#qi-hardware IRC log for Saturday, 2012-10-06

pabs3http://lwn.net/Articles/518764/02:38
wpwrakpabs3: makerbot's fused deposit modeling systems are rapidly becoming obsolete. technology is moving elsewhere, e.g., to stereolithography / solid imaging03:13
wpwrakfaster and more precise. still a but pricy with those projectors, but people are beginning to figure out how to use lasers. so prices should come down. one obstacle are the secret resin formulas. but the sooner or later someone will figure out how to make generic resins, too.03:16
wpwrakhere's an example: http://b9creator.com/03:22
wpwrakhere's another one: http://3dhomemade.blogspot.de/  quite closed, but has nice videos, specially this one: http://www.youtube.com/watch?v=IPmP0u1HuUM03:35
rohwpwrak: just wait... stl iss still expensive by a factor of 3-10 compared to fdm03:51
rohsimply because of the special chemicals.. and their availability and colors still suck. we'll see03:52
rohwpwrak: you know about form1?03:52
rohbasically a stl attempt with laser as lightsource03:54
pabs3wpwrak: do you have an lwn account to reply there?03:55
rohwpwrak: they have support from the founder of solidworks and the mit, so there is a chance ;)03:56
rohnot for opensource tho03:56
larscviric: It was actually mth who sent the patch upstream07:01
wpwrakpabs3: no, no lwn account10:13
wpwrakroh: yes, laser sounds much better than projector. i didn't mention them because they're also among the less open ones.10:14
wpwraki'd also expect lasers to make the chemicals easier. more energy -> easier formula10:15
rohwpwrak: we'll see... for now its very limited (colors)10:17
rohbut i must say that fdm has come a long way now.10:17
rohmaybe we can't do small flimsy noses and clamps like in usual plastic cases, but there are different, more stable mechanics one can use10:18
rohhttp://raumfahrtagentur.soup.io/10:20
rohwe are at 3 printers now atm. one of which is in a suitcase and portable.10:25
wpwrakroh: popfab looks nice :) i think fdm may eventually develop into something similar like ink printing, which will give you good resolution. but you'll have all the problems of ink printers, particularly proprietary ink that's intimately linked with the delivery system. stl doesn't have this inherent dependency, which should make it much easier to make generic "ink".13:41
rohwpwrak: well.. on fdm there are no inkjet-like problems anymore14:51
rohthe parts are known and documented and the material is nothing magic and expensive. abs and pla are cheaper than anybody who is able to use a printer by far (or the cad/cam part)14:52
rohmeh. now i even did a modular version of my muxing board and still can't afford to go on. sucks14:56
wpwrakyeah, fdm has these advantages. but it's low-resolution and is doesn't scale too well, being a one-dimensional method to solve 2-3 dimensional problems14:57
rohits ok for functional prototyping. for production its way to expensive, same as stl14:58
wpwrakfdm also needs a reasonably strong 3-axis system. not quite a mill, but similar. stl is much simpler in that regard. so i'd expect hw prices to come down quite a bit.14:59
rohthe mechanics funnily isnt the issue14:59
wpwrakoh, mass production is a different issue :)14:59
rohit seems software has way more influence than one would think14:59
wpwrakheh, that too :)14:59
rohbasically the progress in the last 2 years were mostly software. the mechanics were only refining to more reliable versions, no real new concepts15:00
rohalso not really needed. stratasys also does abs-fdm and there is still a bit of stuff to optimize15:01
rohwe just updated the firmware on the arduino of one of our machines last night and that helped a lot to reduce resonances and vibrations15:02
wpwraksee. stl only has to worry about such things in one dimension. and there are only one-way movements there. very simple.15:03
rohheh. one would think. i bet there are also weird pros and cons15:03
rohi could imagine that on a tilting boat stl would be at a disadvantage to fdm (which works impressivly well up to 45deg)15:04
wpwraki think one should try to build an stl printer using a blue-ray burner. relatively cheap mass-produced hardware. and the beam has a high energy (high frequency), which should help a lot with the photochemicals15:04
rohgiven you have oversized mechanics like we do15:04
wpwrakhmm. not sure. the active area i stl is still 2D, just at the bottom of the tank15:05
rohthe form1 machine had quite some complex mechanics for the laser.. galvos i guess... not sure how that works15:05
wpwrakif you want, you can close the tank. then you could even reverse the thing ;-)15:05
wpwrakthat's typical for designs coming from academia. they're used to "first principles" instruments15:06
wpwrak(which makes sense)15:06
wpwrakif you goal is "cheap", things change15:06
rohtrue. i guess on the long run dlp could be cheaper (still with limited resolution)15:07
rohsimply because one doesnt need expensive galvos for the laser15:07
rohonly one simple focus plane and done15:07
wpwrakdlp ? you mean projectors ?15:23
wpwrakmy ideas go in the direction of combining a blue-ray drive with a mirror. use the spindle to rotate the mirror, basically like in a laser printer. use the radial drive to move the laser and reuse the laser. that should give some 2D coverage, one axis by moving the radial drive, the other by selecting the time according to the position of the mirror (needs some photo detector for calibration)15:27
wpwrakthe problems would be mainly the shaping of the mirror and the shaping of the beam. while maintaining a sufficient energy level for the photochemical reaction.15:29
rohquite some mechanical details to solve15:37
qi-bot[commit] Paul Cercueil: menu.cpp: Don't use stat() to know if a FS entry is a directory (master) http://qi-hw.com/p/gmenu2x/407064415:40
larscfunny we just yesterday had a discussion if there is a better way to ignore subdirectories when walking a directory using readdir() than using stat()15:42
pcercueiI don't know15:45
pcercueiif you care about re-entrancy, readdir() is quite a pain to use15:46
pcercueiwell readdir_r()15:46
pcercueiah sorry15:47
pcercueiread wrong15:47
pcercueiwhen I do C++ string = const char *16:01
pcercueiis the string duplicated?16:01
lindi-pcercuei: well the docs say "initialized to a copy of .."?16:03
pcercueilindi-: ok, thanks16:06
mthpcercuei: accepting only DT_REG means that symlinks are no longer accepted?16:43
larscthe check is or != DR_DIR16:44
larscis for16:44
pcercueimth: correct16:44
pcercueibut that was not introduced on purpose16:45
pcercueilarsc: at the bottom there's a check for != DT_REG16:46
pcercueias well16:46
larscah16:46
lekernel(stereolithography) now it gets interesting18:58
viricmth: thanks to you too then :)19:13
--- Sun Oct 7 201200:00

Generated by irclog2html.py 2.9.2 by Marius Gedminas - find it at mg.pov.lt!