#qi-hardware IRC log for Sunday, 2015-02-15

DocScrutinizer05watch out! there's a pitfall with this: HNP defines active DIScharge of VBUS to basically transfer data over power. So huge buffer capacitors might be problematic even behind a 44R series resistor00:35
DocScrutinizer05when in a chinese hub some big caps are not equipped, I'd first think of rithless optimization, but in this particular case maybe they even thought having them was nice and then noticed they actually spoil stuff00:37
DocScrutinizer05ruthless*00:37
arossdotmewow01:31
arossdotmedata over + never would have thought...01:31
arossdotmethx01:31
wpwrakxiangfu: heya, did you see what i wrote on adjusting the threshold ? ("maybe start with 500 [...]")04:12
wpwrakin any case, things will be more fun once you have a real case. it changes quite radically how the device feels.04:15
Action: xiangfu check the irc log04:20
xiangfuwpwrak: how to reflash the anelok-2014.bin. do I have to reflash the boot-bin first. then using dfu?04:22
wpwrakyou should be able to use dfu directly. you already have a good boot-2014.bin, from the combined binary04:23
wpwrak(the combined binary is really just what the flash of anelok looks like when you first flash boot-2014.bin and then dfu anelok-2014.bin)04:24
xiangfuwhen I plug the anelok. the led was on. like for 1 sec. then off. from the dmesg. the usb got disconnect.04:26
xiangfuso I should using dfu when the led was on? right?04:26
xiangfuYEAH04:27
wpwrak;-)04:27
wpwrakyou can use "make update" to make this more comfortable: "make update" prepares the transfer, then waits for anelok to enumerate. then you can reboot or reconnect anelok, and the dfu will start automatically.04:28
xiangfugreat.04:34
wpwrakjust added two new parameters to devcfg: "scale", which is the scaling of the touch values, and "offset", which is an offset that's added to the scaled value04:56
wpwrakso if the slider doesn't want to move all the way to top and bottom, increase "scale" (default: 96). if it moves too quickly, decrease it. and if the center seems to be way off, adjust offset (default: -10)04:58
wpwrakand a twist: if you set the offset to 0, then it'll pick the default value since 0 is used internally as "not set". fortunately, nobody would notice a small difference in the offset, so one can just use 1 or -1 in this case.05:08
xiangfuthe cursor flickers quickly. update the thresh from 400 to 10005:29
xiangfunow ...05:29
xiangfuhmm... the anelok take like 2 mins to boot.05:30
wpwrakyou mean the time between a reset and until it responds touch input ?05:31
wpwrakresponds TO touch input05:31
xiangfuYes.05:33
wpwrakyeah, that's the time it takes for the calibration to settle. the rest of the system is up and running almost instantly, but it takes a while until it understands the touch input05:35
wpwrakand the time should be more like 30 seconds. just make sure you don't touch the sensor during that.05:36
wpwraklemme try to add a fast initialization. probably won't work (i've tried it before), but let's see ...05:36
xiangfuwe can make the led one until it finish setup the touch input.05:44
xiangfuhmm... threah = 50000 right now. the cursor still blinking.05:44
wpwrakwow. that's weird.05:45
wpwrakat 5000 i would expect the thing to be quite dead :)05:46
wpwrakbut a blinking cursor (if there is a finger) means the threshold is too high05:46
xiangfuok. I made a mistake. dnow. changed to 5.05:48
xiangfuhmm... still blinking when there is a finger.05:48
wpwrakwith 5, it should detect the finger when it's ~1 cm away :)05:49
wpwrakdid you put some plastic on top of the sensor ? if yes, how thick is it ?05:50
xiangfuno. nothing on PCB.05:50
wpwrakah, you should add some ~0.5-1.5 mm of plastic05:51
wpwrakelse, the thing will be far too sensitive05:51
xiangfuok.05:51
Action: xiangfu finding some plastic...05:51
wpwraki think i found a fast calibration that works :) not it should be ready after just a few milliseconds06:13
wpwraknoW06:13
xiangfuI found a plastic. 06:13
xiangfuit works fine at thresh = 25006:14
wpwrakif the fast calibration didn't work, then anelok may turn on and off on its own for a while (until the regular calibration corrects the issue)06:14
wpwrakwhee ! :)06:14
xiangfubetter on 350.06:15
wpwrakin general, if the slider acts weird, let anelok rest for ~30 seconds and the problem should correct itself06:15
xiangfuok. now I am ready to write some C code. :)06:15
xiangfuok.06:16
wpwrakfor the threshold, higher is better since you then only get samples where the finger is already firmly pressing against the plastic06:16
wpwrakif the sensor triggers when the finger it still away or only lightly pressed, then the finger will deform when completing the push, and this will look to the sensor like a change in position. so it's best to avoid that.06:17
wpwrak(code) great ! :) and welcome to the exclusive club of anelok users ! ;-)06:21
wpwrakah, regarding the fat fs. best to call the directory of the file system driver fat/ or fatfs/. ff/ is a little too terse and generic, even for my taste :)06:25
xiangfuok. got it.06:27
xiangfunow we have 4 full working board.06:54
wpwrakkewl :) until two hours ago it was 2 world-wide, one hours ago it was 2+1 = 3, now 2+4 = 6. if we keep this up, the planet will overflow with aneloks in a day :)06:55
xiangfuwill upload more pics later.06:56
xiangfu:)06:56
wpwrakexcellent !06:56
xiangfuI think I can start the case work after Chinese New Year.06:56
xiangfuand making more boards.06:56
xiangfuI have bought 50 boards components. but I only have 10 PCB.06:57
wpwrakthat's good. it seems that we'll need to change the power supply scheme anyway, then you can fix that directly in the new boards06:59
xiangfuYes.06:59
wpwrakwikipedia says cny is feb 19 this year. so when is the holiday period ? the next two weeks ?07:02
xiangfu18 ~ 24 Feb.07:10
xiangfu18 Feb, 12:00, 19 Feb, 00:00 is the time for New Year.07:11
xiangfu:)07:11
wpwrakinteresting that the vacations aren't aligned to the week. but the timing is good - i'll probably try the rework next weekend, then we'll know whether it works or not.08:05
DocScrutinizer05wpwrak: you should reach 7 at least! :-)  http://pmav.eu/stuff/javascript-game-of-life-v3.1.1/11:15
xiangfuDocScrutinizer05: I got those 4 N900 phone.11:22
DocScrutinizer05xiangfu: wow, you guys are fast there :-)11:23
xiangfuit is like 70%-new, 80%-new, 90%-new and brand new.11:23
DocScrutinizer05well, could we seel those parts to customers?11:23
xiangfuthe brand new is not real brand new. but it really good great.11:23
DocScrutinizer05sell11:23
DocScrutinizer05or would they frown?11:23
xiangfuI mean: but it really looks good (the taobao shop declear it is brand new)11:24
xiangfuthe 'brand new': Yes.11:24
xiangfuothers: hmm.. looks bad. I will upload picture tomorrow.11:24
DocScrutinizer05:-) thanks11:24
DocScrutinizer05you seen chanlog? you were not here when I posted some answers to your question11:25
DocScrutinizer05regarding what disassembly needed11:25
xiangfuyou will see the different. the 'brand new' phone have a box. battery, usb cables. just like the realy new one.11:25
xiangfuDocScrutinizer05: Yes. I saw them.11:25
xiangfuabout the QA.11:26
DocScrutinizer05(new) we are most interested in display and digitizer/touchpanel11:26
DocScrutinizer05the case is cheap11:26
DocScrutinizer05we'll need own boxing anyway, and cable and charger are cheap as well, if we add some to the product at all - probably we will though, since this shall be a convenient complete product11:27
xiangfuyes.11:28
xiangfuI will disassmble them tomorrow.11:28
DocScrutinizer05until "picture #16" omly 11:28
DocScrutinizer05only11:29
xiangfuthose 4 phones have three different layout. one is English, one is Franch(I guess). the other one I don't know.11:30
DocScrutinizer05hehe11:30
DocScrutinizer05yeah, maybe that's actually great since you can't get those "exotic" layout keymats as spare parts anymore11:30
DocScrutinizer05and some of our customers want those exotic ones nevertheless11:31
DocScrutinizer05the 15bucks spare housing kits come with QWERTY keymat only11:31
DocScrutinizer05now if this pans out, we solved a huge part of headache in Neo900 component sourcing. One problem left over: spacerframe11:34
DocScrutinizer05xiangfu: maybe you can help there as well. This would be a part that must get built to order, it doesn't exist yet11:34
DocScrutinizer05smal plastic frame11:35
DocScrutinizer05(or steel, zinc, alu, whatever)11:35
xiangfuwhat is that?11:36
DocScrutinizer05http://talk.maemo.org/showthread.php?t=9419111:36
xiangfudo we have a design file for it?11:36
Action: xiangfu clicking...11:36
DocScrutinizer05http://neo900.org/stuff/joerg/spacerframe/sketch.pdf11:37
DocScrutinizer05(design file) not yet11:37
DocScrutinizer05the CAD dudes are reluctant11:37
DocScrutinizer05werner did the great scans exactly for that purpose to start 3D CAD11:38
xiangfusomething under the 'QWERTY COVER ASSY' on your neo900 phone right?11:42
DocScrutinizer05yes11:42
xiangfuok.11:43
DocScrutinizer05replacing the small frame around the keymat with a slightly larger one that also adds a 2.8mm to the height of the bottom case shell11:43
xiangfuok. understand now.11:44
xiangfuI don11:44
xiangfuI don't now. I need ask my friend, he is doing this very well.11:44
xiangfus/now/know/11:44
DocScrutinizer05great! :-D11:44
DocScrutinizer05for now we have ~400 preorders, every customer will need one of those11:45
DocScrutinizer05we might want to build a 1000 already11:45
DocScrutinizer05more if the project maybe gains pace11:46
DocScrutinizer05(we'd also need more mech parts then)11:46
DocScrutinizer05(LCD, digitizer and slider mech etc)11:46
DocScrutinizer05estimations about ratio between those who are willing to preorder and those who would just order when product gets available is in the range from factor 3 to factor 1511:47
DocScrutinizer05but well, that's dreams about future. for now we need to buold a 500 devices (20% for yield, spares etc)11:48
DocScrutinizer05build*11:49
xiangfuok. 11:54
xiangfuI think 1000 is easy and fast by looking around in Shenzhen. for large amount. we need find a factory or supplier doing that.11:55
xiangfuneed fix the spacerframe first. :)11:56
wpwrakDocScrutinizer05: (game of live) reach 7 ? nice implementation12:43
DocScrutinizer057 is critical mass it seems12:43
DocScrutinizer05for <7 it won't grow12:44
DocScrutinizer05or at least I don't know of a pattern that would grow and consists of <=612:44
DocScrutinizer05((game of live)) hehe, same typo I did and thus had a hard time finding any program or webpage12:45
wpwrak(cad guys reluctant) reminds me of when i did the ben scans. there were also a lot of people said they had fancy equipment, would do better scans, even CAD, etc. i wondered whether it would even make sense to do my very slow scans then. wolfgang said not to worry, i'll still be the only one who actually does something. he was right :)12:46
DocScrutinizer05finally the search term "conway" and googles fuzzy search helped me out12:46
wpwrak(typo) the joy of just having gotten up :)12:46
wpwrakwow, gitorious sends me the whole set of my (!) changes when someone updates their repo that's clones from mine ;-)13:06
wpwrakit's a nice backup, though. just in case :)13:07
--- Mon Feb 16 201500:00

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