General 3D Printing Thread


 
Bryan Mayland said:
I was playing with some Slic3r settings today and stumbled across some pretty interesting relationships.

I'll start with the default settings, 0.40mm layer height, 0.35mm first layer, 200% first layer extrusion width.
Calculated extrusion width: 0.53mm
First layer extrusion width: 0.70mm
First layer extrusion volume: 20% more than other layers

Makes sense to me, right? You put a little more down on the first layer to help it stick right? Well what happens when we change parameters to a 0.3mm layer height for both?
Calculated extrusion width: 0.70mm
First layer extrusion width: 0.60mm
First layer extrusion volume: 13% less than other layers

It gets even worse as the layers get smaller, 0.2mm layer height:
Calculated extrusion width: 0.85mm
First layer extrusion width: 0.4mm
First layer extrusion volume: 59% less than other layers

The reason for this is the percentage is a multiplier on layer height, but the extrusion volume cross section is a square function (actually an ellipsoid). If you want 20% more plastic to be laid down on the first layer regardless of layer height, there is no value you can put in the "first layer extrusion width" that does that.

What's really odd about the first layer extrusion volume, if you set it to 0 for "default" you don't get the same extrusion width or volume you do on the other layers. You actually get about <1% less than other layers. I think this might be due to rounding errors, and this error decreases as the layer height gets smaller-- <0.1% at 0.2mm layer height.
Click to expand...

But I bet the number of first layer passes is also increasing. If not, you would have gaps in between adjacent passes. I would think that the software is essentially increasing the packing density so that there is more surface contact between the bed and the first layer.

Just my theory, but you would have to check that.
 
Ah that's a good point. I'll have to actually look at what comes out instead of just analyzing the GCode. Still, if you're considering the cross section of what it is laying down per segment, it is significantly less using "200%" at lower layer heights than it is at the default. Just something to keep in mind when diagnosing adhesion changing layer heights.

I replaced my Z-endstop with an optical version! The interrupter passes through the optical gate and can travel roughly 10mm past the endpoint if it needs to. I like this much better because I can home the Z axis then adjust the extruder position with a numeric Z-offset with 0.004mm precision. The microswitch worked ok but to adjust the home you had to use adjust where the screw sat between two nuts and adjusting it could be inaccurate, and you couldn't go past where it triggered.
 
FFFFUUUUUU... I built my optoendstop trigger in 123d beta 9 last night and was doing "Save as..." then selecting STL. When I went to tweak the design a little tonight, the .123d file is still the original first version I made. Apparently once you save it as STL it keeps saving to the STL file rather than the .123d file. I can't edit this stl file because it basically just has 2 solids that I can't use any tools on. Is there any way to convert an STL file back into .123d so I can modify it, or am I totally screwed into rebuilding it from scratch?

If it is from scratch I think I'm going to try using another program instead of 123D because I am learning to hate this program so very much. 123D Design is pretty awesome in that I'll be in the middle of building a part and it just disappears (the whole app)
 
Yah, that was my impression.... free but buggy, which can be frustrating when it crashes and you lose your work. If you find something better let us know.
 
Unfortunately, I think you are SOL. It definitely has a few really annoying bugs like the ones you described, plus some others. It doesn't automatically update the save as directory to the recently opened directory. Also, when you are signed in you can't use the open or save icons located in the menu bar. On the other hand, it is free and it works. I originally started off using blender and it was a nightmare.

Brian, since you are a software guy, you probably would love openscad. You should give it a look.
 
I feel for you, tried 123D for about 10 minutes, even after thinking browser based design was a bad idea. I'd think that the people who invented AutoCAD & basically have the same GUI for the last 20yrs would be able to figure out how to make something intuitive... I wonder if my copy of AutoCAD LT R14 (circa 1998) can output .stl files? I'm sure my floppies are around here somewhere :)

Cheap, good 3D design software doesn't exist - If it did, they couldn't sell if fast enough. In college (a whopping 10 yrs ago, ouch), we basically alpha tested IronCAD for a class. I think I made the same design 5 times due to crashes-I still cringe when I see an inside corner with a raduis. Professor had no mercy, just laughed. I got a pirated copy of the final release, still didn't work worth a damn. I saw it on a list of options a company I worked with was trialing. I have a feeling that's what they probably went with...

You all have really cooled off my interest in DIY 3D printing, for now. My wallet thanks you.
 
I ended up using the latest SketchUp to rebuild the part using the last version I printed and calipers. Using the STL export plugin, I was able to generate a worthless STL file that didn't slice correctly. Running it through netfaab fixed it but adding 8 more steps to my pipeline is a real pain in the kiester when trying to do iterative development. Can netfaab tell you which polygons have the problem so you can fix them in the source? It does say I have a non-closed volume or something but it can only fix it automatically, not show me which piece is wrong.

EDIT: Aha! Before you apply the repair, there's checkboxes for "Highlight holes" and "Show degenerated faces" that draws a little rectangle around the area. Went back to my sketchup model and removed a line where two faces adjoined and model is fixed at the source. I've also modified the sketchup exporter to always just export the selection or whole model in millimeters to ASCII stl named the same as the model's filename without 4 popup dialogs along the way. You can even add a hotkey to it. Much better!

It's like the state of game development 10+ years ago, where the artists needed to run their assets through half a dozen different applications manually before launching the game to be able to see what it looked like. I hope the CuraEngine comes along a little further and then maybe it can be integrated directly into applications to skip all the STL export steps in the middle.
 
Last edited:
Oh man SketchUp with Ruby plugins is the greatest thing ever. I added a menu item that exports to STL then runs slicer on it to generate the gcode. I bet I could even get it to upload the gcode to OctoPrint if I spent the time on it.
 
Bryan Mayland said:
.....It's like the state of game development 10+ years ago, where the artists needed to run their assets through half a dozen different applications manually before launching the game to be able to see what it looked like. I hope the CuraEngine comes along a little further and then maybe it can be integrated directly into applications to skip all the STL export steps in the middle.
Click to expand...

That's what I meant when I said the state of 3D printing/apps is kinda primitive at this point... I'm not into game development but I assume the process is a bit more polished these days? I'm looking forward to the day when 3D apps are a bit more polished and can output quality G-Code directly from the app or maybe even print direct from the app, wouldn't that be nice.....
 
Yeah the game industry is a lot lot better now, mainly because most companies just buy their engines then use that extra time to make it easy to get assets in game. Most engines have direct integration with the modeling tools and big asset warehouse software as well, or at least that's where it was years ago when I stopped doing it. I'm sure now it is even more tightly integrated.

The reason 3D printing software can't put out good gcode is because the printers all require so many options to print just right, and one set of options isn't right for every print. They'd have to put all the slicer knobs into their UI which means you can only use whichever slicer they integrate. There's just probably not enough money in it to create their own because >99% of makers probably wouldn't want to pay for it, especially not at the $500+ range that big 3D design/model software companies think their software is worth for the cheapest version.

Autodesk is trying to do this, they have a slicer and pronterface-style uploader built into 123D, but it only supports one GCode dialect. But it is free so I hope it gains some traction so they keep working on it and don't abandon it.

I'm just super pleased Sketchup has plugin support, which sort of makes it my one-stop shop. Design, slice, upload to printer (octoprint, not directly) all from one application just by kicking off various command line programs like Slic3r or curl.
 
I just finished building my aluminum bed for my Prusa i3.... YOWZA, what a difference! I've never seen a first layer go down so well. Finally the first layer is laying down evenly all the way around rather than being squashed here or having poor adhesion there....

I used a 9X12 plate of 5052 Aluminum 3/16 thick, made one cut to get it to a size that would work. I used the bushings I had designed and printed previously to mount the linear bearings to the bed. For mounting the heated bed I tapped M3 threads into the four holes on the Aluminum bed, then used stiff springs between the heated bed and the bed. After leveling I put nuts on the bottom side of the 4 screws to lock them in place. Seems to have worked wonderfully, will have to see over time if they come loose.

I was worried a bit about the mass before I cut the plate to size, but after it was cut to size it didn't seem to weigh too much more than the wooden bed. I was going to drill some holes in it to lighten it up but decided against it because I didn't think it would be necessary. I haven't gotten to any infill layers yet so I'll have to wait and see how that goes. If all goes well I think its time to crank up the infill speed a bit in my Slic3r config. If it seems like the extra mass is a problem I can always pull the bed and drill some holes to lighten it up.

I'm really happy the bed leveling and first print went so well, I love it when things work out!
 
I have worked out that my extruder can only feed between 13-15mm/s no-load maximum without locking up. Even with the stepper removed from the printer, if I try to feed to quickly, the motor just vibrates rather than rotates. Calculating this out, it corresponds to right about 14,000 steps per second where it drops off (13,933 works 14,666 does not). Adjusting the driver current has no effect.

Ralph, will your extruder do a 30mm/s feed? The default Slic3r retraction setting is 30mm/s which doesn't work at all for me.
 
Well, I haven't messed with a ton of settings so I would have to figure out how to make that happen, but would be glad to test and compare...
BTW, been printing with the new aluminum bed installed, and the frame mounted to a wooden base, man this thing is LOT more solid now! When I first built the printer (stock) the bed would move up and down with the slightest touch, (and always was low on the single bearing side), with the aluminum bed is is surprisingly solid and the single bearing side no longer dips down. Before I mounted the frame to the base if I touched one of the Z-towers the frame would wiggle and I could see the extruder head move up and down quite a bit, with the frame mounted to the base it is now rock solid. I am VERY pleased with these two mods on my printer... If you haven't done these yet give your printer a wiggle and consider it...
 
Last edited:
I just powered up the printer, loaded Pronterface and did some extruding to test.....

It extrudes well up to about 1230mm/min and above that it goes off the rails, at 1240 it will extrude but I can hear some vibration and know it's not extruding fully. Above that the extruder gear just vibrates rather than turning... I do have my filament clamped down pretty tight, probably too tight, cause in my initial printing I had some slippage and almost clogged the hotend (because my slic3r settings had the hotend going down to 200C rather than 230C).

I guess that means my extruder is maxing out at about 20mm/s or so....
 
Last edited:
Bryan Mayland said:
....The default Slic3r retraction setting is 30mm/s which doesn't work at all for me....
Click to expand...

So are you suggesting perhaps retraction isn't working because the extruder stepper is choking at 30mm/s?

I did a reverse of 2mm at 1800mm/min and it did reverse, but not sounding real smooth. It is easier for it to pull back on the filament than it is to push it through the head. Although I find that I can even get it to extrude 2mm at a time at 1800mm/min, but again, not smoothy, but not a complete choke either. It does choke on 5mm extrusion at that speed however. I find it will retract smoothly up to about 1575mm/min.... Perhaps we should back the retraction speed down to about 25mm/s?
 
Last edited:
Thanks for testing that for me. It is so nice to have someone with the same printer so we can compare notes!

With my motor unattached with just the small gear spinning, I can't get it past 950mm/min. I can't find specs on these parts so it is hard to say what their limit is supposed to be. That's only 70RPM though which seems a little low. It might be a moot point though because I don't seem to need to retract that much (1mm) to keep ooze from forming. Even at 0.5mm @15mm/s I get fairly clean prints which is 350ms. I might try replacing the stepper because I might have damaged it trying to get it installed which I was being pretty rough with it.
 
Try swapping the pololu driver from the extruder with another. I've had a little trouble with my shield board (I assume) sometimes the X axis didnt want to move. If I pulled the pololu and reinstalled it it would move again (most of the time). Once it was working it never stopped during a print thankfully. I had to do this a few times and grew tired of it, so I got the bright idea to swap it with another (I swapped with the least used Z-axis) and since that day I haven't had a problem with the x-axis again (or z axis). Can't explain it, but that's what happened. I tried adjusting the pololu and everything before I swapped them around...
At any rate, try driving the extruder motor with a different pololu and see how she flies....
 
RalphTrimble said:
Try swapping the pololu driver from the extruder with another. I've had a little trouble with my shield board (I assume) sometimes the X axis didnt want to move. If I pulled the pololu and reinstalled it it would move again (most of the time). Once it was working it never stopped during a print thankfully. I had to do this a few times and grew tired of it, so I got the bright idea to swap it with another (I swapped with the least used Z-axis) and since that day I haven't had a problem with the x-axis again (or z axis). Can't explain it, but that's what happened. I tried adjusting the pololu and everything before I swapped them around...
At any rate, try driving the extruder motor with a different pololu and see how she flies....
Click to expand...

Sounds like your drivers are overheating. Setup a small fan to blow over the heatsinks.
 
I'll give that a try next time I have it apart. Now that I've figured out the trick to getting the idler screws to tightening it is much easier. One of the nuts wasn't going down far enough so I could never get it to screw it and it took 10 minutes of knocking it about to get it to be in the right position. A good push with some pliers once it was off the x carriage straightened it right up.
 
You must log in or register to reply here.

 

Back
Top