Loosing steps and hair

[halopend]

So, I have an older model Rostock Max and thought it was time to upgrade some components on the printer and... it completely borked everything.

I've faced missed steps before when I was printing too fast on the printer, but even printing slow I'm loosing steps now. The issue Im having is I made several changes all at once so narrowing down whats causing it is turning out to be a nightmare. I have new carbon fibre arms, replaced the acetal coverings on the bearings, have a new E3d v6 hotend and replaced the extruder with an ezstruder.

I figure it must be firmware related (I've upgraded to 0.91 and made my own config) but the problem is eluding me since I've combed through it matching what my old working config file used to be. The only other thing I can think of is because the carbon fibre arms are so stiff, there's much less "flex" if the printhead clashes with anything making the motors have to overpower the movement more than they used to.

Any ideas where I should focus my attention? I've attached a copy of my firmware if anything stands out in it to you.

[DavidF]

you might be on to something with the new carbon fiber arms. What does the print itself look like? lots of support or anything like that? Try printing something like a vase (perimeters only) and see if it acts up.

[rpress]

Yes, I too believe that the stiffer carbon arms can cause missed steps. On my Wantai motors I have the current at 210 for xyz and that seemed to solve it. You might also need heatsinks for the RAMBo.

[geneb]

It's very important that you cool the RAMBo in the v1 chassis. Orange Menace will skip like an epileptic hopscotch player if I don't have a fan going on the electronics bay.

You might also want to make sure your belts are not too tight. Another thing to check is the Cheapskates - if they're too tight on the towers, the stepper motors will skip steps trying to overcome the friction.

g.

[Jimustanguitar]

I'd say go with Gene's tip. There's no way that CF arms are adding mechanical resistance like that. The arms are never under a load greater than throwing around the inertia of the effector platform itself, and they definitely don't cause additional load. Unless you have a siezed u-joint (which happens with the metal to metal contact of the stock aluminum ones but not the balljoints on the CF arms) you shouldn't see a force anywhere near what would cause skipped steps.

[rpress]

No, it's not the arms themselves binding. It's when the nozzle hits a curled edge, especially on a travel move. The stock arms would just pop out of joint and pop back in. The CF arms hold tight and the steppers skip.

[halopend]

Well, I tried loosening the cheapskates as much as possible + loosening the belts a hair to give a little more flex and it seems to of helped as there is no sudden "jumps" on layers signifying lost steps, but I have an odd issue where prints skew as they move up.

Walls which should be straight up and down are angled slightly, drifting away from my z tower as they move up. If I didn't know any better I'd say I was loosing very small steps in one direction but that seems unlikely. Stretched belt maybe? Note it did this before loosening things up to prevent the sudden jumps.

[img]http://s9.postimg.org/kib958ban/IMG_0001.jpg[/img]

[halopend]

Actually, it is loosing steps consistently in one direction!! I tried printing again a calibration part and its very clear that there is little shifts in 1 direction.

[img]http://s23.postimg.org/mu01mcrkb/IMG_0002.jpg[/img]

I have no idea why its so sensitive compared to before.

[teoman]

Are your motors hot?

Or the driver chips on the rambo.


Also try powering down the printer and manually moving the hotend.

[mhackney]

+1 million

I'd say go with Gene's tip. There's no way that CF arms are adding mechanical resistance like that. The arms are never under a load greater than throwing around the inertia of the effector platform itself, and they definitely don't cause additional load. Unless you have a siezed u-joint (which happens with the metal to metal contact of the stock aluminum ones but not the balljoints on the CF arms) you shouldn't see a force anywhere near what would cause skipped steps.

The "flex" is in the joints and the joints on the CF arms are smooth and freely moving. Much less friction than my original V1 arm joints.

I suspect (in this order):
1) cheapskate adjustment. you replaced the acetal covers so I know you messed with it! I bet they are too tight.
2) current settings, again, you upgraded firmware so it is possible you missed this one
3) RAMBo cooling (i had the same issue and solution) lower down on the list since you were fine before but it could be a culprit

You have something seriously wrong going on if you hit a curved edge and have the arm joints pop out! How is the part attached to the bed, with super glue? That is an incredible force you are talking about there.

I have to believe it's 1 above. Make sure the cheapskates run smoothly with little resistance. Double check your wiring too, stranger things have happened when a connector loosens up.

[rpress]

You have something seriously wrong going on if you hit a curved edge and have the arm joints pop out! How is the part attached to the bed, with super glue? That is an incredible force you are talking about there.

How did you know? I've actually used super glue on a few prints, heh.

The stock arm joints just aren't very stiff. I've accidentally crashed the machine and the arm joints just fall apart, and the steppers did not miss steps. And yes I've crashed the machine with the CF arms and the steppers did miss steps for sure! So it's a logical conclusion that if the steppers miss steps with the CF arms, the stock arms joints would have popped out before that.

Maybe most people don't know this, but with all the Ardunio based firmwares each single movement of the machine is broken down into a series of moves. This means that most movements on the machine actually go down in Z even though that's not desired, especially the faster the move. So a rapid move will quite likely drag against the part even if there is nothing to catch on.

I kept having a missed step problem with a particular print, over and over it failed until I upped the stepper current. Yeah the motors are warm but big whoop. For some reason this print was much more likely to do it, I guess because it was large and had a lot of rapid moves.

[DavidF]

Actually, it is loosing steps consistently in one direction!! I tried printing again a calibration part and its very clear that there is little shifts in 1 direction.

Can you track the shift back to a particular tower?

Its not that the CF arms are adding any friction, its that they eliminate the flex the stock arms had and when getting hung up on a print something has to give.
My printer would tear itself apart before dislodging a print off the bed so Im curious as to why I have heard a couple of times now that it should have nocked the print free from the bed. That just does not happen for me. Im printing the same part for the 3rd time that I suspect had gotten snagged up and shifted the print. I changed from .2mm layers to .3mm layers on this print and it seems to be doing ok so far, but hasnt hit the point where the first two prints shifted yet.
Im watching it closely today to see what happens. Nothing on my printer has changed in the last 3 months of printing. I mean nothing!!! not even a z height change. Its just something with this print and its support of the part where it transitions from the support to the part.

[halopend]

It seems that layers with more material produce a greater shift, which makes sense as the longer it runs the more chance there is of missing a step. The really odd thing is it's doing it consistently in the same direction.

I turned up the current and lost the "jaggies" i.e. obvious signs of a missed step, but the drift is still there and it is pretty well EXACTLY the same. it's like I loose 1 step for every 100 in a certain direction on my Z tower motor. I can't see it being a overheating issue cause I have a fan on the Rambo for cooling and this "drift" was happening even when I had the current turned way down. I'll try reverting the firmware, maybe it's some kind of rounding error?

[jdurand]

Missing tooth on belt?

[rpress]

It seems that layers with more material produce a greater shift, which makes sense as the longer it runs the more chance there is of missing a step. The really odd thing is it's doing it consistently in the same direction.

I turned up the current and lost the "jaggies" i.e. obvious signs of a missed step, but the drift is still there and it is pretty well EXACTLY the same. it's like I loose 1 step for every 100 in a certain direction on my Z tower motor. I can't see it being a overheating issue cause I have a fan on the Rambo for cooling and this "drift" was happening even when I had the current turned way down. I'll try reverting the firmware, maybe it's some kind of rounding error?

Check your tower squareness with a large carpenter's square against the bed. Also check your belts where they meet the pulleys to see if there is any damage to the belt teeth.

[halopend]

It seems that layers with more material produce a greater shift, which makes sense as the longer it runs the more chance there is of missing a step. The really odd thing is it's doing it consistently in the same direction.

I turned up the current and lost the "jaggies" i.e. obvious signs of a missed step, but the drift is still there and it is pretty well EXACTLY the same. it's like I loose 1 step for every 100 in a certain direction on my Z tower motor. I can't see it being a overheating issue cause I have a fan on the Rambo for cooling and this "drift" was happening even when I had the current turned way down. I'll try reverting the firmware, maybe it's some kind of rounding error?

Check your tower squareness with a large carpenter's square against the bed. Also check your belts where they meet the pulleys to see if there is any damage to the belt teeth.

I thought something along those lines earlier, but the way the angle suddenly "corrects" it-self when it shifts to a layer with much less material suggest it can't be tower alignment or belt stretching and must be missing steps.I'm trying the firmware right now (probably not the issue but the easiest to test) and will have to try switching the motors next if that doesn't work.

[rpress]

It's possible that the belt is messed up just near the bottom and as you move up you're moving to a better area of the belt.

So if you have a part with equal material all the way up (like an open box) it just keeps moving to one side?

Hot hot are the stepper driver chips, too hot to touch?

[halopend]

It's possible that the belt is messed up just near the bottom and as you move up you're moving to a better area of the belt.

So if you have a part with equal material all the way up (like an open box) it just keeps moving to one side?

Hot hot are the stepper driver chips, too hot to touch?

Imagine it looses a step every time it reverses from clockwise to counter clockwise (but not vice versa). Little by little they accumulate and shift the part ever so slightly sideways. The more moves there are in layer, the more the next layer will be shifted over.


Anyway, to report back on my efforts, reverting the firmware nearly completely removed the effect. I've discovered in my searches that mis steps in one direction can be an issue of motor timing so perhaps they made changes to the code in this area. There is an variable in the firmware to adjust the timing, Stepper High Delay, which I've tried adjusting to promising effect. I suspect one more ago and I'll have it perfected (at least on the old firmware, may take some play for the newer one).

Unfortunately, each test takes considerable time because as I've said, the amount of shift is dependent on the amount of moves performed. I suppose I could of programmed a move back/forth 5000 times and return to center test to see the drift, but this lazy test lets me visualize it physically.

[img]http://s12.postimg.org/jlli2ms7h/IMG_0003.jpg[/img]

[rpress]

Anyway, to report back on my efforts, reverting the firmware nearly completely removed the effect. I've discovered in my searches that mis steps in one direction can be an issue of motor timing so perhaps they made changes to the code in this area. There is an variable in the firmware to adjust the timing, Stepper High Delay, which I've tried adjusting to promising effect. I suspect one more ago and I'll have it perfected (at least on the old firmware, may take some play for the newer one).

It really shouldn't be that sensitive to timing. If it were a problem then others would be having it too, assuming you were running the standard values in Configuration.h. My guess is that one of the A4982 stepper drivers is failing. You could move the suspect axis to the E1 driver by changing the pins in the firmware, and move the motor cable over.

[halopend]

Anyway, to report back on my efforts, reverting the firmware nearly completely removed the effect. I've discovered in my searches that mis steps in one direction can be an issue of motor timing so perhaps they made changes to the code in this area. There is an variable in the firmware to adjust the timing, Stepper High Delay, which I've tried adjusting to promising effect. I suspect one more ago and I'll have it perfected (at least on the old firmware, may take some play for the newer one).

It really shouldn't be that sensitive to timing. If it were a problem then others would be having it too, assuming you were running the standard values in Configuration.h. My guess is that one of the A4982 stepper drivers is failing. You could move the suspect axis to the E1 driver by changing the pins in the firmware, and move the motor cable over.

That certainly is worth a try, it would help me find out if its the motor or the driver that is causing the issue.

[Earthbound]

So, I have an older model Rostock Max... ...(I've upgraded to 0.91 and made my own config)

What version is your RAMBo? I think it is silkscreened on the PCB near the power connector (just below it, down being toward the bottom of the printer.)

[halopend]

So, I finally have the answer to all my woes, and it wasn't what I thought at all. After more fiddling with the firmware, testing for a failing step driver, testing for a failing motor and just generally waisting my time getting nowhere I noticed a very interesting comment on the repetier site.

"Always check endstops. Only enable if you have no cross talk from your motors, which could trigger wrong signals causing the print to skew."
Hmm, that sounds awfully familiar.

So I disabled it and...

Spoiler:

[img]http://s13.postimg.org/74ort8kjr/IMG_0559.jpg[/img]

So why now and not before the part upgrade? Because I have an EZ struder mounted on top the Rostock now and ran the wire to drive it down the tower that has been acting up! The endstop wire is picking up feedback from the extruder motor wiring and the effect is the more material printed, the more the part skews!! Grumble.

[rpress]

Ahh, good sleuthing, that makes sense. I don't know why the firmware doesn't fault when an end stop is triggered when not homing. At least with that option enabled it won't allow the carriages to crash.

You can probably fix the noise problem by soldering a small 0.1uF cap on the back of the RAMBo across the switch leads.

[Mac The Knife]

Ahh, good sleuthing, that makes sense. I don't know why the firmware doesn't fault when an end stop is triggered when not homing. At least with that option enabled it won't allow the carriages to crash.

You can probably fix the noise problem by soldering a small 0.1uF cap on the back of the RAMBo across the switch leads.

The build manual for the Max 2 covers this. I think GhostBusters did also,,, something about crossing the streams,,,,