Elusive issue: Nozzle jumping based on direction

[radicaldev]

I've been trying to expand the size of things I've been printing, but I'm having a hell of a time getting it right, and it wasn't until last night that I finally found something that would explain why I've been having such a hard time. Unfortunately, I have no idea what to try next to resolve it. I did a masterful job painting the issue, which I hope will help to illustrate my description.

NOTE: The issue is experienced all over the bed, but gets worse at points where only two axes are doing most or all of the work to move the nozzle.

Using the manual positioning controls in the firmware UI, if I move over to the right edge of the bed somewhere, the following steps result in this:
Move to the right -> Nozzle height decreases
Move to the left -> Nozzle height increases SHARPLY, and then begins to decrease

Essentially, I can draw a circle in the air with the nozzle, just by going right, then left, or vice versa, in either the X axis or Y axis.

[img]http://i1151.photobucket.com/albums/o637/radicaldev/anomaly_1.png[/img]

Specifications:
Orion Delta (black)
Repetier 0.92
Tricklaser carbon fiber arms with metal ball rod ends, 178mm each
V2 effector platform
Towers parallel and equidistant? True, as far as I am able to measure
Belts tight but not too tight, and same for all? True, but I also tried not tight enough and too tight
Eccentric cams tight but not too tight, and same for all? True, but I also tried not tight enough and too tight
Bed level? True, and nozzle positions at center, X, Y, and Z towers are all extremely close to one another.
Horizontal Radius: 74.3 (through experimentation, and measurement)
Rod length: 178mm
Acceleration (movement): 3000 (also tried very high, and very low values here)
Jerk: 35 (Also tried higher and lower values here)


When I get home I can set up a web cam for anyone interested in assisting me.

[radicaldev]

No guesses as to why the nozzle would go down one way, and up the other way?

I'm thinking it has to do with one of the carriages moving too little or too much, and maybe that's a current issue, or a steps/mm issue, or something along those lines.

[bot]

This happens to me, and lots of other people too, to varying degrees. I believe it is caused by over-sized holes on the arms where they connect to the u-joints. Take one axis as an example. As it raises rapidly on the tower, the play in the hole near the joints is at its "maximum" when it lowers rapidly it's at its "minimum."

[radicaldev]

This happens to me, and lots of other people too, to varying degrees. I believe it is caused by over-sized holes on the arms where they connect to the u-joints. Take one axis as an example. As it raises rapidly on the tower, the play in the hole near the joint is at its "maximum" when it lowers rapidly it's at its "minimum."

I have Tricklaser arms, and there's pretty much 0 play in them. I also had the issue with the original arms, though. Actually, there's pretty much 0 play anywhere in the system anymore, except for when forces exceed the capabilities of the components to resist them, which are pretty high as far as I can tell.

The thing is tough now. Touching the effector doesn't permanently adjust position any. I used to be able to alter x/y/z coordinates permanently just by touching the effector. I really can't think of what else would cause that issue besides some incorrect math going on with respect to driving the motors.

[bot]

Very interesting. That throws my theory out the window.

Previous theories of mine have included:

- acceleration settings being incorrectly applied to the tower movement,

- slack in the belts,

- the carriages tilting slightly forward as they move up,

But really, I'm just guessing randomly.

[radicaldev]

Hmm.... I cranked up the acceleration to 10k for print and moves.seems to have lessened the effect, but it's still there.

On a hunch I measured the spacing between my arms, and they vary a bit, like half a millimeter in places. Tried to adjust it, but loosening the nuts introduces slop.maybe tricklaser should make a laser cut effector platform.

[bot]

Yeah, my arms are wonky right now too because I've been experimenting with (too much) tension added to the stock arms. They all got bent haha. Could be the culprit? But I seem to think this happened before the arms got wonky.

[radicaldev]

Odd a thing as it is to consider, pressing down on the effector with about half a pound of pressure eliminates the issue. I may try weighing it down and see what happens. Just need to figure out how to balance 13 255 grain 45cal semi-wadcutters on the platform.

[bot]

You are a genius. Just put the extruder motor on the effector, and solve like 10 problems at once!

[radicaldev]

You are a genius. Just put the extruder motor on the effector, and solve like 10 problems at once!

Not liking that idea so much... What about when retracts occur? The forces generated by that could cause some positional shifting.

..... unless I used a different motor .... Maybe you're the genius!

[bot]

I've seen people put the extruder on the effector... it's probably a lot more weight than necessary to fix the "zig zag sag" problem. That said, if you can find a compact, lightweight stepper motor with the right amount of torque, I may be convinced...

[Mac The Knife]

http://www.ultibots.com/pg35l-geared-stepper-motor/

[radicaldev]

Oh dear... I see that in the Configuration.h for the old Orion firmware, https://github.com/seemecnc/Orion_Delta ... guration.h, motor current is defined like so:
#define MOTOR_CURRENT {175,175,175,200,0}

In my firmware, I do believe, mine is set like so:
#define MOTOR_CURRENT { 135,135,135,135,135 }

I wonder what difference it would make if I changed that....... Now I'm all excited to go home and try some stuff. Damnit!

[KAS]

Would something like this work? Max torque of 70 ounces, .38 degree per step.

http://www.omc-stepperonline.com/gear-r ... p-131.html


or the worm drive: https://www.youmagine.com/designs/modul ... -worm-gear

[bot]

Some of those Nema8 could be promising. Who wants to test it out!?

[radicaldev]

=( Setting current and microstep settings to Orion defaults did not fix the issue in Configuration.h.

It did lessen the issue, though, but I haven't tried that in combination with higher acceleration to see if the lessening is compounding.

I'm going to do some crazy stuff tomorrow, including printing a new effector platform which is more conducive to the tricklaser arms. Also going to try binding the axles to their carriages, and adding weight to the effector.

I think if I can nail this issue, then good looking big prints will follow.

UPDATE: Ok, maybe I am the genius... I was trying to print a ~bed-size circle which started off with all the usual trouble spots, but then I was like... what happens when I pull on the belt of the Y tower where I was having a big dip that just scratched away the hot glue instead of putting down a nice fat bead of ABS? (one of two trouble spots making that task really tough to do)

To my surprise, the Z-height there increased, and a nice fat bead was produced! So, being lazy and a little inebriated, I shoved a PLA fan shround in between the belt and the aluminum extrusion, and let it run for a bit... Good stuff! Very good stuff! I think I'm on to something.

The other trouble spot is right past that one, about 65mm away, where Z height is too high. Bed shims help some, but not completely. The first 10-15mm of the print have crappy adhesion there. From what I know about tower math, if X carriage is too low when x, y is between Z and Y, then z will be too high, but currently if X carriage goes any lower, the bead right there won't be perfect. Seems like bed shims may be my only option.

DOUBLE UPDATE: dang, 1:30am. I found something I didn't like....all the cheapskates had points where they would bind on the towers, even though that wasn't the case the last time I set the eccentric cams. In my 80% awesome circle,I had some ripples which might have been a result of that.... maybe my belt tension and cam adjustment came undone....

[Eaglezsoar]

=( Setting current and microstep settings to Orion defaults did not fix the issue in Configuration.h.

It did lessen the issue, though, but I haven't tried that in combination with higher acceleration to see if the lessening is compounding.

I'm going to do some crazy stuff tomorrow, including printing a new effector platform which is more conducive to the tricklaser arms. Also going to try binding the axles to their carriages, and adding weight to the effector.

I think if I can nail this issue, then good looking big prints will follow.

UPDATE: Ok, maybe I am the genius... I was trying to print a ~bed-size circle which started off with all the usual trouble spots, but then I was like... what happens when I pull on the belt of the Y tower where I was having a big dip that just scratched away the hot glue instead of putting down a nice fat bead of ABS? (one of two trouble spots making that task really tough to do)

To my surprise, the Z-height there increased, and a nice fat bead was produced! So, being lazy and a little inebriated, I shoved a PLA fan shround in between the belt and the aluminum extrusion, and let it run for a bit... Good stuff! Very good stuff! I think I'm on to something.

The other trouble spot is right past that one, about 65mm away, where Z height is too high. Bed shims help some, but not completely. The first 10-15mm of the print have crappy adhesion there. From what I know about tower math, if X carriage is too low when x, y is between Z and Y, then z will be too high, but currently if X carriage goes any lower, the bead right there won't be perfect. Seems like bed shims may be my only option.

DOUBLE UPDATE: dang, 1:30am. I found something I didn't like....all the cheapskates had points where they would bind on the towers, even though that wasn't the case the last time I set the eccentric cams. In my 80% awesome circle,I had some ripples which might have been a result of that.... maybe my belt tension and cam adjustment came undone....

Could you share where you found the stl file for the new effector platform? Thanks!

[radicaldev]

I made it.I'll uplOad to my thingiverse profile tonight or tomorrow.

[Eaglezsoar]

I made it.I'll uplOad to my thingiverse profile tonight or tomorrow.

Thanks!