Eradicate backlash

I have two Prusa printers both are plagued by backlash. One has the machined timing gears and the other uses timing gears i printed myself. On Y axis i can not get below .008" (.2mm) backlash on the machine with machined timing gears. The belts are drum tight, its only 1 year old so i cant imagine the belts are worn out already.

What is the best solution to eliminate backlash on these printers? What have some of you came up with for your own printers? Is there a better alternative to the timing gear and belt?

I thought of making a tensioner pulley but how would that be any different than just making the belts as tight as possible? Ive fought this to the brink of madness, im stumped.

Belt type? MXL, GT2?

---

- akhlut

Just remember - Iterate, Iterate, Iterate!

[myhomelessmind.blogspot.com]

A couple of things to look at. Look very closely at how the belt fits the teeth on the pulleys. If the pulleys were made correctly the teeth of the belt should fit tightly between the teeth of the pulley. I have seen pulleys where this was not true and they had lots of backlash. Have you considered that the backlash is due to twisting of the driven member on its bearings? On lightly loaded drives just friction, no teeth, should have taken care of the backlash. Are you sure that the axis you are pushing against has no mechanical slop in it? That would look like drive backlash, but tightening the belt would have no effect.

Hope this helps,
Gary H. Lucas

It is impossible to mechanically remove backlash 100% in a reversing mechanical system

I think i have narrowed it down to the Y axis carriage. It is twisting side to side just a hair when reversing direction just enough to cause the backlash.

can you switch the geat with a printed hearingbone style? like what is found here: [www.thingiverse.com]

it is parametric, so it is adjustable in size

Quote

It is impossible to mechanically remove backlash 100% in a reversing mechanical system

It is possible. In fact, it's a requirement, because removing it by software at non-zero speeds is impossible.

I guess you have a reason to write this. What makes you think zero-backlash is impossible?

---

Generation 7 Electronics	Teacup Firmware	RepRap DIY

GITRDUN Wrote:
-------------------------------------------------------
> I think i have narrowed it down to the Y axis
> carriage. It is twisting side to side just a hair
> when reversing direction just enough to cause the
> backlash.

This is the kind of problem I was expecting you to find.

I am pretty sure the machine I am building has zero backlash. The screw is tensioned between two bearing blocks and the ball nut is preloaded. I put a tenths indicator on top of a slide and pulled, pushed and twisted and could not get the needle to even move!

Gary H. Lucas

@Traumflug

Backlash Wiki

backlash can be minimized, but not eliminated mechanically. A zero-backlash mechanical system would not be movable.

Dirty Steve Wrote:
-------------------------------------------------------
> @Traumflug
>
> Backlash Wiki
>
> backlash can be minimized, but not eliminated
> mechanically. A zero-backlash mechanical system
> would not be movable.

Backlash is just a fact of life. But software can compensate effectively if used right. Its not as much of a problem when making rectilinear moves but when you go to make a circle it will show up like an ugly sore. Ive been adjusting backlash comp settings on CNC machines for 15 years. Which is why i dont understand why my Sprinter software doesnt have anything to compensate for it.

What is the solution to the sloppy Y-axis carriage? I currently have a 3 bearing setup as in the basic Prusa 2 and I can feel just a slight wiggle in the Y-axis. Is there a good way to prevent this? Do I need a 4 bearing setup?

I have a 4 bearing setup on both machines and they both have a little slop. One much worse than the other. Id like to know the bullet proof design to eliminate the slop as well. Im thinking mine is coming from the printed Y carriages. Im printing a new one now that will be much more rigid so i will know hopefully by the end of the day.

Also i noticed my X carriages have only about .001" ~ .0015" (.038mm) backlash. Id like to know why that is.

To have zero backlash you need zero friction, or infinite torque, so there will always be some. It can be compensated for on subtractive machines but not on FFF because the extruder would need an infinitely quick response. It just has to be made as small as possible.

Bearing slop isn't backlash but it gives similar looking results. LM8UU bearings have preload so should not have any slop as long as you use at least 2. They seem to have angular slop, which is why you need two. You do need precision rods to get the correct preload though and they eventually wear out and lose it. Bushings on the other hand always have some slop as they need some clearance or they would seize. The rolling balls in LM bearings avoid this problem.

---

[www.hydraraptor.blogspot.com]

FINALLY someone that actually knows about backlash, and that doesn't say 'tighten the belts'!!!!

This gives me a warm fuzzy.....

Have had several arguments about this, and usually you can't convince people of something they didn't find on Google.

I'm a French model.....

I made a new Y carriage much more rigid at the point where the belt ends connect. It helped quite a bit and stopped the Y carriage from torquing left and right when reversing. Still couldnt get it totally resolved so i shifted the motor and belt as close to one of the Y axis rails as possible and it helped more again. Still not below the .001" target but getting closer.

One of my machines has the LM8UU self aligning bearings on ground and polished rods with the metal machined belt gears and the other has bronze bushings on non ground steel shafts with printed belt gears. I used to get perfect prints from both machines, then i had to keep hacking away and here i am.

Quote
Dirty Steve
Backlash Wiki

Wikipedia is wrong on this one.

Quote
Dirty Steve
A zero-backlash mechanical system would not be movable.

Nicely they describe a zero backlash system in the same article: [en.wikipedia.org]

With a double gear tensioned against each other, both sides of the teeth are in contact at the same time. And it's movable.

Quote
nophead
LM8UU bearings have preload

This is another example of having zero backlash. "Backlash", like in "the maximum distance through which one part of something can be moved without moving a connected part" (citation from the Wikipedia article).

The stiffness of a zero backlash system is a different topic. For example, ball bearings are much weaker than bushings, because they have a point contact, only. With load, the balls sink into the rods surface and on high loads you can get to a point where the sink-in of ball bearings is bigger than the backlash of a bushing. That's why non-hobbyist milling machines usually have bushings instead of bearings.

---

Generation 7 Electronics	Teacup Firmware	RepRap DIY

Traumflug Wrote:
-------------------------------------------------------
> Backlash Wiki
>
> Wikipedia is wrong on this one.

Yes, but not for the reason you state.

> The stiffness of a zero backlash system is a
> different topic. For example, ball bearings are
> much weaker than bushings, because they have a
> point contact, only. With load, the balls sink
> into the rods surface and on high loads you can
> get to a point where the sink-in of ball bearings
> is bigger than the backlash of a bushing. That's
> why non-hobbyist milling machines usually have
> bushings instead of bearings.

Preloaded or Spring tensioned nuts and other systems that reduce play like 2 gears angulary slided, even with 0 effective play can have backslash. This is caused by the hysterysis caused by the different point of application of forces when moving in reverse. This is typically a magnitude of order smaller, but still important to take in account with precise movements.

In fact, to eliminate play, you need some preload, and the more preload the more hysterysis (hertz pressure theory).

Backslash is compensation of (internal+external) plays plus hysteris. While the former can be mechanically eliminated (within limits), the latter can only be compensated in software.

i think there is confusion over slop versus backlash.


also has the original issue in this topic been resolved yet?

You might be able to have zero backlash gearing but you can't have zero backlash with an open loop stepper motor unless the friction is zero, or the torque infinite.

Stepper torque is proportional to displacement, so if static friction is non zero it can come to rest with a displacement that gives enough torque to match it.

---

[www.hydraraptor.blogspot.com]

I guess here are different opinions on what exactly "backlash" ("backSlash"?) is. Myself I'm a mechanical engineer, but not an english native, so I have to ask dictionaries sometimes. Also, this Wikipedia article mentions "play" as a synonym for "backlash". A bit contradictionary to this quoting:

Quote
alj_rprp
even with 0 effective play can have backslash

---

Generation 7 Electronics	Teacup Firmware	RepRap DIY

... yes, you can have 'backlash' without mechanical 'play' - it's then mostly caused by stick-n-slip or friction and elasticity of the whole mechanical setup.

When you stop and change direction, most driving systems will show this symptoms - you can avoid (or reduce) it with high rigidity, pretensioning of ballbushings or adding ultrasonic oscillation to counter the stick-n-slip ...

---

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

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

My understanding is backlash is the specific problem of reduced movement when there is a change of direction involved. It can be caused by having some play in the drive linkage but play in bearings causes things like the carriage to rotate when it changes direction. I don't personally class that as backlash.

---

[www.hydraraptor.blogspot.com]

... with tooth-belts, some friction and without any 'play' you'll get some measurable backlash simply through the elasticity of the belts.

When moving in one direction, the belt is tensioned, so when stopping and changing the direction, you first have to reduce this tension... then, with friction and/or sticking (as in stick-n-slip) you'll need some force to start the mover again -- so this releasing+retensioning is the amount of 'play-free backlash' ... could be some hundred microns, what's visible as flats/'quadratures' when printing small circles ...

---

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

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

Backlash is specifically generally caused by loose fit (or deformation) between two or more drive components, and it's unlikely you'll ever remove 100%, although you can get close. (personally, I don't usually consider friction based problems to be "backlash" and deal within them as different degrees of freedom from coupling issues)

Using machined metal gears, high precision on mounting, properly tensioned belts, using belts with steel wire strengtheners, replacing general purpose threaded rod and general purpose nuts with precision lead screw, etc will help immensely though.

Edited 1 time(s). Last edit at 04/10/2013 11:50AM by xiando.