Missing steps/shifted layers on Y axis - have tried everything

So im having a bit of trouble trying to get rid of missed steps/shifted layers on my Y axis... I built a RepStrap, printed all the parts I needed for a Prusa, built it, have everything working great, except for these missed Y steps.

I'm running the Sprinter firmware (5D before that), RepSnapper, SF41 and SFACT and something similar to Adrian's Pololu Electronics (same schematic, just built myself on perfboard).

The weird thing is that the missed steps are always only on the Y Axis. I feel like I have tried everything to eliminate it including...

-tightening the belt to eliminate all backlash
-putting a flat on the stepper shaft and really tightening the pulley to ensure it's not slipping (it's definitely not)
-increasing and decreasing the current to the motor
-heatsyncing w/ fan, the drivers & the motor to even push more current through it
-replacing the Y axis motor with a higher torque one (I now have the same high torque motors on X, Y, & Z)
-Swapping the X & Y Pololu drivers (missed steps still stayed on Y axis)
-Swapped the X & Y step, dir, enable and min pins in firmware and plugged X motor into Y plug and vice versa (missed steps still on Y)
-Slowing down speed to 20mm/s (helped, but still get plenty of shifting layers)
-Generating the G-code with SF41 & with SFACT
-Pushed Y axis back and forth by hand to check for binding, just slight friction from PLA bushings, but no binding
-Changing back to my 5D firmware which was working fine on my RepStrap

So after all of this, I don't think the problem is my electronics, I don't think it's the motor not having enough torque, I don't think it's a mechanical problem like backlash or the Y axis binding/sticking....

That really only leaves me to think it's got to be something in the firmware or g-code generation...

Anyone have any thoughts or something i could be missing?

Attached are some pics to help illustrate the problem...

I have found out that if I use something that needs a lot of power like a power saw in my garage even though it's plugged into a different socket the initial draw down of the tool makes my mendel skip and shift the layers over always on the X axis

---

__________________________________________________________________________
Experimenting in 3D in New Zealand

Hey daufhammer,

Sorry to hear about your troubles. It's interesting to hear that it is only on your Y axis, because I had a similar experience. I would check/replace your Y endstop. The problem that I had was that my microswitch had gotten tired and the contact would occasionally open due to vibrations, so the software would think its home and could go no further. That would be my suggestion.

Hopefully this helps.

Stay calm out there.

Ben

To check if its endstop there, maybe replace it with another and see, or use a firmware which doenst care much for endstops like teacup - could use that with no endstops at all, g28 goes to 0,0,0 position where printer first initialized / powered on.

Can you post a better picture of extruder and a better one with the entire X belt? I have a slight impression that X belt is not parallel to x bar in all segments e.g. picture right side. I am under the impression that the idler bearing is too big, so the lower belt segment is at an angle instead of being parallel to the bars. This is not mechanically correct for at least two reasons: belt tension varies, and engaging vector is not along to move direction so among other stuff, it increases friction. I think i also seen this issue in few more prusa pictures too.

If it has to be a mechanical problem, at first sight I am inclined to believe that bushings and low motor torque, coupled, can do that layer shifts, but i saw you excluded these. I have never used pla bushings like those, but i would rather suspect those, especially if bed is heavy than increases the friction directly. Maybe try run the bushings for some time to wear them down just in case, e.g. make a gcode file manually with long moves along that axis and do alot of copy-paste and run that file a few times. Dont try to move the bed by hand, try move it by the belt and fast back and forth like in a print and see how that feels instead.

About motors, if increasing current doesnt help, try lower the it. I think motors with current set too high might cause problems on their own.

Edited 3 time(s). Last edit at 11/13/2011 08:27AM by NoobMan.

My experience after reading many similar posts is that it is always hardware. Consider the friction may be different at the bushing once the print gets going and starts to heat things up. Ditch the PLA bushings for some brass.

Thanks everyone, good suggestions... I will give those things a try later today and let you know how it goes...

@NoobMan, you wanted me to take a look at a better picture of my X belt, did you mean my Y belt? I don't have any problems with my X axis, i have never missed a step in that direction and it's the same exact motor as the Y. You are right though about my belts not being exactly parallel to each axis due to the size of the bearing... it's just the normal 608 bearing that's called for in the bill of materials though, so i would think everyone's is like this.

Also, i don't think it's too much friction from my PLA bushings (anymore than is normal anyways), I have probably run them for 5 hours or so and they move very freely in X, Y and Z directions. I also used Nopheads trick, to take a piece of scrap 8mm smooth rod and heat up the end, then slide the bushing from the cold end to the hot end, then back, this sizes the bushings perfectly.

The easy way to narrow it down from mechanics to (hardware/software) is to flip flop the x and y axis.

ie; plug the x axis stepper into the Y axis driver plug, and y axis stepper into the x axis plug. Then change the pin definitions in sprinter to reflect your changes.

If the exact same problem occurs, it is a mechanical bug.

If the same problem occurs on the other axis, it is a hardware or software issue. 99% of the time the issue is mechanics. Electrical noise is possible, but I have experienced it only once, and that was on a point to point wired breadboard.

---

www.Fablicator.com

Yes i was asking about x carriage. I dont have a prusa but thats the image i got from random pictures, so i thought to check it. I think prusa doesnt care much for belts trajectory, but that could be fixed. Could try with a 606 or if you want to keep the 608 i would rather alter the place it fits on carriage so its parallel instead of leaving it like that. Like it is now its like belt tension is low to one end and high to the other. But again if you have no trouble with it, then you could leave it like that too, its probably not significant in prints although could affect belt life.

Hey daufhammer, I am having similar trouble, though skipping is not restricted to a single axis (occurs at least on x and y, usually separately, not always).

Have tried many things to resolve but have not found the solution. Sometimes I can print for hours without trouble but very rarely can get through a 3 hour print without at least one skip.

Am currently looking at this - [forums.reprap.org] but my hot end had some other trouble before I could fully test. I have my end stops disconnected during testing and have installed an sd module so I am thinking firmware as well.

Better carefully calibrate my mechanics too given all the suggestions above.

Good luck - be interested in your progress.

John

Ok, so I have been working on this all day and still not luck!

@bdcannon and @NoobMan, I tried swapping my endstop switches and the missed steps still stayed on the Y axis.

@NelsonRap, I tried switching to another power outlet on different circuit in the house and unfortunately im still missing Y steps

@Andrew Diehl, I had already tried swapping the X and Y axis motor plugs and pin definitions and the missed steps still stayed on the Y axis.

@NoobMan, I attached pics of my X and Y belts, both are perfectly parallel with the Axis if you look down from the top, but both are not totally parallel if you look from the side due to the size/placement of the 608 bearing (but that's default to the Prusa design). (I don't really think this is the issue, especially since other Prusa users don't seem to have missed steps due to that)

I also tried slowing down my feedrate to 10mm/s and there were actually more missed steps/shifted layers!

Below is the updated list of everything I have tried to eliminate the missed steps on my Y axis....

-tightening the belt to eliminate all backlash
-putting a flat on the stepper shaft and really tightening the pulley to ensure it's not slipping (it's definitely not)
-increasing and decreasing the current to the motor
-heatsyncing w/ fan, the drivers & the motor to even push more current through it
-replacing the Y axis motor with a higher torque one (I now have the same high torque motors on X, Y, & Z)
-Swapping the X & Y Pololu drivers (missed steps still stayed on Y axis)
-Swapped the X & Y step, dir, enable and min pins in firmware and plugged X motor into Y plug and vice versa (missed steps still on Y)
-Slowing down speed to 20mm/s (helped, but still get plenty of shifting layers)
-Generating the G-code with SF41 & with SFACT
-Pushed Y axis back and forth by hand to check for binding, just slight friction from PLA bushings, but no binding
-Changing back to my 5D firmware which was working fine on my RepStrap
(Today) - swapping X & Y endstops
(Today) - slowing down feedrate to 10mm/s
(Today) - double checking for mechanical problems (excess friction, rubbing belts, slipping pulleys, skipping belt teeth, etc)
(Today) - Switching to a power outlet on another circuit in the house.

So it seems like everyone thinks the missed Y steps are from a mechanical problem.... the weird thing is that, like I said above I have the exact same high torque stepper motor on the X & Y axis and same current setting on both X & Y pololu drivers, but even though there is actually more friction and more mass/weight (from the Hot end & Stepper motor) on the X axis compared to the Y axis, and my X axis has never missed a step.... So basically the Y axis has the least amount of friction/binding and weight/mass and it's the one with the problems.... Go figure.

One thing i just though of was that the Molex connector that supplies the 12v power to my electronics goes right by the wires to my Y motor... could that cause that Y motor to jump around a bit? I would think if that was a problem, it would actually go the other way around... motor wires causing noise on the 12v power lines.

Any other thoughts/suggestions..

Edited 2 time(s). Last edit at 11/14/2011 12:02AM by daufhammer.

Ok, so i just got done trying a few more things....

- I re-routed all my wires to keep my signal and power wires totally separate and on opposite sides of my Prusa... they were fairly separate before, but crossed in a few places, but now I completely separated them. After a test print, i still had missed steps in the Y direction and so I think I can rule out electrical interference as a cause.

While doing the above, I had another idea to try to narrow down where the problem was...

I decided to swap my X & Y motor plugs along with the X & Y min limit switch plugs, but this time I DIDN'T SWAP PIN DEFINITIONS in the firmware.... So in my X axis literally became my Y axis, and my Y literally became my X....... so doing this actually rotated my prints 90 degrees.... When the gcode told the Y axis to move, the extruder moved back and forth... when it told the X axis to move, the print bed moved back and forth.

After making these changes and another test print, my missed steps were STILL on the Y axis (i had thought for sure it would switch to the X this time)...

So I think this confirms that my problem is DEFINITELY MECHANICAL in nature....

The only problem is, I have done everything i can think of to eliminate friction and binding in the Y direction....

Anyone have any suggestions... or do I just have to bite the bullet and get some LM8UU bearings and some aluminum pulleys. (the only problem with this is I don't have a working printer to print out the Y axis holders for the LM8UU's)

Have you tried lubricating the bushings? I use Vaseline and it helps a lot.

Do your motors get warm? If no you can turn up the current a little. I always run mine warm.

---

FFF Settings Calculator	Gcode post processors	Geometric Object Deposition Tool Blog
Tantillus.org	Mini Printable Lathe	How NOT to install a Pololu driver

Thanks Sublime. Yeah my bushings are lubricated with white lithium grease (thought i read somewhere that was best). Also, my motors are warm and my A4983 chips on the Pololu drivers are hot to the touch, but not scalding... I have been thinking about putting heat sync's on the Pololu's and cranking the current up a bit more, but i think im already at the sweet spot for my motors.

The drivers are really easy to overheat and they go into protection mode. When this happens they turn off for a few steps and then turn back on for a few seconds and go into protection again, etc, etc, etc.

Also I see you have it wired from the Arduino to the board using the pin out from Adrians Pololu shield. I wonder if the steppers are on the correct timers to use acceleration correctly. I know teacups config mentions which timers are used for the motors and which ones are free to use for other uses like PID and PWM. It may be worth rewiring it to match RAMPS.

---

FFF Settings Calculator	Gcode post processors	Geometric Object Deposition Tool Blog
Tantillus.org	Mini Printable Lathe	How NOT to install a Pololu driver

I knew that about the Pololu drivers, but i don't think they are getting quite hot enough to go into protection mode yet... I should probably throw some heat sync's on them just to be safe though.

And good point about my wiring... yeah it is wired the same as Adrian's Pololu electronics... And now that you mention I also remember reading something (can't remember where though) about pins and timers and the importance of them being matched correctly. I don't think that's causing my problem though because the missed steps stay on the Y axis even after swapping X & Y pin assignments in the firmware, and also literally swapping the X and Y axis' like i described above.... Probably a good idea to update my wiring and pin definitions to match ramps though just to make sure.

Hopefully I can get to these things next weekend (or sometime during the week if i'm lucky!)

Looking at this picture. [forums.reprap.org]

I notice your belt is not parallel with the smooth rod. This is making your belt ride against the hub of the pulley. I would try turning the pulley around so the hub is on the outside. You may also need to space the bearing side out as it looks like the belt is rubbing against the fender washer.

---

FFF Settings Calculator	Gcode post processors	Geometric Object Deposition Tool Blog
Tantillus.org	Mini Printable Lathe	How NOT to install a Pololu driver

Thanks, yeah, I actually can't turn the pulley around because my motor shafts are only 15mm long so there would be no shaft for the set screw to grab onto if i turned it around... I actually was planning to space the bearing out a bit more by adding a few more washers, but I haven't gotten around to it yet because that's my X axis which has never missed a step, so i haven't felt any urgent need to get to that yet... trying to solve my Y missed steps problem first.

I have some motors I need to run at really high current and required me to put heatsinks on the motors to keep them cool.

---

FFF Settings Calculator	Gcode post processors	Geometric Object Deposition Tool Blog
Tantillus.org	Mini Printable Lathe	How NOT to install a Pololu driver

daufhammer,

Im up and printing 2 hours without a skip after increasing the buffer as per the link I posted above. Worth a try for you I would think. I wouldn't say my probs are solved but it is looking a bit better than it has been for a while ... and if your skipping is frequent and persistent you would be able to implement this and see if it helped in a matter of minutes.

scratch that - 2 hours in and my print skipped several times in the space of half an hour - both x and y axis :-(

No heat sinks huh? I'm almost 100% sure your problem is there. Put a heat sink on each pololu, and a fan, don't forget the fan.

Thanks evans2481, I didn't have time to look into what you had suggested until now... It makes sense that increasing the size of the Serial Buffer could help if repsnapper is sending more lines than the firmware can handle. Bummer you still had some missed steps, but i will probably try this anyways to see if it helps.

jcabrer, I'm definitely planning to heatsync and cool my Pololu drivers with a fan this weekend... I had heatsyncs and a fan on them when the electronics were in my junk built RepStrap, but that was just because i was pushing a lot of current to my motors to try to squeeze more torque out of them. I took the heatsyncs off and took the fan away once I was using more normal motors with higher torque that i didn't have to push as much current through on my Prusa. Im still not 100% convinced that is the issue because even when i swapped the X & Y drivers (leaving everything else, including current pot, the same) my problems still stayed on the Y axis... but again.... couldn't hurt right, so i will give it a go!

Like Sublime says, your pictures make it look like the belt is rubbing on the fender washers on each side of all the idler bearings, probably on both axes. I had this problem; it's easy to diagnose. With the power off, centre the belts on each bearing and move the axis back and forth by hand. It should move easily, until the belts start dragging on the fender washer - it's like applying the brakes. Putting more washers in to the space between bearing and fender washer won't help, the belt can just more easily fall off the bearing. You really want some belt guides that go over the bearings, like [www.thingiverse.com]
If you can't print these, you can cut out some thin, circular plastic shims from a milk carton, or anything similar which can freely rotate between the bearing and the fender washer, on top of the small washer that is already there. This will stop all the drag on the edge of the belt. Without this additional load on the motors, you may be able to turn down the power and reduce the heat they produce!

If you print pulleys to go over the bearing you have another source of positioning error.

Anybody tried simply using two bearings side by side? The belt can only wander so far, so it should give a frictionless solution.

---

[www.hydraraptor.blogspot.com]

nophead Wrote:
-------------------------------------------------------
> If you print pulleys to go over the bearing you
> have another source of positioning error.
>
> Anybody tried simply using two bearings side by
> side? The belt can only wander so far, so it
> should give a frictionless solution.

I have this exact problem; those bearing guides make the layer alignment problem worse since they're not gonna be perfectly round. I'm planning to try the dual (or triple) bearing idea today.

Edit: This works, for two reasons: first, the larger contact area obviously means a wandering belt isn't as inclined to reach the washers. Second—and this isn't what I had expected—pushing two 608 bearings up next to each other eliminates the slight side-to-size rotation you see on a single bearing which has a tendency to let the belt wander off to the side and keep it from getting back to the middle. Two 608s next to one another rotate together along only one axis. It's a clear win. …as long as you don't compare it to, you know, real toothed idlers.

Edited 1 time(s). Last edit at 11/15/2011 03:47PM by Pointedstick.

nophead Wrote:
-------------------------------------------------------
> Anybody tried simply using two bearings side by side? The belt can only wander so far, so it should give a frictionless solution.

I did this on my repstrap, works fine. However, if the belt wanders further out on a bearing that isn't supported on both sides, like on the x-idler-end, there's more force trying to pull the x-idler-end around, and the belt just keeps wandering further out. Should be okay on the y axis though, where it's on a rod supported on both sides.

I haven't tried the bearing guides myself, but I've seen them work fine on other repraps, so long as the tolerances and fit are tight. That's why I couldn't make them for my Prusa, my repstrap didn't print accurately enough. I printed PLA washers instead; see picture below of how the idler over the central bar is set up (I've set my Prusa up a bit oddly - on purpose - it's just the easiest one to photograph). My other bearing idlers are set up the same. The white washers move freely, around the small washer trapped between the fender washer and the bearing, to stop the belt binding.

get a can of spray white lithium grease and spray it on all the bushings and both rods.

the gcode should help break in the bushings.

I also noticed skipping and lowered the current until it stopped. (and added a fan+heatsinks)

Nozzle catches/drags in the cold plastic? Does it happen more if you print multiple parts as opposed to a single part?

what are you timing pulleys printed out of? abs or pla, how are they secured?

Do you use the two large washers to guide the belt on the idler? I had exactly the same problem some months ago. The y carriage was smooth when moved by hand but the frictions between the washers and the belt caused always missing steps. After I added a second bearing and removed the washers from the idler, I never had this problem again. Also check if the tension of the belt increases as you move the y carriage to the max/min position as this could be another source of failure.