#homecmos IRC log for Wednesday, 2011-08-10

berndji thought pretty much any electrical copper is pretty darn pure, because its conductivity is so super sensitive to impurities00:10
azonenbergYeah, same here00:12
azonenbergI did my last evaporation run with wire i cut out of a random AC power cord lol00:12
azonenbergthat i found on the loading dock in a dumpster :P00:12
berndjalthough i guess sometimes you get beryllium00:14
berndjwow, they have copper in dumptrucks where you live??00:14
azonenbergno lol00:14
azonenbergthis is the loading dock behind the matsci building00:14
azonenbergthey have a giant e-waste bin there00:14
azonenbergfull of scrap for the taking if you get there before the collection day00:14
berndjah, nice.00:14
azonenbergQuote of the day - apparently a character in some TV show once said "it's impossible to do microelectronics with gloves on"01:47
azonenbergAnd of course anyone who's actually spent time with them knows it's impossible to NOT do it with gloves on :P01:48
azonenbergAt least not if you expect the device to work by the time you're done with it01:48
CIA-67homecmos r110 | trunk/lithography-tests/ (34 files in 18 dirs) | Pushing latest changes to 600dpi test pattern database06:26
Last message repeated 1 time(s).06:35
smedingazonenberg: haha, have those writers ever seen a clean room12:40
azonenbergsmeding: no idea lol12:40
azonenbergThey were admittedly talking about PCB assembly rather than IC fab, i think12:41
azonenbergBut still - i often wear gloves for that12:41
smedingnothing micro about that12:41
azonenbergfingerprints on copper can mess p solder joints12:41
azonenbergAnd agreed12:41
smedingi don't, but i tend to build fairly simple low-speed digital circuits on protoboard :p12:42
azonenbergYeah, those are less criticial12:42
smedingi was going to try my hand at etching PCBs this week but the people at revspace took all the etching equipment to CCC :(12:42
azonenbergcurrently working on final design review for my latest project - 32 channel FPGA based PWM controller12:42
azonenbergAs you can probably guess this board is not being homebrewed12:42
smedingyeah... anything with vias gets annoying12:43
azonenbergI tried them - they're OK as long as they're out in the open12:43
azonenbergBut most of my boards use the majority of vias under ICs :P12:43
azonenbergAnd when you have thirty vias under a single TQFP, well...12:43
smedingyeah... i'm running into issues designing a board with an 8DIP and a 20DIP :p12:44
azonenbergIf nothing else, the via plating process i've used (or tried to, anyhow) is too tall to fit under SMT ics12:44
azonenbergthey have bumps on top12:44
smedingwhat plating process did you use? did it require any tooling?12:44
azonenbergTried several simple ones12:44
smedingi wish there was money for a proper PCB manufacturing line at revspace, but alas12:45
azonenbergconductive paint (didnt work), wire + solder (worked but left a bump and some of them didnt work)12:45
azonenbergEnded up sending the board out for professional fab12:45
azonenbergI still do single-layer ones in house for simple breakouts etc12:45
azonenbergbut i've come to the conclusion it isnt worth it for complex designs12:45
smedingwire+solder is what we've been doing... but plated through-holes won't work with that, and it's something i would like to have12:45
azonenberguntil i get via plating working12:45
azonenbergI want to try it some time, i have ideas12:46
azonenbergSome crazy ones ;)12:46
azonenbergfor example - start out with a blank copper board12:46
azonenbergDrill all of the holes12:46
smedingbut yeah, i want to etch like four small (maybe 10cm²) boards, it's not worth ordering a whole panel12:46
azonenbergSputter coat a thin layer of something conductive (probably wouldnt work though, i dont think fr4 is vacuum safe)12:46
azonenbergthen electroplate up12:47
azonenbergand etch the wiring layer12:47
smedingyeah... i don't think it is12:47
azonenbergWhich means i'd need to use a glass or ceramic substrate12:47
azonenbergand then the costs start rising etc12:47
smedingi think the commercial process uses some annoying chemicals to deposit copper, and then electroplates the rest on?12:47
azonenbergThey try to do additive as much as possible from what i understand12:47
azonenbergbecause every gram of copper you etch is a gram of copper wasted12:48
azonenbergso they start really thin, pattern that12:48
azonenbergand then plate up12:48
smedingsounds pretty sane12:48
azonenbergRe vias... http://i.imgur.com/Quh56.png is the simplest of the two boards i'm currently working on12:48
azonenbergThe other one is the same size but has like six more ICs, four layers instead of two, and has some really tiny components like 0402s12:49
smedingwhat do you use to design those?12:50
azonenbergExpresspcb, but i'm moving away from it12:50
azonenbergEven ignoring the fact that its not using an open file format, there are more serious issues12:50
azonenbergLike the single-level undo12:50
smedingi kind of want to use gEDA12:50
azonenberglack of an autorouter12:50
azonenbergYes, thats what i'm looking at too12:50
smedingbut it's prety bad12:50
azonenbergOnly reason i've stuck with express is that the fab is dirt cheap12:51
smedingi'm thinking of rallying the people to fix it12:51
azonenbergand beats every other one i've found12:51
smedingdon't they take gerber or something?12:51
azonenbergin terms of being cost effective for a one-off design12:51
azonenbergOnly thier format12:51
azonenbergif they took gerber there would be no reason for people to use their proprietary cad tool :P12:51
azonenbergThough i'm sure it gets turned into gerber internally12:51
smedingwell, yes, but they could probably get more sales12:51
azonenbergI only use them for one-off designs that i know i will not need to mass produce12:52
azonenbergLike a prototype that i already plan to completely redo12:52
azonenbergso i'd do rev1 in expresspcb and then based on what i learn from that do rev 2 at a different fab with more layers etc12:52
smedingso what are those boards for, anyway?12:53
azonenbergThis one is a 32-channel RC servo controller for a research group on campus12:53
azonenbergdesigning a prototype robotic arm and couldnt find a MCU with enough outputs12:53
azonenbergSo they had me do an FPGA based design12:53
smedingah, yeah12:54
azonenbergEven the 50k gate i'm using is probably overkill12:54
azonenbergi fit six channels into a large CPLD12:54
smedingyeah.. i wrote a PWM controller for a uni project12:54
azonenbergbut there's nothing in between the 3k gate CPLD it didnt fit in and the 50k fpga on this board12:54
smedingthey're not complicated12:54
azonenbergi really need like a 15 or 2012:54
azonenbergAnd yeah, this is my first actual FPGA based design12:54
azonenbergwanted to start simple12:54
smedingwell, for our project it was part of a little robot thing12:55
azonenbergWas yours in software or hardware though?12:55
smedingVHDL on an FPGA12:55
smedingso hardware... at least sort of :p12:55
azonenbergbitbanging in an arduino is a bit different than custom softcores12:55
azonenbergand yeah, that counts as hardware in my book12:55
smedingwe didn't use a softcore12:55
smedingit was a big, ugly FSM12:55
azonenbergI'd still consider that a softcore12:56
azonenbergas opposed to a hard IP block in an ASIC12:56
smedingwell, several - they had us implement the PWM as an FSM for some daft reason12:56
azonenbergDoesnt make too much sense12:56
smedinginstead of like a single comparator12:56
smedingwell, counter and comparator12:56
azonenbergThats what i did in my design12:57
azonenbergcant see any other sane way to do it for a 10-bit duty cycle12:57
azonenbergnot gonna do a 1024 step FSM12:57
smedingwell, the design was weird12:57
smedingbut they wanted it to do it that way...12:57
azonenbergAnyway so my other board is a slightly larger fpga (200k gate) and an embedded MIPS microcontroller12:58
smedingthankfully they left us alone for most of the project12:58
azonenbergintended for general purpose embedded development12:58
smedingoh, yeah, the pic32 thing?12:58
azonenbergthe fpga has 3bpp (8 channel) vga out12:58
smedingwith the ringosc :p12:58
azonenbergLol yep12:58
azonenbergCoudlnt resist sticking some art on that one too12:58
smedingart is good12:58
azonenbergyou can see the smiley face on this one12:58
smedingyeah i noticed12:58
Action: smeding goes back to coding a little driver program for a hp7550a plotter from '8413:00
azonenbergI stick that one on all of my boards unless i decide to do something more elaboate instead13:00
Action: azonenberg goes to take shower before work :P13:00
azonenbergsadly life isn't all fun and games lol13:00
smedingplan on converting the plotter into a multifunctional thing for vinyl cutting / something with PCB development13:01
azonenbergAnd the professor i work for has stuff she needs done before i finish up my summer research assistantship13:01
azonenbergand start working on my PhD stuff13:01
smedingplan on looking at drawing etch resist directly onto a PCB - if that doesn't work i'll try a photoplotter13:01
azonenbergI want to make a photoplotter13:01
azonenbergBut my design was going to be small13:01
azonenberglike, using steerable mirrors and a bluray laser with a 1cm^2 field or something13:02
azonenbergthe goal is to improve my in-house mask shop :)13:02
smedingwell, the 7550a is a very nice plotter - though it might be complicated to make it accept media sizes other than a3/a413:02
azonenbergnot familiar wit hthat particular model13:02
azonenbergbut like i said i want to make a photoplotter good to 20 microns or better lol13:03
smedingyeah.. that might be a bit ambitious for this13:03
azonenbergSo it needs either minimal reduction (2-4x projection) or nothing (contact mask)13:03
azonenbergrather than the 10x i'm doing now13:03
smedinganyway, go shower, man13:03
azonenbergi'm having issues with line edge roughness since i'm operating this printer so close to its limits13:03
azonenbergeven my 5-lambda design rule is pushing it for this printer13:04
--- Thu Aug 11 201100:00

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