#qi-hardware IRC log for Monday, 2013-12-23

nicksydneywpwrak: got delivery of the new copper board i ordered...will test with that today :)01:19
wpwraknice. new toys for xmas ;-)01:25
nicksydneywpwrak: very much :)02:35
nicksydneywpwrak: got few stuff for my son also to play with 02:36
wpwrakthat's a clever way to hide things. son gets the train set but it's daddy who plays with it until dawn :)02:45
apeletegood morning08:58
DocScrutinizer05hey hardware hackers, and comment on my idea as sketched in http://talk.maemo.org/showthread.php?p=1399859#post1399859 ?10:19
DocScrutinizer05(basically about having the cut edge of a PCB gold plated and using it for contacts. With LEDs embedded in between the contact pads, separating them)10:20
whitequarkDocScrutinizer05: an image is worth a thousand words...10:22
DocScrutinizer05and how do you think I could create such an image?10:23
DocScrutinizer05I'm not a gifted painting artist10:23
whitequarkwell, I bet you can draw that on paper, rougly. you're an engineer after all :p10:24
whitequarkdo the same in gimp10:24
DocScrutinizer05actually I hardly can do that10:25
whitequarkthen, do you understand your own idea?10:25
whitequarknot meant to be offensive or something10:26
DocScrutinizer05but do you understand it? it seems it's not that complex to understand10:27
whitequarkI'm not, that's why I asked for a picture10:28
whitequarkit would probably be easier if I was immersed in your project fully10:28
DocScrutinizer05that's not at all related to any project. The idea is simply to have the cut edge of the PCB plated by gold (pretty usual thing)10:29
whitequarkoh so only the lowest reply in the message is related10:29
whitequarkis it?10:29
DocScrutinizer05http://talk.maemo.org/showthread.php?p=1399859#post1399859 Kommentar 10  since #9 is about exposing the PCB edge to the world, not covering it inside the plastic case10:31
whitequarkok, whether it's lack of sleep, context or intelligence, but I've lost you10:32
DocScrutinizer05I asked if any comments about having edge of a PCB gold plated and using it for contacts on outside of a device10:33
DocScrutinizer05I can't think how to put it any clearer10:34
whitequarkok, it's clear now10:34
whitequarkthough I'm not quite sure how exactly do you intend to use LEDs10:35
DocScrutinizer05I plan to have resections/bays in the outer edge of the PCB where I embedd the LEDs pointing away from PCB in plane of PCB10:36
DocScrutinizer05the LED is ~ 0.8mm*0.8mm*1.2mm10:37
DocScrutinizer05the PCB is 0.8mm high10:37
DocScrutinizer05you cut out a rectangle of 0.8*1.2mm from edge of PCB end embedd the LED there, so LED's shining surface is flush with the edge of PCB10:38
whitequarkwhat would you have that LED there for?10:40
DocScrutinizer05for a lot of funny things10:41
DocScrutinizer05and simply for design reasons 10:41
DocScrutinizer05much like http://wstaw.org/m/2013/12/23/plasma-desktopLW3389.png this switch got embedded10:41
DocScrutinizer05http://wstaw.org/m/2013/12/23/plasma-desktopxY3389.png in this bay10:43
wpwrakLED "height" is really LED "width" ?10:43
DocScrutinizer05though the LED has the same "thickness" like the PCB10:43
DocScrutinizer05no, the height is probably smaller10:43
wpwraki.e., the LEDs would be soldered on the side of the PCB ? and the PCB is 0.8 mm and the LEDs have a width of exactly 0.8 mm, too ?10:44
DocScrutinizer05ok, so when LED is 1.2*0.8*0.4, then the bay is 1.2*0.4 in a PCB of 0.8 thinkness10:44
wpwrakmmh, you need some tolerances, already for the cutting10:45
DocScrutinizer05also for solder etc10:45
wpwrakand soldering will of course be fun. i guess you're looking forward to that already :)10:46
DocScrutinizer05yep, sure10:46
wpwrakis it a problem if the LED sticks out by, say, 0.2 mm at the top or bottom of the PCB ?10:46
DocScrutinizer05well, we *could* handle that10:47
wpwrakthen i wouldn't put these leds. else you'll have the mother of all fights getting them there with superhuman precision10:47
whitequarktechnically those machines already have superhuman precision...10:48
DocScrutinizer05don't see that. When the PCB sits flush on a bench then the LEDs sit flush in the PCB10:48
DocScrutinizer05generally we need component positioning/placing with sub-0.1mm precision anyway10:49
wpwrakwhitequark: do you have a machine that will solder LEDs on the side (!) of a PCB ? i see nimble chinese hands ...10:49
whitequarkhm, yes, that's an interesting question10:50
DocScrutinizer05and given the mech design the LEDs will auto-center easily10:50
wpwrakexcept in the Z axis :)10:50
DocScrutinizer05solder surface tnetion will do the trick10:51
DocScrutinizer05huh, why not in Z axis?10:51
wpwraknaw, the LEDs sounds like a production nightmare in the making. you could design them in and see how it goes. but i'd expect you to run out of patience with them quickly. especially since you require very high precision.10:51
wpwrakalterntively, see if you can get LEDs that are less than 0.8 mm.10:52
DocScrutinizer05the precision is immanent in the PCB design10:52
DocScrutinizer05no way10:52
DocScrutinizer05they need to have height of PCB or we will not get a proper closed outer surface 10:53
wpwrak0.5 and 0.6 mm exist10:53
DocScrutinizer05don't care10:53
DocScrutinizer05the NEED exactly same height as PCB is thick10:54
wpwrak(closed surface) why not ? and who cares ? would the LEDs also bear mechanical load ?10:54
DocScrutinizer05huh? it's the outer surface of the device10:54
wpwrakthe PCB sticks out of the case ?10:54
DocScrutinizer05no, it doesn't stick out, it is fliush with outer surface10:55
wpwrakstill, that's pretty unusual. well, calculate some 0.1 mm tolerances there10:55
DocScrutinizer050.1mm is a damn lot10:56
wpwrak0.1 mm aren't a big deal. the case will still seem smooth. and especially if you have contacts there10:56
wpwrakof course you'll also have to ESD-protect them, etc.10:56
DocScrutinizer05but I don't want LEDs stick out a 0.1mm from our case surface10:57
wpwrakno, i mean in the LED's XY plane, which would be the PCB's XZ or YZ plane.10:57
wpwrakyou should in fact recess them a little. else, of something flexible slides past, it will catch on the LEDs10:58
DocScrutinizer05I rather have a 0.1mm gap on the "bottom" side of the LED to the contacts on the PCB edge, and fill that with solder10:58
wpwrakand you have two big holes on the sides of the LEDs anyway10:58
DocScrutinizer05no way10:58
DocScrutinizer05i rather plan to have the bay in PCB to fit on pressing10:59
wpwrakdo you want to make the PCB go "around" the LED ? in a 0.3 mm space ? no way :)10:59
wpwrakmaybe you really should draw it :) from what i understand, you plan to mount regular top-facing LEDs on the side of the PCB. and it seems that this won't be very nice11:00
DocScrutinizer05I think it is very nice and beautiful11:01
wpwrakmaybe if you try to draw it, you'll notice some problems yourself. i often do :)11:01
DocScrutinizer05I don't need a drawing11:01
DocScrutinizer05I see it in front of my inner eye to the last atom11:01
DocScrutinizer05and I think our PCB manuf can cut PCB to a 0.01mm precision11:02
wpwrakokay, you see something beautiful that is easy to make. i - and i think whitequark too - see something a bit ragged that will be a QA PITA :)11:03
DocScrutinizer05yeah, just like the switches we need to use anyway11:03
wpwrakand if i understand the design correctly, you couldn't just omit the LEDs if they don't work out, because you'd then have a hole in the device's outside, right ?11:04
DocScrutinizer05I again sense this "we can't use this object since it can't get mounted to the wall using a hammer. You need to *turn* it in!" concerns11:05
wpwrakswitches go into case openings. so there is usually a bit of wiggle room. but yes, i hate them. too bloddy big, expensive, messy.11:05
wpwraki think you really should make a drawing. top view, side view. nothing fancy 3D.11:06
DocScrutinizer05there is zero "wiggle room" for the switches mounted like http://wstaw.org/m/2013/12/23/plasma-desktopLW3389.png into http://wstaw.org/m/2013/12/23/plasma-desktopxY3389.png11:07
wpwrakthere is ALWAYS wiggle room ;-) are you an engineer or a theoretical mathematician ? ;-) (a joke in german, from a mathematician: "As gibt zwei Zweige der Mathematik, die angewandte Mathematik und die abgewandte Mathematik.")11:08
DocScrutinizer05our switches are ~0.2mm too high ATM11:09
DocScrutinizer05which already defeats their proper operation11:10
wpwrakah, dremel time ?11:10
DocScrutinizer05huh? we're talking PCB layout here, not DIY11:10
wpwraki mean if they're too high, the bump into something, right ? so you'd have to make room.11:11
DocScrutinizer05and no, you can't dremel anything here, except maybe the switch's moving part11:11
DocScrutinizer05since when you dremel the PCB you remove the soldering pads at edge of PCB that the switch is soldered to11:12
wpwrakno no, i don't mean dremel'ing the PCB. that would be silly. well, unless you tweak a prototype. then anything goes.11:13
DocScrutinizer05no, THIS doesn't11:13
wpwraki mean that "too high" would suggest a conflict with something else.11:13
DocScrutinizer05http://projects.goldelico.com/p/neo900/issues/518/#ic1580 *211:14
DocScrutinizer05see the gap between the two halfmoon shaped things in center?11:15
wpwrakyou mean the switches ?11:16
DocScrutinizer05that's the operation pushdown our switches already operate on, and it mustn't be there11:17
DocScrutinizer05I mean the thing in the center11:17
DocScrutinizer05in the middle between the two switches11:17
wpwrakthat looks like an axis11:17
DocScrutinizer05and the rocker hoovers ~0.2mm above that axis11:17
wpwrakah, so there should be no gap between the plastic and the PCB (at the axis)11:18
DocScrutinizer05gap starts at top of http://wstaw.org/m/2013/12/23/plasma-desktopCz3389.png11:19
wpwrakwell, or filter out the incorrect use by software: buffer rocker input by T, then check if both are activated. if yes, cancel the volume up/down event(s) in the buffer.11:19
DocScrutinizer05the upper thing with concave D is the rocker, the huge thing in center is the axis11:20
wpwrak(gap) ah, i see11:20
DocScrutinizer05between both there's a gap of ~0.2mm11:20
DocScrutinizer05thus when you push on middle/center of the volume rocker, *both* switches operate11:21
wpwrakso it seems that you can't change that. welcome, software solution :)11:21
DocScrutinizer05how am I supposed to not be able to change that?11:21
wpwrakisn;t the rocker switch a single mechanical unit ?11:21
DocScrutinizer05again, we are talking PCB prototyping here11:21
DocScrutinizer05next prototype, the switches sit 0.2mm deeper11:22
DocScrutinizer05into the PCB11:22
wpwrakso it isn't a single item then11:22
DocScrutinizer05this is PROTOTYPING for Neo90011:23
wpwrakrelax :)11:23
DocScrutinizer05isn't exiting/upsetting me what you aim at?11:24
wpwrakyour descriptions are particularly difficult to understand today. maybe it's time for that weekly hour of sleep ? :)11:24
wpwraknaw, i'm struggling with comprehending what exactly you're describing11:24
DocScrutinizer05nah, had ~20 in the last 3011:24
DocScrutinizer05I'm describing existing problems of Neo900 first mech prototype PCB, and ideas for the final product11:25
wpwrakyes yes, the high-level view is clear. it's the details where things get fuzzy. and alas, the details matter, especially when it comes to mechanical things.11:27
DocScrutinizer05and I think we can embed LEDs with about same precision we need for those switches11:27
DocScrutinizer05so I don't see why I have to make LEDs protrude a 0.1mm when we can do better anyway11:27
DocScrutinizer05we even MUST do better11:28
DocScrutinizer05e.g. for switches11:28
DocScrutinizer05witch are mounted quite similar to the way I want to mount those LEDs11:28
wpwrakwhat kind of LED do you intend to use ? not the normal T-shaped one, i suppose ?11:29
DocScrutinizer050.2mm is about as much you press down a button to make the switch operate11:29
DocScrutinizer05I never seen T-shaped LEDs11:30
DocScrutinizer05normal 0804 or 1206 or whatever11:30
wpwrakT-shaped in the XZ plane11:31
wpwraklike these: http://media.digikey.com/photos/Lite%20On%20Photos/LTST-C190KGKT,LTST-C190KRKT,LTST-C190KSKT,LTST-C190KYKT.jpg11:32
DocScrutinizer05when they need to be 0.8mm wide then it's pretty easy to determine the oh-x-oh-y SMD package, though i'm lazy now11:32
DocScrutinizer05My LEDs look like a brick11:32
wpwrakthey are 0.8 mm wide11:32
wpwrakaha !11:32
DocScrutinizer05well, maybe they have "feet" but that's not any problem since such feet can go into holes in the corners of the bay11:34
DocScrutinizer05yeah, actually they may look like the photo you have there11:34
DocScrutinizer05excellent, since then we can push them in to fit from inside to outside (or from bottom, relative to the LED), thanks to the slightly conical shape11:35
wpwrakit's those feet that i'm worried about. you have only about 0.5 mm between top surface of the LED and the top of the feet. that means you need a very narrow cut of the PCB and you end up with a fragile structure there11:36
wpwrakyes, if you can get it to work, it would hold them in place rather well11:36
DocScrutinizer05how's that structure "fragile"?11:36
DocScrutinizer05it's not only 0.5mm high, it's also just 0.3mm wide11:36
DocScrutinizer05again, see http://wstaw.org/m/2013/12/23/plasma-desktopxY3389.png11:37
wpwrakit's quite small and it's on the outside, exposed to mechanical forces. 11:37
wpwrakwhat's the radius of these dogbones ?11:38
DocScrutinizer05it's not worse mechanical than any arbitrary 90° corner of a PCB11:39
wpwrakthat looks like the mill diameter11:39
wpwrakhmm, not sure if you're familiar with dogbones: http://downloads.qi-hardware.com/people/werner/anelok/tmp/11:40
wpwrakif you want to make a "bay" into which something with "sharp" corners fits, you have to make a dent because the mill can only make round inner corners11:40
wpwrakthat's what you see in this picture11:41
wpwrakso you can't make things arbitrarily small in this sort of concave structures11:41
DocScrutinizer05we'll see11:42
wpwrakif things are convex, you can make your corners as pointy as you like, no problem11:42
DocScrutinizer05I don't care how round the concave corners get, we simply make them wide enough11:42
wpwrakyes, but that also means that the remaining material gets thinner and more fragile11:43
DocScrutinizer05after all, the whole structure doesn't have to be exceptionally rugged anyway, since the surface is flusg and plain and inly 0.8mm high11:44
wpwrakalso, a mill has usually a depth to diameter ratio of about 2:1. so for your 0.8 mm PCB it would be about 0.4 mm11:44
DocScrutinizer05flsuh* even11:44
DocScrutinizer05anyway, thanks for sparring11:45
wpwrakheh, always fun :)11:45
DocScrutinizer05seems those LEDs have contact pads on upper side of their 'feet', eh?11:45
wpwrakthe gold goes around, yes. so you could solder on both sides11:46
wpwrakthey're not intended for upside-down mounting, though, because of mechanical tolerances11:46
DocScrutinizer05mount them "from below2 then11:46
DocScrutinizer05hmm, good point11:47
wpwrakthe ones for upside-down have larger legs11:47
DocScrutinizer05know what, maybe I use lightguide plastic and mount the leds 'remote'11:47
wpwrakand they recommend HUGE holes in the PCB for them. but that may in part also anticipate those dogbones (since a hole is a concave cut, you get them here as well)11:48
wpwrakthat sounds like a good idea11:48
wpwrakmore work, but also a lot more flexibility11:48
DocScrutinizer05with a bit of nifty thinking we ight be able to abuse the PCB as lightguide11:49
wpwrakheh :)11:49
wpwrakinstall a side-facing LED upside-down ? ;-)11:49
DocScrutinizer05though, that would drastically reduce efficiency11:49
wpwrakthey may not like you at the SMT fab when you come with that idea ;-)11:50
wpwrakwhat kind of distance between LED and PCB edge do you have in mind ?11:50
DocScrutinizer05heh, told you about that embedded-components PCB technoligy lately ;-)11:50
wpwrakstill sounds scary :)11:52
DocScrutinizer05you know what's scary? the damn LEDs that emit sideways to illuminate the kbd domesheet lightguide11:59
wpwrakwhy are they scary ? just because they face sideways ? or do they have other odd properties ?11:59
DocScrutinizer050.6mm high (guess)12:00
DocScrutinizer05maybe less12:00
wpwrakthat's the LED to PCB edge distance (i.e., the "light guide") ?12:03
DocScrutinizer05sorry for dust on lens12:06
wpwraksoldering looks a little strange. well, side-facing leds are nothing new. we even had them in openmoko :)12:06
DocScrutinizer05not these critters12:06
wpwrakbtw, what kind of camera do you use ?12:07
wpwrak(leds) no, the openmoko ones had a different shape12:08
DocScrutinizer051.8mm long12:08
DocScrutinizer051mm wide12:08
DocScrutinizer050.4mm high??12:09
DocScrutinizer05err, 0.4mm is the domesheet you see next to the LED in the second picture12:10
DocScrutinizer05*incl* the silver foil and two protective films12:10
DocScrutinizer05so the LED might be 0.3mm12:11
wpwrakwow, very flat12:11
DocScrutinizer05I wonder what we shall use for those12:12
wpwrakbtw, you may want to invest some of that neo900 money into getting a better camera. you n900 camera seems quite good for a phone cam, but the graininess and what looks like artefacts are pretty bad in those close-ups.12:16
wpwrakfor comparison, some LEDs snapped with my canon powershot SD880 IS: http://downloads.qi-hardware.com/people/werner/tmp/closeup-sd880is.jpg12:16
wpwrakthat's not a very fancy camera but it has good macro capabilities12:19
DocScrutinizer05I can't convince users or Nikolaus to get me a camera for Neo900 snapshots12:19
wpwrakof course, now you'd have to get a model some 4 generations newer :)12:19
DocScrutinizer05well, maybe the crappy shots at http://projects.goldelico.com/p/neo900/issues/541/ help ;-P12:20
wpwrakhehe ;-)12:20
DocScrutinizer05artifacts are due to max digital zoom12:20
DocScrutinizer05no other way to get a semi-decent shot of that small component details12:21
wpwrakwell, ask them: you do want THIS in the future, or would you rather have THAT ? :)12:21
wpwrakwell, look at my picture. that's without digital zoom.12:21
DocScrutinizer05nah, i need to sort that on a more radical level12:21
DocScrutinizer05project management has to move to me12:21
wpwrakhehe :)12:21
wpwrakwell, it can have advantages if someone else takes care of the paperwork12:22
DocScrutinizer05about to get stuff sorted to set up a business/company/UG/GmbH12:22
wpwrakbut he should understand that you need decent tools to do your work12:22
DocScrutinizer05yes, that's the problem, I can't do paperwork (not even draw) ;-D12:22
wpwrakand especially in this kind of project good pictures are important. also for the community.12:22
DocScrutinizer05good schematics are also essential12:23
wpwrakkinda, yes :)12:23
DocScrutinizer05not something where you don't find signals among all GND and VDD lines12:23
wpwrakhmm, don't draw so many GND and VDD lines ?12:24
DocScrutinizer05that's the problem, I'm not _allowed_ to draw anything (at least not based on GolDeliCo eagle schematics)12:24
wpwrakhmm ?12:24
DocScrutinizer05Nikolaus defines my tasks a tad different12:25
DocScrutinizer05one project one project file one editor/developer12:26
DocScrutinizer05and that's him12:26
DocScrutinizer05when I want changes in anything, i please send sketches (you know i hate sketching on paper) or better yet lyrics12:26
wpwrakthat may not be the smartest division of work ...12:27
wpwrakmaybe he wants you do get better at sketching ;-)))12:27
DocScrutinizer05and my argument that my schematics review and debugging skills get stifled by the schematics style... well, i'm basically not supposed to review schematics12:29
DocScrutinizer05particularly since i'd possibly need a professional license for Eagle eventually, to do that. And (here he's right) Eagle doesn't really support concurrent edits12:31
DocScrutinizer05but then, he plans to have some 2 weeks holiday now afaik. I could completely redraw the schematics in that time and no concurrent editing would happen. But he doesn't like to even consider such approach, probably he likes his own style too much12:33
wpwrakin gta02-core several people worked on the schematics. (kicad, but it doesn't know about concurrent edits either) wasn't a big deal. you just need to communicate a bit.12:33
wpwrak(2 weeks of revolution) pity12:33
DocScrutinizer05maybe kicad differs here12:33
wpwrakwe checked everything into SVN (now it would be git)12:34
DocScrutinizer05won't fly for eagle12:34
wpwrakif you planned to work on some sheet, you told the others on IRC12:34
DocScrutinizer05you can't work on a single sheet of a project, in eagle12:35
DocScrutinizer05when the project is multisheet12:35
wpwrakah, one file to rule them all ?12:35
wpwrakthat sucks12:35
DocScrutinizer05one .brd and one .shm12:35
wpwrakin kicad each sheet is a separate file12:35
DocScrutinizer05and forward/back annotation between the two12:36
wpwrakboards are just one file, though. but i guess for boards you don't want to parallelize too much anyway.12:36
DocScrutinizer05and of course ERC/DRC across all sheets and the board12:36
wpwrakin kicad i always use a forward flow. i think it has some backannotation but i'm not even sure what it actually does.12:38
DocScrutinizer05well, basically first thing you want is to click a component in one notation and see it highlighted in the other notation12:45
DocScrutinizer05sucks to search for N1140 on board12:45
DocScrutinizer05when you spitted something about N1140 to check, while reading schem12:45
DocScrutinizer05spotted even12:46
DocScrutinizer05no matter what, maemo users donated to that project just because they know *me* and my EE background, and because of me telling them I will check the device for any bugs to make sure it works before we ship. So Eagle and proofreading is mandatory12:49
DocScrutinizer05and I won't content with anything less in quality than the GTA02 schematics, when it comes to publishing those12:51
DocScrutinizer05which already gives me headache, since in eagle it's obviously impossible to have proper searchable strings in a pdf printout12:52
DocScrutinizer05not to talk about component placement, but well, we didn't manage that with PADS either 12:52
DocScrutinizer05Eagle schematics are quality like PADS component placement printouts12:53
DocScrutinizer05seems everything is just bitmap in pdf12:54
wpwrak(highlight in both) kicad does that. alas, at least in the version i have, there's a bug, so it sometimes hangs when both schematics and pcb are open at the same time. should check if they've fixed it since ...13:11
wpwrak(you're having edit rights) yes, doesn't make sense to make you go through nikolaus. that's weird.13:12
wpwrakah, my camera has an image stabilizer. didn't even know that :) now i understand why pictures with very long exposure look implausibly good ;-)13:19
ysionneauw/win 4717:58
kristianpaulhmm i tought there will be instrucions to flash nanonote from owrt, but they point to qi wiki18:24
wpwrak_DocScrutinizer05: btw, if shopping for a camera, this one looks like a nice choice: similar characteristics as mine but slightly higher resolution, you can get as close as 1 cm (mine: 2 cm), quite good light sensitivity, and also a nice display. also, it's dirt cheap for al it can do.22:53
wpwrak_in .eu it's called IXUS 255 HS22:53
wpwrak_it doesn't have a lot of manual controls but there's always CHDK .. :)22:54
nickoewpwrak_: adding the "allow holes" after the "unit mm" will make fped segfault.23:13
wpwrak_are you using the latest version from git ? could be that it didn't work in the past23:18
wpwrak_at least the example you posted yesterday or so worked with "allow holes"23:24
wpwrak_ah, and fped can sometimes crash. there's a weird bug *somewhere* that's really good hiding from me. it's somehow connected to deleting things. so save often.23:26
nickoewpwrak_: I am on 23a334823:41
nickoe(it says so in the window title)23:41
nickoenot exactly the latest, rebuilding23:42
nickoenewest seem to not sfault23:43
wpwrak_yes ! victory ! :)23:44
wpwrak_23a3348 was loong before "allow holes". interesting that it crashed, though23:46
wpwrak_syntax error handling seems to be fine, though23:48
nickoewpwrak_: works now it seems23:52
--- Tue Dec 24 201300:00

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