#qi-hardware IRC log for Sunday, 2014-01-19

whitequarklook: http://hackaday.com/2014/01/17/1-coin-cell-charger/?utm_source=feedburner&utm_medium=feed&utm_campaign=Feed%3A+hackaday%2FLgoM+%28Hack+a+Day%2910:31
whitequarkso apparently LR* cells are chargeable! I always wondered about that, their internal structure seemed pretty damn similar to chargeable Li batteries10:31
nicksydneywpwrak: fancy of adding a charger for the batter as per the hackaday.com link :)11:43
whitequarkactually, now that I think about it, the RTC battery on old-ish motherboards was just such a coin cell and it certainly was being charged11:47
whitequarkso apparently they can survive such mode of operation for years11:47
whitequarklarsc: why does your hostname refer to a bear?11:50
whitequarka... sea bear, apparently?11:50
whitequarkmy latin is rusty11:51
wpwrakah, polar bear. in german, eisbaer ("ice bear"). maybe it's a reference to a song that at its time was quite famous: http://www.youtube.com/watch?v=cTuTc_liKS411:54
wpwrak(battery charger) sounds like a nice little standalone project :)11:54
wpwrak(polar baer) the lyrics are "i'd like to be a polar bear in the cold arctic. then i wouldn't have to shout/scream anymore. everything would be so clear." and "polar bears never have to cry." (supposedly because the tears would freeze, which would be kinda inconvenient)11:56
whitequarksounds like something DocScrutinizer05 would say11:57
whitequarkthat being said, the freezing point of tears can be considerably lower than 0°C, since they're salty11:58
whitequarkand bears certainly have to keep the mucous membrances from freezing over, either way11:59
nicksydneywpwrak: another USB design http://ctrlv.in/285034 based on this schematic http://www.555-timer-circuits.com/police-siren.html :) ... not pretty 12:02
wpwrakthey also can go for a swim at times. so yes, they are able to deal with water.12:03
nicksydneywpwrak: oh and also laying out the track by hand not auto-routing :)12:03
whitequarkwpwrak: well, if water is already liquid, why would it freeze on them?12:03
wpwrakhere in argentina they even get to spend a lot of time in water if they live in captivity. we have one left in a zoo where it is exposed to peak temperatures of up to 40 C. of course without an overly fancy cooling system. one year ago, the polar bear in the zoo of buenos aires died, due to too much heat and getting irritated by the christmas/new year fireworks.12:05
wpwrakwhitequark: (liquid) if it loses heat to the environment quicker than thermal leakage of the body can warm it, then it could reach freezing temperature. e.g., consider windchill.12:06
wpwraknicksydney: circuit is starting to look good :) but your footprints are weird - they have silk screen on the pads. if a factory uses that "as is", the boards will be unusable12:10
wpwrak(of course, in diy, you typically  just ignore the silk screen)12:10
nicksydneywpwrak: i'm testing the footprint from this site http://smisioto.no-ip.org/elettronica/kicad/kicad-en.htm to see how it looks like12:10
whitequarkwpwrak: right. but also consider this: how would it lose heat?12:11
wpwraklet's hope the others are better :)12:11
whitequarkit can lose it by radiation, convection and evaporation of water12:11
whitequarkand I'm pretty sure the latest effect would be the strongest in short timespans. but! evaporation of water makes the solution more concentrated, hence lowering the freezing point even more12:12
qwebirc37386hi there, i have  short question regarding 3-Way CrossFireX and if it's in any way related to CPU12:13
qwebirc37386could somebody help e about it, aparently it's an obscure search for google12:14
nicksydneywpwrak: learn something new when doing layout for a single side board like that using +5V's and GND's pad on the side really help on the routing12:14
nicksydneywpwrak: was thinking of the concept (when doing the routing) 1-to-many in database world :)12:14
qwebirc37386i think i asked in the wrong chat room right ?12:17
whitequarkqwebirc37386: yes12:18
qwebirc37386uups, sorry about that then, i'll be more carefull, anyway thanks12:19
qwebirc37386nice day to you12:19
whitequarkhe was quite nice12:20
apeletelarsc: hi there12:31
apeletelarsc: wondering which value should be assigned to dst_maxburst member of struct dma_slave_config:12:35
apeletelarsc: looking at datasheet, I can't figure out if that should be half the fifo size (8 words ?) or something else12:37
wpwrak(http://ctrlv.in/285034) hmm, why do the vias (!) have silk screen rings around them ? did you put components for them ?12:44
wpwraknicksydney: also, your vias do noting ;-) a) you board is single-layer, you there's nothing they could connect to12:45
nicksydneywpwrak: it's not vias it's a pad .... 12:45
wpwrakb) even if it was double-sided, you wouldn't need all these vias. you probably copied the concept from RF boards. there, they're needed because of RF. but your board is much much slower ;-)12:45
wpwrakah, that's why12:46
wpwrakwell, anyway. not needed :)12:46
wpwrakjust draw a straight trace and you're good12:46
nicksydneywpwrak: was having difficulty running the track between the 2 chip and was wondering how to make it easy so that i can run the track nicely...so i put a pad vertically and put them as +5V and GND so I can run the trace neatly instead of going around too long as it was super duper ugly :)12:47
wpwrakfor components that have a specific orientation, e.g., U1 and U2, you may want to put some sort of marker on copper. e.g., i often put a little triangle (made of traces not connected to anything else) a bit outside the corner of pin 1.12:48
nicksydneywpwrak: when using the pad I can run the track neatly and nicely otherwise it will look like a sphagetti :)12:48
wpwrakthat way, i can see how to turn the part when soldering and i'm less likely to mis-rotate it12:49
nicksydneywpwrak: (marker) good point will do that12:49
wpwrak(pad) but you still don't need them :) you can just branch off a trace. when drawing the trace, you have to begin somewhere the trace connects to so that kicad will know which net you're routing, but that's all.12:50
wpwrakand if you find that restriction bothersome (it can sometimes be), then you can just turn off DRC while doing that12:51
nicksydneydidn't think about turning off DRC...will try that 12:51
wpwrakbut in general there's no issue. you can also end a trace when you're in the area you want to go to and then add more things later12:52
wpwrakturning off DRC is usually a bad idea. but when you really need to, then it's nice that you can.12:52
wpwraktypical scenarios are when you have an orphaned trace that no longer has a net associated with it and try to connect it again12:53
wpwrakDRC won't let you do that. so you can turn it off, make the connection (be very careful :), then turn it on and run a DRC check (the bug symbol). kicad will then notice the connection and update the trace's net information. so you'll see net name on the trace in small print and can check whether it's what you intended to get.12:54
nicksydney(marker) to do this do you use the "Add text  on copper layer..." right ? .... i came out the same layer as the copper....can you change the colour ?12:55
wpwrakanother useful feature is "delete segment" if you want to rearrange a trace without deleting all of it, you delete the last segment of the bothersome part, before it reaches the good part12:55
wpwrakthen you can "delete" the entire bad part(s) without affecting the good part12:55
nicksydneyok found it need to add keepout area and add the text in :) ..duh !13:18
wpwrak(marker) you don't want to change the color ;-) after all, you want copper. since this is what you'll already "print" with toner transfer. so it's no extra work. you could of course also transfer a silk screen (where you already have such markers), but that's messy.13:20
wpwrakyou could also just go over the board after etching and manually mark things. but that's lame :)13:20
wpwrakand i wouldn't use text for that. text needs a lot of space. if you look here: http://downloads.qi-hardware.com/people/werner/anelok/tmp/ybox-draft-20140118.png13:21
wpwrakthe chips have little copper triangles in the corner with the triangle on the silk screen, just a little farther from the center (so that they remain visible when placing the chip, and also to make sure they don't make any accidental contact)13:22
nicksydneyi put an 'o' http://ctrlv.in/28504913:23
wpwraki'd put it somewhere where you can still see it when the chip is placed. but the O shape is good.13:24
nicksydneythis is how it looks like when plotted http://ctrlv.in/28505113:27
wpwraknice :) what trace width did you use ?13:27
nicksydney0.45 mm :)13:29
wpwrakah, i mean for the "o". that's about 0.15 then ?13:30
nicksydneybut you see some traces are still small as it's not even in sizes .. don't know why13:30
nicksydneyfor the '0' it's 0.1mm13:30
wpwrakah, i was wondering about that. maybe you did something strange in the design rules. that's not something that normally happens.13:30
nicksydneythis is what i set in my DRC ... http://ctrlv.in/285054 .. http://ctrlv.in/28505513:32
wpwraksetting look safe enough. maybe you drew part of it with other settings, then changed them13:34
nicksydneylet me redraw them and see13:37
nicksydneyinteresting project http://www.indiegogo.com/projects/ice-xpc-a-modular-computer-as-small-as-your-phone13:42
nicksydneywhat I like about Neo900 is the form factor is very much suitable for any other project/product :)13:45
wpwrakyeah, it'll be a good project to scavenge from :)13:50
wpwrakpity they're quite constrained in their choice of chips. but if the thing flies well, maybe they'll make a neo1000 where they can take more chances :)13:51
nicksydneywpwrak: you know looking around I haven't found any open source project in similar kind of form factor as neo900....it is an attractive proposition if another non-mobile product can be created out of neo900 13:53
wpwrakyeah, neo900 takes a very smart approach: do something that seems overwhelmingly hard by reusing already existing complex parts (case, sw)13:56
wpwrakhehe ;-)13:58
nicksydneyOpenPhoenux Neo900 Progress Update - January (showblockers resolved) --> http://www.reddit.com/r/linux/comments/1via7r/openphoenux_neo900_progress_update_january/14:00
wpwrakthe world is full of naysayers who are all to eager to projects their own limitations on others ;-)14:05
nicksydneythat's why the world is not a safe place to live in :)14:13
apeletelarsc: made a few changes in jz4740_mmc.c, in particular requesting the mmc memory region for dma operation (instead of requesting it for void __iomem *base member of struct jz4740_mmc_host) in probe(): http://paste.debian.net/77217/15:50
apeletelarsc: seems to boot fine -> http://paste.debian.net/77218/15:51
apeletelarsc: but what do you think about those changes ?15:54
apeleteI'm asking before replacing jz4740_mmc_{read,write}_data() with the dma equivalent functions since I guess I may experience data corruption if something goes wrong :)15:55
larscdevm_ioremap-resource does the request_mem_region interally16:13
larscotherwise looks good16:14
larscfor max_burst you need to make sure that max_burst + DMA request threshold <= FIFO size16:15
larscanyway need to go, I'll be back tomorrow16:22
qi-bot[commit] Paul Cercueil: Sort links once they have all been added to the menu (master) http://qi-hw.com/p/gmenu2x/8bb58f622:11
qi-bot[commit] Paul Cercueil: Change sort order so that OPKs are listed after normal links (master) http://qi-hw.com/p/gmenu2x/1792e6322:11
qwebirc96208anywhere I can get info about upcoming qi-hardware projects?22:23
qwebirc96208is there anything in the works?22:23
whitequarkwpwrak has anelok22:23
wpwrakthe main information source is the qi-hardware mailing list22:29
wpwrakas far as code is concerned, anelok has its repos here: https://gitorious.org/anelok22:29
nicksydneywpwrak: code and schematics to be precise inside the repo :)22:43
wpwrakyeah, almost all that matters :) pictures are here: http://downloads.qi-hardware.com/people/werner/anelok/tmp/22:44
nicksydneywpwrak: well you don't have your picture in the repo though :) ... hahhaah22:45
nicksydneywpwrak: do you know how long it took for the Neo900 to be fully laid out and routed ?22:54
--- Mon Jan 20 201400:00

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