smoother and faster printing VS accuracy

[yoavmil 0]

Hi all.

we are building a large and fast printer, and we are willing to reduce the printing resolution and accuaricy to achive faster printing.

the issue is that at each line segmant, the printer try to reach XY axsactly, and only then continues to the next XY, and sometime, EX at corners. the printer will stop completley, and only then continue to the next point.

If the firmware would make corners rounded, the acceleration will be easier, and the printing will be faster.

any idea how to achive this?

 

[illuminarti 18]

it seems like this is something that needs to be taken care of in the modeling process, perhaps with the addition of smart acceleration management in the firmware, as Marlin has to carry as much velocity through corners as possible, but in a controlled way, so as to avoid print artifacts.

Arbitrarily rounding corners in the firmware/slicing process seems like a recipe for disaster if you're expecting to be able to design printed parts that can interact predictably. I can see reducing the resolution of movement, and layer height in order to allow for faster printing but whatever tradeoffs are made, it needs to be totally predictable, the actual positional accuracy of the system needs to be maintained within that resolution, else you just end up with a mess where layers don't line up and you get problems with overhangs etc.

Furthermore, bear in mind that the practical speed limits of deposition printers are more to do with the extruded material characteristics than the printer mechanics. Most high-end 3D printers can reliably deposit plastic faster than will actually create a usable print - you have to slow them down to allow time for the layers to cool and harden sufficiently.

 

[gr5 2,229]

Are you asking about Ultimakers specifically? Or any printer with Marlin? Because if you are then you can tweak the movement settings - specifically XY Acceleration and XY Jerk. If you increase these paramters your printer will print faster at sharp corners.

But Ultimaker uses stepper motors (not servo) so there is NO FEEDBACK if a step is missed. Once you start missing steps the part comes out pretty bad (each layer horizontally shifted in a bad way). So you have to play with these settings until you start losing steps and then back off a bit. It helps to put lots of oil on the rods and keep the belts loose (tight but not very tight). Loose belts are easier. It also helps to increase the current to the steppers (there is a tiny potentiometer on the stepper drivers to control this) but you can damage the steppers if they get too much current so this is only if you are very comfortable with lots of careful testing and reading of the specifications for how hot the motors can get and how much current they can handle.

 

[gr5 2,229]

Another way to print much faster is with thicker layers. The defaults is .2mm but .3mm is fine also. I recommend using a nozzle with a larger diameter. The Ultimaker uses .4mm hole. Illuminarti who posted up above uses .65mm nozzle and where .4mm thick layers is okay quality. The larger the nozzle hole, the easier it is to manufacture (easier to drill out the nozzle) and the faster you can print. At some point you will be printing so fast you can't get the heater block to melt the plastic fast enough.

Illuminarti will likely sell you a .65mm nozzle that fits the UM original if you ask him privately.

 

[gr5 2,229]

If the firmware would make corners rounded, the acceleration will be easier, and the printing will be faster.

any idea how to achive this?

 

If you can write your own slicer, you can slice corners into 10 straight line segments and the printer will print this MUCH faster because Marlin looks at the angle between each line segment and the next. The closer that angle is to zero the less Marlin slows down.

 

[jumpmobile 0]

So if i print a round object a high resolution of my model will not only make my part look better but also speeds up the printing process?

Or is this resolution something i have to set in the firmware?

 

[gr5 2,229]

I can print a full size cup with zero infill and .4mm (one pass) walls in 20 minutes - so I guess so - roundish objects tend to print fast.

Print it again with 20% infill and same print speed and I'm going to guess 2 hours. Infill has lots of corners where the print head has to slow down almost to a stop. An algorithm for printing curved ending infill might be tricky. It might even be impossible to get it to be fast and gently touch the walls but I think you want a path similar to what the Zamboni does on an ice rink.

 

[illuminarti 18]

Speed is a balancing act though... 'round' objects print quickly because the velocity change at each corner is small, so the head doesn't have to slow down and speed up. However, if the source mesh is too dense, then you end up with lots of very very tiny segments, and it takes longer for the firmware to process each segment than it does to print it. The end result is that you then get buffer-underruns (even printing from SD). This causes the head to have to keep stopping while it gets told what to do next. This slows down the print, and results in a blobby mess that also sounds awful as the motors keep stuttering.

So higher resolution meshes are faster up to a point, but then it starts to get worse. Also, for any given volume of container, a square (or better yet, triangular) cross section is probably the absolute fastest. Yes, the head has to slow down for the corners - but the long straight runs allow it to accelerate to the full amount in between.

Short segments with sharp corners (like hex infill) are probably the worst - it has to slow down a lot at the sharp corner, and never gets to accelerate fully between the corners.