#qi-hardware IRC log for Saturday, 2014-02-15

wpwrakheh, i think i have the perfect connector problem to drive joerg mad: 0.7 mm pitch ;-) (we don't even have to go into any other details. that alone makes it impossible :)00:08
wpwrak(been thinking of a simple way to make the OLED in anelok a little more modular. seems that this would require a little PCB then)00:09
DocScrutinizer05ok, oo draw sucks00:19
DocScrutinizer05placing small circles on a larger one, you can't resize the larger circle and the small ones would stay same radius but move their center points accordingly to the large circle00:20
wpwraktry fped :)00:20
wpwrakapt-get install fped   may work00:21
DocScrutinizer05prolly not00:21
wpwrakahm you're of course on suse ...00:21
DocScrutinizer05lemme try something. You got a URL of .deb ?00:23
wpwrakyour epicycles simulator would require the use of trigonometry in fped. so it's not really easy (it's not a common pattern/operation in electronics), but doable00:23
DocScrutinizer05ouch, no thanks00:23
DocScrutinizer05I'd be absolutely happy when small circels' center point would persistently snap to epicircle00:24
DocScrutinizer05or mastercircle00:24
wpwrakhmm, here are some .deb: http://packages.ubuntu.com/precise/amd64/fped/download00:24
wpwrakwell, you'd make a vector r*sin(a*i), r*cos(a*i)00:25
DocScrutinizer05amd64? really now?00:25
wpwrakwhere i is a loop from 0 to N-1 and a is 360/N00:25
wpwrakwhat do you have ? ENIAC ?00:25
wpwrakZuse ? :)00:26
DocScrutinizer05NFC?00:26
DocScrutinizer05x86?00:26
wpwrakyou're still on 32 bit ? wow00:26
DocScrutinizer05nope00:26
DocScrutinizer05but neither on AMD00:26
wpwrakthen amd64 ...00:26
wpwrakamd64 is the architecture. intel copied it.00:26
DocScrutinizer05uhuh :-P00:27
DocScrutinizer05LOL http://wstaw.org/m/2014/02/15/plasma-desktopQK3389.png00:29
wpwrakmaybe it's easier to build it from sources ... ;-)00:31
DocScrutinizer05now, is that thanks to .deb gets forwarded to that unspeakable "apper" that doesn't know .deb, or is it just because the .deb contains a amd64 ?00:31
wpwrakor maybe suse has it ?00:31
DocScrutinizer05tbh I don't think I want to go 100% parametric as of wpwrak's definition00:31
wpwrakit's easy00:32
DocScrutinizer05as long as there isn't a GUI that does elementary stuff like resize, drag, snap etc for me00:32
DocScrutinizer05sounds like writing poems in tex00:33
wpwrakit does all that. but you have to define the underlying geometry in a parametric way00:33
wpwraki.e., there's no grid00:33
DocScrutinizer05and I mean: *pure* tex00:33
wpwrakfped has a gui :)00:33
wpwrakbut you can also "program" it. with CPP and all that ;-)00:34
wpwrak(but don't mix the two. if you want mix modes, don't use CPP because it doesn't know how to preserve that information)00:34
DocScrutinizer05how would I create two circles that touch each other?00:34
wpwrakyou set the two radii, make a a vector (r1, 0) and attach to it a vector (r2, 0)00:35
wpwrakthen you draw a circle with center at the first vector to its end, and another one from the end of the 2nd vector to its beginning00:36
wpwrakexact details depend a bit on how the geometry of your circles is defined, but that's the basic approach: you define the geometrical structure with vectors, then you put lines, circles, rectangles, pads, or holes on them00:37
wpwrakif you're just using it as cad, forget about the pads and holes00:37
Action: DocScrutinizer05 feels like in school, messing around with an enormous clumsy compass and a set square and chalt on the chalkboard00:37
wpwrakyou can also turn circles into arcs00:38
wpwrakwell yes, but your drawing tools are pretty alive this time :)00:39
DocScrutinizer05no, they are pretty formal it seems00:39
wpwrakthe workflow is optimized for translating footprint drawings in data sheets. there, you usually have "this many mm up from here", "this many mm from the center", etc.00:40
wpwrakfped knows a bit of trigonometry so you can do also the more common other operations.00:41
DocScrutinizer05hmm, it seems to not know about snapping00:41
wpwraksomething else it does it to let you vary parameters. you can have tables with sets of parameters. so you you, say, have five different circle diameters, you can make a table with them00:42
wpwrakfor the circle, drag from the start to the end00:42
wpwrakor are you at the vector ?00:42
wpwrakyou also drag the vector00:42
wpwrakthen click on it to edit the parameters00:42
DocScrutinizer05look, I draw a circle. How would I create a vector that starts at circle's center?00:44
DocScrutinizer05how would I find the point where the vector cuts the circle?00:44
DocScrutinizer05how would I start a new vector there that has same direction like the first one?00:44
wpwrakthe vector is your radius00:44
wpwrakyou attach v2 at the end of v100:45
DocScrutinizer05how would I create a 2nd circly at end of the 2nd vector?00:45
wpwrakfor the direction, you set your parameters accordingly. depends on what you want to do00:45
wpwrakyou click and drag from the end of v2 to the beginning of v100:46
wpwrakerr, beginning of v2 / end of v100:46
DocScrutinizer05"set parameters accordingly" sounds like I'd be better off with a HP41C doing the calculations for me00:46
wpwrakshall i walk you through it ?00:46
DocScrutinizer05why do I need a CAD then?00:46
wpwrakdo you have fped running ?00:46
DocScrutinizer05no, I think I'm not that interested00:47
wpwrakwas der bauer nicht kennt ... ;-)00:47
DocScrutinizer05as long as the thing doesn't know a "pick from drawing" mode for arbitrary parameters, where you click near a significant point and it picks the closest*highest-significance point next to where you clicked and picks the (usually) X,Y from it...00:51
wpwrakactually, i lied. it has a pseudo-grid it uses when you create new vectors. but it's only real purpose is to let you drawn vectors at 90 deg angles (so that you only need to edit one parameters, saving you some 0.5 seconds ;-)00:53
DocScrutinizer05so I'd draw a circly anywhere, then hit F37 or whatever to enter "parametric mode" which pops up a window with all parameters of the circly, I click into "center x" and into "center y" then hit F38 to "pick from drawing" and click near the corner of a square...00:54
DocScrutinizer05it still wouldn't solve the touching-circles problem in one step, but for sure in two or three00:55
DocScrutinizer05I'd also like to have the option to either "pic by value" or "pick by reference"00:56
wpwrakbut all that's not parameteric00:56
DocScrutinizer05"pick by reference" would glue the circle's center to the square's corner00:56
wpwrakit's merely a way to use either mouse or keyboard to enter a number00:56
DocScrutinizer05^^^00:56
wpwrakah, so if you resize the square, it stays at the corner ?00:57
DocScrutinizer05sure, when circle#s center is by reference to the wquare00:57
DocScrutinizer05square even00:57
wpwrakokay, that's a big step in the right direction00:57
DocScrutinizer05mind you, I'm just picking the value for a textfield (center X,Y) in circle's parameter dialog00:59
wpwrakbut that would break the reference, right ?01:00
DocScrutinizer05by reference would probably result in something like "$object14.(origin+Y)"01:00
wpwrakokay, so they have a parametric mode01:00
wpwraknot just numbers but expressions01:00
DocScrutinizer05who says a text input field can only hold numbers?01:01
wpwrakand yes, my vectors are "origin" or such. only that i made vectors first-class objects, not things you extract from the things you actually draw01:01
wpwrakit's much tidier that way01:01
wpwrak(and if you really hate them, you can also hide them :)01:02
DocScrutinizer05sorry, you lost me 01:03
wpwrakhere is what your cycles look like: http://downloads.qi-hardware.com/people/werner/tmp/fped-cycles.png01:05
wpwraknote that i actually made three drawings: one with the circles on the outside, one on the inside, one on the big circle. you can switch between the three01:07
wpwrakor if you like chaos, have them all at the same time: http://downloads.qi-hardware.com/people/werner/tmp/fped-cycles2.png01:08
wpwrakif you do't like the big circle, just delete it. since the geometry is derived from the vectors anyway, the rest still works01:10
wpwrak(note that the 2nd picture isn't quite clean since the big circle is draw three times. you'd have to add another frame to avoid that. of course, the visual result is the same in this case. it would be a problem if you defined pads/pins, though, since they would overlap)01:13
DocScrutinizer05nice, but quite hard to grok, and looks more like writing programs than like drawing01:13
wpwrakyeah, nearly as intuitive as  $object14.(origin+Y)  ;-)01:13
wpwrakNOT nearly even01:13
DocScrutinizer05tzzz01:13
DocScrutinizer05the little difference being that I wouldn't type a single char of all that01:14
wpwrakyou can actually get a look at the "code" too: http://downloads.qi-hardware.com/people/werner/tmp/fped-cycles3.png01:14
wpwrakwell, i just clicked for most things, too ...01:15
wpwrakand do you actually have a primitive for angular iteration ?01:15
DocScrutinizer05well, maybe in "copy" under "rotation" ?01:16
wpwrakokay. now a challenge: change the number of circles :)01:17
wpwrakin fped, you basically have to world views: "objects" and "instances". objects are what you draw, instances are what you see. by making this distinction explicit, controlling what happens is very straightforward01:19
DocScrutinizer05actually oo draw calls that "duplicate"01:19
DocScrutinizer05and it has a parameter "number of copies"01:19
wpwrakof course you don't really think of it as object/instance but you just have fped "replicate" things for you01:19
DocScrutinizer05change numbver of circles? easy! click on one of the epicircles, select "show tree", go one up in tree to find "duplicate", click on it to open parameter requester, change "number of copies" to whatever you like01:21
wpwrakah, nice. so they maintain the tree as well. good.01:21
DocScrutinizer05what's ow? again not good enough for a hacker that doesn't use a mouse?01:22
wpwrakcan you "fuse" items ? that's an operation that often breaks such trees (it theory it wouldn't have to, but in many cad system that's what happens)01:22
DocScrutinizer05fuse?01:22
wpwrakso something that creates something new from two different objects. e.g., line between the centers of two circles01:23
DocScrutinizer05sorry01:24
wpwrak(that "loss of tree" is particularly infuriating in 3D cad systems)01:24
wpwrak(i guess they do it because their tree would then become a loop-free directed graph. still not really difficult but of course getting messy)01:26
DocScrutinizer05vector between two circle centers? easy: draw a vector, clicking close to center of first circle, drag til near center of second circle01:26
wpwrak(in fped, the vectors form a tree. and the visible objects (lines, etc.) connect "leafs" of that tree. so the structure can be kept a bit simpler.)01:27
DocScrutinizer05select by-value or by-reference mode for each the start end end point of your vector01:27
wpwrakokay. now change the number if circles in the duplication01:27
wpwrakstill okay ?01:27
DocScrutinizer05well, when you picked circle numer 8 as one of your vector's points, and it's by reference, and you then change duplication multiplier so that circle doesn't exist anymore, then you run into an exception that you need to think about how your program is supposed to handle it01:28
wpwrakdoes it tell you that ? or what does it do ?01:29
DocScrutinizer05your could: a) highlight the error/problem, b) convert the by-ref into by-val, c) make "duplicate" be aware it needs to create a virtual 8th circle which would move your center point of your vector to 360 * 8/701:31
wpwrakno, i mean what das ooffice do ? does it pop up a dialog to offer you these choices ?01:32
DocScrutinizer05d) highligt the problem already on editing "duplicate" parameter01:32
DocScrutinizer05oo?? knows shit about all that01:32
wpwrakerr, about what program are we talking ?01:33
DocScrutinizer05oo can't even link objects, except by "grouping" placing them which simple locks their parameters01:33
DocScrutinizer05to each other sub object of group01:34
DocScrutinizer05err, we're talking about...01:34
wpwrak(fped simply ignores things that don't get instantiated. they "come back" if/when the parameters are such that they get instantiated / "exist")01:34
wpwrakjoergCAD ;-)01:34
DocScrutinizer05<DocScrutinizer05> no, I think I'm not that interested. as long as the thing doesn't know a "pick from drawing" mode for arbitrary parameters, where you click near a significant point and it picks the closest*highest-significance point next to where you clicked and picks the (usually) X,Y from it...01:35
wpwraki thought you said it had references ? or was that also already joergCAD ?01:35
DocScrutinizer05joergCAD, of course01:36
wpwrakman, you are a piece of work ;-)01:36
DocScrutinizer05oo draw sucks01:37
wpwrakwell, rejoice - fped can do pretty much all joergCAD would do ;-)01:37
wpwrakthat's more like what i expected :)01:37
wpwraki was rather surprised by all the impressive features you mentioned. it's not that they would be all that far-fetched, just that nobody seems to get them right (if they even try at all)01:38
DocScrutinizer05I'd try ;-)01:38
DocScrutinizer05if I were to code such thing01:38
wpwraki mean, that's why i use fped as cad. it's not really meant to be a general-purpose cad but it still beats the "competition" so badly that it's almost ridiculous01:39
DocScrutinizer05I'd even try to come up with a way to pick closest point of a circle aka find vector intersection01:39
wpwrakyeah, for classroom-grade geometry you'd basically want a "puzzle solver". freecad has that. it gets very confusing very quickly, though01:42
DocScrutinizer05usecase: I draw a circle (watching dynamic update of bottom status line mouse-X, mouse-Y for center, status line radius for size). then I draw a "infinite" line (click next circle center for origin, watch status line for angle). then I draw a second circle like the first one, just this time I click next to the line way out of circle1 for center of circle2. Then I pull circle2, clicking next to intersection of circle2 with line and 01:45
DocScrutinizer05pull that towards circle1 until it snaps01:45
wpwrakin freecad you draw the elements, then you specify constraints. e.g., beginning of this line is coincident with the center of this circle, this line is parallel to that other one, this line has a length of 100 mm, etc.01:47
DocScrutinizer05then I remove the line (effectively making it just virtual/invisible)01:47
wpwrakthen freecad tries to find an arrangement of things that meets all the constraints. and it tells you how many degrees of freedom you still have left.01:48
DocScrutinizer05funny01:48
wpwrakthe problem is that this very graphical approach gets messy very quickly because you have little constraint symbols all over the place01:48
wpwrakalready a rectangle with rounded corners is messy01:49
DocScrutinizer05yeah, useless01:49
DocScrutinizer05ooh, of course you need cartesic as well as polar coords for your lines/vectors01:53
wpwrakfped has sin and cos for that ;-)01:53
DocScrutinizer05karthesic?01:53
DocScrutinizer05yeah, taht sucks 01:53
wpwrakcartesian01:54
DocScrutinizer05thanks :-)01:54
wpwraksin/cos are nice to have if you need to go deeper into geometry. as i said, these are already things that are rarely needed01:55
DocScrutinizer05when you have to start with tan() and arctan(9 when it comes to length of line, that really sucks01:55
wpwraktan maybe. atan shouldn't be needed. i think i don't even offer it01:56
wpwraknaw, no atan. you have sin, cos, sqrt, and floor (for rounding)01:57
DocScrutinizer05I think of the whole thing like a very special form of spreadsheet01:58
wpwrakdunno01:59
DocScrutinizer05and in this speadsheet each "object" (row?) must have a length for example01:59
wpwrakmy approach is "progressive" - you build the thing step by step, rarely "looking back"01:59
DocScrutinizer05so you easily can mark several adjacent objects and do sum(length)01:59
wpwrakwell, in fped you can   sum = a+b+c   is may be more readable at the end of the day ;-)02:00
DocScrutinizer05wtf is a, b, c?02:01
wpwrakyour three lengths02:01
DocScrutinizer05uhuh, so you got them for the sum() function but not for directly reading them?02:02
wpwrakif get them directly from the parameters that define them, not indirectly from the objects that use the parameters :)02:02
wpwrakby the way, here is a "no frills" variant of your circles problem: http://downloads.qi-hardware.com/people/werner/tmp/fped-cycles4.png02:03
DocScrutinizer05tell me how that look for a closed polygon you grouped from 53 vectors. Can you write the furmula in one max line length of IRC?02:04
wpwrakwhile there are some equations, they're pretty intuitive02:04
wpwraknow you're asking for a diagonal ?02:04
DocScrutinizer05yay, typo galore02:04
DocScrutinizer05well, talking about polar coords I of course allow all possible angles02:06
wpwrakand indeed, fped doens't have a sum(field:field) function. it doesn't FFT, complex math, non-Euclidean geometry, prime number search, etc., either ;-)02:06
DocScrutinizer05you'd probably write a true program with for-loops and all, to get the circumference of a polygon02:07
wpwrakit's not even turing-complete :)02:07
DocScrutinizer05too bad02:07
wpwraki think what you want is autocad. that's loaded with all sorts of crap as well ;-)02:08
wpwrakit may take you a day to find the function that does what you want, but it's there, somewhere, amidst a gazillion of other once-in-a-lifetime functions :)02:09
DocScrutinizer05well, no. I for now just would insist on the program showing me the length and angle of arbitrary lines/vectors02:09
wpwrakif i really needed that sort of stuff, i'd add some include mechanism (besides cpp #include) and generate the weird stuff with a script ;)02:10
wpwrakoh, fped does that of course02:10
wpwraksee the example02:10
DocScrutinizer05didn't you say your fped comes with outo-metering?02:10
DocScrutinizer05I wonder how you can get the length of vectors when it only works for vertical and horizontal ones02:11
DocScrutinizer05is that really useful?02:11
DocScrutinizer05how do you rotate vectors?02:12
DocScrutinizer05do you pull out your HP calculator?02:12
DocScrutinizer05I mean, allowing both polar and cartesian coords for vectors can't be that hard?02:14
DocScrutinizer05and maybe even length(vector)02:15
wpwrakactually, it doens't show you the angle. auto-metering .. yes, here: http://downloads.qi-hardware.com/people/werner/tmp/fped-cycles5.png02:15
DocScrutinizer05maybe even - DUH - *show* length and angle while drawing a vector (if fped can do such stuff like "drawing2 a vector interactively, with the mouse)02:16
wpwrakyes, you can but you shouldn't02:17
wpwrakfped is not for artistic drawing. things are very precisely defined dimensions. so you just enter these numbers. no silly trying to hit the right spot with the mouse.02:18
DocScrutinizer05comsidered rather writing postscript by hand? ;-)02:18
wpwrak(rotate vectors) here we meet our old friends sin and cos again :)02:19
wpwraknaw, writing all the postscript generation in fped was bad enough :)02:19
DocScrutinizer05can't have been that bad, to me it looks like it's not much more that a simple syntax translation02:20
DocScrutinizer05than*02:20
Action: DocScrutinizer05 really woonders how wpwrak woud solve the following task: create a FPGA footprint for 800 balls, with the usual left out pads every now and then. in an angle of 45°, like f02:24
wpwrakfirst page of http://people.openmoko.org/werner/gta02-core/gta02-core-modules.pdf02:27
DocScrutinizer05hmm?02:27
wpwrakif you want to rotate them, feel free to do that, too :)02:27
DocScrutinizer05that doesn't look like rotated-right-45°02:28
wpwrakthat file is a catalog of fped-drawn components02:28
wpwrakwell, footprints02:28
DocScrutinizer05hmmpf, i closed the oodraw 02:29
wpwrakif you want more, here's the current catalog: http://downloads.qi-hardware.com/people/werner/tmp/kicad-libs-modules.pdf02:30
wpwrakbut the old one had more intersting BGAs. we never used them in real life, though. if we ever do, i'll move them over02:30
DocScrutinizer05http://wstaw.org/m/2014/02/15/plasma-desktoprP3389.png02:34
DocScrutinizer052 of those 4:30 were eaten by starting up oodraw and by making screenshot and copy it to wstaw02:35
wpwrakfped doesn't currently have a mechanism to "knock out" things. but you could just make a list of strips/blocks of pads02:37
DocScrutinizer05the pitch is 0.50cm02:37
wpwrakHUGE02:37
DocScrutinizer05well, in 150 seconds I create one with smaller pitch. Unless you accept that pads and gaps shrink by same factor, then it takes 15s02:39
wpwrakwhat does the pitch matter when drawing ?02:39
DocScrutinizer05huh?02:39
DocScrutinizer05that's meant to be a FPGA footprint02:40
DocScrutinizer05FPGA *have* a certain pitch02:40
wpwrakyes, but why would the pitch affect how long it takes you to draw it ?02:41
DocScrutinizer05not at all02:41
wpwrakexactly :)02:41
DocScrutinizer05but shrinking the whole thing is one single action, while creating the footprint were like 10 actions02:42
DocScrutinizer053 for the 15*15 grid of pads02:42
DocScrutinizer05N for removing some pads02:43
DocScrutinizer051 for rotate-right-45°02:43
wpwraknow, let's assume you counted wrong and removed a pad that was meant to stay ... ;-)02:44
wpwrakwell, i you still have it in "undo", you'll be fine. else ;-)02:44
wpwrak#s/ i / if /02:44
DocScrutinizer05else I do what you'd have done for whole thing02:45
DocScrutinizer05I copy the pad left-upper of the missing one, down right by sqrt(0.50cm)02:45
wpwrakooh. i thought you'd precision-drop it with the mouse ;-)02:47
DocScrutinizer05tzz02:47
wpwrakbut you've almost convinced me to add an "else" item to tables :)02:48
DocScrutinizer05I could do that as well, copying pos-X parameter from pad above, and pso-Y from left or right02:48
DocScrutinizer05but calculating cartesian coords for "duplicate2 is probably faster than copying absolute values for X,Y from two other objects' parameter dialogs02:50
wpwrakyeah, it does seem to get messy-ish ;-)02:51
DocScrutinizer05almost near as messy-ish as fped when designing that footprint02:51
DocScrutinizer05not even near02:52
DocScrutinizer05oops, i forgot one step in the virtual 10 steps I claimed it took to design that: group the whole thing, for rotating it02:53
DocScrutinizer05to elaborate: 1) I created a square of arbitrary size for my master pad (I could have used parameter dialog to make sure it has e.g. 0.38mm side length) 2) I picked "duplicate" for the already selected pad, entered the qty "15" and a X skew of 0.5cm (random guess, I could have written 0.5mm as well) 3) I selected the 15 pads and "duplicated" them 15 times with Y skew 0.5cm. 4) I clicked and "del"eted some of the pads, to make it 03:06
DocScrutinizer05look more real-life-y 5) I selected all pads and "group"ed them 6) I highlighted the group and picked "rotate", selected center for rotation center and the 45° preset radiobutton. 7) done, snapshot and pastebin03:06
DocScrutinizer05actually the 2.5 minutes this took me were quite slow, since I never before worked with oo draw and thus had to find my way around03:10
DocScrutinizer05ooh, you know OpenOffice (aka libreoffice) can run macros written in basic, python, whatnot else?03:12
DocScrutinizer05hmm, could you use such stuff? http://maemo.cloud-7.de/hidden/ooFootPrint.svg03:20
DocScrutinizer05ohwow, the pads even have rounded corners. wonder how that got introduced there03:23
DocScrutinizer05http://wstaw.org/m/2014/02/15/plasma-desktopIT3389.png03:24
wpwraki think i figured out how to do a nice "else" :)05:22
wpwrakbeen wanting that for a while already05:22
DocScrutinizer05it just occurred to me that oo draw might know to import objects into drawings by reference05:28
DocScrutinizer05which in the end whould be a nice way to do "macros"05:29
DocScrutinizer05meh, of course not05:44
mthI think rounded corners are simply a consequence of having a round-tip pen for the drawing09:08
mthwhich might even be the default09:08
nicksydneywpwrak: http://www.indiegogo.com/projects/metrix-create-space-bluetooth-low-energy-hardware-workshop09:35
nicksydneywpwrak: another BTLE project http://www.indiegogo.com/projects/rise-the-wearable-sit-tracker-that-motivates-you-to-sit-less09:49
CYB3Rxiangfu: Hi! I tried using usbboot and xbboot from xburst tools, but it keeps telling "no XBurst device found"11:38
CYB3RAnd I've got a "JZ4750 USB Boot Device" (601a:4750)11:40
CYB3RWhat may be wrong here?11:40
whitequarkthe mystery of package from iceland: http://mailpile.is sent me my crowdfunding reward12:11
larscwell, that deescalated quickly13:14
whitequarkslowly* :p13:45
whitequarkit was not nice of them to specify my nickname on the package. I had to convince post office I was whitequark.13:46
whitequarkquote: "we deliver to individual, not address"13:46
wpwraknom de plume :)13:47
whitequarktold them "it is a part of my email, idk why sender wrote it like that"13:48
whitequarkwas not in mood to explain what exactly "nickname" means, to a 50-something lady in the post office.13:48
wpwrakyou could have told them you're an artist and everybody knows you by that name, so it's understandable that they'd make that mistake. e.g., if you sent a letter to "madonna" with the correct address, i kinda doubt she'd have difficulties receiving it (well, her "explosives, antrax, and other unpleasant-surprises-from-weird-fans filtering team would")13:51
whitequarkactually you're right, the word "pseudonym" predates internet age for a long time13:58
whitequarkguys, I think I've nailed 4 mil traces: http://i.imgur.com/SX9XY0x.jpg14:29
wpwrakdidn't know they make nails that small14:33
wpwrakand it only counts if they survive etching. which i think they're not very likely to do :)14:34
wpwrakor is that milled ?14:35
whitequarkno, that's photoresist on copper14:35
whitequarkI don't think endmills that thin even exist14:36
wpwrakokay. so i'd say the 0.10 mm will not survive etching14:36
wpwrakthe 0.15 mm might14:36
whitequarkwe'll learn at monday14:36
wpwrakmake sure you have a thin Cu layer so that you don't need to etch very long14:36
whitequarkCu is .035mm on that board14:37
whitequarkoh, tuesday, actually14:38
whitequarkI wonder why the first try ever I made with this piece of PCB had the best photoresist adhesion14:40
whitequarkit was likely the worst cleaned one. hmm.14:40
wpwrakit's like making semiconductors they work best if you add a little dirt ;-)14:41
whitequarkhehehe14:41
whitequarkthough, the later tries don't pass the water break test14:43
wpwrakwater break ?14:43
whitequarkreally clean copper is hydrophilic. contaminants are usually hydrophobic (or otherwise you could wash them off trivially).14:44
whitequarkso, you pour some water on copper, and if it distributes itself in a really thin continous layer, your copper is clean14:44
wpwrakphew. invited corrosion. interesting concept :)14:46
whitequarkwell, the whole process is wet anyway...14:48
lekernelwtf, is this a hoax? http://www.youtube.com/watch?v=zrv78nG9R0416:29
larscyummy16:31
whitequarkbleargh16:48
Action: whitequark stopped watching it immediately after opening shot16:49
whitequarkwpwrak: http://i.imgur.com/gsUNFrD.jpg17:33
whitequark2 mil one almost works, everything else is pretty much perfect!17:34
whitequarkit's a bit underdeveloped; I expected 2% developer to work slower than 5% one but it actually was faster due to some bizarre mechanism17:34
wpwraklekernel: good old propaganda ... :)20:12
whitequarkwpwrak: I found my perfect process, with which I can do 2mil traces20:39
whitequarkhttp://lab.whitequark.org/notes/2014-02-16/optimizing-for-best-photoresist-adhesion/#try-920:39
whitequarkhttp://lab.whitequark.org/images/photoresist/2014-02-13/try-9-developed.jpeg20:47
wpwraknice. let's see how much of that will be left after etching ;-)20:47
whitequarkoh, sure20:48
whitequarkI don't actually want to etch 50µm traces, it's more a matter of having enough room for error20:50
whitequarkand here I have plenty20:50
DocScrutinizer05whitequark: wow, that's damn impressive21:53
whitequark\o/21:53
DocScrutinizer05whitequark: did you improve your exposure lighting?23:03
whitequarkDocScrutinizer05: nope, same old UV flashlight23:04
whitequarkthe defects on previous pictures can be attributed to flawed process23:05
whitequarkthe defects on this one are entirely present on photonegative, it maps perfectly to what's on copper23:05
whitequarkthat junk on the right, and .3mm line being wider than .35mm one -- all on the negative23:06
whitequarkshitty printer.23:06
whitequarkand I'm now really at a loss of what printer to use. :/ :/23:07
DocScrutinizer05laserplotter?23:21
DocScrutinizer05maybe only a question of focusing?23:21
DocScrutinizer05no more printing but direct curing of photopositive on PCB with laser23:22
whitequarkDocScrutinizer05: that's a rather radical change23:23
DocScrutinizer05only for the flawed half of process23:23
whitequarkI mean if only this printer printed exactly what it is told, that'd be *great*23:23
whitequarkwhy is it that nothing with processors ever works as it should23:23
whitequarkoh, hrm. idea: instruct inkscape to print it as a bitmap23:23
DocScrutinizer05you'll need a new one, incl new high quality toner and photodrum23:23
whitequark(toner and photodrum) toner is fresh, I bought a new cartridge23:24
whitequarkphotodrum is not, I've cleaned it in the past; it was covered with spots of hardened toner23:24
DocScrutinizer05then remove it and give the rest a *good* cleaning23:24
whitequarklooked quite awful23:24
whitequarkI've cleaned it already, probably not good enough23:25
DocScrutinizer05a new drum will be essential23:25
whitequarkyou think so? it appears that drum surface was not damaged at all23:25
DocScrutinizer05drums degrade a lot23:25
whitequarkat least not visibly23:25
DocScrutinizer05new printer cartridge types come with drum integrated23:26
DocScrutinizer05the photosensitive coating is not exactly durable23:26
DocScrutinizer05and of course the rest of the whole paper feeding mech also needs to be absolutely clean23:28
DocScrutinizer05just like any carbon brushes etc for electrostatic charge23:28
whitequarkthis one has cartridge and its "frame" with drum separate23:28
whitequarkwhat do you clean drum with? water? isopropanol?23:29
whitequarkvendor website says "dry cloth" but it's impossible to get it clean using just dry cloth, not when it was as dirty as it was23:30
whitequarkback then I used water and q-tips23:30
DocScrutinizer05drum? never touch it23:30
whitequarkeh?23:30
DocScrutinizer05is all I ever heard, frequently23:31
whitequarkwell, the vendor says I can clean it if the print is shitty23:31
whitequarkso I guess it's not that awful23:31
DocScrutinizer05wow, weird stuff23:31
whitequarkok. now it's not really very dirty. so I think dry cloth would work fine.23:31
DocScrutinizer05the photosensitive coating is some special gue or nobody-knows-what23:32
whitequarkselenium and... someshit23:32
whitequarkbut I know for sure there's a lot of selenium23:32
DocScrutinizer05and any ditches or fingerprints or whatever usually are irrecoverable23:32
whitequarkditches sure. but it's not damaged23:33
DocScrutinizer05at least that's what I heard, no experience here23:33
DocScrutinizer05yeah, Selen sounds about correct for old ones. New ones are probably some hightech stuff pased on polymers or whatever23:34
DocScrutinizer05or even organig stuff23:34
DocScrutinizer05organic23:34
DocScrutinizer05maybe wikipedia knows more23:34
whitequarkok, some stuff found on web says it's ok to clean drum with isopropanol23:35
whitequarkworst thing happens, I need a new drum--guess what with its current state I need it anyway23:35
whitequarkso gonna give it a try23:35
DocScrutinizer05hehe, sounds sensible reasoning23:35
whitequarkshould remove any fingerprints / toner23:35
whitequarkoh crap23:38
whitequarktried to use DUSTER on insides of printer23:38
whitequarkit spewed out a bunch of liquid and there was basically ice everywhere23:39
whitequarkalso it basically emitted such a powerful stream it detached a spring from some part of it23:40
whitequarkwtf23:40
whitequarkshould be called DISASSEMBLER23:40
whitequarkDocScrutinizer05: anyway. it still works and test page is perfect. so... I win?23:44
DocScrutinizer05well, yes. seems so23:46
Action: whitequark idly plays with that spring23:46
DocScrutinizer05for sure you won some experience ;-P duster kills stuff23:46
whitequarkon a second thought I think the spring detached while transportation or by previous owners23:46
whitequarkand duster just picked it up23:46
whitequarkgood thing: it may have jammed it or something.23:47
DocScrutinizer05you're aware that tones is microdust that's extremely hazardous?23:47
DocScrutinizer05toner*23:47
whitequarksure23:47
DocScrutinizer05you're also aware that printers are filled with it to the label23:48
whitequarkthere aren't any clouds of toner around, I just wiped whatever adhered to the drum so badly I wouldn't get it off without scrubbing and alcohol23:48
whitequarkoh. you mean that23:48
whitequarkyes it was not very smart23:48
whitequarkhowever what's even worse is that the air outside my apartments is filled with same microdust from longdistance haul trucks going by all night23:49
whitequarkso that is probably what's going to kill me rather than toner23:49
DocScrutinizer05right23:49
wpwraksuicide by toner - to avoid getting killed by chlorine ;-)23:49
whitequarkhehe23:49
DocScrutinizer05maybe even one catches the other and they neutralize 23:49
whitequark"fight fire with fire", eh?23:49
whitequarkalthough toner on fire sounds extremely bad23:50
whitequarkno23:50
whitequarkit sounds awesome23:50
whitequarka volumetric explosion of plastic nanoparticles23:50
DocScrutinizer05anyway, next time you'll use a vacuum cleaner with hepa-filter23:50
whitequark(hepa filter) right. thanks for the idea23:50
whitequarksee, dust is really a problem for me in this apartment. as you may have noticed all my photos of pcbs are extremely dusty23:51
whitequarkthis is as clean as it ever gets in here. I can clean everyday and it will make no difference23:51
DocScrutinizer05try find out what's that spring for. It might be total nonsense, or essential for proper functioning23:51
whitequarkI bet most of it is pulled out from outside23:51
whitequark*nod*23:51
whitequark(dust) I have A/C which claims to have dust filtering and its filters are regularly chock full with dirt23:52
whitequarkbut that doesn't really seem to reduce its amount in air either23:52
DocScrutinizer05hmm, that makes me think... move your lab to bathroom maybe? mist is an excellent dust remover23:53
DocScrutinizer05fog, whatever23:53
whitequarkeh23:53
whitequarkyou germans, how big exactly bathrooms do you have?23:54
DocScrutinizer05or get 10 pots with boiling water in your lab23:54
DocScrutinizer05mine is tiny here23:54
whitequarkmine has like 4 sq. m. of space23:54
DocScrutinizer05mines has less, probably23:54
whitequarkwell, and less than 1 of that is even free, after plumbing, washing machine and others23:55
DocScrutinizer05but there's also a shower that has almost 2 with nothing obstructing it23:55
whitequarkbut you have pointed me to an idea, get an air humidifier23:55
whitequarkfor my living room / lab23:55
DocScrutinizer05humidifier is of limited usefulness23:56
whitequarkwell, doesn't it basically make mist? with an ultrasound emitter23:56
DocScrutinizer05only real fog takes out 99.9999% of dust particles from air23:56
whitequarkyou can really make enormous amounts of mist with that thing if you pump a few watts in it23:56
DocScrutinizer05hmm, you don't really want those 90% humidity plus fog in all your lab23:57
whitequarkand you told me to move it to bathroom!23:57
DocScrutinizer05but you can run your shower for 15min in your bathroom, and them move the printer and laminator to there23:58
whitequarkoperating mains electronics in a bathroom right after running shower there. not sure if serious or trying to kill me :p23:58
DocScrutinizer05it's a proven procedure to apply screen protectors etc23:58
whitequarkoh?23:59
whitequarkinteresting23:59
DocScrutinizer05well, even while there's probably a recommendation to not run the printer in humid air or even fog, I think it#s not really dangerous23:59
--- Sun Feb 16 201400:00

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