RINGING the electromechanical challenge

Any topics that don't fall into above categories. Please note that your thread may be moved to other discussion boards.
MDVolle
Posts: 56
Joined: Fri Oct 20, 2017 9:00 pm
Location: Fullerton CA

RINGING the electromechanical challenge

Postby MDVolle » Mon Oct 30, 2017 8:18 pm

This is really a MOVE of an interesting discussion that was happening (out of place but fun) over in a printing polyflex thread.

The issues surrounding ringing in the print quickly branches off to resonance in the mechanical system, from vibration to step loss due to shock loading and jerk rates.

The solutions range from mechanical to firmware and g-code - it seemed like a really valuable thread to continue in a more accurately labeled location - NEITHER a software or a hardware only problem...

I have seen isolators (viscoelastic) between steppers and the machine recommended (I think primarily to reduce stepper produced noise radiating out throughout the frame).

The article referenced in the previous thread from University of Michigan focused on g-code modification of accelerations - very interesting

I would suggest that there is even a discussion potential around the stepper couplers - Raise3D uses a "spiral cut" style while there are several other options also available - like the bellows type - which is "supposed to be" more "accurate" - interestingly, especially at high jerk rates.

AGAIN - I think that this is a HIGHLY complex set of issues - and the N-series printers seem to have a very interestingly balanced solution - just "upgrading" any given part could easily upset the balance and start some new and (maybe) interesting round of issues and solutions.

If there were simple, cost effective changes we could uncover to improve these specific machines, I am sure that company and community would find it very helpful - even if our discussion may look like a foreign language to most!

I could see several "priorities" in this thread;

Printed objects that cause repeatable and easily visible "defects" that we can associate with a specific mechanism within the system

Printed objects that demonstrate a specific "limit" - like onset of missed steps/layer shifting or extruder thermal capacity -

Printed anomalies no-one can (yet) explain but are repeatable

Discussions, experiments, proposals, tests, dats, etc surrounding any and all of these more complex issues


Remember - impossible just means that is hasn't been done yet - and those seem to be the most fun things to do...

Mark

User avatar
walshlg
Posts: 1346
Joined: Wed Mar 23, 2016 12:01 pm

Re: RINGING the electromechanical challenge

Postby walshlg » Thu Nov 02, 2017 8:39 pm

Yes, that article sounded great but when you read it there is nothing specific, "just we can do this trust us". What I found fascinating was that it was done based on first principles (theory) rather than on measurement. I am a big fan of measuring things!

I propose rig up a little accelerometer and glue it to the head and "tune" the slicer to the real head movements - essentially do a form of "PID" tuning to iteratively test the second order effect of change in acceleration (or some other measure of ringing) in a trade off with print speed. If you wanted to, you could run this tuning on any specific stl every time you wanted optimized spatial resolution.

User avatar
Julia Truchsess
Posts: 233
Joined: Mon Mar 21, 2016 1:30 pm
Location: Connecticut
Contact:

Re: RINGING the electromechanical challenge

Postby Julia Truchsess » Fri Nov 03, 2017 12:05 am

walshlg wrote:I propose rig up a little accelerometer and glue it to the head and "tune" the slicer to the real head movements.


Now, THAT is a cool idea.

Tinkerer
Posts: 176
Joined: Mon Oct 09, 2017 8:30 pm
Location: Germany

Re: RINGING the electromechanical challenge

Postby Tinkerer » Fri Nov 03, 2017 10:44 am

One thing I noticed:
The severance of the ringing-effect differs from one material to the other!

The least (nearly none) ringing I got when printing with Polymaker Polywood.
Polymaker Polyflex did show the most ringing, I'd say.
Since they both print quite slow I did a test with PLA at only 30mm/s as well.
Here I got the same ringing than in the test I printed at higher speed & PLA.

So my conclusion is the material used plays a role in the whole dynamics as well.

I now this effect from milling machines, where you get different artifacts depending on the material used, as well.
But I would not have thought this to play such a big role with 3D-printing.

One thing I can imagine that could help would be to make the print head mounting more stiff.
At least this did help with the milling machine.
After stiffening the portal we used, we could mill high density aluminum with nearly no artefacts,
which was impossible before.
I checked the print head stiffness at the N2.
Just by pushing the print head with one finger to move, it gets tilted quite a bit.
The dial I had mounted (to measure bed leveling) showed some 0,2+ mm of difference.
...there's nothing like the smell of fresh ABS in the morning...

User avatar
walshlg
Posts: 1346
Joined: Wed Mar 23, 2016 12:01 pm

Re: RINGING the electromechanical challenge

Postby walshlg » Fri Nov 03, 2017 1:23 pm

On first read, Tinkerer, I thought that didn't make much sense but now I suspect that what you noticed was the interaction of the nozzle head on the previously printed material. That is another whole area we just wave our hands and ignore unless its bad enough to make a stepper skip or peel the print off the bed. Nice observation, thanks

zemlin
Posts: 353
Joined: Sat Oct 21, 2017 2:02 pm

Re: RINGING the electromechanical challenge

Postby zemlin » Fri Nov 03, 2017 2:52 pm

There is another factor I've seen on my Makergear printer which is also a Marlin machine. I can provide some images later if there's interest.
Some apparent 'ringing' appears to be due to mismatched stair-stepping of velocities between the motion and the extruder. This manifests in the width of the extruded bead varying rather than the position.

oOoOoO ... if you will.

I haven't noticed this on the N2. The predominant ringing I've seen on this one is wavy lines, not bulgy lines.

On my N2+ I'd put most of the blame on the cross-rods that support the print head. Those are pretty thin for the length and it wouldn't take much force for them to spring a bit. It all adds up though.

I noticed a significant reduction in ringing when I swapped the stock extruder with the Bondtech dual direct.

Vice Chief
Posts: 88
Joined: Wed Aug 23, 2017 5:04 am
Contact:

Re: RINGING the electromechanical challenge

Postby Vice Chief » Fri Nov 03, 2017 4:18 pm

To me, it makes sense to start by measuring it. We are seeing ringing - are we all seeing the same ringing? What is the fundamental frequency of the ringing?

If money was no object, the right way to measure it would be to put accelerometers all across the motion hardware (definitely agreeing with walshlg here). One at the extruder tip, some up the extruder assembly, at the ends of each rod, and back to the motors. You could literally watch the vibrations build up during a print, and then go back through the G-code and see what/where the cause is. Here's a page that describes the use of accelerometers in characterizing vibration: http://www.ni.com/white-paper/3807/en/

There are probably other, simpler ways as well. Here's a shot of an old print of mine, showing bad ringing. You can see that the period of the fundamental frequency of the ringing is between 1.2 and 1.6mm (caliper is not the right instrument here, really). Given that my print was printed at 60mm/s, there is probably some very basic math to do here to figure out the resonant frequency of the system. Would probably have to account for acceleration, if there was any in play.

We could make print designed to induce ringing and have people print it, and then measure the period of the ringing. It's less direct, but it would give us more data, and most everyone here has a caliper and a Raise3D printer.

RingingVC.jpg
Last edited by Vice Chief on Fri Nov 03, 2017 4:36 pm, edited 1 time in total.

Vice Chief
Posts: 88
Joined: Wed Aug 23, 2017 5:04 am
Contact:

Re: RINGING the electromechanical challenge

Postby Vice Chief » Fri Nov 03, 2017 4:20 pm

Another thought I had, I have removed the second extruder from both of my machines, but it's not difficult to bolt the second stepper back on. I could print one print with both steppers in place, and one with just one stepper in place. Since the whole system gets lighter, one would expect that the fundamental of the ringing would shift up.

User avatar
walshlg
Posts: 1346
Joined: Wed Mar 23, 2016 12:01 pm

Re: RINGING the electromechanical challenge

Postby walshlg » Fri Nov 03, 2017 5:03 pm

Vice Chief wrote:To me, it makes sense to start by measuring it. We are seeing ringing - are we all seeing the same ringing? What is the fundamental frequency of the ringing?

If money was no object, the right way to measure it would be to put accelerometers all across the motion hardware (definitely agreeing with walshlg here). One at the extruder tip, some up the extruder assembly, at the ends of each rod, and back to the motors. You could literally watch the vibrations build up during a print, and then go back through the G-code and see what/where the cause is. Here's a page that describes the use of accelerometers in characterizing vibration: http://www.ni.com/white-paper/3807/en/

There are probably other, simpler ways as well. Here's a shot of an old print of mine, showing bad ringing. You can see that the period of the fundamental frequency of the ringing is between 1.2 and 1.6mm (caliper is not the right instrument here, really). Given that my print was printed at 60mm/s, there is probably some very basic math to do here to figure out the resonant frequency of the system. Would probably have to account for acceleration, if there was any in play.

We could make print designed to induce ringing and have people print it, and then measure the period of the ringing. It's less direct, but it would give us more data, and most everyone here has a caliper and a Raise3D printer.

RingingVC.jpg


so I think that the biggest ringing problems we are actually capable of seeing are in the 0.5-2mm range, what frequency range would that be? It looks like 30-120 Hz BUT you don't get many ringing problems when going flat out at a constant speed, they are mainly on decelerations!

Tinkerer
Posts: 176
Joined: Mon Oct 09, 2017 8:30 pm
Location: Germany

Re: RINGING the electromechanical challenge

Postby Tinkerer » Fri Nov 03, 2017 6:11 pm

walshlg wrote:On first read, Tinkerer, I thought that didn't make much sense but now I suspect that what you noticed was the interaction of the nozzle head on the previously printed material. That is another whole area we just wave our hands and ignore unless its bad enough to make a stepper skip or peel the print off the bed. Nice observation, thanks


I think it makes sense to keep this in mind.
Because there seems to be an interaction between the nozzle head and the PRINTING material, yes.
(we are talking about the printing nozzle, not the second one.)

Something to keep in mind when talking about this issue, I think.

Take a look - both printed at 30mm/s, 2x2cm wide...
DSC_4002.JPG

(PolyWood)

DSC_4003.JPG

(PolyFlex)

Printed right one after the other.

PolyFlex is much denser than PolyWood - so the more dense material carries more energy/resonance back into the nozzle?
That leads to more ringing - at least that's what I assume.

Btw. - the distance is 0.6 mm in my case.

P.S.:
As I wrote earlier, I observed the same ringing - in means of distance between the ghosts as well as the severance -
in two prints with PLA at different speeds...
...there's nothing like the smell of fresh ABS in the morning...

Vice Chief
Posts: 88
Joined: Wed Aug 23, 2017 5:04 am
Contact:

Re: RINGING the electromechanical challenge

Postby Vice Chief » Fri Nov 03, 2017 8:38 pm

Although it's not really instrumentation grade, I may pick up one of these: https://www.phidgets.com/?tier=3&catid= ... &prodid=31 to get started.

Vice Chief
Posts: 88
Joined: Wed Aug 23, 2017 5:04 am
Contact:

Re: RINGING the electromechanical challenge

Postby Vice Chief » Fri Nov 03, 2017 9:39 pm

walshlg wrote:so I think that the biggest ringing problems we are actually capable of seeing are in the 0.5-2mm range, what frequency range would that be? It looks like 30-120 Hz BUT you don't get many ringing problems when going flat out at a constant speed, they are mainly on decelerations!


Maybe I'm a little confused -- I think I might recall seeing ringing on accelerations from a stop. Just to be sure, I am going to print the attached file in ABS tonight and see what's going on.

Cube.JPG


This cube is just a first attempt at something that induces ringing. The idea is that after you've printed up a ways, it induces ringing around the edges of the rectangular insets in each side face. The top of the cube is labeled, and the XY origin corners are labeled.

I'll see how this works and make improvements. My feeling is that I'll probably want to label every face, just to make sharing what we learn a little easier.
Attachments
Ringing_Test_Print-export.stl
(72.64 KiB) Downloaded 23 times

Vice Chief
Posts: 88
Joined: Wed Aug 23, 2017 5:04 am
Contact:

Re: RINGING the electromechanical challenge

Postby Vice Chief » Sat Nov 04, 2017 12:39 am

OK, well this turned out to be fairly interesting.

Conditions: 1 year old N2. ABS plastic. .15mm layer height.

I printed my test object twice. The first print was with a single stepper on the extruder carriage assembly. The second was with both steppers in place.

Ringing4.jpg


To make these images, I scanned the "X" faces of each of the test cubes @300DPI, and cranked the brightness and contrast way up to exaggerate the surface features. You can clearly see that with more mass on the extruder carriage assembly, the ringing frequency goes up and amplitude goes down.

With my very crude measurements, I get 7px period for the 1 stepper version, and 10px for the 2 stepper version. So, a quick fix to reduce your apparent ringing by 30% - remove your secondary stepper when you are not using it.

Here's an animation that makes the effect very obvious. I can't get the forum to display it directly - I think you have to click on it to see the animation.
Ringing.gif


AGH, somehow the animation is left-right reversed. Here's a photograph of the actual parts to prevent any confusion. Sorry about that.
Ringing2.jpg


Also, regarding the accel-decel issue, these parts were printed from left-right as scanned. So the ringing evident in the images somehow comes from when the extruder carriage assembly makes a 90 degree change in direction and accelerates.

Vice Chief
Posts: 88
Joined: Wed Aug 23, 2017 5:04 am
Contact:

Re: RINGING the electromechanical challenge

Postby Vice Chief » Sat Nov 04, 2017 4:01 am

walshlg wrote:Yes, that article sounded great but when you read it there is nothing specific, "just we can do this trust us". What I found fascinating was that it was done based on first principles (theory) rather than on measurement. I am a big fan of measuring things!


Press releases usually suck. I found the actual paper here. They measure the printer and the paper is full of detail. The mathematics of their control algorithm are pretty heavy, but the general idea is not. They measured the printer to create a model that predicts the motion and unwanted vibration for a given input. They look into the future of their motion path a little bit, predicting the motion of the printer based on that model. If those predictions contain unwanted motion, their algorithm can null out some of the unwanted motion.

To make that kind of compensation possible, they remove the brain of their lulzbot and replace it with controllers capable of sending commands at 1000hz, which makes sense given the frequencies at which they would need to compensate. That probably rules out using their technique on our RAMPS-based system. Maybe we could catch the lowest-frequency stuff. How fast can RAMPS update a stepper driver?

In any case, if you check out their videos on the topic, none of the prints look great. There must be an accuracy cost to some of this.

The most interesting section is 4.1, in which they describe their characterization setup. They use a PCB Piezotronics 393B05 on the nozzle and a Kistler 8704B100 on the print bed. Those should be good starting specs for any accelerometers we'd like to use.

In the paper they wrote:The FRFs are measured by applying swept sine acceleration signals (with amplitudes ranging from 1.5m/s2 to 3.0m/s2) to the printer’s stepper motors and measuring the relative acceleration of the build platform and printhead using accelerometers


This is probably a good range of accelerations to characterize the ringing we are seeing. Anyone want to whip up that GCode? We'd probably just want to do it one axis at a time, and probably only for X and Y to start. I can test it on my machine(s) with the accelerometer I ordered.

It's also interesting that they measured the build platform vibration. I don't think we would need to do that, since the build platform is fixed in X and Y. But I could be wrong here.

Vice Chief
Posts: 88
Joined: Wed Aug 23, 2017 5:04 am
Contact:

Re: RINGING the electromechanical challenge

Postby Vice Chief » Sat Nov 04, 2017 5:39 pm

I'm starting to think I have a pretty good idea of what's happening here. Here are some images that tell a story. In the images of this print, the motion goes from left-to-right. These were all shot with my photomicrography rig.

1. I printed my test cube, but stopped the print halfway. That way, I can see the relief profile of the plastic where it is being extruded. All images are from the area circled in red:

Overview.jpg


2. This shot is looking at the corner straight-on. You can see that at the corner, the extrusion is "turned up", consistent with the top of the extruder carriage assembly tilting rightward:

Ringing_CornerView_XFace.jpg


3. This shot is looking at the same area from the top. You can see the rippled extrusion. To my eye, it looks like the extruder is tilting left and right along the motion path as it proceeds from the corner.

Ringing_CornerViewTop.jpg


4. Here is a panoramic view of the side. You can see that the initial tilt has the highest amplitude, and it drops off as the carriage continues to travel to the right.

Ringing_CornerView_XFace_Panorama.jpg


Although I can't say conclusively, it looks like the whole extruder assembly is tilting as it's being "pulled" by the timing belt and accelerated from a low velocity. Why would it tilt like that? Two reasons. Slop in the bushings is one, but I also have the strong feeling that it could be from stick-slip friction in the bushings themselves. In fact, it looks like a classic case of stick-slip chatter.

Vice Chief
Posts: 88
Joined: Wed Aug 23, 2017 5:04 am
Contact:

Re: RINGING the electromechanical challenge

Postby Vice Chief » Sat Nov 04, 2017 9:05 pm

Here's some 1000FPS footage of ringing as it happens.
https://www.youtube.com/watch?v=Gqbj1IrtdpM

Here's a plot of ringing extracted by tracking features of the video.
RingingPlotCrop.jpg

User avatar
Julia Truchsess
Posts: 233
Joined: Mon Mar 21, 2016 1:30 pm
Location: Connecticut
Contact:

Re: RINGING the electromechanical challenge

Postby Julia Truchsess » Sat Nov 04, 2017 11:34 pm

Nice work! The X-Y ringing seen in the video is exactly what one would expect when executing abrupt changes in speed or direction of a mass suspended on long rods :) Reduce mass and/or acceleration/deceleration and ringing should decrease proportionally.

Regarding the "rippled" extrusion of apparently varying width when viewed from overhead, might this not be a consequence of the toothed wheel and stepping motor feed system? Or does that stuff get "averaged out" by the fluid melt inside the nozzle?

Re PolyWood, this is a really special material - the only one I know of that has a foamed internal structure. That foamed structure tends to hide layer lines and may well hide ringing in the same way; I'm skeptical about reduced visible ringing being due to its lower weight, as that weight is miniscule compared to the weight of the extruder.

Vice Chief
Posts: 88
Joined: Wed Aug 23, 2017 5:04 am
Contact:

Re: RINGING the electromechanical challenge

Postby Vice Chief » Sun Nov 05, 2017 1:14 am

Julia Truchsess wrote:Regarding the "rippled" extrusion of apparently varying width when viewed from overhead, might this not be a consequence of the toothed wheel and stepping motor feed system? Or does that stuff get "averaged out" by the fluid melt inside the nozzle?


If this were the case, wouldn't it be all over the part instead of just the regions with visible ringing?

The photomicrograph above doesn't show the rest of the print, but looking at it, large flat regions of walls appear flat and smooth.

Vice Chief
Posts: 88
Joined: Wed Aug 23, 2017 5:04 am
Contact:

Re: RINGING the electromechanical challenge

Postby Vice Chief » Sun Nov 05, 2017 1:16 am

Julia Truchsess wrote:Reduce mass and/or acceleration/deceleration and ringing should decrease proportionally.


Right! You can see that in this gif download/file.php?id=5422&mode=view - removing one stepper motor reduces the amplitude and increases the frequency of the ringing. Still a long ways to go, tho. IMHO.

firesped
Posts: 853
Joined: Mon Mar 21, 2016 9:23 pm
Location: Tucson, AZ

Re: RINGING the electromechanical challenge

Postby firesped » Sun Nov 05, 2017 4:19 am

the N-series is not a RAMPS, it is a RUMBA. also I have looked into switching out the board and it really can't be done due to all the inputs. No one else really makes a board to handle 3 to 4 thermal couples. Though I'm not sure if extruder 3 is on thermal couple or thermistor.

Overall, if we have ringing, then likely our acceleration is not calibrated correctly on the printer? Overall as well, we would want have to configure it for the specific setup the end user has. 1 stock stepper, 2 stock steppers, or the various other bondtech upgrades.

It is a mix of hardware and software configurations.

I think I saw something where printers like the Prusa I3 were using things to dampen vibration on the steppers. Not sure what can be done on the N-series for it though.
RL name: Michael Nolen
printers:
raise3D N2 kickstarter Early Bird
Trinus Deluxe (running smoothieware on Azteeg X5 GT board)
Monoprice Maker Select v2


Return to “Others”

Who is online

Users browsing this forum: No registered users and 3 guests