so what causes this?

[truly_bent]

From Dancope's reference:

When a step motor makes a move from one step to the next, the rotor doesn’t immediately stop. the rotor actually passes the final position, is drawn back, passes the final in the opposite direction and continues to move back and forth until it finally comes to a rest (see interactive diagram below). We call this “ringing” and it occurs every single step the motor takes. Similar to a bungee cord, the momentum carries the rotor past its stop point, it then “bounces” back and forth until finally coming to rest. In most cases, however, the motor is commanded to move to the next step before it comes to a rest.

This is the "ringing" of the motor shaft as it comes to rest. Then there's the "ringing" pattern that shows up on some sharp edges of a printed part. It could be that the ringing of the motor shaft is exacerbated (amplified) by the inertia of the extruder's mass overshooting the stop position.

Here again, the delta type printer has an advantage because most of them seem to be using the Bowden style extruder, which mounts most of the mass on the frame above, instead of being on the shuttle itself. Of course, that begs the question of which type of extruder is better, but that's for another thread.

When a delta makes a sharp change of direction, it involves at least two of the three motors driving in tandem. Most likely, none of the motors come to a complete stop (if ever), as is more likely the case in the Cartesian coordinate system. If none of the motors ever come to a complete stop, there's less opportunity for motor shaft "ringing" to occur.

Just a theory mind. Someone with a stronger math background would have to verify this.

[printbus]

...The 'Vibration Problem' section on this page seems to describe the problem we have. It kind of seems it's just a limitation of these motors.

But remember that for at least the i3v printers, we microstep the motors with 80 steps per 1mm of linear movement in the XY axes. The ringing described in the article would occur every step. At 80 steps per mm movement, this implies ringing would occur every 0.012 mm of linear movement. The article doesn't convey any magnitude to the position ring, but it'll be far less than the size of the step, right? So, every 0.012 mm movement we'd have some ring effect far less than that. I'm not sure we'd be able to see the effect of that.

[DanCope]

The ripple we see in the print is very tiny in terms of depth, so it could be possible that it is a fraction of 0.012mm. Maybe the ringing amplifies as the motor continues to move and overshoots by more when it finally comes to rest?

I am hesitant to say that the weight of the print head has much to do with it because then it would be much less noticeable in the Y axis due to the printer moving the bed for this axis.

[AbuMaia]

But remember that for at least the i3v printers, we microstep the motors with 80 steps per 1mm of linear movement in the XY axes. The ringing described in the article would occur every step.

*If* the motor came to a complete stop after every step.

I'm not sure we'd be able to see the effect of that.

Yes, the ringing isn't very deep, but remember it's being stretched out by the other axis moving while the first is coming to a stop. If it was a stop and return movement with very little movement of the other axis, I doubt it'd be noticeable. It's a ripple, happening the same way in the same location over many layers. It doesn't have to be very large to become visible.

[printbus]

*If* the motor came to a complete stop after every step.

Yeah, it's hard to gauge how the momentum of the rotating mass could mess with things, but I'm not ruling out the possibility that the motor does come to a stop between steps. These motors aren't moving all that fast. We know the XY motors can be rotated at the equivalent of a 250 mm/sec axis speed. When I run a test print at 50 mm/sec, the steps are occurring five times farther apart than at the 250 mm/sec.

Yes, the ringing isn't very deep, but remember it's being stretched out by the other axis moving while the first is coming to a stop. If it was a stop and return movement with very little movement of the other axis, I doubt it'd be noticeable...

I'm trying to go where observations from test prints lead. It's easy to grasp the possibility of mechanical axis ringing affecting major direction changes like a corner. Where the mechanical axis ringing theory seems lacking in my test prints is in small perturbations like shallow lettering. Does it make sense that the "ripple" from a shallow recess with only miniscule movement in the axis perpendicular to the primary printing plane is just as pronounced as with a 90-degree corner? If Marlin move planning is doing its job, we should be getting only minor momentum built up in that perpendicular axis movement.

Pointing a finger at the jerk setting causing problems with those miniscule perpendicular movements is natural. But one of my diagnostic prints set XY jerk to 2 mm/sec instead of the MakerFarm default of 20 mm/sec. The test object has identical holes, notches, and dimples on all four sides. I did see some reduction in ripple on the surfaces parallel to the X axis, but there's no discernible difference to surfaces parallel to the Y axis. In other words, in this test case the degree that the X and Y axis should have being slowing to a stop before even a minor offset and then creeping away from that offset didn't change the ripple observed on the Y plane. At all.

EDIT: Yes, the weight of the X-carriage is higher than the Y-bed, but I was still surprised to see no reduction in the Y-plane ripple with the low jerk setting.

[TopJimmyCooks]

I hate to say it but until the science and physics is figured out on why this happens, I'm just being careful to design or modify parts so they don't have 90 degree outside vertices. Even a 1mm radius roundover on any near 90 degree turn seems to take care of it.

I would also be interested in results of ringing test objects that are oriented in a 45 degree angle to the bed.

[printbus]

I would also be interested in results of ringing test objects that are oriented in a 45 degree angle to the bed.

Good idea. I'll add that to the list of diagnostic prints for my test model.

[printbus]

...I would also be interested in results of ringing test objects that are oriented in a 45 degree angle to the bed.

Of course, I think sniffle essentially already touches on this by his picture provided in the first post. The ripple caused by the shallow text continues throughout a rounded corner on the object.

[DanCope]

Hey printbus, any luck with your tests?

[printbus]

Fair question. I had even made promises elsewhere to publish test results by the end of last week. So far I've evaluated 16 test prints. Through these, I've been able to rule out at least a few possible sources of the ripple. I've identified a few ways to reduce the ripple or at least the extent that it is noticeable, but most also have other negative effects on print quality. Following a lead I read about elsewhere, the results from print 16 were very promising and I spent the time figuring out some complex translation and rotation math in openSCAD for a more elaborate test model that I hoped would be more useful in exploring it. The model didn't work out. I haven't quit digging into this - I've been easily distracted by just about anything else that comes up. I may try another print or two on the original test model later today. Perhaps that'll get me to the point where I can compile info on what I've tried and what I've observed.