#qi-hardware IRC log for Thursday, 2014-03-20

--- Thu Mar 20 201400:00
whitequarkwpwrak: no, it absolutely does get in the way05:07
whitequarkfirst, the raised edge cuts resist, means the soft underbelly of metal-plated holes is not protected from etchant05:08
whitequarksecond, when it's high enough, it even warps resist, lowering quality substantially05:08
whitequarkthird, there's some weird-ass interaction with plating bath, which results in huge copper crystals growing on all discontinuites05:08
whitequarkI *think* it's somehow related to current density05:09
whitequarkI mean, like corona discharge. don't know if I'm right, though05:09
whitequarkhmmm, I think endmills worked much better for drilling holes than drills, might want to try that05:15
nicksydneywpwrak: http://hackaday.com/2014/03/19/meet-lynx-a-costly-offline-password-keeper/06:49
wpwrakhmm. the hardware doesn't sound all that exciting. sw .. hmm, xor with (fixed ?) key stored in the MCU ...08:13
larscthey just wanted to be the first to monetize the idea08:23
whitequarkwow. just... wow. I've read recently that persulphates are good for preparing a microporous surface08:28
whitequarkso I decided to check whether photoresist has better adhesion on such a surface08:29
whitequarkI put the board in persulphate for literally five seconds, the results are amazing. it's like using an entirely different process.08:29
wpwrakand, how's the new apartment ?08:29
whitequarkif before, I've had to be extra careful to not tear off the fragile layer, now it's entirely opposite08:29
whitequarkit adheres so fucking well I'm actually having problems stripping the resist layer after I don't need it08:30
whitequarkawesome08:30
wpwrakah, the non-explosive kind of surprise. how boring.08:30
pcercueiwpwrak: it's pretty good, thanks08:30
whitequarkoh and the look and feel of the PCB+resist is entirely different too.08:30
wpwrakpcercuei: hehe, that was actually for whitequark. but i'm glad that you're happy with yours as well :-) how's .de ? first signs of spring yet ?08:31
whitequarkit really does remind me of photos from industrial processes. the photoshopped ones they show you in marketing leaflets :p08:32
whitequarkvery smooth08:32
pcercueiwpwrak: actually yes, we're getting like 18°C today and 20°C tomorroz08:33
pcercueitomorrow*08:33
wpwrakoh, nice. time for having beers outside :)08:33
pcercueitoo bad that when I come out of the office the sun is almost gone >.>08:34
whitequarkwpwrak: http://imgur.com/ZVFNu8B,J07VwDw08:34
wpwraklunch break ?08:35
whitequarksome parts are actually fubar. that's simply because I was using old values for my process08:35
whitequarkand goddamn printer is printing bullshit AGAIN08:36
whitequarkI'm starting to really hate this devil's spawn08:36
wpwraklooks nice (apart from the offsets)08:36
whitequark(nice) well, not quite. look closely at edges.08:36
wpwrakGDI crap ?08:36
whitequarkthey're not very clean. but this is easily fixable. and I never, ever, had such amazingly good adhesion08:37
whitequark(GDI) hm, no08:37
whitequarkit's not postscript but I think not GDI too08:37
wpwrakedges tend to be difficult under any circumstances08:37
whitequarkwell, "difficult" is the second name for my job here :p08:39
ysionneauwpwrak: I guess you are aware of http://hackaday.com/2014/03/19/meet-lynx-a-costly-offline-password-keeper/09:06
ysionneauah ok already pasted09:06
ysionneausorry09:06
larscpcercuei: chat less, work more, then you can leave earlier ;)09:12
larscmaybe09:12
pcercueiD:09:21
whitequarkwpwrak: https://pbs.twimg.com/media/BjKjICeCUAEZAFm.jpg:large10:55
whitequarkmy notebook hinge broke and I had to make a thing10:55
wpwrakcnc-milled, i hope ?12:16
whitequarkof course12:17
whitequarkhttps://gist.github.com/whitequark/966123912:17
whitequarkI finally figured out how cutter compensation works, yay12:18
whitequarkwidth*12:18
whitequarkwpwrak: I'm absolutely unable to cut anything harder than paper in straight lines.12:20
whitequarkthere's probably some trick to that. I'm not aware of it.12:20
wpwrakmaybe you12:22
wpwrakre going too fast ?12:22
wpwrakor you're using drills as endmills12:23
whitequarknonononono12:23
whitequarkby hand.12:23
whitequarkCNC works just fine12:23
wpwrakoh :)12:23
whitequarkwell, sometimes it does not, but that's shitty fixture mainly12:23
wpwrakwell, then don't worry. and welcome to augmented reality ;-)12:23
whitequarkaugmented reality?12:23
wpwrakusing the term loosely ;-)12:24
whitequarkyou mean cutting stuff up with CNC? well, you see12:24
whitequarkI have to pre-cut material to smaller pieces for my CNC to chew on12:24
whitequarkand that is a *nightmare*12:24
Action: whitequark casts a glare full of vicious hate on a hand saw12:24
wpwrakuse a jig saw ?12:25
whitequarkI'm a bit afraid for my fingers12:25
wpwrakjust strictly apply pauli's exclusion principle :)12:27
whitequarkhehe12:27
whitequarkstrictly saying, it happens *because of* pauli's exclusion principle12:27
wpwrakbesides, as long as you're cautious, you're rather unlikely to get hurt. these machines aren't particularly unsafe per se and most accidents happen when people get careless12:28
whitequarkwell, let's say I'm not particularly strong. unless you mean fingers. :). I'm kind of afraid that the machine will want to go somewhere and I'd be unable to stop it.12:29
whitequarkit packs quite a bunch of power for that matter.12:29
wpwrakif it goes out of control, turn it off :)12:31
whitequarkmhm12:31
whitequarkright :)12:31
wpwrakif it jumps and dances around, then you probably have an unsuitable blade12:31
wpwrakor you didn't fix the piece properly.12:32
whitequarkunrelated: by a chance, I stumbled upon another reagent that can be used for surface activation for copper plating12:34
whitequarkit's copper(II) tartrate12:34
whitequark*that* I'm sure is not present on any precursor list, and pyrolysis t° seems to be under 200°C (140°C for sodium tartrate, but copper will likely be different)12:35
whitequarkalso, should be easily obtainable12:35
whitequarkoooh neat, this time I went much much slower with the plating process itself (limited current to 10mA), and also added Cl¯ ions to bath and made it slightly more acidic12:36
whitequarkseems a nice smooth layer of copper is growing inside holes12:37
wpwraki think the drug designers need to work a little harder and always point out when they involve water in their process. that should get that pesky liquid banned, too :)12:37
whitequarknext try I'll need to dry the activator more carefully. it boiled on PCB, and as a result I have a very flakey layer of copper deposited12:37
whitequarkso what I really need is to first dry it under 100°C and *then* decompose at 150°12:37
Action: whitequark ponders refitting the oven he has with a thermo-profile-able PID controller12:38
wpwrakthat often seems to be necessary12:41
whitequarkhmm. I just realized what the black residue after pyrolysis is12:42
whitequarkthat's CuO12:42
whitequarkseems there's a race between decomposition to CuO and to Cu12:42
wpwrakhmm, mateallic, heck. yellowish, check. alas, still not gold. back to the alchemy books ...12:43
whitequarkhehe12:43
DocScrutinizer05inert gas ?13:26
DocScrutinizer05or is the oxygen from one of the chemicals involved?13:27
DocScrutinizer05seems quite natural to me when elementary copper that just pops into existence from a decomposition, at a high temperature, thends to grab some O2 when some is around13:28
DocScrutinizer05tends*13:28
whitequarkDocScrutinizer05: well yeah13:35
whitequarkinert gas is too hard13:35
whitequarkprobably. hm. maybe not, if it's N213:35
whitequarkstill a lot of hassle13:35
whitequarkanyway take a look? http://i.imgur.com/FxDXLnR.jpg13:35
whitequarkok, I confirmed: there's both electrical conductivity and capillary effect for solver13:49
whitequark*solder13:49
whitequarkso... I did basic plated throughhole board? this was really much easier than I thought13:49
wpwrakhmm, something seems to move around a lot. maybe it's not the board but the head ?14:55
wpwrake.g., look at the "exit holes" where the mill was inserted or retracted14:55
wpwrakthat would also explain all the burr14:55
whitequarkhm, yes, exit holes15:01
whitequarkit's a bit weird15:01
DocScrutinizer05whitequark: nice!20:02
DocScrutinizer05but those ends on the long holes are really weird. Like if Z axis movement causes whole spindle motor to tilt20:05
DocScrutinizer05maybe some of your bearings have some play/gap?20:13
DocScrutinizer05((<whitequark> inert gas is too hard)) heard you own some magic soup called Galden?20:32
DocScrutinizer05or is that Cu-oxyde forming in a liquid phase of the process?20:33
DocScrutinizer05((http://i.imgur.com/FxDXLnR.jpg)) I still don't grok it. I just thought it might be tangential force created by spinning mill vs PCB which in turn bends the mill itself. But that would mean that some of those and serifs should be other way round than they actually are22:23
DocScrutinizer05whitequark: how about starting and ending toolpath in the middle of a linear cut instead of at the end?22:25
DocScrutinizer05you should mill 8 squares, with their particular toolpath starting at 12o'clock 3, 6, 9, and in the 4 corners at 1:30 4:30 7:30 10:3022:27
DocScrutinizer05wait, how did you fix your PCB? thick doublesided (white) sticky tape? Then this might have some elastic/plastic properties, causing the PCB to slowly move away from the mill due to pressure the mill applies to PCB laterally22:31
DocScrutinizer05so when you start toolpath, you mill into "upper" direction (in http://i.imgur.com/FxDXLnR.jpg, above the 5 holes) further than when you end at that point after cutting the upper line which moved PCB upward slowly in the elastic sticky22:33
DocScrutinizer05this would mean your toolpath started exactly left upper right of http://wstaw.org/m/2014/03/20/plasma-desktopOx3389.png and went counter-clockwise until it came to same point from right and ended there22:36
DocScrutinizer05...while the other long hole left of this would have clockwise toolpath22:37
DocScrutinizer05you should check if width of the trench is exactly what you specified and expected, or if only the serif has the correct height resp "width" in reference to the "lower" wall of trench22:39
--- Fri Mar 21 201400:00

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