#qi-hardware IRC log for Friday, 2014-06-27

wpwrakbecause if one side was completely flat, it would be relatively easy to mill. if both sides need to be milled, then it gets hard, though, since perfect alignment is hard to archive with lesser mills00:00
DocScrutinizer05well, the upper side is completely flat except for the rounding of edges00:35
DocScrutinizer05particularly where the kbd sits00:35
DocScrutinizer05this however could get done even manually00:36
DocScrutinizer05if everything else fails00:36
DocScrutinizer05or you do it with a "atari-sign"-shaped mill00:37
DocScrutinizer05the inverse function of a half ball00:37
DocScrutinizer05so you can undercut00:38
wpwrakthat sounds good. a bit of manual smoothing is quite acceptable (and normal) for prototyping00:38
wpwraki guess there is no real-life prototype of that structure yet, is there ?00:39
DocScrutinizer05or tooling like http://wstaw.org/m/2014/06/27/plasma-desktopP27898.png00:54
DocScrutinizer05there's the original kbd frame though00:54
DocScrutinizer05which we will replace00:55
DocScrutinizer05I also pondered to add a structure that takes the existing frame, but that's probably cumbersome00:55
DocScrutinizer05I dunno if we can do casts once we got a proper prototype00:57
DocScrutinizer05in resin00:57
DocScrutinizer05I guess for a few hundred it's better to mill every single one from proper surdy acryl or whatever00:58
wpwraklow-quality injection molding should cost only a few kEUR. but then, i'm not sure what quality to expect.00:58
DocScrutinizer05luckily we need no "concave" corners00:59
wpwrakyes, undercuts have a tendency of making things considerably more interesting :)01:00
DocScrutinizer05we have a few L structures but only Z axis01:00
wpwrakyou mean L-shaped slots in the XY plane ?01:01
DocScrutinizer05exactly not01:01
DocScrutinizer05in XY plane we got not a singe one01:01
DocScrutinizer05we got steps01:01
wpwrakso the Ls are "on their back" ?01:02
DocScrutinizer05but thise are easy to do with e.g. a cylindrical mill01:02
wpwrakyes, steps are no problem01:02
wpwrakwhat can be a problem is if there's not enough space for the mill to pass01:03
DocScrutinizer05actually everybody does steps all the time when milling01:03
wpwrake.g., wall, 1 mm wide 5 mm deep valley, another wall01:03
DocScrutinizer05not on spacer frame01:04
DocScrutinizer05it's a pretty simple thing to CNC, I guess01:04
wpwrakwell, when i get some parts, i can see if my mill can handle it. would be the first proper DIY spare part :)01:04
DocScrutinizer05what's nasty is that we have a smal 3..5mm high wall at the outside, and an rather huge area in the inside that needs to get milled down to a material thickness of <1mm01:05
wpwrakthat only takes a bit of time :)01:06
wpwrakand 3.5 mm isn't all that tall anyway01:06
DocScrutinizer05but that's not much differend to your transparent acrylic "pots" you did01:06
wpwrakyes, these were much worse01:06
DocScrutinizer05actually it's quite identical, coming to think about it01:07
wpwrakgood :)01:07
dos1"project ara" prototype was demoed at google io a few hours ago01:07
DocScrutinizer05just larger, and with a apperture or two cut into the "top"01:07
dos1it was shown booting android01:07
dos1and that was actually it, cause it hanged at the end of boot :)01:07
wpwrakah, what are the total dimensions ?01:08
DocScrutinizer05whatever might be project ara01:08
dos1motorola/google phonebloks01:08
wpwrakAffiliate Registration Authority01:08
wpwrakwell, no too far off ;)01:09
DocScrutinizer05(dimensions) pretty much identical to the PCB, a tad larger so the PCB firs in01:09
wpwraknumbers, please :)01:09
wpwraki.e., < 10 x 15 cm ?01:09
wpwrakah yes, the whole critter isn't all that long. excellent.01:10
wpwrak(10 x 15 cm is the work area of my mill. so subtract about 5 mm on each side and you have the maximum size i can mill)01:11
wpwrakelse we'd have to enlist whitequark ;-)01:11
DocScrutinizer05I could carve a carton model01:12
wpwrakthat's another approach: stacked 2D panels01:13
DocScrutinizer05but probably not to scale, at least for the rim01:13
DocScrutinizer05the step we need at the rim will be a tad nasty01:14
wpwrakthe vertical wall ?01:15
DocScrutinizer05as in: sub-mm01:15
wpwrakhow thin is the thinnest wall ?01:15
DocScrutinizer05that's pretty much up to us01:15
DocScrutinizer05thick enough to be stable01:16
wpwrakthere must be some constraints from the existing case.01:16
DocScrutinizer05yes: the step to take the wall of existing case needs to be 1.2mm deep01:17
DocScrutinizer05since that's the thickness of the existing wall01:17
DocScrutinizer05an d that existing wall has a step as well, where the kbd frame sits. The wall mecomes 0.8mm there01:18
DocScrutinizer05so our step needs a step of 0.4mm01:18
wpwrakis that 0.4 mm part exposed on the outside ?01:19
DocScrutinizer05actually two steps. one 0,4mm deep, one 0.8mm deep01:19
wpwrakoh, depth isn't much of an issue01:19
wpwraki'm more interested in wall thickness01:20
wpwrakor ridge/rim01:20
DocScrutinizer05"deep" as "into the wall"01:20
wpwraksteps (in any direction) with a precision around 0.1 mm are usually no problem01:21
wpwrakthe problems start when the walls get too thin01:21
wpwrakand then there's Z accuracy. so the thickness of the bottom may have some variation. with my mill, which isn't very rigid, that variation can be significant. so if the bottom doesn't have to be super-thin, that helps.01:22
wpwrakanyway, there are a lot of ways to cheat when prototyping. for production you'd want to find a place nearby with better equipment. maybe roh knows something. or nik. worst case, you'll have to make friends at a local hacklab :)01:24
wpwrakoh, that looks quite easy01:25
DocScrutinizer05the steps have to be super-level01:25
DocScrutinizer05they sit on a level existing case and are visible01:25
wpwrakthat's what mills do :)01:25
DocScrutinizer05I got a place nearby, university spin-off. they do such CNC stuff01:27
DocScrutinizer05they take pride in doing absurdly complex things01:27
wpwrakat thing on top of "arbitrary" would be the bottom, correct ? what sits on the 1.2 mm rim, what on the 0.4 mm rim ?01:27
DocScrutinizer05didn't talk to them yet since I have no model or data01:27
DocScrutinizer05that's a negative copy of what the case rim looks like01:28
DocScrutinizer05the wall is 1.2mm thick01:28
DocScrutinizer05and ahs a step of 0.4mm for kbd frame01:28
DocScrutinizer05the thin end of this spacerframe wall (ARBITRARY) sits inside the case01:29
DocScrutinizer05the lower right side vertical is visible01:29
wpwrakhmm, looking at sketch,pdf, the "arbitrary" part looks unused. i.e., could be 0.01:30
DocScrutinizer05somewhere below the "lid" would go to the left01:30
wpwrakunless it has to push against anything on the inside of it all01:30
DocScrutinizer05nope, but we need it for mechanical rigidity01:31
DocScrutinizer05the kbd frame alone isn't sturdy01:31
DocScrutinizer05the step, that is01:31
wpwrakhmm. we'll see :) there may be better ways01:31
wpwrak(e.g., perpendicular supports)01:32
DocScrutinizer05the wall of case is only 0.8mm there, and a ...01:32
DocScrutinizer051.1mm high01:32
DocScrutinizer05we want to add an additional 1 or two mm down (now up) into the case01:33
DocScrutinizer05at least every now and then01:33
DocScrutinizer05might be notches, 2mm wide01:33
wpwrakthe only bit that has to be perfectly level is the first step. seems that the rest can be bumpy since it's on the inside and has no mechanical function that would require perfection. so you'd add a vertical gap of, say, 0.1mm and you're good.01:33
wpwrakhow thick is the top surface ? what goes around keyboard etc.01:36
DocScrutinizer05or we accept a eyecandy gap on the outside and make the wider inside the mech fit01:36
DocScrutinizer05one mm, or two01:36
DocScrutinizer05there I found it, the friggin kbd frame01:37
Action: DocScrutinizer05 eyeballs it01:37
DocScrutinizer05some funny structures there01:37
wpwrak(eyecandy gap) naw, so far all this looks quite doable. i work with similar dimensions for my stuff. and there are also limits to what nokia can have done without ending up with something the crumbles when you just look at it ;-)01:38
wpwrakthe funny structures tend to be the worrisome part ...01:38
wpwrake.g., i wouldn't want to have to make something like the openmoko case. that one was full of nasty undercuts and thin flexible walls01:39
DocScrutinizer05ok, in ~23min http://neo900.org/stuff/joerg/kbdframe/ will be complete populated02:06
DocScrutinizer05the recangular "pot" one one side is actually part of what forms the mic acoustic chamber02:10
DocScrutinizer05we might not neet it02:10
DocScrutinizer05there are 3 thin "horns" on top, and right to them a 18mm wide "table". The horns go into apertures of shielding cans on N900, the table has even a U-shaped steel structire soldered to N900 PCB. All to thold the kbd fram in position, dunno what would make it jump around when those things were not there02:14
DocScrutinizer05on the "lower" thin side you see three pairs of latches protuding ~3mm into the case, those are somewhat holding this flimsy thing plastic string in place under the kbd and between kbd and case02:16
DocScrutinizer05we might not heed those either, obviously02:17
wpwrakhmm, lots of "horns" ...02:20
wpwrakgood if we don't need them :)02:20
DocScrutinizer05err 8min02:21
DocScrutinizer05once you got a 16Mb downstream, a 1Mb up can really suck02:22
wpwrakthe "table" has an undercut. so part of it has to go for basic CNC02:24
DocScrutinizer05yes, I think we can forget about it, we won't get the steel U anyway02:24
wpwrakthe screw "barrels" where the screws go need a tool change. ah, *-08 is very nice02:25
DocScrutinizer05and actually I got NFC why it's needed02:25
DocScrutinizer05the barrels will have to live with getting cut by the screws, eh?02:26
wpwrakyeah, that too.02:26
DocScrutinizer05those are FR02:26
DocScrutinizer05TX6 FRdunno*402:26
wpwraki hope nothing depends on that "bay" on the left side on  neo900.org/stuff/joerg/kbdframe/2014-06-27_02-46-06.jpg  to be exact02:27
DocScrutinizer05that looks pretty bogus02:28
DocScrutinizer05I thinks there's nothing on PCB there (yet)02:28
wpwrakneo900.org/stuff/joerg/kbdframe/2014-06-27_02-47-24.jpg shows the microphone chamber, right ?02:29
DocScrutinizer05though on Neo900 I want a lot of LEDs there ;-)02:30
DocScrutinizer05yes, that's the acoustic chamber02:31
DocScrutinizer05we will not have that thing - probably02:31
DocScrutinizer05since it needs to go down to the tube which ends on that tiny hole left of AV jack02:32
DocScrutinizer05NB it's connected with the smaller rectangle left of it, the start of that tube02:32
DocScrutinizer05the whole thing is a rubber part anyway02:33
DocScrutinizer05you can see the two holes in rubber and the plastic bolts fixing it at extreme left and right02:34
DocScrutinizer05(left almost covered by screw barrel)02:34
DocScrutinizer05btw upload finished02:35
DocScrutinizer05uploading some more pictures of kbd frame in case02:47
DocScrutinizer05(chamber) we either keep microphone in that position (on UPPER) and male the chamber have a direct aperture/hole to the environment, or we place the mic on LOWER, so where it originally been, and make the chamber by a rubber frame between LOWER and UPPER02:49
wpwrak(more pix) maybe snap a few of the things it attached to as well02:50
DocScrutinizer05I somehow prefer solution #1 since it allows for symmetrical second mic on left side02:50
wpwrakand it would be nice if you did a little triage first. e.g., if you have N pictures of the same area, keep only the best. it's very tedious to sift through a large collection over the net02:51
wpwrak(more pix) remove kbd critter, then take pictures of that's underneath02:51
DocScrutinizer05yeah, will weed out a few of them02:51
DocScrutinizer05the first three "identical" pictures are of case without frame, a lot are actually bogus shit with frame places randomly on top02:53
DocScrutinizer05for what's *actually* underneath the frame in N900, refer to highres scans of PCB02:54
DocScrutinizer05the frame obviously sits on top of the hige white domesheet02:54
whitequarksomeone pinged me?06:53
wpwrakwe just thought about recruiting you for spacer milling :)07:00
wpwrakwith great CNC mills comes great responsibility ;-)07:35
whitequarkI could try it09:01
wpwrakyou even have an N900, don't you ?09:20
Textmode...and I've completely forgotten how to navigate the nn's filesystem.10:01
Textmodewait, didn't it have some limitations on what partition formats it supported?10:02
Textmodemight explain why I can't find my card...10:04
whitequarkwpwrak: nope I don't10:13
wpwrakah, right, that was a gta0410:22
whitequarkI only preordered neo90010:44
wpwrakwhee, seems we'll have an **UPGRADE WEEKEND**. colorful bug in LZO found, article in german: http://www.heise.de/meldung/Zwanzig-Jahre-alte-Luecke-in-Lempel-Ziv-Kompression-gefaehrdet-Linux-Nutzer-2240442.html15:46
Textmodefun for the whole family?15:53
wpwrakwell, it seems that it's less bad than it looks: fastcompression.blogspot.de/2014/06/debunking-lz4-20-years-old-bug-myth.html15:53
wpwrakof course, still a good idea to fix it before someone finds an exploit that can do serious damage despite the problem being considered "harmless". wouldn't be the first time something like that happens.15:54
whitequarkyeah. exploits never grow weaker, only more powerful.15:58
ysionneaureal security issues are just playing lego with a sum of inoffensive exploits, when summing them it ends up deadly16:08
Action: DocScrutinizer05 figures an ICBM built of Lego blocks18:45
astrI want to get a led that is as bright as the ones in smartphones for camera flashes/touch. what sort of brightness would you say there are and do they have a special name?21:46
wpwrakmaybe search a parts catalog for "flash" ? that should find some specifically designed for such use22:35
wpwrakwell, or at least specifically marketed :)22:35
astrI guess I look based on brightness, I don't plan to use them for flashes specifically23:08
dos1flash leds on phones can be very bright, but only briefly23:09
dos1if you use flashlight mode, it's not as bright as on short flash23:10
--- Sat Jun 28 201400:00

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