After a very long time sticking with Sprinter Firmware and Sfact for my Gcode generator I decided to upgrade a little bit. I moved to Marlin Firmware (Which has became all the rage in the RepRap IRC Channel), and Slic3r for Gcode generation (Does same job as Skeinforge).
Marlin is a litle more confusing to set up than Sprinter (Configuration.h is modified by commenting in and out things instead of setting variables). Only real change I have found with Marlin is acceleration has a different sound to it, and my rounded parts are a bit more round.
Slic3r is amazing. STL that would take 30-40 minutes to convert to Gcode it can process in 2-4 minutes (The .01 layer height woman only took around 8 minutes to slice and would have been over an hour with sfact/skeinforge). Also, setup is dead simple. Just enter your nozzle size, filament diameter, and extrusion multiplier if you e/steps are a bit off and your good for a 1st print. Only downside of Slic3r is it does not support single wall prints, or support material yet (But if you like to play with code I am sure Sound would like the help on hit Github, or the slic3r IRC).
For ultrafine printing with Slic3r the only parameter you change is the layer height, So both the failed Yoda at .04 and Pink Panther at .01 only required a change from .25 layer height to .04/.01 respectfully. The Yoda failed because at such fine layer heights I discovered that bridging is completely broken, so any small break in the outer skin of a part will never get fixed by the bridging of the next layer and just get worse over time.
After the failure of the Yoda I needed to find a model that looked nice in fine printing, and had no bridging, of course my mind went straight from Yoda to the naked female form.... don't ask.
Printing at this layer height really shows off that I need to upgrade from standard Prusa Z couplings to a Nophead or aluminum Z coupling, but beyond that the print turned out wonderful. at .01 layer height as you can imagine the 1st 2 layers where a bit sketchy, took me 1 hour to get the bed THAT level (1/25 your normal layer height really shows you how not level your bed is).
From talking to some friends in the RepRap IRC I think that with 3mm filament .01 layer height will not be possible, just because the feed rate would actually be too low, and the retractions would cause you to repeatedly go over the same piece of plastic, causing the plastic to shred. But if your running 1.7 plastic and Slic3r repeating this stunt should not be that hard. Hope you have a lot of time, it took 9 hours to get of to the upper chest on this print before I ran out of Z. :)
Equipment/Software used:
Makergear Prusa Hybrid
1.7 Red ABS Plastic from Ultimachine
Marlin Firmware v1
Latest experimental branch Pronterface in Linux
Slic3r GCode Generator v5.7 (Fully stock settings besides layer height at .01)
3 perimiters, 0 infill
I think you'd just need to step 2.9x more finely to match the same feedrate with 3mm. Right? Just (3^2)/(1.75^2)?
ReplyDeleteFor repeated retractions, maybe alternating cams or cycloid gears could force the filament along with less deformation?
I remembered that the last time I checked out slic3r, I was using a Makerbot in RPM mode. I never got a chance to play with it before realizing that it wouldn't work for me because it was 5D-only. I'm checking it out now. There are a lot of Skeinforge features that I'm going to miss, but it looks like it might have just enough functionality for most things.
ReplyDeleteBTW, Slic3r omits redundant dimensions in its G1 codes, which is allowed by the spec, but gcode visualization programs tend not to be prepared for it. I've added support for these omitted dimensions to my Processing gcode examiner: https://github.com/whosawhatsis/gcodeExaminer
ReplyDeleteFor more skeinforge speed. I've build a skeinforge version around PyPy which makes skeinforge not just a bit faster, but LOADS faster.
ReplyDeleteThe topic about it can be found at: http://forum.ultimaker.com/viewtopic.php?f=8&t=153
It's tailored around 5D printers (because a step motor on the filament drive makes sense). And only the SF45 works (included SF41 is bugged)
would it be possible to have the software put down 2-3 layers at .25 then switch to the .01 layer height? that way the bed would not have to be so critically leveled, you could slap down the first few layers without much worry then have the high resolution for the rest of the print.
ReplyDeleteI've been asking for that for a while. It would make slicing a bit trickier, but shouldn't be too hard to implement. If I was familiar with how skeinforge plugins worked, I'd have written one to do just that a long time ago.
ReplyDeleteYou can just slice the object with two (or more) different heights then combine the gcode manually in a text editor at the layer-change you want, I do this especially for low layer printing.
ReplyDeleteCan I confirm that this is 0.01 of a mm? ie One Hundredth of a millimeter layer height?
ReplyDeleteI know this is an old post but just in case someone is reading, I can print just fine in 0.02 layer height with 3mm filament and a Hyena hobbed bolt from arcol.hu.
ReplyDeleteI haven't tried 0.01 yet but nothing happened at 0.02 that would suggest I was on the verge of tripping my filament.