Suspect something wrong with motion planning

gnydick

This photo shows a slow moving print with an acceleration setting change about half way up. You can see the ripples disappear. The change was M201 X2500 Y2500 to M201 X500 Y500. My printer can easily handle 6000 acceleration, I don't suspect there is vibration moving at 15mm/sec, no matter what the acceleration setting.

I suspect that the extrusion motion and XY motion are not well synchronized. I'm guessing, and this is a complete guess, that the XY is accelerating along each segment in micro spurts, but the extrusion is not accelerating in the same manner or possibly just can't match, which is totally reasonable.

Thoughts?

DocTrucker

We no nothing about your machine at this stage.

You could be hitting a resonance, too little motor current, too much instant speed, too much friction on axis, non-geared extruder issues, v0.6/v0.8.5 vs duet2?

I've been struggling with similar wobbles and have found many minor contributors, but not the route problem yet.

https://forum.duet3d.com/topic/9222/part-wobble-formerly-need-gates-gt2-belt/36

Given the numbers of duet users if this were a motion planning issue we'd all be suffering.

gnydick

@doctrucker sorry, I'm so active, I forget to list my config because dc42 probably knows it by now.

It's not resonance, low current, friction, or any of the other things. It's all linear rails that move smoothly. Appropriately high current, genuine GT2 belts (love them), tension is perfect, etc. I can print at .02mm and create prints that look quite nearly like they're injected molded. No matter what the jerk setting is, at super slow speeds, it has no impact on a well built, rigid machine, because jerk is speed dependent. Nobody would set it so low to not be able to instantaneously change direction at such low speeds.

But, just because others haven't reported it doesn't mean it's not there. For example, what is your acceleration settings? If it's low, could you increase it to 2500 and see if you get ripples?

It is possible that it's a configuration issue, maybe the extruder motion settings don't match well with high XY acceleration, but to me, that makes me still think it's a motion planning problem.

I've experienced mismatched configuration where the slowest link in the chain caused the entire motion system to slow down, which is correct. Which is why I suspect motion planning.

M566 X400 Y400.00 Z150.00 E500.00:500.00
M203 X48000.00 Y48000.00 Z1200.00 E6000.00:6000.00
M201 X500.00 Y500.00 Z150.00 E5000.00:5000.00

DocTrucker

I have already said I'm suffering wobbles and have linked to the thread with pictures. Increasing acceleration to levels that will cause ghosting doesn't help proove anything.

There are others running 3000 and more on acceleration. I can run more than 3000 acceleration on my cantilever before the steppers will stall, but the parts look majorly ghosted. I tend to drop acceleration on perimeters to 500 to get a better part.

I'm not ruling out a motion planning issue, I'm merely suggesting it's unlikely when the symptoms closely mirror ghosting.

gnydick

@doctrucker I mean, maybe acceleration causing ghosting is not supposed to be happening at this low of acceleration.

gnydick

@doctrucker also, if your phone does decent slowmo video, try taking some closeup video of your movement. I did that and saw crazy swings that were not visible to the naked eye. It was amazing.

deckingman

@gnydick said in Suspect something wrong with motion planning:

.................I'm guessing, and this is a complete guess, that the XY is accelerating along each segment in micro spurts, but the extrusion is not accelerating in the same manner or possibly just can't match, which is totally reasonable.

Thoughts?

That's a perfectly reasonable assumption IMO. But it has nothing to do with firmware or motion planning, and more to do with laws of physics. Where there is change in carriage acceleration, there is very little lag between the demand and physical change. However, when you change the acceleration of the extruder, there is a physical lag for the filament coming out of the nozzle because the hot viscous filament acts a bit like a sponge. Firstly you get an increase in pressure as you accelerate the incoming filament but it takes time for that pressure increase to translate into more filament coming out of the nozzle.

Or to put it another way, the acceleration due to applying a force to a free moving but rigid lump of mass, will be different than when applying a similar force to a lump of spongy filament which has to be melted then forced through a small hole.

DocTrucker

Reading through the active ripple damping on the Duet2s there is quite a tie between instant speed, acceleration, and print speed.

I wonder if when one of the situations occur your system becomes more susceptible to wobble/ripple/ghosting.

Duration of acceleration is an exact multiple of major stepper size.
Actual 0-Instant speed time is an exact multiple of major stepper size.
Step rate an integer multiple of resonant speed.

Next time I buy steppers I'll be getting double ended so I can try some mass dampers.