X, Y not to move bed in X, Y

So leadscrew nut could float in connection to bed.
or
If leadscrew nut is fixed to bed -- then leadscrew must float.

The leadscrew is easily OVER CONSTRAINED (3 or 4 bearings) leading to binding and X, Y movement.

Many have bearing near motor taking radial / axial movement, bed resting on leadscrew nut
weight of bed holding it down and loaded and other end not constrained at all.

Load in a leadscrew nut (Pre-load) is a method to minimize backlash
--- to force leadscrew threads to always ride on the upper OR lower nut threads and not move back and forth

But hard to question ikcl's mechanical engineering excellence
and answering his own questions at great length with pHD theisis

???? [web.mit.edu]

[www.roton.com]

??? [ocw.mit.edu]

Quote
cozmicray
answering his own questions at great length with pHD theisis

thanks ray - i'm amazed and appreciative of the depth of knowledge demonstrated and available at your fingertips

i think i have an idea, which would do preloading on a pair of bearing hubs, as well as stop the lead screw from travel if friction on the central hub(s) are insufficient to stop it travelling: a sandwich of housing-bearing-collar-bearing-housing. the collar will have grubscrew(s) and the top and bottom housings will have screws to keep them together (loosely but enough to do the preloading on the bearings).

colour-coding: bearings, blue, collar green, top housing cyan, bottom housing red, rod purple, nuts and bolts black



key factors which fulfil the requirements:

• the bearings sit inside the bottom housing under friction but tight fitting is not essential
• the collar has protrusions either end that make it sit against each bearing's central hub NOT the walls of the bearings or the outer hub
• the top housing's purpose is to perform pre-loading and to keep all internal components pushed down, preventing them from moving along the axis of the rod

yeah i have the rod poking through the whole assembly, just realised that, it's not necessary but possible. thoughts appreciated.

grasshoppa

?Why? constrain leadscrew mount motion that doesn't effect anything?

?Does the motor and coupling constrain leadscrew from up/down motion?

? What is a "bearing hub"? an ikclism for something created by ikcl
"stop the lead screw from travel" traveling where which way? is a passport necessary?

You have "a Rod" in the middle of your drawing ? rod on end of leadscrew (unthreaded section)

Use a bearing properly --- inner race connected to inner piece
-- outer race connected to outer piece
let bearings do the work -- and use a bearing designed for the loads you wish to constrain.

got to get the dry ice and torch out to fit them!

What are you calling lateral movement

1: of or relating to the side
2: situated on, directed toward, or coming from the side
3: extending from side to side
4: produced with passage of breath around the side of a constriction formed with the tongue <\l\ is lateral>

This is not complex enough -- perhaps an active system sensor servo system with
micro controller to sense nano lateral motion and compensate
with an IMU and 6 servos would do it.

Quote
lkcl

Quote
cozmicray
answering his own questions at great length with pHD theisis

thanks ray - i'm amazed and appreciative of the depth of knowledge demonstrated and available at your fingertips

i think i have an idea, which would do preloading on a pair of bearing hubs, as well as stop the lead screw from travel if friction on the central hub(s) are insufficient to stop it travelling: a sandwich of housing-bearing-collar-bearing-housing. the collar will have grubscrew(s) and the top and bottom housings will have screws to keep them together (loosely but enough to do the preloading on the bearings).

colour-coding: bearings, blue, collar green, top housing cyan, bottom housing red, rod purple, nuts and bolts black

[attachment 77437 Untitled.png]

key factors which fulfil the requirements:

• the bearings sit inside the bottom housing under friction but tight fitting is not essential
• the collar has protrusions either end that make it sit against each bearing's central hub NOT the walls of the bearings or the outer hub
• the top housing's purpose is to perform pre-loading and to keep all internal components pushed down, preventing them from moving along the axis of the rod

yeah i have the rod poking through the whole assembly, just realised that, it's not necessary but possible. thoughts appreciated.

Quote
cozmicray
grasshoppa

yetth maarthturrrrr?

Quote

?Why? constrain leadscrew mount motion that doesn't effect anything?

it just feels... right.

Quote

?Does the motor and coupling constrain leadscrew from up/down motion?

no. it's belt-driven. there's a GT2-6mm-40tooth 30mm up the leadscrew.

Quote

? What is a "bearing hub"?

a bearing comprises an outer ring, a set of bearings, and a central inner hub. what's the usual phrase for that central inner hub?

Quote

"stop the lead screw from travel" traveling where which way?

along the direction of the axis of the lead screw. so if it's mounted vertically, stop it moving up and down.

Quote

You have "a Rod" in the middle of your drawing ?

rod, lead screw, etc. etc. yes.

Quote

Use a bearing properly --- inner race connected to inner piece
-- outer race connected to outer piece
let bearings do the work -- and use a bearing designed for the loads you wish to constrain.

nice suggestion.. but i can't. spent a whole stack of money very very fast, i've got to go with what i've got, and i'm also running out of time.

Quote

got to get the dry ice and torch out to fit them!

eek! that's beyond the requirements of "must be built without a workshop and without special tools". but i know what you mean - use thermal expansion of metal to get 2 pieces to fit and then stay. requires close tolerances on the parts. also too much to expect here.

Quote

What are you calling lateral movement

1: of or relating to the side
2: situated on, directed toward, or coming from the side
3: extending from side to side
4: produced with passage of breath around the side of a constriction formed with the tongue <\l\ is lateral>

4, definitely.

Quote

This is not complex enough -- perhaps an active system sensor servo system with
micro controller to sense nano lateral motion and compensate
with an IMU and 6 servos would do it.

niiiiice! yyyeah, bit beyond the budget i feel

8mm Mounted Block Cast Housing Self-aligning Pillow Bearing $2.10
[www.ebay.com]

Housing bolted down, leadscrew locked to inner race with two set screws



probably will constrain radial and axial motion

nuff said

Quote
cozmicray
8mm Mounted Block Cast Housing Self-aligning Pillow Bearing $2.10
[www.ebay.com]

looks great, ray: i mentioned i encountered those in my reply to digital_dentist, earlier. except as i said i'm now constrained by time and budget. i'm leaving the premises i am currently in, within the next 3-4 weeks, and will be moving to a different country each month EVERY month for a YEAR. i cannot wait for random ebay suppliers to deliver in a random amount of time to a location where i do not know if i will be at the destination address. and what i now have, i have to work with.

what i like about the design that i came up with, *at a later point* i can drop in some thrust bearings into the bottom housing, and, if using a second housing at the other end (say if the arrangement were used horizontally), then the axial motion is constrained at both ends by the thrust bearings, and that "colllar" becomes redundant, as does preloading the 608 bearings.

anyway, what you found - the KFL08 - is great for anyone else who is not constrained by time and budget.

Quote

what i like about the design that i came up with, *at a later point* i can drop in some thrust bearings into the bottom housing, and, if using a second housing at the other end (say if the arrangement were used horizontally), then the axial motion is constrained at both ends by the thrust bearings, and that "colllar" becomes redundant, as does preloading the 608 bearings.

Yes, that's the direction I'll go next time. For my first attempt I did a halfway house with thrust bearings in a block at the bottom with the top end free floating and I'm not happy with it. Next time I'll use a longer lead screw with turned ends to run in the radial bearings and vertical travel limited by thrust bearings at both ends. I plan on making a complete Z module with lead screw, bearings, smooth rod and frame elements combined, hopefully avoiding the hideous alignment issues I had setting up the first build. The one thing I have concerns over is differential thermal expansion between the steel lead screw and the aluminum frame. If that turns out to be a problem I'll have to substitute a steel frame element.

Quote
JamesK

Quote

what i like about the design that i came up with, *at a later point* i can drop in some thrust bearings into the bottom housing, and, if using a second housing at the other end (say if the arrangement were used horizontally), then the axial motion is constrained at both ends by the thrust bearings, and that "colllar" becomes redundant, as does preloading the 608 bearings.

Yes, that's the direction I'll go next time. For my first attempt I did a halfway house with thrust bearings in a block at the bottom with the top end free floating and I'm not happy with it.

interesting. ok so what you're saying is, relying on gravity *even* with the lead screws vertical simply wasn't good enough. i kinda intuitively thought that might not be a good idea, i just couldn't back it up

Quote

Next time I'll use a longer lead screw with turned ends to run in the radial bearings and vertical travel limited by thrust bearings at both ends. I plan on making a complete Z module with lead screw, bearings, smooth rod and frame elements combined, hopefully avoiding the hideous alignment issues I had setting up the first build. The one thing I have concerns over is differential thermal expansion between the steel lead screw and the aluminum frame. If that turns out to be a problem I'll have to substitute a steel frame element.

eek! that's getting serious, that is, if you're concerned about differences in thermal expansion - what kind of tolerances / accuracy are you looking at? i ask because if you're concerned about it, i should evaluate it as well.

Quote

what kind of tolerances / accuracy are you looking at? i ask because if you're concerned about it, i should evaluate it as well.

When I made these for my first build I tightened them up so that there wasn't any discernible vertical play in them



However, they're only about 50mm long. The new design will be about 300mm so a 30C delta could give 0.2mm of change for aluminum and about half that for steel, so a .1mm differential. I suspect that it won't be a problem in practice, but it's in the order of magnitude that it's worrying. I suppose it's a good thing that the fit gets looser as the temperature rises, so it shouldn't bind.