#qi-hardware IRC log for Sunday, 2014-02-02

qi-bot[commit] Werner Almesberger: cameo/fped2d2z.pl: hack: maintain decimal point across flipping (master) http://qi-hw.com/p/cae-tools/453f64e05:39
Action: whitequark just made his first isolation milled board10:18
whitequarkI'm starting to really like this process10:18
wpwrakwow. the mill must be quite precise if you got it to work so quickly10:38
whitequarkwpwrak: oh? what problems did you expect to arise?11:04
whitequarkit's not quite "works" yet, because my engraving bit is a tad too big for tqfp (0.5mm tip), but otherwise I'm satisfied with the result11:04
wpwrak(problems) pcb not perfectly level, bad toolpaths, many tries needed to find a good speed11:06
whitequark(pcb level) the pcb I've bought was pretty flat. so I just used a lot of two-sided duct tape and it laid out perfectly11:07
whitequark(toolpaths) eagle's pcb-gcode ULP generates very nice toolpaths11:08
whitequarknot that generating toolpaths for pcb milling is rocket science11:08
whitequark(good speed) dunno, I just spun it to the maximum speed and went with the default feedrate. guess I got lucky :)11:09
whitequarkI guess this will be more of an issue with the proper 0.2mm endmill.11:09
wpwrakprobably :)11:12
whitequarkhttp://pcbgcode.org/ <-11:12
nicksydneyseems like there is a chinese company that is making Bitcoin chip
nicksydneychip is called BM138011:30
larscthere are probably a dozen chonise companies making bitcoin chips11:34
whitequarkreally, mining is not interesting anymore11:39
larscjust start your own $COIN12:00
pcercueiS2hey larsc12:14
wpwraklarsc: in the last century you'd start your own religion / sect to get money and girls; these days, you start your own currency to get money; i guess in the next century you'll start your own own language and ... get nothing12:34
whitequarkwe're already here12:35
whitequarkprogramming language :)12:35
nicksydneyinteresting discussion of hacking and putting electronic in your body http://discuss.biohack.me/discussions12:43
nicksydneywpwrak: in the next century you start selling cells :)12:44
nicksydneywhitequark: human cells :)13:11
nicksydneymy blue sky prediction :)13:12
whitequarkwhy do you need to sell them?13:14
whitequarkI mean, it's (relatively) trivial to cultivate them however you want13:14
whitequarkeven today, the bigger problem is the ethics board, not the technicalities13:14
lekernelanyone knows http://www.openscad.org/ ?13:29
larschm, pcercuei already gone again13:32
paul_boddielekernel: The site (openscad.org) or the tool (OpenSCAD)? The latter I know a little bit.14:14
lekernelthe tool14:15
lekernelhow good is it?14:16
paul_boddieWell, that's a rather open question!14:16
paul_boddieYou probably want to know if it produces usable results, right?14:16
paul_boddieAnd whether I don't hate it just by having to use it, perhaps.14:17
paul_boddieIf you prefer working with code to trying to aim at pixels with the mouse, it's better than other things, certainly.14:17
paul_boddieI'd say it's not noticeably less usable than other CAD tools for Linux, though, at least for my simple applications.14:18
paul_boddieI published a design here: http://hgweb.boddie.org.uk/AcornElectron14:19
paul_boddieMaybe it can tell you more than I can manage to describe.14:19
paul_boddieIt's very simple by most CAD people's standards, but given that I hadn't used OpenSCAD before, I think I figured it out.14:20
paul_boddieThe task of making anything from that model has yet to be attempted, though.14:21
wpwraka long time ago i did some experiments with it and had problems with coincident faces. basically had to add \epsilon to all the coordinates to make them overlap. not fun.14:46
wpwraksystems based on opencascade don't have that problem14:51
paul_boddieYes, that's a known problem and perhaps the only infuriating thing I can think of.14:54
paul_boddieYou get told about bad 2-manifolds, or something, and it doesn't tell you where they are!14:56
paul_boddieBut it's less of a problem than one might think. Maybe they did some fixes.14:56
larscpcercuei: hi15:20
pcercueiI'm in Munich15:21
larscwant to grab a beer or something?15:22
pcercueisure, why not15:25
pcercueibut we're sunday, is that a problem?15:26
wpwrakthen they'll serve the good sunday brew :)15:28
DocScrutinizer05((<nicksydney> wpwrak: in the next century you start selling cells :))) nah, virus that modifies cell dna the way you want, in only the cells you want, and is easily stoppable and even revertable (your blue skin prediction)15:36
wpwrakbrought to you by Monsanto and TEPCO. Companies you can trust.15:43
larscside effects may include swelling, nausea and death15:48
whitequarklife: side effects include death15:57
wpwrakthe final cure :)16:06
ysionneaunice, even a test rig16:17
larscwpwrak: if you see it that way the fear of dying is kind of irrational, cause once you are dead you don't care anymore. One the other hand though if your genes didn't program you to avoid death you probably wouldn't be here in the first place16:24
larscin shool the teacher once asked what we think the purpose of life is. What I wanted to say is that purpose of life is life itself, what I said in my attemnt to condense it down to one word was 'reproduction'.16:42
larsche didn't like that answer ;)16:42
lekernelhttp://ehsm.eu/ << just announced the first talks16:51
wpwraklarsc: i can kinda see how all the girls in your class mentally moved some 10 m away from you :)17:14
wpwraklarsc: and no, the fear of dying makes perfect sense. but it also means that you shouldn't be afraid of it. well, not more afraid than of any other end of a universe.17:15
rjeffriesIs this similar to what wolfspraul was working on before he was abducted by aliens?17:32
rjeffriesYosys is the first step towards a fully open FPGA/ASIC toolchain, and is currently used with vendor place-and-route tools.17:32
larscit's one or two layers above what wolfgang did17:33
larscthis is synthesis, wolfgang was working on configuring the fpga17:34
larscin the middle of those two is par17:35
rjeffriesDo we know is wolspraul is still working in that area?17:35
larscthe project is on ice for an undefined amount of time17:35
rjeffriesthanks larsc17:38
rjeffriesUnrelated: the new RADXA dev board [uses Rock SOC] by Tom Cubie is shipping in small quanties. Early reports are positive. Impressive what one gets for $100.17:40
paul_boddieDoes it have more than 1GB or will we have to wait for that Opteron A-series thing?17:45
rjeffries2 GB RAM, quad core. Not bad at all17:49
rjeffriesa small grace note is it ships with a decent looking plastic case. Also, placment of i/o ports is more sane than  a certain high-volume, low-cost ARM board. In other words, i/o is on front and back, NOT all four sides.17:53
paul_boddiePresumably still arm32 and not arm64, then.17:59
rjeffriespaul_boddie yes, correct. Not sure that matters a lot, but yeah, in another year or so that can be a valid next step.18:10
rjeffrieshas RTC, that's a nice touch. http://wiki.radxa.com/Rock/FAQ18:12
nicksydneywpwrak: any pic of this http://downloads.qi-hardware.com/people/werner/anelok/tmp/ybpgm-0-rig.jpg without the casing ?21:12
nicksydneywpwrak: what solder / station you use ? man your solder is 'clean' and smooth :)21:12
wpwraknot much else to see on the pcb ;-) it's completely passive, just traces and pogo pins21:16
wpwrakwell, and two spots where i over-etched and had to patch the traces :) the acid didn't want to work today. took forever to remove the large open spaces ... and attacked the traces while slooowly etching away at these areas21:18
wpwrakjust used a lot of flux. now mainly the one from chipquik. 21:34
wpwrak(the ones i used before is no longer available so i have to transition soon to something else)21:35
wpwraknicksydney: pictures of just the pcb: http://downloads.qi-hardware.com/people/werner/anelok/tmp/ybpgm-0-pcb-angle.jpg and http://downloads.qi-hardware.com/people/werner/anelok/tmp/ybpgm-0-pcb-top.jpg21:36
wpwrakthe nasty-looking contaminations are non-conductive. i used a bit of solder paste for the pogo pins and that leaves such residues.21:36
wpwrakthe "vias" are in this case wires running at the bottom. didn't feel like doing the full two-sided process for just two traces21:37
wpwrakthe black "scorch marks" in the middle are toner residues. would probably need some very nasty solvent to remove these. (i use paint remover for the toner)21:39
whitequarkwpwrak: acetone removes toner easily21:40
nicksydneywpwrak: thanks ...the black smudge ( http://downloads.qi-hardware.com/people/werner/anelok/tmp/ybpgm-0-pcb-top.jpg  ) is that because of the toner ? i have the same problem after applying acetone lots of black smudge21:40
whitequarkpaint remover sounds like acetone, or maybe a shitty version of acetone. get a non-shitty one ;)21:40
nicksydneywpwrak: didn't read your reply :)...yes i have the same residue problem with my toner too21:41
whitequarkinteresting. so it seems to be dependent on the kind of toner. my toner was easily removed completely.21:41
nicksydneyyour non-copper board colour is different than mine...mine is like orangy-brownish kind of colour21:42
nicksydneyyours like yellowish kind of colour21:42
whitequarkwpwrak: by the way, why don't you mill your PCBs?21:42
wpwrakwhitequark: it does remove the toner. but it also seems to help it get a little into the FR4. i've enhanced the contrast in the images so it looks a little worse than in real life.21:42
whitequark(toner removal) I usually left most of the copper on the PCB to spare etchant, so maybe that's why I didn't notice the smudge21:43
nicksydneywhitequark: i have tried 3 different kind of acetone that i bought from pharmacy and none really helped to clean the toner smudge completely21:43
whitequarksounds quite likely21:43
whitequarknicksydney: from pharmacy? is it pure acetone?21:43
whitequarkor something like nail polish remover?21:43
nicksydneynail polish remover i should say21:47
wpwrak(mill) insufficient Z precision: http://people.openmoko.org/werner/cnc/pcb-1st/21:50
wpwrak(pcb color) pcbs can have all kinds of colors. this one is quite yellowish. i also have some unclad pcb that's basically white. i used that for the front cover of http://downloads.qi-hardware.com/people/werner/labsw/web/21:52
whitequarkwpwrak: (mill) wow, your one sucks.21:53
wpwrakthat's not for the mill but for the piece21:54
whitequarkhrm. I fed it at 250mm/min = 4mm/s21:54
whitequarkah, the bit21:54
wpwrak0.3 mm, and for the last one i used 0.5 mm/s21:55
whitequarkg-wizard's feed and speed calculation suggests about 200mm/min for the 0.2mm endmill. we'll figure out soon whether that will break it21:55
wpwrak3 mm/s ? i see dead endmills ;-)21:56
whitequarkI'm going to be using a TiAlN carbide endmill with 0.2mm dia and 0.4mm stickout21:56
whitequarkperhaps the very low stickout contributes to the high feedrate calculated21:56
whitequarkactually, wait, I think it's not the right term21:57
whitequarkd=0.2 l=0.4 D=3 L=5021:58
whitequarkso it really doesn't get a lot of opportunities for breakage21:58
wpwrakfamous last words :)21:59
whitequarkI've also been assuming I can take out just the copper, making a cut 0.04mm deep22:01
whitequarkwell, 0.05mm. it is theoretically within the machine's repeat positioning accuracy.22:01
whitequarkthe highest possible resolution based on stepper and ballscrew parameters seems to be 0.0025-0.00125mm, but I guess it's not very reliable22:04
wpwrakhow deep (nominally) did you mill ?22:06
whitequarkthis time? 0.2mm deep22:07
whitequarklemme take a picture22:07
wpwrakah, that may explain why you succeeded. 0.2 mm gives you a lot of reserve.22:08
wpwrakof course, with boards going as thin as 0.4 mm, hrmm ...22:08
wpwrakyeah, quite deep. soldering may get a little inconvenient with those "deep valleys"22:13
wpwrakyou also seem to have a fair amount of burr. maybe that's because of the speed. or maybe a 2nd pass would help.22:14
whitequarkhm, burr?22:14
wpwrakthe copper that got folded up onto the trace22:15
whitequarkthere's none22:15
whitequarkif you see anythong on the traces (as opposed to linuxcnc letters or sides of pcb), that's imaging artifacts22:15
wpwrakhmm. those white edges then aren't copper that was moved upward ?22:17
wpwrake.g., the 3rd (center) large elliptical pad seems to have a lot of it22:18
wpwrakand the trace right of it, too (on its left edge)22:18
whitequarkoh, yes, I see. it looks like a bevel. but it's deception. the actual surface is perfectly flat.22:20
whitequarklemme try to take a better picture.22:20
whitequarkI really need to stop taking these on a webcam.22:23
wpwrakah yes, aggressive compression / contrast "enhancement" i guess22:24
whitequarkit actually takes great pictures for a webcam, even HD and very nice colors22:26
whitequarkbut compared to a proper camera... just no22:26
wpwrakyeah. you couldn't tell just from the picture that it's a webcam22:30
--- Mon Feb 3 201400:00

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