Makerware erratic motion when printing small circles

[andre]

Anyone have issues printing small tight circles using Makerware?  Even at a lowly 30mm/sec, the bot exhibits rough motion when going around.  It looks like the tangential velocity is too large, but I can't tell for certain.  There are parameters for "doDynamicSpeed" of which I toyed it, but it make the behavior different, yet never smooth.

Mind you I can slice with Rep G or Cura and it cruises around tight bends nice and smoothly.  Thoughts?

[jason carlisle]

I've noticed it also, and it doesnt seem slicer specific, I see it with files from makerware, repG and simplify3d. I kinda thought it was just something I hadnt noticed before or maybe badly designed stl's, and I've been running my prints really slow lately to mitigate the roughness. I'm using sailfish 7.5 r1135. After seeing your post though, I realized I never noticed it before 7.5 (I believe I saw it in both Sailfish and MBI 7.5)  and checked replicatorG and sure enought there's now version 7.6 out, I'll have to update and see if it makes a difference. By the way, I remember it being especially bad trying to print the replicator filament dust filter at http://www.thingiverse.com/thing:52203

[Whitebat]

Hi Both,

I was wondering if either of you had any luck fixing the issue? I noticed it myself last night... basically any tight circle starts off fine (carriage moving from left to right), but as the carriage starts to head back right to left there is a 'judder' from the machine. There is nothing terribly wrong with the print but that judder and the associated sound cannot be good for the print or longevity of the machine.

I had thought it was as I had just installed an alucarriage, and aluminium x-ends on my Rep 2x. However, on checking these run smooth and true in all directions. I ran the system in jog mode and no problems. I also checked the tension of the belts; but this seems okay. It only seems to happen at standard and high print speeds.

Any ideas or information would be gratefully received

kindest regards

Simon

[Joseph Chiu]

When installing the alu carriage, did you verify that it traveled smoothly without the belts attached and made sure that there is very little resistance?  In the alu-carriage installation that I did for a client, I found that when I had the set screws for the bearings tightened down, there was just enough compression of the rods to cause some binding - it would show up about 30% of the time in the form of not completely moving to the X-starting position at the start of the job.

--
You received this message because you are subscribed to the Google Groups "MakerBot Operators" group.
To unsubscribe from this group and stop receiving emails from it, send an email to makerbot+u...@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.

[Dan Newman]

On 07/02/2014, 5:30 AM, Whitebat wrote:
> Hi Both,
>
> I was wondering if either of you had any luck fixing the issue? I noticed
> it myself last night... basically any tight circle starts off fine
> (carriage moving from left to right), but as the carriage starts to head
> back right to left there is a 'judder' from the machine. There is nothing
> terribly wrong with the print but that judder and the associated sound
> cannot be good for the print or longevity of the machine.

This is often discussed: small tight circles need to be printed slower.
Use a slicer which can slow them down (e.g,, RepG + Altshell plugin with
slowing down for all loops enabled). The issue is that the firmwares
control for linear acceleration but not centripedal. That's because
the firmware has no clue that a small, tight circle is being drawn.
And for centripedal acceleration, it's square(velocity) / radius. Since
you cannot do much about the radius, you're left with reducing the velocity
and that's the job of your slicer.

Dan

[Whitebat]

Hi Joseph,

That was my one of my first thoughts - indeed, when I test installed the alu carriage the first time I had just this problem. However, once I reread the instructions and realized I had overtightened the grub screws that went away and the operation was smooth. There is still a very slight bit more drag at the very ends of the rods, but nothing major...and no binding I can tell.

Kindest regards

Simon

[Whitebat]

Hi Dan,

This is what I feared... the only weird question though is why did I only just notice it? I haven't upgraded Makerware or Sailfish in a couple of months. Maybe it was just the model; and once you notice these things you tend to always 'notice them'.

I'll tone down the makerware default settings and see what happens... actually it gives me a good excuse to try to get to grips with simplify 3D, which I purchased a couple of days ago :)

Kindest regards

Simon

[Ryan Carlyle]

MW has a "dynamic acceleration" function (disabled by default) that slows down as it approaches a turn. You can set the radius that triggers the slowdown, in terms of degrees turn per mm circumference. I use it on exterior shells and it helps. But my circle issues are caused by backlash and dragging the filament around the corner, so I haven't really tried using the dynamic acceleration for circle quality.

[Jamesarm97]

I have always noticed this with small circles. My machine looks and sounds like it is going to rattle apart when doing them, other than that it is smooth and quiet.

[jason carlisle]

I actually had worse jerking behavior with small circles when I was using "dynamic acceleration" in Makerware. It was a very wierd behavior. I ended up turning it back off, and just slowing everything down in general if any of my prints had circular areas a little smaller than a dime. Also I just updated to sailfish 7.6, it didnt make a difference. :(

[Dan Newman]

On 07/02/2014, 8:54 PM, jason carlisle wrote:
> I actually had worse jerking behavior with small circles when I was using
> "dynamic acceleration" in Makerware. It was a very wierd behavior. I ended
> up turning it back off, and just slowing everything down in general if any
> of my prints had circular areas a little smaller than a dime. Also I just
> updated to sailfish 7.6, it didnt make a difference. :(

The acceleration planner and motion control (stepper interrupt) in MBI's firmware
*is* Sailfish. So, you'll see the same behavior in this regard with either.
Marlin, grbl, Sprinter, etc. as well. These all control only linear acceleration
and are blind to centripedal acceleration and it's this latter which you are
being hit with (v^2 / radius). On a much faster processor it would make
sense to watch the incoming segment end points and predict/recognize that
a circle of radius r is being traversed and then kick in centripedal
acceleration control. But our 90's 8 bit, 16 MHz processors just aren't
up to it, at least not when also running ~90 steps/mm stepper motors at
80+ mm/s. Just not enough compute cycles to do everything that needs
doing.

Now, with quite small circles, you're often better off in your models
using 5 or 6 sided polygons instead anyway. (Give you more dimensionally
correct holes as well.) For larger circles (6mm), you just want the slicer
to slow it down. With Skeinforge you can use the Altshell plugin and tell
it to slow down for shells AND loops. Some of the other slicers have
controls as well, but I'm less familiar with them and how they work.

Dan

[funBart]

Now, with quite small circles, you're often better off in your models
using 5 or 6 sided polygons instead anyway.  

Some slicers are capable to change the 'resolution' of the model, by making it more simple (Slic3r has it now, I think to remember, and Makerware in the 'coarsness' setting.) Probably more a setting to play with when you have repeatedly have to print relative simple objects with tiny circle segments as well.

For example: Kisslicer has a mode to define that 'oversample' resolution. Of course it depends on the model and result wanted, but I was able in the past to eliminate that erratic sound (at a give perimeter speed) by slicing with an decreased resolution (so not by 0.8mm grid, but a 1.5mm grid as 'mask' for the stl.)

[jason carlisle]

I noticed today, using simplify3d, that if I print over usb with the software controlling the printer, that my small circles printed without the jerky motions, then I transferred the same file to the sd card and printed it again and the small circles printed with jerky motions. I didnt save the prints to take a photo of, but the print from the sd card had a horrible surface finish compared to the other print

[Dan Newman]

On 09/02/2014, 10:54 PM, jason carlisle wrote:
> I noticed today, using simplify3d, that if I print over usb with the
> software controlling the printer, that my small circles printed without the
> jerky motions, then I transferred the same file to the sd card and printed
> it again and the small circles printed with jerky motions. I didnt save the
> prints to take a photo of, but the print from the sd card had a horrible
> surface finish compared to the other print

That's little tiny pauses being caused by delays of some sort with data
going over USB. Generally not a good thing, but if you're happy with it
for some prints then run with it.

Dan

[andre]

I did resolve this issue.  Thanks to tips from Dan...  basically, under acceleration settings, I lowered the X and Y Max Speed Change (mm./x) from the default 40 to 5.  This essentially will prevent any slicer from attempting to change the speed of an X or Y move (hence extrusion move) too quickly, as is what happens on very small circle moves.  Worked like a charm!

[andre]

And another update...

After toying a bit more with their  "doDynamicSpeed", I think I finally have it working well, for now:

    "doDynamicSpeed": true,
    "dynamicSpeedCurvatureThreshold": 30,
    "dynamicSpeedDetectionWindow": 16.0,
    "dynamicSpeedSlowdownRatio": 0.7,
    "doDynamicSpeedGradually": false,
    "dynamicSpeedTransitionWindow": 10.0,
    "dynamicSpeedTransitionShape": 0.2,
    "doDynamicSpeedOutermostShell" : true,
    "doDynamicSpeedInteriorShells" : true,

Seems to now slow down nicely on small circles without the stuttering motion.  Originally, I may have been using the SpeedGradually, which was introducing some strange motion behavior.  My normal extrusion speed is 120, so this will slow down small circles down to 84mm/sec.  Also, I have slowed down the X and Y acceleration values from 2000 to 1200.  This further slows down very small travel lengths, but allows the bot to reach full speed on many typical motions such as infill (the most repetitive and time consuming motions). 

I was a bit frustrated since RepG/Skein50 was doing small circles very smoothly, yet fast everywhere else.  I went back and forth to understand why, and after visiting the gcode, it was obvious that Skein was being smart about it.  In otherwords, it wasn't a bot hardware/alignment/firmware issue if anyone suspected...