Stuttering during large complex print

[swindelljd]

I've done a few small items in PLA and the fan shrouds in ABS. Thus far, I am very happy with the quality of my Rostock Max v2.

This was my first "big print" that ran more than an hour or so and also was a more complex object than the fan shrouds. I downloaded this object http://www.thingiverse.com/thing:576209 and used the default settings for MatterControl printing PLA downloaded from the SeeMeCNC site. I did have a 2mm raft and create support material enabled. Though I don't think that it generated and extra supports.

While this print was running I noticed that the printer stuttered/jumped around quite frequently as if the MatterControl program / computer that was feeding the g-code was not doing it as a smooth data stream, but instead would work in fits and starts. Also, I found that MatterControl had a lot of trouble rendering this model for preview.

As you can see, the right arm ended up printing in two sections and eventually broke off. I noticed that the arms were very unstable and saw the PLA fan shroud even bump into the left arm a couple of times during the print process.

So I am wondering
1) Is MatterControl really only capable of printing smaller or simple objects. Is there a different program that is recommended?
2) While the laptop running the print isn't brand new, it does have 8gb of memory and a dual core processor that was not pinned out. Is the pc the more likely culprit of the stuttering during print than MatterControl itself?
3) Is it better/more reliable to push the g-code to a SD card and let the RAMBo handle the streaming of the data?
4) Has anyone else seen the PLA cooling fan shroud bump into parts of the print? Yes, the fan was running 100%, is this a good idea or really necessary?

Any suggestions or advice is greatly appreciated.

tia,
John

[RegB]

My GUESS is that the PC can and does keep up, although by putting it on the SD card you could eliminate any doubts you may have on that front.
Matter control has a fairly good rep here, that too could be taken off the suspect list by using slic3r or cura.

I suspect a mechanical problem, e.g. belt slip, slack skate eccentric, etc.
A quick and easy check of belt tension and cheapskate fit will probably show something.
Jumping and bumping COULD be due to the skates being too TIGHT on the columns, i.e. SQUEEZED bearing covers get temporary flat spots.

Among my many failings I have a tendency to assemble things slack until ALL the bolts, nuts and screws are in place - although this is generally a GOOD thing I sometimes find places where I never went back to torque them all down.
I am most likely to do it where things are supposed to be "snugged lightly" while adjustments are made, then tightened fully after the adjustment, e.g. the cheapskate eccentrics.
I am likely to do the test run and not go back for the final tightening - YMMV.

[BenTheRighteous]

A few thoughts...

When I first printed my fan shroud a couple months ago, I noticed that mine was bumping into prints too. I realized that as-designed, it doesn't fit onto the effector 100% perfectly - part of the shroud body collides with the effector body and forces the tip of the shroud downwards before you can seat the shroud fully and screw it into place. My solution was to remove the shroud and whittle away at the problem area with a pocket knife (on the shroud, not on the effector! Much easier to replace the shroud if you screw up).

Once I'd removed sufficient material, I had plenty of clearance between the fan shroud and my in-process prints.

The other thing was that I've noticed stuttering as well when printing "very curved" surfaces. In other words, surfaces that are composed of lots of tiny polygons. You might already know this, but computers generally suck at dealing with genuinely round objects, so they're usually approximated to a number of small flat faces instead. The number is up to the designer - less is easier for the computer to handle, but is a crappier approximation also. Same in the other direction.

So if you're printing a round object that's got a high number of faces, then your host computer needs to send lots more commands to the printer in the same timeframe. (Think of it like, 20 short moves vs. 5 medium moves, in order to approximate that curve). Sometimes the interface can't keep up. Usually the computer can send commands much faster than the printer can execute them (which is great; the printer has a buffer), but when that situation is reversed, you get stuttering.

It's been my experience that when that's the case, running the print job from the SD card completely eliminates the stuttering behavior.

Your model looks pretty organic so I'm going to guess that's what you're seeing. Lots and lots of high-poly count curves, which are lots and lots of short little moves and your printer is outrunning the interface.

[BenTheRighteous]

I would also be wary of the speed that you're printing this model at. Those supports look pretty insufficient to actually hold the arms at that angle all the way up until they connect with the main body.

Make sure you have excellent bed adhesion, and printing slowly will put less stress on the contact area making them less likely to pop off prematurely.

[swindelljd]

Thanks for the feedback!

I'll have to check the print speed since it's the default from SeeMeCNC. The default settings also ran the PLA @ 210C. Would running it cooler help or hurt a print like this?

Since I have had a bunch of successful prints (+/- 1st layer issues) I think I am leaning towards the idea that this model with a lot of non-linear surfaces is the issue and will try running it from the SD card. I also think I am going to file down the bottom corner of the shroud to give it just a bit more clearance.

I had made a bunch of calibration settings for the hot end and bed level that I entered into the EEPROM settings area in MatterControl. I am assuming those are written/persisted on the RAMBo and not something that MatterControl sends every time so I should just be able to load the g-code on an SD card and hit "go".

[BenTheRighteous]

You would probably want to print this a bit cooler, as you will likely get a lot of stringing between the legs and arms/supports otherwise.

You are correct in that EEPROM settings are saved to the RAMBo. Printing from SD is as simple as exporting the gcode, then selecting the .gcode file from the LCD menu on the machine.

[DanHall]

I'll second the the thought of printing from the sd card. I have had the same problem on other printers when printing via usb. I'm not sure if there is noise or a commo speed issue somewhere. I was printing from a hexacore Xeon workstation so it shouldn't have been a processor speed issue.

On the Rostock, I did have an issue with the layer fan hanging too low for my liking. A minute or two with some fine sandpaper and I'm happy with it.