#qi-hardware IRC log for Saturday, 2016-06-18

arossdotme-planbwhen ya look at this, how good or bad do you think it is? i think some people on diyaudio.com forum said it was good :/. it does have the sleep and mute pins exposed via connector which is the attractive feature :) http://www.aliexpress.com/item/TPA3116-digital-class-D-amplifier-50W-50W-with-silent-sleep-design-with-heat-sink/32654679975.html?ws_ab_test=searchweb201556_0,searchweb201602_2_10037_10017_507_10032_401_10040,searchw03:10
arossdotme-planbeb201603_3&btsid=c43559b8-b84a-4ded-b95c-a11eb830efa103:10
wpwrakheroic rework of he day: unsoldered a worn-out micro USB connector (from anelok) and replaced it. amazingly, the board still works. but my disdain for micro USB only deepens ...18:15
DocScrutinizer05well, prolly for anything micro-*18:22
DocScrutinizer05that's the problem with all micro things: they are really small18:23
EDTwpwrak: you can't hear it, but I'm clapping for you18:25
wpwrakEDT: thanks ! ;-) the trick was to let "chip-quik" do its magic. tossed a huge drop on the connector (on the shield), heated it at 400something C for a minute or so, until it felt unwelcome and slid off the board.19:40
wpwrakDocScrutinizer05: naw, it's the specific suckiness of USB micro. they somehow messed up the angular tolerances, by trying to make them a bit too short. mini was fine and USB C also seems to correct this problem.19:42
wpwrakthere's probably just 0.5 mm or so missing, but that's enough ...19:42
whitequarkwpwrak: chipquik si fucking awesome19:55
whitequarkI've reworked complete boards with chipquik two times or even more, with no lifted traces19:55
DocScrutinizer05which product of chipquik?20:09
DocScrutinizer05there is quite a number of different chipquik products  http://www.chipquik.com/store/20:10
DocScrutinizer05ooh I see, alloy with a melting point <100°C20:13
DocScrutinizer05wood's metal20:14
DocScrutinizer05rather Roses metal I hope20:15
DocScrutinizer05https://de.wikipedia.org/wiki/Roses_Metall20:17
EDTSo I just flashed my nanonote to the most recent version, and it doesn't have emacs installed or available in the repos20:24
EDTanyone have experience with emacs and the nanonote?20:25
whitequarkno20:26
whitequarkDocScrutinizer05: its an indium alloy20:26
DocScrutinizer05indium? wow20:26
whitequarkPb-free20:26
whitequarkI think it's BiInSn but exact composition is a trade secret20:26
whitequarkindium is not that expensive anyway20:28
whitequarkI have a kilo of InSn20:28
DocScrutinizer05Tin, Silver, Bismut, Antimony, Indium, Copper20:33
DocScrutinizer05http://www.chipquik.com/msds/SMD4.5NL.pdf20:33
DocScrutinizer05hmmmm :-)  Solder Paste no clean Sn42/Bi57.6/Ag0.4 Low Temp 138C 20:40
DocScrutinizer05though.... will prolly fall off the PCB before chip reaches max operating temperature ;-P20:41
DocScrutinizer05sth completely off topic: I recently learned cochlear implants have 12 to 28 electrodes, also because they can't get more wires into the whole thing without it getting too rigid. How about one-wire serial setup with active component for each electrode?20:52
DocScrutinizer05they also said that more electrodes would be useless since crosstalk gets too large, but I guess that could get fixed by carefully modelling a matrix that creates compensation current on neighbor electrodes20:53
whitequarkhaahha cochlear implants20:54
DocScrutinizer05haha?20:55
whitequarkfun fact: for all the techie talk about 'uploading' we still have no idea how hearing works20:55
DocScrutinizer05sounds about right20:55
whitequarki'm sometimes chatting with a friendly neuroscientist who does a masters in that area20:56
DocScrutinizer05they recently implemented MP3-alike audiophysiological data reduction to the cochlear implant data, improving legibility of the created sensation drastically20:57
whitequark*nod*20:57
DocScrutinizer05(crosstalk)  even when we limit the count of concurrently active electrodes to maybe 24, I guess it still would make a huge difference when you could choose the position of those active electrodes from a say 500 existing electrodes21:00
DocScrutinizer05a simple one-wire daisychain design (actually maybe 3 or 4 wires then) should facilitate a way higher number of electrodes21:02
whitequarklemme ask him21:02
DocScrutinizer05higher than 2821:02
DocScrutinizer05the needed node chips are prolly as small as 0.2*0.2*0.05mm or somesuch21:03
DocScrutinizer05they basicaly could integrate the electrode already21:04
Action: DocScrutinizer05 should file a patent ;-P21:05
DocScrutinizer05but like always, I bet somebody else already did21:05
DocScrutinizer05I already think I know the electrical macro design: GND wire, V+, V- as bus. Data as daisychain from electrode to electrode, where the electrodes actually are chips21:09
DocScrutinizer05maybe even 'in plane switching' with two adjacent electrodes (integrated into one chip) could work to reduce crosstalk21:12
DocScrutinizer05whitequark: (ask him) yes, please do. Let me know what he thinks21:15
DocScrutinizer05maybe we could even start a project21:15
whitequarklol nope, have you seen the approval time for medtech?21:29
whitequarkhow does that count against your life expectancy? how does that count against *my* life expectancy?21:30
DocScrutinizer05yeah21:37
DocScrutinizer05however an intriguing topic21:37
Action: DocScrutinizer05 reads about Anisotrope Conductive Adhesive, used in flipchip mounting21:38
DocScrutinizer05(approval time) the idea for sure wouldn't be to start a medical supply company. Rather to sell the concept to such company21:39
whitequarkthat doesn't really happen because concepts are worthless21:41
DocScrutinizer05prolly21:41
whitequarkask a grad student who really doesn't want to study21:41
DocScrutinizer05I might try to contact the university scientists that do R&D on that stuff21:42
whitequark" Error: Multiple definition of the extension constructor name Data_source_error.21:43
whitequarker21:43
whitequarkwrong paste21:43
DocScrutinizer05btw the topic is somewhat similar to the also really pathetic retina implants they test nowadays21:44
DocScrutinizer05which have a resolution of 8*8 or somesuch21:45
DocScrutinizer05for the latter they are evidently still in prototype state, with different issues than resolution21:46
DocScrutinizer05cochlear implants are well established21:46
whitequark"afaiu the major restriction on electrode counts is actually crosstalk between electrodes - plus more electrodes at the same stim current means more charge transfer which is bad for both the electrodes and the body. Plus these active devices would have  to operate inside the cochlea if I understand the intended design - complicates. Pictures: http://www.medel.com/blog/close-up-with-cochlear-implant-electrode-arrays/"21:47
DocScrutinizer05yes, that sounds all right but not to the point21:48
DocScrutinizer05in that picture imagine those 12 electrodes with their separating spaces in between would each be broken up into - say- 20 segments of 'partial' electrodes of which an arbitrary consecutive sequence of 10 would get activated as one 'macro' electrode, acting exactly on same physical parameters like the recent discrete design, just the virtual position of such virtual electrode could get shifted in 10 steps back and forth21:52
DocScrutinizer05then there's also the idea of "in plane switching" where the electrode potential isn't vs "GND" but vs twin electrode 't 21:53
DocScrutinizer05>>active devices would have  to operate inside the cochlea<< is an issue which gets solved on a technical level, unrelated to medical concerns21:54
DocScrutinizer05I think flipchip will just work fine when embedded into that transparent silicon21:57
whitequarkthe majority of issues with implants are technical though...21:57
DocScrutinizer05sure, I know21:57
DocScrutinizer05it's an endless nightmare of "damn this electrode material corrodes" etc21:58
DocScrutinizer05but those issues are not really entangled with the core idea of daisychaining and having more and smaller electrodes21:58
DocScrutinizer05IPS could help a lot with crosstalk, and I'd bet nobody considered that so far since it wouldn't help anyway as long as you need to double the number of electrodes for it and you can't get more than 28 of them because of technical reasons22:00
DocScrutinizer05I know the stimulus current must not exceed rather low levels or there will be long term damage to the nerves22:02
DocScrutinizer05but that's a concern that gets handled on a calibration level, not at design level22:03
DocScrutinizer05s/ on / at /22:03
DocScrutinizer05IOW more electrodes doesn't per se mean higher stimulus current22:04
DocScrutinizer05au contraire IPS might mean you get away with way lower stimulus current22:06
DocScrutinizer05and crosstalk outside of the area of the twin electrodes should be drastically reduced22:07
DocScrutinizer05umm this http://i2.wp.com/www.medel.com/blog/wp-content/uploads/2013/10/Hüttenbrink-011.jpg?w=1140 already looks like they do IPS22:14
DocScrutinizer05no?22:14
DocScrutinizer05yes, the video clearly shows pairs of electrodes22:16
DocScrutinizer05ok, didn't know they already use IPS22:17
DocScrutinizer05http://i1.wp.com/www.medel.com/blog/wp-content/uploads/2014/01/Top-of-electrode.jpg?resize=600%2C450 however doesn't suggest they actually use 24 wires for that 12 electrode pairs22:20
DocScrutinizer05whitequark: could you ask your contact what's the deal with thiose 24 electrodes in pairs of 2, vs seemingly only 12 wires at the end where the processor connects?22:43
--- Sun Jun 19 201600:00

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