Play between carriage and timing belt on replicator 2x

[Bas van Deursen]

Hi there!

At our office we bought a replicator 2x a few weeks ago and we were having problems printing nice circular objects (see attached picture: "before"). In the end we found a solution to this problem. I thought it might be nice to post here how we solved it so that people who are having the same issue know how to solve it. Besides this I am quite curious how many others are having this problem.

We noticed that we had quite some play (about 0.6mm) between the carriage and the timing belt, which caused quite some backlash on the x-axis. I contacted makerbot support to explain the issue that we were having and sent them a video showing the play between the carriage and the timing belt (see attached video: "play carriage - timing belt").

Makerbot support first asked me to adjust the tension of the timing belts. When this did not resolve the issue, they decided to send me a new carriage and timing belt. The new carriage and timing belt combination had just the same amount of play (see attached video: "play on new carriage").

I decided to try and solve the issue myself by improving the connection between the carriage and the timing belt. I made a set of clamps to clamp the carriage securely onto the timing belt (see attached pictures). This solved the problem. We are now able to print circular objects (see attached picture: "after").

In my opinion the connection between the timing belt and the carriage is badly designed! I have explained to makerbot support how we solved the problem and told them that if they redesign the carriage that I would appreciate it if they would send me an upgraded version.

If you are also having this problem with your replicator 2x, please let me know! I'd like to know if this is really a design flaw or that I was just unlucky receiving two bad carriages in a row.

Cheers!

Bas

P.S.: the videos were too large to include here, I'll try to post them in a separate post and if that does not work I will put them on youtube and post a link here.

[Bas van Deursen]

A video of the play we had initially on our x-axis.

[Bas van Deursen]

A video showing the play on the replacement carriage and timing belt set sent to me by makerbot.

[Daniel Reetz]

Thank you for sharing your solution, Bas. I had the exact same problem on two Replicator 2X printers (one at work, one at home). My solution was simpler, but uglier. I used very high strength hot glue to glue the timing belt at the sides of the carriage. That stopped all play and allowed me to print objects that are closer to circular.

I was surprised that few people report this problem - circular parts (including the standard traffic cone print) show the defect immediately. To me, this says that most 2X users are not printing parts that need to be dimensionally accurate. I know for me, the non-circularity was absolutely dealbreaking.

Thanks again,
Daniel Reetz

[Ming Kawaguchi]

this is really suprisingly to me because my traffic cones were perfect ootb.. however, i think i may have something to add. the videos shown are ex-situ and not representative of the statics or dynamics in the machine. in order to install the belt into the x-gantry, the belt actually has to be stretched. i did a complete rebuild of my x-stage after about 300 build hours and it took about 12 hours before the belt was stretched enough to move the stepper completely into place and torque it down. in the meantime, i would push on it a fraction of a mm every few hours, and then torque it down again to let the belt stretch until it hit op tension. it's not _that_ surprising that an untensioned belt would have play in the abs carrier (note that my rebuild changed very very many variables, including a new belt, carl's x-carrier, new case hardened very high precision support rods, new very high precision linear sealed ball bearings and a complete relube of all moving parts). 

however, like i said, ootb, i had near perfect cones coming out. the moldings have obviously not changed -- that tooling has been set and will be used for a long time. the belts as far as i know, have not changed. they are using high quality high precision gates belts. i am curious what happens if you try properly stretching the belt as i did on my last rebuild. it wouldn't surprise me if the current factory construction regime does _not_ wait for the belts to stretch and ships them with play considered acceptable. 

[Jetguy]

Timing belts do not stretch. What is happening is a more likely answer is that the X axis rods were not fulled seated into to the X axis plastic ends. This would then cause the distance between the motor and the idler to become closer resulting in a loose belt.

[Jetguy]

Again, it's a timing belt. If it stretches, then the teeth do not line up in the pulleys and it jumps. This is why they use kevlar, fiberglass, steel, or nylon reienforcing fibers in the belt.

http://www.gates.com/facts/documents/Gf000284.pdf

"Synchronous belts do not slip, need retensioning or require lubrication."

"Studies by Gates power transmission engineers reveal that synchronous belts may be the most efficient form of power transmission short of a direct drive."

[Ming Kawaguchi]

i believe they were fully seated, but what you said is certainly possible. i was literally moving a couple hundred microns between each torque down. also, belts do stretch, especially long narrow belts like these. it's possible that in order to hit optimal tension while still meeting tolerances of the gears, the belt is manufactured and sent slightly shorter than what is required and expected to get an initial tension and yielding elongation meet center of tolerances.

again, it's impossible to have a belt under tension which does not stretch. in fact, this is why belts need to be cut and refitted, because as you said, once it stretches past tolerance, it will not mesh properly with the gearset and you will get skipping. everything, including timing belts, _especially_ timing belts, are built to tolerances and expected to have elastic and post-elastic (elongation) behavior. when one wants more precise timing, one uses chains, and even then, you have the wear phenomenon known as 'chain stretch', which is just the individual chain bearings widening in diameter with rotation.

[Jetguy]

I'm not trying to be a pain, but your inkjet printer, probably the scanner too, in nearly every house on the civilized planet uses a timing belt for the X axis. Many are even narrower, I know the pile  of HP ones I took apart use 3mm wide belts.

They don't have a spring tension mechanism and seriously, I doubt the manufacturer stretches the belts before mass producing these. Some of them were 10 years old, the same belt, no tension lost.

 

I'm just trying to show a real world scenario the debunks your theory.

It's not logical because my Replicator 1 which I got used is fine. ONLY 2 series owners are seeing this problem. Knowing the belt is the same 424 tooth version on my bot that means the ONLY logical explanation is the axis distance is shrinking or moving.

[Jetguy]

Again, lets get this right. The carriage has NOTHING to do with the problem of belt tension. It only clips onto the belt loop. The belt tension is set by adjusting the mounting position of the motor a distance away from the idler at the other end.

That distance is mechanically created by the X rods bottoming out in the plastic ends. If the plastic ends molds have somehow changed over production or they got a batch of short rods would be a logical explanation.

Another reason why Carl's aluminum X ends are nice in that they lock onto the ends of the shaft with setscrews but are adjustable.

[Jetguy]

There is one more plausible explanation and there is a report and thread of it somewhere in this group. A user noticed the idler was no longer square to the axis. Upon further inspection, lack of lubrication on the pulley to shaft interface of the X idler then caused the shaft to spin against the slot in the X end bracket. This quickly damaged and wore the slot in the direction of the belt tension. If you are in a situation where the belt is loose and you moved the motor as far as it would go, either your X rods are short or the idler may be wearing and damaged the X end bracket.

 

This "fix" you guys are doing tells me you have a much bigger problem. You really need to inspect these other areas, determine the real cause why your belt tension cannot be set normally.

[Carl]

When I look at the video that Bas uploaded... the play he is experiencing has nothing to do with the timing belt tension... it looks as if the clip on the stock carriage is the source of the problem.

Somehow the profile that the timing belt slides into is allowing the carriage to move left and right before being stopped by the teeth on each side... this could not be solved by tensioning the timing belt - if anything... it might make the problem more pronounced. 

The strangest thing is that he has had the problem on both his original carriage and timing belt and then again on the replacement carriage and timing belt... maybe it is time for MBI to have a new mold made for the ABS carriage!

I never had this problem on my stock carriage... if anything the clip was too tight! :-)

[Ming Kawaguchi]

yes carl, i alluded to the tooling being old and out of tolerance (hah!) in my original post. unless they smash it with a titanium hammer between every run or do something incredibly ridiculous to it, i find this explanation to be nearly impossible.

however, think about the three objects in mechanical play in that video. there is a reinforced elastomer belt, an upper tooth profile molded in ABS and a lower tooth profile molded in ABS. the mechanical connection between the upper and lower tooth profile we can model as a cantilever with force equal to the elastic modulus of the abs that connects the two profiles.

now, elastomers stretch. the reinforcement laid into the elastomer is meant to control that tendency to elongate and to lower the elastic modulus of the elastomer material, but it can't eliminate it or else the belt either not be able to bend around the toothed gears or otherwise work as a timing belt connecting things with any degree of mechanical tolerance (ie, anything).

so anyway, the point is that if you're designing this abs part you don't design it so that it's "loose"; ie, within 3 sigma of mean you're slightly larger than the geometry of the belt, but you design it so that it's extremely tight (which jives with your experience carl), and so within 2-3 sigma of mean you're fitting a squished belt. now, imagine a belt that is a couple sigma thicker than the mean and a the cantilever detailed above. the action of the cantilever is such that the gap between the two tooth profiles could actually widen at the ends and lead to the the play that the video is showing.

hence, my advice, again, is to get a new belt (because your rig has stretched the crap out of your current one), and reinstall by moving the motor gradually to the right (looking from the front), then, once you have reached a point of maximum tension and the forces seem static (ie, the belt is not actively elongating), move to the carriage to the right endstop and fit the belt into the tooth profile at the rightmost point. at this point you should be able to get a feeling for what the play, if any is. use a very fine erasable pen mark or something if you want, make some prints, check for smudging, etc. your belt and profile play test is invalid because it neglects to factor in the rest of the mechanical system.

anyway, the x-y stage of the any printer of this style is going to have a huge number of error components everywhere, and doubly so if you use abs in load bearing parts. it's not anyone's fault, it's just physics.

[Ming Kawaguchi]

speaking of the cantilever design, a much more efficient way to reduce play in the mechanical connection between belt and carriage is just to squish that cantilever together. your gigantic brass things add a bunch of mass in a place you don't want it. coincidentally, squishing the belt is what carl's alucarriage design does, as machining little tooth profiles is a pita. i've had quite good success with the squishing approach so, worth a try *shrug*

[Bas van Deursen]

In the two carriages that I have (the one that was originally in my machine and the replacement sent to me by makerbot support) the belt is not squished at al in between the clip. In both carriages I can slide the belt very easily in and out of the clip. The clip just doesn't fit well with the belt. I think it is a problem in the design of the carriage. Daniel had the same problem on two other replicator 2x printers.

I also believe that the safest solution would be to actively clamp the belt onto the carriage, as is the case in the aluminum carriage.

Furthermore I agree that it is best if the printer head is as light as possible (that's why the design of the Ultimaker is so nice; because they use a bowden cable they don't have the heavy stepper motor on the carriage), but I don't believe that adding two brass clamps will make a significant difference in this case. Their mass is negligable compared to the mass of the stepper motors.

Anyway, the brass clamps solved my problem. Since I mounted them I have printed for about 60 hours and the problem did not recur.

-Bas

[nickythegreek]

When I first received my rep2x, I had this issue.  All my circles had 2 points of backlash.  If I held my belt, I could easily wiggle the carriage left and right.  The issue was the plastic mold injected carriage was not tight on my belt tight.  I shimmed a business card piece between the belt top and carriage and the issue was resolved.  Quick and easy solution.  I've checked on it a couple times now and it has held up fine.  Still think it was pretty insane that issue existed in the first place though.

[nickythegreek]

Where I found my original answer:

https://groups.google.com/forum/#!topic/makerbot/80fTHxXKFp0