#qi-hardware IRC log for Thursday, 2014-02-13

nicksydneyjust read the blog from Pyra and it is 8 layer board damn !00:08
nicksydneythe same like Neo90000:08
nicksydneyanything more than 2 layer for me is awesome 00:10
DocScrutinizer05hey, N900 is 10 layer, it seems (I didn't disclose the photos I _tried_ to shoot from my cut out bare PCB where I slightly cut into the PCB active area and it seemed under mag glass like a two times 5 layers showed on the edge)01:13
nicksydneywow..10 layers...blown away :)01:19
nicksydneywhat do they have on each layer....i know that one of the layer must be a ground layer01:19
nicksydneythe components are soldered on top and bottom right ? so that's 2 layer taken and 1 layer for gound ..so that's left with 7 layers ..am i correct ?01:20
DocScrutinizer05I might have missed to count the two outermost layers ;-)01:21
wpwrakwith so many layers, that board will be hell to produce ...01:32
wpwrakvery limited choice of factories, high cost, endless lead times, ... i have fond memories of when openmoko went crazy on their pcbs as well ... and then the one fab that could actually make them closed ...01:34
wpwraki suppose you have buried vias and such as well ?01:34
DocScrutinizer05please reread01:39
DocScrutinizer058 layer used in N900 is pretty much standard01:39
DocScrutinizer05at least for 0.8+ mm FR401:39
nicksydneyDocScrutinizer05: so it's 8 layers not 10 layers ?01:40
DocScrutinizer05please reread01:40
DocScrutinizer05do you think i'd *guess* about the number of layers for neo900?01:41
DocScrutinizer05>><DocScrutinizer05> hey, N900 is 10 layer, it seems...<<01:41
DocScrutinizer05Neo900 is 8-layer01:42
wpwrakah, good :)01:43
wpwrakit sounded as if you had ended up with 12 ...01:43
DocScrutinizer05the heck, it sounded *to you* like that01:43
wpwrak(guess) well, if the number gets large enough you could use some just to rounding errors ... ;-)01:44
DocScrutinizer05and no, actually I counted the suface layers on that N900 board01:44
wpwrakgood :)01:44
DocScrutinizer05good? why?01:44
DocScrutinizer05does it matter in any way?01:44
wpwrakgood for them and their sanity01:45
DocScrutinizer05I don't give a flying F about that01:45
DocScrutinizer05or rather, my notion about that latter is already quite dedicated01:46
wpwrakyou should be nice to them. if you want, think of them as the little monkeys neo900 evolved from :)01:46
DocScrutinizer05they had some good EE staff and a real crew of douchebags for project managers or higher up01:47
wpwrakwhat's new ? ;-)01:47
DocScrutinizer05nothing really01:47
wpwrakactually, that's something i liked when openmoko took on the dash project: fic shut down fiwin, fired all the management that had screwed up things to royally, and only retained (a good number of) the engineers01:48
wpwrak#s/ to / so /01:49
wpwraknice :)01:59
DocScrutinizer05I damn need a camera that can do microscope shots where the 0.8mm fill the picture02:00
DocScrutinizer05I hardly can see that stuff with a 10x mag glass02:01
DocScrutinizer05I'm still in doubt if it's really 10 layers02:02
DocScrutinizer05and no, I have no damn clue if Nokia has buried vias02:02
DocScrutinizer05but I guess for a SoC/footprint the class of a OM3530 you don't get away without them02:03
DocScrutinizer05you definitely need in-pad vias02:04
DocScrutinizer05which Nokia is using everywhere - you can tell by the microscopic craters in the pads which have one02:05
wpwrak(in-pad) mmh, a bit suckish but i can see why02:06
DocScrutinizer05see http://maemo.cloud-7.de/Gallery-N900-exploded/platine_01.jpeg02:07
DocScrutinizer05you'll easily spot them craters02:07
wpwrakxiangfu: heya ! question: do you still have any contact with wolfgang ?02:08
DocScrutinizer05in roundabout 90% of solder pads02:08
wpwrakDocScrutinizer05: well, if you have them, you may as well use them ;-)02:08
DocScrutinizer05hey, a totally on-topic question: how would you design a very simple detector for 2GHz / 5mW@50Ohm?02:17
DocScrutinizer05or rather: 1GHz..3GHz range02:17
wpwrakhmm, something like bandpass + diode + comparator ?02:19
DocScrutinizer05no real need for bandpass02:19
DocScrutinizer05diode has a Ufwd threshold, even a Schottky02:19
DocScrutinizer05fwd-bias it?02:20
wpwraksounds like an idea02:20
wpwrakor just use a photodiode ...02:21
wpwrak... attached to the screen of a nice spectrum analyzer ;-)02:21
DocScrutinizer05or would I rather use a HF-transistor and bias that to the point where it just starts opening02:21
DocScrutinizer05C-E should draw noticeable current that can get smoothed by an e.g 1nF in parellel02:22
DocScrutinizer05only when RF is present02:22
DocScrutinizer05a darlington with a 100MOhm from basis to collector should do, eh?02:23
DocScrutinizer05wpwrak: you are always cheerful when you can play with spice ;-)02:24
DocScrutinizer05input -> 10k -> 10pF -> (A) darlington base; (A) -> 100MR -> darlington collector -> (B) -> output; (B) -> 47k -> +Vio; Darlington Emitter -> GND; (B) -> 10nF -> GND02:28
DocScrutinizer05input: sweep 600MHz..2800MHz, 5mW@50R02:29
DocScrutinizer05output: high-Z02:30
Action: DocScrutinizer05 wonders if darlingtons with Ft:3GHz are commonly available nowadays02:31
wpwrakqucs ... no spice for me, thanks ;-)02:34
DocScrutinizer05that 100M resistor actually should be 47k * amplification_darlington * 2002:37
DocScrutinizer05I'm afraid that the darlington basis acts like a rectifier which would create a negative bias on the 10pF, so the darlington would again operate "on the edge" after first 2 or 3 sine waves from input02:39
DocScrutinizer05mmpf, my bouncer went offline02:40
wpwrakdoes your bouncer kick you out of the lab when you've had too much solder fumes ? :)02:42
DocScrutinizer05no, it makes my N900 create funny clicking noises02:42
DocScrutinizer05wpwrak: maybe using a mosfet would avoid such rectifier effect creating negative bias on the 10pF DC-blocker capacitor?02:49
wpwrakyou're already way beyond my knowledge there :)02:52
DocScrutinizer05it's pretty simple: when input is "high2, the 10pF will charge positive on input and negative on basis side, due to current flowing from input into basis-emitter diode(s). Now ehen input goes GND level or even negative, the capacitor will pull down the basis but no current will flow out of basis thru capacitor and series R to input02:55
DocScrutinizer05so the negative bias on darlington basis will sustain02:55
DocScrutinizer05effectively compensating the positive bias created by the 100M02:56
nicksydneyDocScrutinizer05: what is this http://maemo.cloud-7.de/Gallery-N900-exploded/platine_01.jpeg ...looks sexy03:59
DocScrutinizer05that's what the name says: a N900 bare PCB05:11
DocScrutinizer05the inner 7 layers (4 copper, 2 thin and core thick isolation) make for the flex cable that connects the mainboard to the daughterboard05:20
DocScrutinizer05at least I guess how it's built05:22
DocScrutinizer05at least I guess that's how it's built05:22
xiangfuwpwrak: Hi I just find him on IRC and another email address.05:29
xiangfuhave you try message him by IRC?05:31
nicksydneyDocScrutinizer05: sorry for stupid question but when you say N900 are you referring to Neo or Nokia ?05:32
DocScrutinizer05N900 is Nokia05:33
nicksydneygotcha...that's why i was confused back and forth :)05:33
DocScrutinizer05Nokia never built a Neo device05:33
nicksydneyboth has got the N infront so just want to make sure05:34
nicksydneyand both has the 900 too :)05:34
DocScrutinizer05and both look damn similar05:35
Action: DocScrutinizer05 wonders WTF for are those pads above "E" key05:37
DocScrutinizer05they are unused in N90005:37
DocScrutinizer05err, actzally above "W" key05:38
DocScrutinizer05(3rd from left, upper row)05:38
nicksydneyno idea :05:42
nicksydneythis Nokia pcb design is available ?05:43
DocScrutinizer05err, available?05:46
DocScrutinizer05please rephrase!05:47
nicksydneythe files .. the schematics like neo90005:47
nicksydneythat pic you took looks  like a brand new pcb so was wondering did you produce that ?05:47
nicksydneyi mean fabricate that05:47
DocScrutinizer05otherwise I wouldn't wonder if it's 10 layer05:48
DocScrutinizer05yes, that one05:50
nicksydneydoes this mean that the software than ran on Nokia 900 was all open sourced ?05:52
nicksydneyor only the kernel?05:53
DocScrutinizer05neither of both is true06:46
whitequarkhaha https://pbs.twimg.com/media/BgRs_I-CIAErVKg.png:large06:50
whitequarkheadline translation: "Tolyatti citizen went to Ivanovo, lost his ID, came to visit a retiree, then died and decomposed. / Meanwhile, the retiree continued drinking."06:52
larscsounds like it all happend within a day07:42
whitequarkwithin less than a week07:50
DocScrutinizer05LOL! http://privatepaste.com/a50744fd05 (for German readers)14:55
DocScrutinizer05though, I wonder if that's really German14:56
DocScrutinizer05from: xxxxxxxxx@mail4.doshisha.ac.jp14:58
DocScrutinizer05I wonder what a weird kind of spam is *this* .oO(???)14:59
whitequarkit's just regular scammers. I get that in English, Russian, Japanese, ... all the time15:03
whitequarkChinese also, I think15:03
whitequarkhrm. I can reliably reproduce 6 mil traces with my photoresist process, but not lower.16:29
Action: whitequark wonders if he should keep trying or is that overkill16:29
whitequark4 mil almost works. 2 mil really isn't.16:30
viricwhat is a mil?16:45
whitequark6 mil = .15mm; 4 mil = .1mm; 2 mil = .05mm16:45
whitequarkmil is 1/1000"16:46
viricthank you16:47
DocScrutinizer05I heard you can flatten gold to around 5 atom layers thickness ;-P16:48
whitequarkDocScrutinizer05: is that your friendly way of saying "overkill"? ;)16:50
DocScrutinizer05hmm, looking at that sample, it's probably not16:50
whitequarknevermind the defects in it, it's random dust and not really related to process16:51
whitequarkdust and/or shitty negative. neither matters for what I'm working on now16:51
DocScrutinizer05aah, then16:52
whitequarkbut the .1mm line *is* broken in a lot of plces16:52
whitequarkand .05mm is simply absent16:52
whitequarkI've actually had it on earlier samples, dunno why that stopped working while everything else improved16:52
DocScrutinizer05well, in that range you already need to compensate for undercutting 16:53
whitequarkI still have a few ideas to try; soaking board in NaOH for 10min to remove all organic crap from it; using lowest grit sandpaper for better surface finish and contaminant removal16:53
whitequark(undercutting) yeah... copper thickness is .035mm16:53
whitequarkso .05mm actually hardly makes sense there16:54
DocScrutinizer05photoresist thickness16:54
DocScrutinizer05light diffuses in photoresist as well16:54
whitequarkdiffuses, hmm...16:54
DocScrutinizer05gets reflected from copper16:55
whitequarkyou know what, let me measure the actual line width I have there. the photo is almost perfectly parallel to camera sensor so it's easy16:55
DocScrutinizer05doesn't come in ideally parallel16:55
DocScrutinizer05toner tends to build heaps16:56
DocScrutinizer05so isn't plane16:56
whitequarkno, that was taken care of with "toner density"16:56
whitequarksee http://localhost:4000/notes/2014-02-13/negative-photoresist-calibration-mask/16:57
DocScrutinizer05though, that probably depends on printer16:57
DocScrutinizer05anyway, when your photoresist is considerably thicker than wide, it will probably simply brake off from copper16:58
DocScrutinizer05there's probably a natural physical limit of resolution you can achive with a particular photoresist16:59
whitequarkDocScrutinizer05: before "density toner": http://lab.whitequark.org/images/photoresist/calibration-mask-printed.jpeg after: http://lab.whitequark.org/images/photoresist/calibration-mask-dense.jpeg16:59
DocScrutinizer05ooh, forgot you're using photonegative17:01
Action: DocScrutinizer05 points his laser through a 0.3mm pitch stencil and watches the funny patterns on the wall17:03
whitequarkwhat's that stencil for? bga?17:04
DocScrutinizer05looks like a monster FPGA17:05
whitequarkDocScrutinizer05: neat17:09
whitequarkby the way, you were right17:09
whitequarkthe actual trace widths are often off by .05-.1mm depending on their position17:09
whitequarkthe centermost trace is really damn wide17:10
whitequarkI need to make a better light source17:10
whitequarkDocScrutinizer05: also, thickness of developed resist is somewhere around .04mm if I'm using my caliper right17:14
whitequarkso .05mm trace means it would be almost rectangular in section! no wonder it doesn't work well17:15
DocScrutinizer05with copper you'll get similar problems17:33
DocScrutinizer05you need to control your etching process a factor 50 better than you're able to do right now, to etch a 0.05mm wide trace into standard FR4 material17:35
DocScrutinizer05copper fpild for up to 1.55mm are generally 17µm it seems17:41
DocScrutinizer05but then you start to have 35 or even 70µm17:42
whitequarkDocScrutinizer05: it's 35µm I have here18:03
whitequarkat least the vendor claims it's 3518:03
whitequark"1 oz" translates to 35µm afaik18:03
whitequarkyeah, it does18:07
roh35 is 'normal' for hobbyists stuff19:14
rohif you produce pcbs professional its much thinner and they make it thicker only after the first round of etching19:14
rohto reduce the amount of needed copper (expensive)19:15
Action: whitequark nods19:15
rohbut afaik the processes needed arent feasible at home.. so.. we have simple 35µ from the start19:15
whitequarkroh: I think they're pretty feasible, I've encountered that while researching whether I can set up electroplating19:15
whitequarkit's just a CuSO„ bath19:16
whitequarkand a lot of current19:16
rohi see... well.. send pix if you done it ;)19:41
whitequarkelectroplating holes or whole boards?19:41
whitequarkI've been thinking to cover everything except vias with resist and then electroplate only to holes19:42
whitequarksimply because this way it's much quicker; I don't need to cover the entire board, just tiny vias19:42
whitequarkI'll document it thoroughly, too19:42
whitequarkwhat I really want is not a quick hack but a reproducible process, if only for sake of understanding how exactly to set up such a process19:43
rohyeah. also what to do with the waste19:55
rohthats what kept me off all this chemical stuff most the time. waste disposal is really tricky to do well19:55
wpwrakwith HCl+H2O2 you produce nearly zero waste19:57
wpwrakand if you want to get rid of some, leave it outside so the water evaporates and you're left with solid residues19:58
whitequarkroh: waste? which waste?19:59
whitequarkCuSO4 bath doesn't have any. it deposits copper ions from electrodes to solution and then from solution to PCB20:00
whitequarkthe electrolyte can be used indefinitely20:00
whitequarkother chemicals I use... NaOH, K2CO3 can be poured down the drain in any part of world20:00
whitequarkpure (NH4)2S2O8 is benign (it'll immediately oxidize some organic and will turn into harmless ammonia and sulphate anions)20:01
whitequarkused (NH4)2S2O8 with copper cations is worse, but honestly that miniscule amount of copper is not going to do anything, no matter how greenpeace wants you to think otherwise20:02
whitequarkalso consider that in Norilsk factories probably dump dozens of tons of copper a year20:03
whitequarkif I was really concerned about the whopping 20g or so of copper I'd need to dump when I use all of this solution, I'd add NaOH to it, after which insoluble Cu(OH)2 will precipitate20:06
whitequarkthat last thing can be thrown to landfill, it's not going to enter water stream.20:07
whitequarkroh: I suspect your local bureaucracy would be tougher than actual safe waste disposal20:07
rohyep... well.. its rather a question of honor i guess ;)20:09
rohatleast in the amounts we use for hobby.20:09
whitequarksure. I think you've actually convinced me to dispose of it properly, considering I see how it is not hard at all20:09
whitequarkjust some highschool chemistry and you're done20:09
whitequarkwpwrak: I've recalled another argument against HCl+H2O220:12
whitequarkit's a good way to get you some chlorine gas and that is going to corrode all iron tools you have around20:12
whitequarkI've actually screwed up like that once and some of my tools are still rusty from that incident20:12
whitequarkthe thing is, you don't even need that much Cl2. tiny amounts still do have impressive effect on anything Fe or Cu or even plastics20:13
wpwrakyes, that's why i store it outside and open the bottle near the window. the etching doesn't produce much Cl gas.20:17
whitequarkcall me weird but I still prefer "no chlorine" over "a little chlorine"20:20
whitequarkhuh. I found a vendor who sells 12kg of H2O2 for $1320:25
whitequarkand they apparently work with individuals20:25
wpwraknice price. what concentration ? 0% ? :)20:30
whitequark30-40% apparently20:31
rohno chlorine is nice. the broken tools really suck20:31
rohit even makes aluminium really ugly20:31
whitequarkwpwrak: I've googled it and it seems to be a common price for H2O2 here20:32
whitequarkanother vendor sells 34kg drums for ~$3020:32
whitequarkroh: it just ruins everything it touches :/20:32
whitequarkhuh. "Metallic Na 98%" $50/kg20:33
whitequark"H2O2 33%" $3/kg (retail price here)20:34
whitequark"Bromine" $20/l20:34
whitequarkI really hope they do check whoever they sell this stuff to20:35
wpwrak30-40% is good. don't forget to wear gloves also when opening/closing the bottle or next summer you'll have to work extra hard to develop a tan :)20:36
whitequarkyeah I'm using gloves all the time. mainly to not leave fingerprints currently, as none of the stuff I use is corrosive20:37
whitequark"Nitric acid, concentrated" $3/kg20:37
whitequark"Aluminum, finely ground" $15/kg20:38
whitequarkthis is a real treasure of a vendor20:39
whitequarkhow is shipping that even legal20:40
wpwrakyou should ask them if the have activated FOOF on tap ;-)20:40
whitequarkthey don't seem to stock nonexistent or absurd stuff. it just seems awkward in wake of recent regulations20:42
lekerneldo they have dimethylmercury?20:42
lekernela few years ago there were plutonium listings on alibaba.com20:43
lekernelthey seem to have disappeared20:43
whitequarkyes, I'm talking about that kind of stuff.20:43
whitequarkthis shop doesn't have anything with mercury at all20:43
whitequark"Disprozium" $5300/kg20:44
whitequark"Europium(III) oxalate" $2400/kg. they seem to stock quite a bit of rare-earth stuff20:45
whitequarkum, "Heavy fluid" $30/kg20:46
whitequarkheavy fluid ?20:46
wpwraktry to order some of their nastier stuff. see if they let you. maybe they just forgot to update their shop. or require clients to show some credentials.20:46
whitequarkwpwrak: nah don't want to. I'll probably get put (and rightfully so) onto some bozo list20:47
wpwrak(heavy fluid) they'll add more details once someone has ordered it and they autopsy results come out20:47
wpwrakwell, then get something difficult but harmless. like a small quantity of HCl :)20:48
whitequark(heavy fluid) I missed the standard#; it's actually concentrated Ba(CdI)4 solution20:49
whitequarkused for ore flotation or something20:49
whitequarkthe standard makes reference to similarly generic properties of fluid. for example, "Colority factor (max) -- 60"20:50
whitequarkWTF is a "colority factor" and why does it not even have units20:50
wpwraknot everything has a unit ;-)20:51
lekernelwhitequark, do you happen to know where to find cadmium-109?20:52
whitequark>HALF-LIFE: 464 days20:52
whitequarkyou must be an international terrorist or something20:52
whitequarktry Libua20:52
lekernelno, I wonder if https://www.youtube.com/watch?v=3YHaWHwJHWo can be reproduced20:54
whitequarkwpwrak: (colority factor) it is actually equal to %wt concentration of I2 in water which would have the same color20:54
wpwraklekernel: moscow tap water ? ;-)20:55
whitequarklekernel: wow. fascinating.20:56
whitequarkbut I bet it's measurement error, somehow.20:56
whitequarklekernel: http://www.ritverc.com/products/detail.php?ID=1682 ?20:57
lekernelbeen looking at it for a while, also asked a couple professional physicists20:57
wpwrakbreaking news: russian scientists reveal: "quantum theory is measurement error"20:57
lekernelno one could point out an obvious flaw20:57
lekernelalso, it seems true that no one has experimented with gamma rays20:58
lekernelthe shortest wavelength reported seems to be in the x-ray20:58
wpwrakwhitequark: your next myth to debunk should then be dark matter. that would leave dark energy and maybe for dessert relativity. once you're done with that, newton can safely rise from his grave20:59
lekernelin http://link.springer.com/article/10.1007%2Fs10043-010-0042-320:59
whitequarkwpwrak: why me?20:59
lekernelwpwrak, quantum physics is often full of crap, no matter who does it20:59
lekernelrespected professors or random guys21:00
whitequarkI'm not saying QM is incorrect, especially since I understand exactly zilch in it21:00
whitequarkand, you know, the whole "device I'm typing on has had QM applied in hundreds of places to great effect"21:01
whitequarkone of funniest sights: people arguing GR/SR is incorrect, while using GPS21:02
whitequarkwpwrak: ooo by the way21:02
whitequarkthey stock Y and Sr salts21:03
whitequarkhigh-temperature superconductors!21:03
whitequarkoh, YBaCu actually21:04
whitequarkI know a guy who claims to have successfully homebrewed one21:04
whitequarkhehehe, "Silicium(II) oxide" $15/kg21:06
wpwrakwhitequark: bah, gps is basic geometry. euclid could have designed it.21:33
whitequarkwpwrak: surely you know it has corrections for both SR and GR?21:38
whitequarkbtw. I ordered the chemicals required for activating the holes for electroplating21:54
whitequarklet's see how it goes21:54
wpwrakwhitequark: (gps) i didn't say that his design would have worked ... at least not without adding a few layers of epicycles and such ;)22:26
whitequarkI wonder if you can model Lorentz contraction with epicycles or somesuch22:31
whitequarkprobably not22:31
wpwrakoh, i'm sure you ca22:32
wpwrakn. at least to any precision you want ... :)22:32
wpwrak("sure") not in the "i've proven it or saw someone prove it" kind of sense but more in the "if someone descends to that level of madness, it's inconceivable that they won't find a way that makes sense at least in their head"23:12
--- Fri Feb 14 201400:00

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