#qi-hardware IRC log for Monday, 2014-02-10

wpwrakhmm, i'm slowly getting the impression that P&E are bastards. it seems that their firmware for the FRDM-KL25Z doesn't support programming the KL26 even though it's virtually identical to the KL2501:59
wpwrakat least i've made it all the way to the AHB on the KL26 by now without encountering any obstacles ...02:00
DocScrutinizer05ok, found FRDM-KL25Z but what's P&E?06:01
whitequarkhttp://www.pemicro.com/ ?06:07
wpwrakthey're the one who make the firmware this critter ships with06:09
wpwraki use the FRDM-KL25Z for flash programming of anelok. but when switching from kl25 to kl26, it seems that i'll either have to get an FRDM-KL26Z, which may now be difficult, of make my own solution06:11
wpwrakwell, i got as far as being able to reliably read the AHB. that's a start. there's more needed for being able to program the flash, but i think what i've covered so far was the worst part06:12
wpwrakone of the ARM buses (one of the fast ones, kinda "CPU bus")06:16
DocScrutinizer05aaah, yes. with context it's easy06:56
DocScrutinizer05didn't know you explicitly read AHB, I always thought that's part of how ARM controls its internals without "user" even getting to know about06:58
DocScrutinizer05like CPU talking to e.g. UART via AHB, but that's happening invisibly when you access the registers of UART06:59
whitequarkDocScrutinizer05: it's exactly like that07:01
wpwraknormally you don't necessarily see which bus things are on (though it appears in some places, e.g., clock gating), but here it's a debug port that's specifically on AHB (and has "AHB" in all the identifiers as well)07:01
whitequarkbut debug interfaces, which AIUI wpwrak uses, allow you to poke it directly07:01
DocScrutinizer05hmm yeah. ETM for example you actaully need to understand about that stuff07:02
DocScrutinizer05and stuff like clock gating etc is exactly why I favor maemo over genuine debian for OMAP3 devices like Neo90007:05
DocScrutinizer05the upstream stuff never is thoroughly tested for proper power management07:05
DocScrutinizer05Nokia's maemo kernel devels didn't invent closed proprietary stuff for fun and to annoy FOSS community, they added or modified FOSS kernel stuff to handle things like shutting down clocks or set suspend timeout for UARTS and the like, and some of that stuff didn't make it to upstream for various (usually stupid) reasons. that however doesn't mean that a system on a OMAP3 platform is better off without those non-mainline patches07:10
wpwraknon-mainline does not imply closed. there's a lot of kernel code out there that isn't in mainline for one reason or another but that it quite open07:12
DocScrutinizer05and I wish FPTF devels like freemangordon would get that07:13
DocScrutinizer05wpwrak: yes, exactly07:13
DocScrutinizer05the problem is that same considerations apply to userland aka middleware and even to apps as well, and those often are not FOSS for maemo07:15
wpwraklet's hope people are motivated to change that (by replacing closed junk). else, you have a growing problem on your hands ...07:16
DocScrutinizer05well, not me exactly07:17
DocScrutinizer05but e.g. mer and Qt-mobility often demonstrated a scaring ignorance about such aspects of an embedded system, when e.g. developing/specifying sensorframework07:21
DocScrutinizer05the whole basic concept been flawed, basing mostly on definition of poll intervals07:22
DocScrutinizer05you hardly can turn a app triggered API into a event driven API07:22
DocScrutinizer05since your API will miss the needed functionality to specify and handle events in the only correct way07:23
whitequarkugh :(07:24
DocScrutinizer05and the general mood about this been "let's deal with power management and all that stuff later. first we need to make sure it works"07:24
whitequarkDocScrutinizer05: qt-mobility iirc doesn't depend on polling. just the default handlers it has poll sensors07:25
whitequarkyou can easily (in like 200 loc) write evented ones07:25
DocScrutinizer05qt-modility at large maybe not. qt-m with sensorfw - as of maybe 2 years ago - yes, they did07:26
DocScrutinizer05and no aurprise since they didn't even get the catch that kernel API been flawed for e.g. the lis302 accelerometer07:27
whitequarkcan't comment on that, I used to work with it when it still was qt embedded07:28
whitequarkdon't think they made the API *worse*07:28
DocScrutinizer05the mainline lis302 kernel driver is basically only for joysticks07:28
DocScrutinizer05and Carsten Munk (mer developer, now sailfish) answered all my concerns with "it's mainline/upstream, and we won't change that"07:29
DocScrutinizer05so lis302 kernel driver only supports polling  --> sensorfw defines an API for accelerometer sensor that's based on the concept of polling07:31
DocScrutinizer05--> qt-m sensors are basically polling07:31
whitequarkok, this is dumb07:32
DocScrutinizer05when I would design such a system, I'd forbid polling and whoever wants to poll nevertheless needs to implement it on app level07:32
DocScrutinizer05what they did was the opposite, they implemented polling on every single layer: in kernel driver, in sensorfw, and in qtm sensor plugins07:33
DocScrutinizer05kernel polls chip at 100Hz, sensorfw polls kernel API at 25Hz, and finally app developer decides the poll frequency for app to sensorfw, iirc07:34
whitequarkthat must absolutely kill battery life07:35
DocScrutinizer05yes, it does07:35
DocScrutinizer05well, 100Hz is the lower of two freq for polling mode where lis302 drives the IRQ line for every new data it provides07:36
DocScrutinizer05the higher freq was 400Hz07:37
DocScrutinizer05joystick mode07:37
DocScrutinizer05the only mode the lis302 upstream kernel driver supports07:37
DocScrutinizer05both Nokia and OM had their own lis302 kernel driver implementations that were magnitudes smarter than the mainline one07:39
DocScrutinizer05but when you do system architecture for a new platform and your 1st commandment is "we use mainline. ALWAYS", then you end up with API definitions that never heard about IRQ/event driven concept07:41
DocScrutinizer05I don't know of a single successful embedded linux device that is using 100% mainline kernel07:42
DocScrutinizer05not even OE kernel07:43
DocScrutinizer05why is qi-bot +o? is there anything it does for channel maintenance?08:39
larsckicks people who ask to many questions08:39
DocScrutinizer05well, the simple answer is:because qi-bot is +O08:42
DocScrutinizer05but who decided qi-bot needs +O in access list. particularly since +O is deprecated by freenode08:43
DocScrutinizer05and why?08:44
DocScrutinizer05hmm, ding leaves me wondering what "crickets" might mean here09:29
DocScrutinizer05chirping insects? birds? something even different?09:30
larscthe sound that crickets make09:32
DocScrutinizer05or playing a game with a ball and driver09:32
larscno, you know when there is a long silence you start hearing the crickets chirping in the background09:33
DocScrutinizer05well, I couuld do a -O on qi-bot09:35
DocScrutinizer05would make me feel less threatened by the machines09:36
DocScrutinizer05I thought somebody could provide a good reason why qi-bot is permamnently opped. 09:37
DocScrutinizer05I guess it's been an administration oopsie then, and i'll remove +O from qi-bot's access list record, to fix the issue. If later on somebody find a good reason why the bot needs +o nevertheless, any chanop with access list permissions can add the +O again09:38
DocScrutinizer05well, or not. 09:40
DocScrutinizer05[2014-02-10 10:40:11] [Notice] -ChanServ- You are not allowed to set -O on qi-bot in #qi-hardware.09:40
-ChanServ:#qi-hardware- DocScrutinizer05 (jOERG_rw) set flags +O on jOERG_rw.09:42
-ChanServ:#qi-hardware- DocScrutinizer05 (jOERG_rw) set flags -O on jOERG_rw.09:42
DocScrutinizer05damn 2manpage" about accesslist management is pretty incorrect or at least fuzzy. It says a chanop on access list may set all the flasg on other access list records that the chanops owns as well. I have +f (for access list modification) and +o which implies +O09:44
DocScrutinizer05I've even set my own record to +O but still can't set -O on qi-bot09:45
larscthe bot is more mighty than you are09:46
larscblessed by wolfspra1l himself09:47
DocScrutinizer05iirc wolfspra1l had no clear idea about access lists and chanop09:47
DocScrutinizer05that's how I came to the burden of having +o in this channel09:48
DocScrutinizer05well, whoever has access to that eggdrop, please consider make him say  /msg chanserv flags #qi-hardware qi-bot -O09:50
DocScrutinizer05please don't mistype that as -o !09:51
DocScrutinizer05(though even that was recoverable since the bot still is +F)09:51
DocScrutinizer05(which is probably the reason why chanserv denies me changing +O flag for qi-bot)09:52
DocScrutinizer05what I probably can do is:09:53
DocScrutinizer05but that won't last09:53
DocScrutinizer05a common nice practice is to:09:53
DocScrutinizer05to mark the bot as approved by chan-owners09:54
DocScrutinizer05on access list that would be flag +V09:54
DocScrutinizer05so make that:09:54
DocScrutinizer05 /msg chanserv flags #qi-hardware qi-bot -O+V09:54
DocScrutinizer05god! I'm chanop of 32 channels10:05
DocScrutinizer05I guess most of them are dead, though10:05
DocScrutinizer05at least stuff like #opie.de or #openmoko-wikireader or #openmoko-de10:06
nicksydneyDocScrutinizer05: http://www.cnx-software.com/2014/02/10/dragonbox-pyra-open-source-handheld-game-console-to-feature-ti-omap5432-soc-or-allwinner-a80-or-intel-bay-trail-or-qualcomm-snapdragon/11:58
nicksydneyreally like this blog post http://pandoralive.info/?p=2808 .. gives an insight of how difficult it is to find SoC :)...never knew it's going to be this hard12:02
DocScrutinizer05yeah, that's pretty nice12:10
DocScrutinizer05and yes, actually nobody knows12:11
DocScrutinizer05almost nobody12:11
DocScrutinizer05and even when you're there and have your SoC, you're in for the next joy: Silicon Errors, i.e. unpleasant errant behaviour of the chip thatnks to bugs in the mask and design12:13
DocScrutinizer05when you're very lucky the SiErr gets documented by the chip manufactuer and even has a workaround ideally in software. When you're less lucky your project needs a massive redesign or fails completely. When you're punished with the usual lack of luck, you finf out about the SiErr when you've built and sold a few 100 k of devices12:14
DocScrutinizer05find* out12:15
DocScrutinizer05I know of such SiErr in OMAP412:15
DocScrutinizer05which seem to not even gotten publicly disclosed, and are conveniently tagged as "WONTFIX in OMAP4" by TI12:16
DocScrutinizer05and sw workaround is difficult or even impossible12:16
Action: DocScrutinizer05 idly wonders if Neo900 already qualifies as steampunk12:25
whitequarkDocScrutinizer05: can you tell more about those SiErrs?12:27
DocScrutinizer05well, we seen the HSI wakeup failing under certain conditions12:28
DocScrutinizer05it just drops off the queue or sth12:28
whitequarkHSI = ?12:29
DocScrutinizer05since HSI protocol doesn't include any recover/repeat from such pending data transfer that doesn't get answered by the destination interface, our modem basically got stuck when talking to the TI Blazer12:29
DocScrutinizer05MIPI HSI12:29
whitequarkah, modem12:30
whitequarkyeah, that sucks.12:30
DocScrutinizer05evetually the modem blew chunks when FIFOs bursted from overflow12:30
DocScrutinizer05then the customers blamed us for a bug that actually been in OMAP4 and not in NovaThor or the firmware12:31
DocScrutinizer05TI shipped a patched HSI linux driver, that allowed recovering from such lockup. but only when the modem actively did retry, which is not a part of the MIPI HSI specs12:33
DocScrutinizer05so, we were able to find a sw workaround, but you can't do that when you don't control e.g. the modem that's conected to the OMAP4 HSI interface12:34
DocScrutinizer05for sure we'd have a hard time talking Cinterion into any such patch for their P*S8 modem12:34
DocScrutinizer05(though those modems don't even have HSI interface)12:35
DocScrutinizer05(sth that still gives me headache - I'm not exactly convinced about USB being a good inter-chip bus for that purpose)12:36
DocScrutinizer05anyway since the bug been suspected somewhere in IRQ/WAKE controller, it wouldn't be limited to HSI at all12:37
DocScrutinizer05fun detail: when you try to debug such stuff and attach a Lauterbach to ETM/ETB, the whole ARM system usually doesn't enter sleep state ever12:38
DocScrutinizer05in the end we needed one of the customers to physically attach a probe to the HSI "IRQ" line (of course the name is different on HSI interface) which definitely wasn't easy to accomplish, to learn that the level of the signal was OK but the OMAP4 didn't act as supposed12:41
DocScrutinizer05damn, sorry for my confusing use of "we" - it's meant for my last employer where I developed modem chips, except 6 lines up where "we" means Neo900 group when I talk about Cinterion not doing patches in their modems for us12:49
whitequarkDocScrutinizer05: (sleep) even low-end stm32l has a secialregister12:49
whitequarkspecial register*12:49
whitequarkfucking asus and their fukking keyboar.d itsedaagain12:49
DocScrutinizer05my siemens laptop kbd recovered from lockups by suspend to disk ;-)12:52
DocScrutinizer05but sometimes the suspend to disk itself failed, when the amount of RAM to save was too large12:53
DocScrutinizer05whitequark: http://www.heise.de/hardware-hacks/meldung/Paperduino-Gedruckter-Arduino-auf-Papier-2106948.html you'll probably like that one12:55
DocScrutinizer05why mill PCB when you can draw then ;-)12:57
DocScrutinizer05YEEEHAAA!!!!! \o/  https://www.sparkfun.com/products/1204213:09
DocScrutinizer05this is not silly alu tape with conductive glue, this claims to connect upside sticky surface with downside sticky surface strictly vertically, thus making a nice alternative for multi-pin connectors13:12
DocScrutinizer05connector granularity seems ~0.5mm13:13
DocScrutinizer05that's soooo cool13:19
DocScrutinizer05contact resistivity: 0.3 Ohm. fine13:20
DocScrutinizer05isolation resistivity in the meg-ohms13:20
DocScrutinizer05that's the ultimate solution for a wide range of problems13:21
DocScrutinizer05whitequark: still alive?13:28
DocScrutinizer05wpwrak: you probably even could glue BGA chips to your PCB with that Z-axis-conductive sticky tape13:30
DocScrutinizer05umm, at least when pitch is >=0.5mm13:30
DocScrutinizer05or 0.4 actually?13:32
DocScrutinizer05>>3MTM Electrically Conductive Adhesive Transfer Tape 9703 is ideal for interconnection of flexible circuits with other flexible circuits (flex), rigid printed circuit boards (PCB) or LCD screens. Applications include polyester flex circuit splicing, keyboard manufacturing, LCD assembly and many others. Tape 9703 is also ideal for EMI/RFI shield and gasket attachment applications. Applications include EMI shields for displays and 13:43
DocScrutinizer05gasket attachment to EMI/RFI cabinets and enclosures.<<13:43
DocScrutinizer05AWESOME! https://www.sparkfun.com/videos#all/K3SPijvXtew/6413:47
whitequarkDocScrutinizer05: yeah still alive. it seems that keyboard malfunctions when it's cold (!)13:58
DocScrutinizer05watch that video! https://www.sparkfun.com/videos#all/K3SPijvXtew/6413:58
whitequarkthis is awesome14:00
DocScrutinizer05it's not perfect but pretty nice. i've been searching for something like this since maybe 20 years14:02
DocScrutinizer05and 3M has ~10 other similar Z-axis tapes with slightly different specs14:03
DocScrutinizer05not size, but type of tape14:04
DocScrutinizer05I'm quite excited about his14:04
whitequark3M makes a lot of neat things14:06
whitequarkperhaps you could discover some interesting ones just by browsing their catalog page by page14:06
whitequarkmaybe ORDER BY price DESC :)14:06
DocScrutinizer05in the 80s I had a HP34(?) calculator which I disassembled, and they used a very unique technique consisting of a flex PCB on foam, and the feet of the DIL chipts were bent forming U contacts that got pressed on that PCB14:06
whitequarkflex PCB?14:07
DocScrutinizer05yeah, 3M are really awesome14:07
DocScrutinizer05sure, never seen flex PCB?14:07
DocScrutinizer05even GTA02 had one, for WLAN14:08
whitequarkwell, I've seen flex PCB: copper on polyimide polymer14:08
DocScrutinizer05or was it BT?14:08
whitequarklike those flex cables in notebooks14:08
whitequarka darkish brown color14:08
whitequarkbut why put it in a calculator? regular pcb would be just fine there14:08
DocScrutinizer05yep, exactly that stuff14:08
DocScrutinizer05they used flex to get some elastic force against the DIL feet14:09
DocScrutinizer05the DIL chips sat in a plastic frame and got pressed against the FPC which itself sat on a PU foam14:09
DocScrutinizer05no soldering14:10
DocScrutinizer05that's when I started thinking about stuff like that Z-axis sticky tape14:10
DocScrutinizer05http://en.wikipedia.org/wiki/HP-34C  >>...  pressure-mounted circuitry (earlier variants 197983).<<14:17
whitequarkDocScrutinizer05: why can't they just solder it?15:24
DocScrutinizer05maybe they thought it's easier this way15:24
DocScrutinizer05ooh, and doublesided through-hole is a tad... annoying to design15:33
whitequarkis it?15:42
DocScrutinizer05derr yes, when your bottom side is filled with contacts for kbd clickdisks16:35
DocScrutinizer05so you'd not only need a PCB to solder your DIL chips to, you also need a 2nd PCB to build your kbd from, and then you need a connector between the two16:36
DocScrutinizer05whitequark: how do you like the paperduino?16:41
whitequarkDocScrutinizer05: seen it already16:45
whitequarkit's cool for learning I think16:45
whitequarkhas a "wow" factor16:45
DocScrutinizer05yeah, and a few nice ideas like that z-axis tape 16:46
DocScrutinizer05or the plotting of PCB with a conductive pen 16:46
DocScrutinizer05I think you could build a complete circuit without soldering in less than half an hour 16:47
DocScrutinizer05your CNC would make for a nice plotter16:49
DocScrutinizer05you even could drill holes for thru hole components and then use copper rivets to create via as well as mechanically rigid solder posts16:51
whitequarkDocScrutinizer05: or I could build a semi-automated machine which would do developing, etching and electroplating for me16:53
DocScrutinizer05hehe, how about multilayer created from sticky foil and an occasional Z-axis patch for vias, and the traces plotted with that pen. then stick on next foil layer16:53
whitequarkjust had that insane idea16:53
whitequarkreally, the only things I would really need to do manually is moving PCB from/to machines and aligning the photomask16:54
whitequarkno kill like overkill :p16:55
DocScrutinizer05what would you use for electroplating?16:55
whitequarkI want to try that process I linked you yesterday16:55
whitequarkusing thermal decomposition of copper hyposulphite (iirc) to deposit copper nanoparticles on exposed areas16:56
whitequarkthen using that as a prime layer for depositing more copper using regular CuSO4 bath16:56
DocScrutinizer05that's only half of the job16:56
DocScrutinizer05aah CuSO416:56
whitequarkwhat else would I use?16:57
DocScrutinizer05dunno, NFC about galvanics16:57
whitequarkwell, there's probably something. but CuSO4 is very simple and reliable16:58
wpwraksays he who wouldn't touch HCl ;-)16:58
DocScrutinizer05I just know it's usually extremely toxic and parcels with a galvanics lab on them are like "the zone"16:58
whitequarkwpwrak: what's up with CuSO4? it's non-corrosive and not really toxic unless you actively ingest large quantities16:59
DocScrutinizer05and even then... I seem to recall it's used for inducing vomiting16:59
whitequarkapparently it's only toxic to fish16:59
whitequarkooooh I remember, in high school some dude told me he tasted CuSO416:59
whitequarksays it was foul and metallic (duh)17:00
whitequarkLD50: 8-30g17:00
whitequarkon the same order as goddamn table salt17:01
DocScrutinizer05LD50 of water is even more ;-)17:01
DocScrutinizer05when you don't inhale it17:02
whitequark15L is definitely not LD5017:02
whitequarkno such thing17:02
whitequarkwell, it would not be very useful, that is17:03
whitequark90g/kg apparently, so... 7L for a grown man17:03
whitequarkwhich is exactly what I recall17:03
DocScrutinizer05quite possible17:03
DocScrutinizer05you could probably take the antidot: your 8g NaCl17:04
DocScrutinizer05I guess should be more17:05
whitequarkfwiw: NaCl LD50 is 3g/kg so 240g for a grown man17:05
DocScrutinizer05yeah, that's also more like what I'd expect is needed to turn 7L water into isotonic beverage17:05
DocScrutinizer05aka "detoxifying" it17:06
whitequarkhm... isotonic is 0.9% NaCl17:06
whitequarkand that is 0.33%17:07
DocScrutinizer05that be 60g then17:07
DocScrutinizer05for 7L17:07
whitequarkoops, missed a zero... that's 3.3% :)17:07
whitequarkyou're right17:08
DocScrutinizer05btw forget about saltless diet for lowering blood pressure. It's a hoax just like margarine and cholesterol17:10
DocScrutinizer05and salad and spinach17:10
whitequarkyeah, a healthy human regulates all electrolytes with impressive precision17:11
whitequarkit really only matters to those with impaired renal function17:11
DocScrutinizer05it seems like 95% of what health care at large is telling us to do is actually rather detrimental to our health17:11
whitequark"95% of health care at large" = "5% of health care, which wants to sell shit to you"17:12
DocScrutinizer05though i wonder who wants to sell which shit when suggesting to use less salt17:12
DocScrutinizer05for poison like NutriSweet it's obvious17:13
whitequarkmaybe it's a bit longer chain here. brainwash people with supposed health benefits => form a social group => exploit social group for great profit17:13
whitequarkha, talking about diets, there's some dick of a manufacturer who makes gummy bears with artifical sweetener aka sugar replacement17:15
whitequarksells them in 1-2kg packages17:15
whitequarkproblem is, that artifical sweetener is a really good laxative. if you eat >10 bears (and they're tiny), you better prepare for explosive diarrhea17:15
whitequarkand that is not an exaggeration17:16
DocScrutinizer05atrifical sweeteners are problematic in many ways17:16
DocScrutinizer05and one of the most significant effects is: increased appetite and constant hunger, resulting it people getting fat from low-shugar diet17:17
wpwrak(salt) well, if your kidneys are busy 'round the clock removing excessive salt you can pretty much count on them developing problems :)17:17
whitequarkwpwrak: there's no medical data indicating so17:17
whitequarkin fact, kidneys are "busy 'round the clock" regardless of what you do with your body17:18
whitequarkbecause kidney works like this: 1) push everything smaller than ~1kDa to urine 2) reabsorb useful stuff17:18
wpwrakso why do diabetics develop kidney problems ?17:18
whitequarksame reason they develop sores on skin, have heart attacks, poorly regenerate and everything else17:19
DocScrutinizer05same reason the develop problems with their eyes, or injured legs17:19
whitequarkit's a systemic illness17:19
DocScrutinizer05I guess17:19
whitequarkit wrecks the whole body pretty hard17:19
DocScrutinizer05yeah, increased sugar level in blood can do much damage17:20
whitequarktake your SoC and plug it into an unregulated power supply with ripple = half of Voutput17:20
wpwrakso you're saying sugar is more damaging to kidneys than salt ?17:20
whitequarkwpwrak: it's not sugar. it's the fact that cells' main energy source is improperly regulated17:20
wpwrak(sugar vs. salt) the "works like this" model suggests it should be pretty much the same ...17:21
wpwrakyou mean ATP ?17:21
DocScrutinizer05sugar even used to obliterate varicose vains17:21
whitequarkhow do you think the cell gets ATP?17:21
wpwrakthat would have been my next rethorical question ;-)17:22
whitequarkyou don't have any ATP in your blood, you have glucose17:22
wpwraki use "sugar" synonymous to "glucose" here17:22
whitequarkso cells convert glucose to ATP. but when there's no insulin, they can't, and that energy starvation is harmful to them17:22
wpwrak(since we're talking about the bloodstream, not the digestive process)17:22
wpwrakoh, but there is insulin17:23
whitequarkit depends on type of diabetes. type I means there's no insulin. type II means cells aren't sensitive to insulin17:24
DocScrutinizer05depends on type of diabetes, no?17:24
whitequarkend result to most cells is the same, no energy17:24
wpwrakhigh blood sugar levels themselves are an issue. insulin is usually present. e.g., in type II isn't often there anyway, in type I (and some type II) is gets added. but there are two categories: fast and slow-acting. the slow-acting one stays around for almost 24 h17:24
wpwrakwhitequark: untreated type I diabetics don't develop long-term complications. they die long before that. long-term complications are specifically a result of a treatment that does allow type I diabetics to survive for more than a few months/weeks/days17:26
Action: DocScrutinizer05 goes get some salty snacks17:26
whitequarksure. long-term complications are the result of still poor regulation of glucose uptake17:28
whitequarkbecause injections or even insulin pump cannot quite regulate it with the same precision17:28
whitequarkunsurprisingly, since they're both essentially open loop17:29
wpwrakwell, in the end, insuling-glucose mismatch and extreme glucose levels are strongly correlated because one rarely happens without the other (it's possible though, but normally not a frequent event)17:31
whitequarksure. your point is?17:32
wpwrakthat observing the two doesn't allow to determine the nature of the causal relationship (beyond suggesting that one exists)17:40
whitequarkuh... you sure aren't saying that everything about diabetes is a lie?17:41
whitequarkI mean, the causal relationship is rather well established with a heap of other facts17:42
larsccan you build a closed loop insulin pump?17:42
whitequarklarsc: absolutely17:42
whitequarkand I believe we'll see them soon, it's a long time coming17:43
larscsounds like a nice idea for a startup, if it doesn't exist yet17:43
larscalthough getting approval for use in humans can be hard to get, I guess17:44
whitequarkI'm sure such an obvious idea is being pursued already for years17:45
larscYou'd think so17:45
whitequarkshit just takes ages for approve... I heard 40 years is current time to market for really innovative tech17:45
larscgoogle autocomplete for "closed loop ins" is "...ulin pump"17:53
whitequarkobviously won't be available for at least 5-10 years17:56
Action: whitequark bought a better camera18:01
CYB3Rmy board is finally delivered http://storage8.static.itmages.ru/i/14/0210/h_1392055763_4671704_68100a828f.jpg18:08
larscI think something is missing there ;)18:09
OrnotermesCYB3R: where do you get the JZ4725?18:15
CYB3RI ordered it from China18:16
CYB3RJZ4725B is $5 for each IC18:19
Ornotermesany particular store?18:19
CYB3RNo, just random supplier on aliexpress18:20
Ornotermesoh, ok... have to look it up. the only decent linux capable chip i've found in the big suppliers that's not BGA are i.MX233 18:21
CYB3RThat is that store: http://www.aliexpress.com/store/91058618:22
Ornotermesthanks, do you have a web page (or something) about your board?18:23
CYB3RNot yet, but it is on github18:24
CYB3RThis is what I ordered from this store: http://pastebin.com/Dqt1EL3T18:24
CYB3RThe package is still passing the customs control18:24
whitequarkCYB3R: you're russian right?18:25
CYB3Rwhitequark: yep18:25
whitequarkso... how much have you ordered?18:25
CYB3Rwhitequark: 5 pieces JZ4725B and 5 pieces JZ475518:26
CYB3R10 NAND Flash and 15 SDRAMs18:26
whitequarkhmmm, and did you write the truth in customs declaration?18:27
CYB3RThe supplier claimed to write lower price18:27
whitequarkthey usually do. but what you needed here is a lower count, so it won't count as "wholesale"18:28
CYB3RAnd I'm currently not in Russia, so this is not russian customs control18:28
CYB3RI'm actually from Moscow, but now I live in Finland18:29
Ornotermesi live in sweden, and all small stuff from places like DX and ebay have never been caught in customs18:31
CYB3ROrnotermes: I only had troubles with DHL18:32
CYB3RThe fees are very jigh18:33
Ornotermesi had to pay castoms when i ordered from digi-key but they do all paperwork properly, and it's from USA18:33
CYB3RI also ordered some free samples (and still didn't use it, shame on me) and everything was okay18:34
OrnotermesCYB3R: what's your github username?18:35
CYB3RHere is a link to repo: https://github.com/CYB3Rhuman/JZ-YKS18:36
CYB3Rnew KiCAD board format is very strange. I just got this:  1 file changed, 4524 insertions(+), 41743 deletions(-)18:38
Ornotermeshmm... looks like i have a too old version18:41
whitequarkhey wpwrak, do you use something for static electricity?18:47
whitequarkmy old chair broke and the new one causes some crazy buildup, I stand up once and BOOM18:47
DocScrutinizer05van der Graaf?18:47
whitequarkDocScrutinizer05: I meant for reducing, but I'm not surprised you understood it the other way round :p18:47
whitequarkI know there are bracelets for that, but where to attach them?18:49
larscto your spine18:49
larscor do you mean the other end?18:49
whitequarklarsc: you would make a great GLaDOS18:50
whitequarkactually, you all guys would :D18:50
larscwe have these funny anti static things you put over your shoes, so they don't isolate you from the floor18:51
larscthe other anti static stuff is typically connected to ground18:52
Ornotermeswhitequark: i have a similar problem whan the weather is dry, and it make my computer drop the connection to my 3d-printer. Some times i spray it with water to prevent static build up18:52
larscas in ground from you ac connector18:52
larscin our office I sometimes feel like one of pavlov's dogs18:53
whitequarklarsc: but my floor is linoleum18:53
whitequarkbet it makes a nice dielectric18:54
larscwhitequark: there are special anti-static mats that you can put under your desk18:54
larscprobably cost some $ though18:54
whitequarkand it actually won't help me since I don't wear shoes at home18:54
Ornotermeswhitequark: you could experiemnt with liquid fabric softener18:55
whitequarkand it is easier to take socks off than to invent some fancy antistatic thing18:55
Ornotermeswhitequark: http://en.wikipedia.org/wiki/Fabric_softener#How_it_works18:56
larscI think you can also connect the bracelet to your radiator18:56
larscor other pipes18:56
larscbasically anything that you can assume as a conductive connection to ground18:57
whitequarkradiator, hmm... but it is painted18:57
whitequarkand iirc there is some regulation that states that radiator must not be used as ground connection18:57
larscI don't think anybody cares about your bracelet19:00
larscit's not like it is high voltage or high current or anything19:00
larscthings might go bad if the radiator is connected to the lightning rod ;)19:01
Ornotermeslarsc: thats why you put a fuse in between them :P19:02
larschere at the appartment complex where I live the lightning rod is connected to a metal fence in front of the building19:04
larscI wonder if that is safe19:04
whitequarklarsc: no, I think it's because radiator might not actually be well-grounded19:06
Ornotermesi'd keep my feet close together if i'm near that during a lightning storm19:06
DocScrutinizer05whitequark: nope, it's most likely to avoid corrosion from phantom currents flowing thru the radiator tubing, and also for gas heating it's an absolute nogo: gas pipes are strictly forbidden to get connected to anything but a central potential equalization19:11
DocScrutinizer05larsc: I guess the fence is connected to the lightning protection ground19:13
DocScrutinizer05mandatory for all huge metallic objects outside19:13
DocScrutinizer05your gutter and downpipe will be as well, if made from metal19:14
DocScrutinizer05basically even your bathtub should be connected to the central potential equalization19:15
DocScrutinizer05and probably you Protective Earth of your mains grid also shall be on Potential Equalization19:18
DocScrutinizer05I'm not absolutely sure about the latter, since OTOH PE and Neutral shall get connected in your central mains distribution19:19
DocScrutinizer05which in turn would create phantom current in parallel to your house's mains Neutral along the Ground of potential eq.19:21
DocScrutinizer05which might corrode the anode19:21
DocScrutinizer05all that is an art rather than a science19:22
DocScrutinizer05in the old days you were allowed to connect 3 mains outlets to a RST-N 4wire in star config. Nowadays houses burn down from Neutral overheating when you connect non-linear loads to all three outlets19:25
larscbut they didn't back then?19:27
DocScrutinizer05since for non-linear loads the currents from R, S and T don't add up to 0 on N, but can rather add up to 3 times the current on either of the life wires19:27
DocScrutinizer05back then you had no non-linear loads19:27
DocScrutinizer05worst thing you could have was an inductive load, and that wasn't so much of a problem19:28
DocScrutinizer05a completely inductive load on R and a Ohm-ish load on S could at best add up to 1.5 maybe on N19:29
DocScrutinizer05and you hardly could create an equally massive 100% capacitive load on T19:29
DocScrutinizer05nowadays with switched mode PSUs you got completely different current shapes that have almost no relation to the voltage applied to them19:30
DocScrutinizer05so you end up with a current on N that's possibly 3*f and the sum of R, S and T currents19:33
DocScrutinizer05whitequark: usually you attach those antistatic wristbands to PE of next mains outlet19:36
DocScrutinizer05via a secure 10M R19:36
DocScrutinizer05secure means 400V rating19:37
DocScrutinizer05possibly higher, so it doesn't "break down" when you touch something that's carrying static charge19:39
DocScrutinizer05mind you: once there's a tiny spark creating an ionized path across the resistor, even 220V can flow across that path 19:40
DocScrutinizer05sth that the resistor is meant to protect you from, to mitigate worst case19:41
wpwrakwhitequark: (lie) no, of course not. just that it's not an argument to support your claim that excessive levels of salt are okay.21:28
wpwrakwhitequark: (new camera) and no exif ? that's lame :)21:30
wpwrak(static electricity) not really needed around here. today, the air was so humid i had trouble breathing when walking outside ... (first time it was *that* bad)21:33
wpwrak"GLaDOS) hah, now i know where that portal meme comes from :)22:00
dos1speaking of memes...22:51
dos1https://www.youtube.com/watch?v=1FTGFpb1f_0 - ncurses-based port of Flappy Bird running on Ben :)22:52
wpwrak"unlisted" ? i think the world deserves to know !22:56
wpwrakyou should post that on heise.de22:56
dos1wpwrak: I've uploaded it few minutes ago ;) it's public now anyway :D22:59
nicksydneywpwrak / DocScrutinizer05 : http://www.designfaves.com/2014/02/stir-kinetic-desk-fit-work ... make one of this and sell 1/2 price :023:08
wpwrakdos1: you should make it executable23:25
wpwrakdos1: also, if it does this: _curses.error: use_default_colors() returned ERR23:26
wpwrakthen you have to  stty sane23:26
wpwrak(trying to run it on my desktop linux)23:26
wpwrakwith TERM=xterm it runs in monochrome ... but doesn't show the obstacles23:30
dos1wpwrak: interesting, here it works without any problems23:35
wpwrakyou mean it finds the colors ? or that it displays something visible if the terminal is monochrome ?23:42
wpwrak(stty sane) naw, that's not what i meant ;-) what happens is that  curses.curs_set(0)  fails and then the program exits, leaving the tty in raw mode23:44
wpwrakso after the failure one has to "clean up" with  stty sane23:44
dos1yep, already had the "aha" moment :D23:44
wpwrakyou don't need it before setup, and you also don't need it after normal exit23:44
dos1wpwrak: it should be now playable on monochrome vts23:55
dos1at least works here with TERM=vt100 :)23:56
wpwrakyeah, here too ! thanks ! :)23:57
dos1some parts of the code were written on ben itself :D23:58
--- Tue Feb 11 201400:00

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