#qi-hardware IRC log for Tuesday, 2014-02-04

nicksydneyeverybody split ?05:56
DocScrutinizer05http://wstaw.org/m/2014/02/04/plasma-desktopwm3389.png 07:52
Action: DocScrutinizer05 headdesks - particularly since you can't purge more than one page worth of items. "Select all" means all *of this page*07:53
whitequarkDocScrutinizer05: write a script!07:55
DocScrutinizer05:nod:07:55
DocScrutinizer05pretty easy since page number is in URL07:56
DocScrutinizer05but nfc how to handle that crappy JS on pages07:57
DocScrutinizer05I heard grease monkey should be able to run scripts on rendered webpages07:57
whitequarkDocScrutinizer05: eh, don't bother with greasemonkey07:58
whitequarklook at the POST query it makes and emulate it with curl07:58
whitequarklikely will involve message id -> get it with your favorite libxml binding, or even grep07:58
DocScrutinizer05eeew07:59
DocScrutinizer05I have not even an idea how auth is handled08:00
whitequarka cookie08:00
whitequark-H 'Cookie: ...'08:00
whitequarklook at what the browser sends08:00
DocScrutinizer05nightmare08:01
whitequarkin chromium I just press F12, Network and it nicely lays out the requests08:01
whitequarkfor firefox you need web developer tools addon or something08:01
whitequarkbeen years since I used it.08:02
DocScrutinizer05midgard, what a crap!08:03
nicksydneytest jigs using pogo pins https://www.sparkfun.com/tutorials/13810:11
nicksydneywpwrak: https://github.com/andete/PySWD SWD programmer with Python using Bus Pirate10:37
wpwraknicksydney: as long a your boards have no bottom-side components, you can indeed get away with such simple PCB-only fixtures :)13:12
wpwraknicksydney: else, you'll see that mine are pretty much equivalent, with the differences being in a) using acrylic instead of pcb (because i need more depth), b) my registration being particularly poor in the case of ybpgm (not sure what went wrong, but it seems that i'm almost 1 mm off between acrylic and pcb. ironically, i made an13:14
wpwrakextra effort here to keep those coordinates perfectly matched. in that older MDF fixture it was all manual ... and quite accurate), and c) the holes for the pogo pins are much wider. that's because i'm too lazy to change tools and/or couldn't be bothered to buy a drill of exactly the right size. the pogo pins survive a bit of abuse, especially if they're as short as the ones i'm using, so not having mechanical support at the top is no13:17
wpwrakt a cause for immediate concern.13:17
wpwraklast but not least, yes, designing production testing is important. i skipped that for my firs "mass-produced" product, UBB. since it's hard to get that one wrong but i did it for my second one, the atben/atusb pair. here's the process:13:19
wpwrakhttp://downloads.qi-hardware.com/people/werner/wpan/prod/13:19
wpwraknote: no proper fixture there. hadn't had time to develop one for it. there's a bit of a pseudo-fixture, though: http://downloads.qi-hardware.com/people/werner/wpan/prod/atusb-programming.jpg13:20
whitequarkwpwrak: hm, I just had an idea13:20
whitequarkwhy does a paste stencil need to be out of brass or copper? or for that matter any metal at all?13:21
wpwrak(this one actually did eventually load the pogo pints to their breaking point. but it took many cycles to get there.)13:21
whitequarkwouldn't much more easily obtainable (and easier to work with) sheets of plastic be just fine?13:21
wpwrakprobably for no reason other than durability. and don't they use steel ?13:21
whitequarkperhaps13:22
whitequarkactually, let me check how plastic will work. the bits I have right now should suffice...13:22
whitequarkhrm. the pcb-gcode script can't generate toolpaths for that.13:23
wpwrakwhitequark: high time to write your own toolpath-generating scripts :)13:24
whitequarkwpwrak: lol, no, I'll just modify it a bit13:25
whitequark... no license for it? darn13:30
wpwraklicense ?13:30
whitequarkfor pcb-gcode script. it's in source form, but without any license info.13:31
wpwrakah, that's handy :) good hunt ! :)13:32
whitequarkhunt ?13:33
wpwrakhunting for licensing information13:35
whitequarkoh. yeah. fortunately the authors are active.13:35
whitequark"The source is provided with no license. Use it as you wish." :/13:36
larscpublic domain then13:37
larscbut I guess by "no license" they mean, no license where you need to sign a contract and pay money or something13:38
whitequarkoh, it actually has a license. I was looking at an older version.15:07
whitequarkwheee, I made it generate toolpaths for stencils19:27
whitequarkhttp://i.imgur.com/JIIKxcr.png19:31
whitequarknow that I think about it, why even bother with that spiralling thing? excess stencil material will just fall out19:39
wpwraka very astute observation ;-)20:11
whitequarkhttp://i.imgur.com/QagZnYg.jpg20:16
whitequarkthe one on the right is incompletely milled20:16
whitequarkthe one on the left is much better. all faults are due to the fact that I only had corrugated cardboard for a sacrificial material and it is not exactly flat20:17
whitequarkwpwrak: how does one, in general, reduce burr?20:37
whitequarkshould I increase or decrease feedrate?20:37
nicksydneywpwrak: http://www.cnbc.com/id/101388620 Rich Chinese buy visas in US and Australia20:43
wpwrakwhitequark: yes, i'd reduce the feedrate or increase the spindle speed. if this doesn't help, let the mill run the path twice (the second time you can go faster since it does almost no work)20:47
wpwrakburr can also be cause by worn bits, but that would be a bit early soon your case20:48
wpwrakgrmbl. the high art of editing. s/early soon/soon in/20:48
wpwrakother things to look for: a loose fixture, an unstable table, and lack of lubrication (if the material is hard)20:50
whitequarkplastic :)20:50
whitequarkit's polycarbonate sheets for laser printer.20:50
wpwrakokay, polycarbonate is actually quite hard. but not quite bad enough for that :)20:51
whitequarkactually, polystyrene, not polycarbonate20:51
wpwrakwith plastic, there's also the issue of finding the right speed. if you go too slow, you may melt it. but you seem to go fast in general, so that's probably not the issue20:52
whitequarkok. 1.4mm/s feedrate is better. let's try 1mm/s, and I'll add another 1/4 pass per pad to mill out the loose bits20:52
wpwrakso as a first step, it's go a little slower. you can also make a test pattern of parallel lines or such that will allow you to test parameters more rapidly20:52
wpwrakaah, still the stencil. i see.20:53
wpwrakflexing of the plastic is then probably the main issue. so yes, minimizing the forces should help.20:54
whitequarkyea, it flexes in a nasty way20:57
wpwraki guess that's why they use steel ;-)20:58
wpwrakmaybe try to glue it down, possibly via adhesive tape. that would restrain the movement. then you could look for a way to chemically weaken/dissolve the glue.20:59
nicksydneywhitequark: Bitcoin CPUs Coming, Says Investor http://www.eetimes.com/document.asp?doc_id=132085721:00
whitequarklolwhat21:00
nicksydneywpwrak: https://community.freescale.com/community/the-embedded-beat/blog/2014/01/28/kinetis-mini-mcus-good-things-do-come-in-small-packages?cid=social18045934 ... too hard to solder :)21:01
whitequarknicksydney: that Thomas guy says the right thing, actually.21:01
wpwrakelse, i'm sure china makes also nice cheap laser cutters ;-) that would eliminate the mechanical wear21:01
whitequarknicksydney: (and the part about CPUs in the headline? the article author pulled it purely out of their own ass.)21:02
whitequarkwpwrak: well, yeah. they start from about $1200 for a ridiculously tiny one21:04
whitequarkwhich is big, requires water cooling, and whatnot. I'd rather fix the CNC machine. :)21:04
wpwrakhmm. maybe i could get you interested in a pick & place machine then ? http://www.alibaba.com/product-gs/955073560/HOT_Selling_TM220A_SMT_Pick_and.html21:05
wpwrakthey also have a larger model with 28 reels, USD 1000 more21:06
wpwraki want, in this order: 1) a major reform of customs procedures in argentina, 2) some more space, and 3) such a device :)21:07
whitequarkwpwrak: you nailed it :)21:07
whitequarkalthough for $5k I'll be perfectly fine with either placing stuff myself or customizing the machine I already have21:08
whitequarkbut, yeah, space is a problem. I'm already out of it.21:08
wpwrak(laser cutter) this one doesn't look too bad. doesn't seem to need external air. dunno how the water cooling works, though. http://www.alibaba.com/product-gs/621992867/60w_Smal_Co2_Laser_Cutting_Machine.html21:10
wpwrakwhitequark: btw, did you ever figure out how to use the fixtures ?21:12
whitequarkwpwrak: nope21:12
whitequarkI guess I'll roll my own anyway, these won't suit my thin, flat material requirements very well anyway21:13
nicksydneyVideo: Building an open source PCB printer - http://atmelcorporation.wordpress.com/2014/02/04/video-building-an-open-source-pcb-printer%E2%80%A8%E2%80%A8/21:14
wpwrak(customizing the cnc mill) it may get a little tedious. you need to add air pumps, an elevated table, a feeder system, a fairly complicated head (with a motor for rotation and also the transport needle, although that needle probably doesn't need to be motorized)21:14
whitequarkyou do have a point21:15
wpwraknicksydney: that's a photo plotter ;-) we've had that technology for just how many centuries now ? ;-)21:21
whitequarkcombines the worst of wet and dry processes?21:26
whitequarkslow, requires chemicals21:26
whitequarkwpwrak: ok, I've figured out the problem with plastic flexing.21:26
whitequarksee, it actually results in burr being pushed *under* workpiece21:27
wpwrakhmm, what do you use as support ?21:33
whitequarkI've duct-taped it to a PCB21:33
whitequarkthe bare side, to be more specific21:34
wpwrakthat ought to be stiff enough21:34
wpwrakmaybe try reversing the spindle ;-)21:34
whitequarkmmhm21:34
whitequarkit actually pushes material to both sides of the sheet, so it won't do anything21:34
wpwraknot sure if this won't just make things much worse, though :)21:34
wpwrakah, i see. lower speed didn't help enough ?21:35
whitequarkI think the problem is that I'm using an engraving bit, but ought to use a real endmill with flutes21:35
whitequarkno point in lowering speed under 100mm/min. no difference in results.21:35
whitequarkmaybe I should run spindle slower?21:35
wpwrak(engraving) that sounds suspicious, yes21:35
whitequarkaccording to datasheet right now it should run around 6000 rpm21:36
wpwrakyou can try that. not sure what it does. mine if fixed-speed21:36
wpwrakhave you tried running it twice on the same path ?21:36
whitequarkthat won't really do anything. see, if you look at the bare sheet from the thin side, it looks like this:21:37
whitequark----------21:37
whitequarkand after milling, it looks like this:21:37
whitequark-----III----21:37
whitequarkwhere III is the material pushed to the sides of the sheet21:37
whitequarkit's not *inside* the holes, meaning the bit won't touch it21:38
whitequarkit's basically a bevel now21:38
wpwraktry it anyway :) at least with endmills, this works wonders21:39
whitequarknah, it's worse this way21:50
whitequark& spindle speed (50%) does not noticeably affect result21:51
wpwrakah, pity21:51
wpwrakhmm, lowering the spindle speed is what i would have suggested next21:51
wpwrakwell, if you want to fix the plastic very solidly, you'd cut another pcb such that it has slightly larger openings, then make a "sandwich"21:52
whitequarkfrom what I understand, material being pushed under the sheet has quite a leverage for pulling it out of the fixture, or flexing the plastic & fixture21:53
whitequarkmainly fixture, PET doesn't stretch at all21:53
whitequark(yes, I was wrong two times, it's PET actually. http://www.amazon.com/gp/product/B005T8WR6I21:54
wpwrakhmm. maybe just wait until you get the endmills.21:55
whitequarkyeah22:08
whitequarkto summarize: https://gist.github.com/whitequark/4cfc6c6df7916605d04022:09
--- Wed Feb 5 201400:00

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