#qi-hardware IRC log for Thursday, 2014-02-27

whitequarkeither it's false memory or something that's only accepted in RU00:00
whitequarknot even info on amphetamine-caused degradation00:00
DocScrutinizer05Dr House will feel a certain degree of relief about that ;-)00:02
whitequarkwell, there's this: http://www.plosone.org/article/info%3Adoi%2F10.1371%2Fjournal.pone.005643800:04
whitequarkbut it is far from accepted knowledge I'd say00:04
Action: whitequark giggles00:17
whitequarkhttp://i.imgur.com/rJThx60.jpg � a mouse00:17
DocScrutinizer05wow, that looks dangerous00:18
DocScrutinizer05we had similar incidents here, for our football stadium floodlight00:20
DocScrutinizer05My dad was responsible for the technical aspects, and once the lights were out of order for >60min since a rabbit got between the contacts just like that mouse00:21
DocScrutinizer05I'm not sure though if that switches were already so called "low voltage" aka 4kV or 15kV00:26
DocScrutinizer05ooh, "low voltage" is up to 1kV only, above it's already "middle voltage"00:28
wpwraknice. you can still see the fur pattern :)00:37
DocScrutinizer05that's indeed weird00:39
wpwrakit would be even more impressive if you could also distinguish feet, tail, head, ... and of course the bulging eyes :)00:42
DocScrutinizer05UPS are evil00:43
wpwrakyeah, pulling the plug won't save you00:44
DocScrutinizer05my personal statistics suggest that UPS cause 10 times more blackouts than they avoid. Thanks to mice, broken relays, defect batteries or electronc defects00:45
wpwrakonly had dead batteries so far. but yes, once the battery goes bad, the UPS makes small problems worse.00:47
whitequarkwhy do UPS batteries tend to inexplicably and quetly die all the time?00:48
whitequarkI thought Pb batteries really like being continuously recharged with tiny current00:48
DocScrutinizer05because the UPS designers have no real clue about treating batteries nicely00:48
DocScrutinizer05actually it seems to me they simply don't care00:49
wpwrakor maybe they do care - about making money with replacement batteries00:49
DocScrutinizer05but that's for tiny UPS only00:49
DocScrutinizer05the size like that one with the mouse is usually fine with batteries for 10..20 years00:50
DocScrutinizer05when a whole datacenter goes down, you can bet on "UPS defect"00:51
DocScrutinizer05which doesn't mean much since when the UPS wasn't defect, the DC wouldn't go sown00:52
DocScrutinizer05down even00:56
nicksydneywpwrak: http://info.tek.com/sg_mdo3000oscilloscope-oneisthenewsix.html11:24
wpwrakwill cost a bundle :)11:26
wpwrakah, the spectrum analyzer is something new. i was wondering how long until they's add that11:27
nicksydneyDocScrutinizer05: http://www.projectara.com/#project-ara11:27
nicksydneywpwrak: i think you can enter competition for the MDO300011:27
nicksydneywho knows you can win :)11:28
wpwrakyou never win in this sort of things ....11:29
wpwrakbesides, it wouldn't pass customs11:29
wpwrakit's about time someone makes a nice open hardware scope and put an end to the "option" mania. there's a clear trend towards marketing very nice devices but selling crippled junk that you have to "unlock" with option packages that cost you more than the original device11:31
nicksydneyopen hardware scope...hmmm......using FPGA ?11:33
wpwrakyou can't build a non-toy scope without an fpga :)11:33
nicksydneywpwrak: this http://www.osciprime.com/ 11:41
wpwrakstill a toy. 6 Msps. that's ridiculous. that's not even enough for arduino projects ...11:44
wpwrakand they want to sell that for 250 CHF ;-))) you almost get a real scope for that11:45
nicksydneyRussian jets patrol Ukraine border http://www.cnbc.com/id/101451132 11:50
nicksydneyflexing muscle eh11:50
wpwrakbtw, if you want low-end but a bit above toy level, you could consider this: https://www.kickstarter.com/projects/751733865/smartscope-reinventing-the-oscilloscope?ref=category11:53
wpwrak100 MSa/s is still very low. (suitable for signals up to around 10 MHz, maybe a bit higher but not a lot). they also have an fpga, not a cpld. so they might even be able to do a bit of protocol decoding.11:54
whitequarkwpwrak: I wonder, how do you determine which MSa/s value corresponds to which MHz?11:55
whitequarkwon't you need 2x MSa/s by Nyquist theorem/11:56
wpwrakdepending on a pc for processing is a major bottleneck, though. you generally can't get the data through fast enough. that's why integrating the display in the scope makes sense - direct path between display and sample memory11:57
wpwrakwhitequark: correct. but that's a sine wave under ideal conditions. there must be another limitation in there, e.g., about phase and phase drift.11:58
wpwrake.g., if i give you a series of all the zero crossings, nyquist would be happy, but you'd still have a hard time reconstructing the signal :)11:59
wpwrakso for reconstruction you need more. as a rule of thumb scope manufacturers recommend a sample rate that's 10-12x signal bandwidth12:00
wpwrak(for real-time sampling. if you use equivalent time, then things change)12:00
wpwraka few hints are here; http://en.wikipedia.org/wiki/Nyquist%E2%80%93Shannon_sampling_theorem12:02
wpwrak"In practice, infinite sequences," ... and " must be strictly less than ½ the sample rate"12:03
wpwrakthat was "B must ..."12:03
whitequarkwpwrak: oooooh sine wave12:04
wpwrakactually, i could imagine that it may work with any not too crazy waveform you agree on12:04
wpwrakbut don't forget that it has to be infinite12:05
wpwrakso it's similar to equivalent-time sampling12:05
wpwrakonly that the latter has a trigger while nyquist doesn't need that12:05
wpwrakof course, if you had to tell your scope what waveform you expect to see, that's kinda cheating ;-)12:06
whitequarkwpwrak: though: any waveform is a bunch of sine waves ;)12:11
whitequarkin fact, *you* are a bunch of sine waves12:12
whitequarkDocScrutinizer05: you'll like this, I've just seen an ad for a smartphone with 3 sim cards16:25
whitequarkexplay someshit16:25
DocScrutinizer05hmm, 3SIM isn't new16:26
DocScrutinizer05even 4 SIM seen occasionally. But no particular use in that automated SIM swapper16:26
wpwrakthe world need smartphones with a mechanical sim card changer. like the cd changers of old16:27
DocScrutinizer05exactly like that16:27
DocScrutinizer05dual-standby is the thing we want16:27
DocScrutinizer05and I haven't seen triple-standby yet16:27
DocScrutinizer05actually that concept pretty soonish runs into technical limits16:28
wpwraknaw, you want "all out" and "insert by gps location / fake identity". also set the IMEI accordingly :)16:28
DocScrutinizer05complete nonsesne16:28
wpwrakgive the nsa something to chew on :)16:29
DocScrutinizer05authorities and carriers have no difficulties whatsoever to track that IMEI/SIM change16:29
wpwrakhow ? if you change location and keep very low duty cycles ?16:30
whitequarkyep, they even readily give that to my bank16:30
DocScrutinizer05one IMEI vanishing and a new one appearing same time and same location? Dang, what the heck might THAT be???16:30
whitequarkeach time I change SIM they get aroused and send me SMSs16:30
whitequarkif I dare to get a new SIM for same number I'm locked out of account16:30
whitequarkand I don't remember signing anything that gives them this privilege...16:30
whitequarkOR reading about that feature at all16:30
wpwrakaroused bankers. now there's an ugly thought16:31
wpwrakreminds me of that 80es (?) song "lawyers in love". that also had a certain creepiness to it.16:31
Action: DocScrutinizer05 feels nausea16:32
DocScrutinizer05hmm, I guess they are running a SIM app to authenticate you against the bank account16:35
DocScrutinizer05that SIM app may send IMSI, IMEI, phone number, whatnot else16:35
whitequarkno, see, the only thing I did is I gave bank my number16:36
whitequarkwell, maybe they can remotely install something on SIM only given number. that makes it even worse16:36
DocScrutinizer05you're doing no eTAN or mobile banking or whatever?16:36
whitequarknever heard of16:37
DocScrutinizer05bank sends TAN per SMS16:37
DocScrutinizer05or wasit iTAN?16:38
whitequarkno clue16:38
DocScrutinizer05first there been TAN, like 100 one time passwords on a paper they sent to you. Then they added index numbers to that and asked for TAN #4216:40
DocScrutinizer05then they came up with TAN via SMS16:40
whitequarkwell yes, they do send me confirmation code for transactions to that phone16:40
whitequarkthough never heard it called TAN16:40
DocScrutinizer05PIN/TAN method16:41
DocScrutinizer05iTAN is Indexed, mTAN is Mobile16:43
DocScrutinizer05see http://en.wikipedia.org/wiki/Transaction_authentication_number#Mobile_TAN_.28mTAN.29 second paragraph16:45
DocScrutinizer05no surprise your bank is taking measures to make sure you are the legal owner of that SIM16:48
DocScrutinizer05however they do it. Prolly by cooperation with your carrier16:48
DocScrutinizer05I don't think bank can send SIM applications right away16:49
DocScrutinizer05but SIMs *can* get updated OTA, no news in that16:49
DocScrutinizer05and your carrier might even diclose more than the usual phonenumber to banks that send mTAN16:50
DocScrutinizer05the carrier knows your IMEI and IMSI16:51
DocScrutinizer05it's just usually not forwarded with SMS or ACK-SMS-replies16:51
DocScrutinizer05banks may negotiate to receive such details, since they have an obvious interest in it16:52
DocScrutinizer05however only after first mTAN sent to you16:52
DocScrutinizer05bank receives a reception-confirmation feedback, and that may contain rich info16:53
DocScrutinizer05SIGH! >>** PROBLEM Service Alert: firewall-b.maemo.org/SSH is CRITICAL **<<16:54
DocScrutinizer05o.O web interface has no alarms16:55
DocScrutinizer05meh, prolly backup16:58
wpwrakfinally fission can begin :)16:58
whitequarkgrmbl. lost all day to figuring out how to get to the warehouse with the reagents sent to me18:07
whitequark*no* public transport whatsoever stops within 1km of that area18:08
whitequarkhowever that didn't stop Google from trying to tell me to board a bus that doesn't even exist18:08
whitequarkon the good side I got aerosol spray paint (not water based, judging by how hard it is to get it off my hands), and bought a *new* thermoregulated oven for $30 from some guy18:40
whitequarkit even has a °C scale (instead of parrots, as it usually is) and the thermoregulation is surpsiginly good, according to my IR thermometer18:41
whitequarkso, I don't even have to mod it.18:41
Action: whitequark ponders cooking something in it before condemning it to an industrial purpose18:42
whitequarknaw, that'll make it dirty with grease. and I could always repurpose it back to food later...19:00
wpwrak... with all the health benefits of the added nutrients, namely lead, flux vapours, ...19:01
wpwrakdegC vs. ... parrots ?19:01
whitequarkoooooh you don't know19:01
whitequarkhang on19:01
wpwrak(warehouse) the 1 km blast radius may mean something. also, the disappearance of the bus may have a perfectly logical if tragic explanation19:02
DocScrutinizer05parrot seems a useful unit for heat19:03
DocScrutinizer05ROTFL @ wpwrak19:03
whitequarkwpwrak: DocScrutinizer05: parrots: watch this: http://www.youtube.com/watch?v=4Lj30QBLDh419:04
wpwrak(parrots) i'm sure there are places where there are a common food source. but russia ?19:04
whitequarkit's 8 minutes but it's totally worth it19:04
DocScrutinizer05wth? what were those authors smoking?19:07
whitequarkhang on, it's not just some drug-fueled children's cartoon. it's a drug-fueled children's cartoon tackling a deep philosophical problem!19:08
whitequarkfor some reason pretty much all USSR cartoons have this, erm, "smokey" feel to them. "what was the author smoking" is a rather commonly used idiom...19:09
wpwrakhmm, i learned a lot about the rational number pi there :)19:13
whitequarkI especially like python's catharsis upon successful measurement19:14
wpwraki was waiting until he got to have lunch. alas, they skipped that.19:14
whitequarknah, they're for preschoolers. strictly no violence19:15
whitequarkto be honest, they're also much better than the pointless mundane crap people usually try to show preschoolers19:15
wpwrakhmm, considering what - thanks to youtube - russians are known to do for weekend entertainment, i'd expect a rather flexible definition of "no violence" :)19:17
whitequarkwell... keep in mind it's from USSR days. state controlled media rather tight, *especially* whatever could have an effect on children.19:18
wpwrakah, only vegetarian pythons and peaceful military parades :)19:20
whitequarktalking about weird animation: http://www.youtube.com/watch?v=xpuWhpV-6-0#t=6619:22
whitequarkor from this point: http://www.youtube.com/watch?v=xpuWhpV-6-0#t=27519:24
wpwrakLSD and vodka don't mix19:26
Action: DocScrutinizer05 is short of proper reply19:28
whitequarkthere also was a certain fragment which is I'm absolutely certain was produced by an LSD-fueled binge and nothing more, it didn't even make sense in context19:30
whitequark(the stuff above kinda does, it's from fairy tales... the author definitely did add an unique touch though)19:31
DocScrutinizer05what's the philosophical problem tackled now? the number of dimensions in a parrot?19:36
whitequarkin the parrot cartoon? well, "what is measurement?" and the fact that you need an ethalon19:38
DocScrutinizer05or how much are 38 parrots in Liters Vodka19:38
whitequarkI mean, that's not obvious to 5-year-old children unless they're DocScrutinizers. I've met 30-year-olds to whom this wasn't obvious :p19:39
DocScrutinizer05ooh, such questions probably never occured to me19:39
DocScrutinizer05probably I always seen measurement as an arbitrary relation between two entities19:40
DocScrutinizer05and it's never meaningful without the proper contect/definition19:41
wpwraki feel some relief that "dict" doesn't know "ethalon" either. google does, of course, but it's a bunch of names19:41
whitequarkalways thought it was a loaned word19:42
wpwrakmaybe it moved straight up north, without stopping in the west19:42
whitequarkoh, fuck.22:05
whitequarkI've realized a rather huge problem with my stencil milling ide22:06
whitequarknamely if I mill out, say, an outline of a chip... guess what it'll fall out.22:07
wpwrakclosed polygons ? :)22:07
whitequarktwo options: 1) not mill it out but engrave rather, then do as DocScrutinizer05 suggested: copper plate printing22:08
wpwrakso you have to leave little bridges22:08
whitequark2) bridges, yes22:08
whitequarkI wanna try both options22:08
wpwrakelse, you can always try the cnc mill + brush + pot of paint idea ;-)22:09
whitequarknah, too unstable22:10
wpwrakor, use an airbrush. but that limits you to about 1 mm minimum width22:10
whitequarkhow'd I?22:10
wpwrakmount on cnc mill, add computer-controlled air valve22:10
wpwraklock the airbrush's own control knob into a suitable position22:11
wpwrakwould probably take about a week to get it right. but hey, the views on youtube ! :)22:11
wpwrakadd some fancy music and dance around the mill (while painting) like a wild dervish and yours is gangnam style 2014 :)22:13
wpwrakyou have to add girls, though. else, it's just some sad geek. but having girls in the picture means, whatever it is you're doing, it's not an evolutionary dead-end.22:15
Action: whitequark sighs22:16
whitequarknow I can segfault eagle22:16
wpwrakthat's an unusual request22:21
whitequarkit's not request, it's a bug in this damn CAD22:21
wpwrakah :) right ... "now", not "how". need stronger glasses22:22
wpwrakwhy use eagle when there's open and free kicad ?22:22
whitequarkI'm used to eagle and practically everyone (sparkfun, dangerousprototypes, etc) use it22:23
whitequarkalso, pcb-gcode CAM22:23
DocScrutinizer05GANGAM STYLE!22:24
DocScrutinizer05segfault eagle? :-o22:25
DocScrutinizer05Send the file to me, I might turn it into something useful! ;-)22:26
whitequarkDocScrutinizer05: it's an ULP. repro: UL_WIRE w[]; int n; B.elements(E) E.texts(T) T.wires(W) { w[n++] = W; }; UL_WIRE W = w[0]; W.layer;22:28
whitequarkeven: board(B) { UL_WIRE w[]; int n; B.elements(E) E.texts(T) T.wires(W) { w[n++] = W; }; UL_WIRE W =  w[0]; W.layer;22:29
whitequarkput any package on board and smash it22:29
DocScrutinizer05ow that's true gibberish to me ;-P22:29
DocScrutinizer05do you think upper and lowercase W/w is a smart idea?22:31
whitequarkwell, that's just the abridged version I posted here22:32
DocScrutinizer05I'd not be surprised when that does havoc to the interpreter, when at one occasion those two vars are considered entities and other occasion does a toupper() or whatever22:32
whitequarknah, it's irrelevant22:33
whitequarkthe interpreters which are that shitty died out at about 2000 :p22:33
whitequark(I've checked)22:33
DocScrutinizer05eagle is older22:33
whitequarkits ULP language is surprisingly good22:33
DocScrutinizer05still looks like dark voodoo to me22:34
whitequarkI mean, it's not painful to use, that's already something22:34
DocScrutinizer05B.elements(E) E.texts(T) T.wires(W)22:34
whitequarkusually when CAD people go and design a language it's torture22:34
whitequarkthat's how you enumerate parts22:34
whitequarkB.elements(E) basically means22:34
DocScrutinizer05create a obj/var "E"22:35
whitequarkin pseudocode: for(UL_ELEMENT E in B.elements) {22:35
DocScrutinizer05that looks decent22:35
whitequarkhm, I should try their beta22:36
DocScrutinizer05ohwell, I wrote COBOL and even SEVAG programs22:37
whitequarknope. still crashes.22:37
whitequarkSEVAG? short for "SEWAGE" ? :D22:37
DocScrutinizer05Siemens Eingabe Verarbeitung Ausgabe Generator22:37
whitequarkooh, no wonder can't google thta22:38
DocScrutinizer05some weird shit almost similar to d-base mask generator22:38
whitequarkd-base mask generator?22:38
DocScrutinizer05each program step was a record in a ISAM DB22:38
whitequarksome relative of subatomic photon torpedoes?22:38
DocScrutinizer05dang, you're too young to even know d-base22:39
DocScrutinizer05dbase mask generator was cool22:39
DocScrutinizer05SEVAG was da shite22:40
whitequarkooooh dBase?22:40
whitequarkI know that thing22:40
DocScrutinizer05printing sttrings to screen was a "statement" that had X, Y, and 17chars text22:40
DocScrutinizer05when I was completely fed up, I wrote a SEVAG editor tool that could do grep and sed commands across the ISAM db22:41
DocScrutinizer05impressed the living shit out of Siemens-dudes who also worked with that crap since decades22:42
DocScrutinizer05nevermind, that was IT pleozaen in 1985+-22:43
DocScrutinizer05we called that stuff http://www.tentacle.franken.de/m80/index.html computers back when22:44
whitequarknot sure if I should look ...22:44
whitequarkhaha, nice bug. that ULP blows through eagle stack22:50
whitequarknow the problem is, how do I make it work? -_-22:53
wpwrak("everybody" uses eagle) billions of flies can't be all wrong about the taste of feces ... ;-)22:57
whitequarkwpwrak: everybody uses eagle means there are designs and libraries available22:58
whitequarklast time I checked you had to do practically all cases AND symbols yourself for kicad22:58
wpwrakdid you ever machine-copy from a design ?22:58
wpwrakwell, yes, but i made sure i spend very little time on such things :)22:59
whitequarkokay... hm. stencils. I wonder if I can position this shit well enough to engrave on film.23:02
whitequarksurprisingly, yes, but the engraving is crap23:13
whitequarkum... it worked, in a weird way23:24
whitequarkI pushed the ink into grooves on stencil *and it stayed there823:24
whitequarkabsolutely no ink transferred to board23:25
whitequarkguess when I spread it in a thin layer, it dries so much quicker that there is no way I can transfer it quickly enough23:25
whitequarkokay... tomorrow I'll try to figure something out23:30
wpwrakapply solvent ;-)23:34
whitequarknot reproducible well enough23:40
whitequarkI think I devised a way to make CAM output bridges23:40
wpwrakmaybe like this: order corners by angle. place bridge on the ones with sharpest angles unless there's another sharp-angled close enough that making a bridge there will do. when done with corners, consider all straight lines that are too long. divide by two until you have the right distance. then add bridges.23:45
whitequarknah, I'll do simpler23:45
wpwraki think that should look reasonably well for most patterns23:45
wpwrakfor text, you may want manual rules23:45
whitequarkthe CAM is really simple: it just asks eagle for list of wires and then outputs those as gcode23:46
whitequarkso what I'll do: I'll make it record the length of the toolpath it wrote since last plunge into material23:46
whitequarkwhen that length exceeds X mm, it takes cutter out of material for Y mm while following same toolpath23:47
whitequarkeg X=5mm Y=0.4mm23:47
whitequarktext to me is just wires. I don't render it23:47
whitequarkwell, "wires", it's really lines when talking about silkscreen.23:47
wpwrakokay. you'll get some funny results that way, but at least it will let you explore the process23:47
whitequarkfunny? I'd think the result would be rather good23:48
whitequarksure, lines would be interrupted in random places, but who the hell cares. it's an inherent defect of this proces23:48
whitequarkrearranging bridges to make it more regular changes absolutely zilch23:48
wpwrakX= 4 mm, line of 10 mm. bridge from 4.0-4.5 and 8.5-9.0, meaning slot 0.0-4.0, 4.5-8.5, 9.0-10.0 :)23:49
wpwrakit'll look odd. don't forget that your patterns aren't necessarily closed polygons23:50
whitequarktook that into account23:50
wpwraki mean for the visual effect23:50
whitequarkdistributing brides evenly is a rather inordinate amount of work I'd rather not do23:50
wpwrakwell, it's the same for a closed polygon. two bridges very close to each other will look odd. but then, you can always optimize later.23:51
whitequarkwhy would they be close in a closed polygon? if lines are arranged in just the right way, the entire polygon would be considered "one line"23:51
whitequarkin CAM23:51
whitequarknot always the case, e.g. whoever drawn the symbol for pinhead arranged wires in a *weird* way23:51
whitequarkalmost random23:52
wpwraki mean like in the 10 mm example, well, actually no. it's the opposite - you'll get larger gaps23:52
whitequarkwell whatever, you'll see it soon enough23:53
wpwrake.g., 0 to 10 mm, closed path. X=3, Y = 1. the slots you get: 0-3 4-7 8-0 -> 8-323:53
whitequarkah yes, at the closing point of path there'll be an irregularity23:53
whitequarkcan't do anything with that23:53
whitequarknot without treating path as proper polygon and not just a bunch of lines23:54
DocScrutinizer05the paint won't magically flow out of trenches and onto your PCB. You need to apply some arguments that convince paint to move23:54
DocScrutinizer05a) wet the copper plate with e.g water with a drop of detergent23:54
DocScrutinizer05b) use very thin paint23:55
DocScrutinizer05c) use any form of knocking. Like maybe ultrasonic23:55
whitequarkwell... let's first look how the bridges idea fares23:56
wpwrakif you can "collect" paths from tool down to tool up, you could just adjust your initial distance value to distribute it evenly. and, to make closed polygons nice, make the initial slot not much larger than X/223:56
whitequarkI know, I just don't really want to implement that in ULP23:57
DocScrutinizer05when you cut V shaped trneches into copper plate, you can count on some water stciking in tip of that V. When paint sits above that minimal amount of water, then laminator heating up the copperplate to >100°C will probably work23:58
--- Fri Feb 28 201400:00

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