#qi-hardware IRC log for Monday, 2014-01-27

DocScrutinizer05anybody knows about pagers and any widely adopted standard those are using (or the lack thereof)?00:13
larscfirst the robots, now this http://techcrunch.com/2014/01/26/google-deepmind/08:58
whitequarkAI craze, round 2 ?08:59
whitequarkfollowed by AI winter, round 2, I guess08:59
larscyou know the saying "In 10 years there will be a breakthrough in AI development"09:02
whitequark"In 50 years we'll have cold fusion"09:02
whitequarkgood chance it's just another acquihire for google. analytics, that's how they call AI that actually works09:03
larscwell, they are supposed to work for Ray Kurzweil at google09:06
whitequark"Kurzweil claims to have forecast the demise of the Soviet Union due to new technologies such as cellular phones and fax machines disempowering authoritarian governments by removing state control over the flow of information"09:09
larsche probably also did also forecast a lot of other things that did not happen09:10
larsc"The best way to have good ideas is to have many ideas@09:11
whitequarkthe statement above is false as well09:11
whitequarkI mean, USSR didn't fall due to that reason, and GCF can be quite effective as well09:12
whitequarkugh, ebay doesn't actually notify when the seller responds to a case, wtf09:17
whitequarkthey responded at 24th and I missed it :/09:17
whitequarkhm, actually it does, the email just got lost.09:19
whitequark".Friend,this item is the top-quality. So during two years warranty time, mostly it will not have any problem. If it has the problem, we will send you the replacement to replace it. Mostly it will be resolved."09:20
whitequarkI like their use of "mostly"09:20
whitequarkat least it's honest09:20
whitequarkhow are cylindrical connectors with 2, 4, ... pins inside called?09:49
whitequarkthey look like DIN, but are much bigger09:49
rohchinese plugs09:51
rohwe got the same on our mill controller09:51
whitequarkwell duh, can't put that into digikey09:52
rohmake sure you get a set with yours, but they are available if not supplied09:52
whitequarkI want to put another connector in a spare hole in the box09:52
whitequarkand it's convenient, not to say already has the proper size09:52
whitequarkbut no clue where to source09:52
whitequarkroh: what mill do you use btw?09:55
rohwe got a syil bf20 vario09:56
whitequarkroh: actually... searching "chinese plug" on ebay worked :D09:59
rohthe old version (no rpm meter, no belts on the motors)09:59
rohwe got it already cnc converted, with ball spindles and big ass steppers directly mounted where the handwheels were (those moved out by the motor length)10:00
rohthe controller was delivered with a mach3 demo, but we use linuxcnc10:01
rohrecirculating ball screw10:02
apeletewpwrak: about soldering wire, better to use lead free or is lead solder ok ?12:28
apeletewpwrak: was going for lead solder before reading that it is prohibited in the EU, but it seems lead free generally melts at higher temperatures12:31
CYB3Rthere are many opinions about soldering wire12:32
CYB3RI think that 63/37 is the best12:32
whitequarkapelete: lead free is shit13:00
whitequarkreally, the only reason people use lead free is because of ridiculously strict RoHS restrictions, if you want good soldering, use leaded13:01
whitequark*especially* for prototypes and such13:01
whitequarkyou will only reach harmful levels if you do leaded soldering 24/7 for years13:01
apeletewhitequark: haha, was wondering how a higher melting point temperature could be a good thing in my case anyway, now it's clear :)13:04
apeletewill look for a good Sn/Pb solder wire then13:04
whitequark63/37 is pretty much standard13:08
whitequarkI guess telling you to not use russian solder would be excessive :p13:09
apeleteonly found 60/40 so far13:09
whitequarkbut, really, don't be too greedy on it. I have ASAHI solder, with flux inside, it's quite great13:09
whitequark60/40 is same thing13:09
apeleteokay, will go for a reasonably priced 60/40, not too cheap though (went the cheap way in the past, painful memory)13:13
wpwrak60/40 != 63/37 !13:16
whitequarkwpwrak: is there much difference?13:17
wpwrakin 63/37 all components melt at the same time. in 60/40 they melt at different temperatures, so you have a distinguishable "pasty" stage13:17
wpwrakif you need that, e.g., to intentionally make solder bridges, get 60/40. if you don't need that, i.e., because solder bridges are exactly what you don't want to see, get 63/37.13:18
wpwrakapelete: regarding lead-free, i'd say: if you plan to sell it, use lead-free. else, use leaded13:19
wpwraklead-free is harder to work with. it's doable but you have to work at higher power levels, increasing the risk of doing damage.13:20
wpwrakwhitequark: the difference isn't huge, but noticeable13:22
wpwrakapelete: and don't forget to get flux :) flux is the smt-solderer's best friend13:23
apeletewpwrak: thanks for the details. I don't plan on doing a solder bridge, I find the pad too small anyway. let's see if I can find 63/3713:25
whitequarkwpwrak: oooooh, that explains some things13:25
whitequarkI should get some eutectic solder then13:25
wpwrakyup :)13:26
whitequarknever thought the 3% difference was significant, fool me!13:26
wpwrakmetallurgy is a direct descendent of alchemy ;-)13:26
wpwrakthere is a bewildering variety of solder formulas: http://en.wikipedia.org/wiki/Solder#Solder_alloys13:27
nicksydneywpwrak: https://github.com/nekromant/antares13:31
wpwrakapelete: another parameter is the thickness of the solder wire. if you have a choice, for smt, thin is preferable since it makes it easier to get a small quantity.(thin: < 0.7 mm)13:31
wpwraknicksydney: nice. might have things worth stealing :)13:33
nicksydneynot steal..borrow :)13:33
larscI only need three things for smd soldering: flux, flux and flux ;)13:34
wpwrakapelete: for the flux, you get three basic types: RA, water-soluble, and no-clean. RA is the most vigorous but it's very sticky and a pain to remove from the pcb. (and you must remove it since it attacks metals)13:34
whitequarkthere's also RMA, isn't it?13:35
wpwrakapelete: water soluble is less powerful but also less sticky. i find it a good compromise. "water soluble" means "hot demineralized water under high pressure". so it's not quite as easy to get rid of at it may sound, but it's feasible. i use a cheap ultrasonic cleaner for that.13:36
wpwrakwhitequark: yeah, in think RMA and RA are in the sama ballpark. not even sure what i had.13:37
whitequarkwpwrak: AFAIK RMA does not need to be cleaned13:37
whitequark(plus I never cleaned it and I have some >5yr old boards which are not corroded little bit)13:37
wpwrakapelete: "no clean" promises that you don't need to clean up but i'd be careful with such statements. it's the weakest of the three and i found that the one i tried didn't help enough with the soldering.13:38
wpwrakwhitequark: RMA is still aggressive. the only one you're supposed to be able to leave on the board is "no clean".13:38
apeletewpwrak: shouldn't the flux be embedded into the solder wire already ?13:38
wpwrakapelete: there is a little bit there but it's not enough for smt13:38
whitequarkwpwrak: hm, " RMA flux is formulated to result in a residue which is not significantly corrosive, with cleaning being preferred but optional."13:39
whitequarkI guess so13:39
wpwrakwhitequark: also, the flux will happily act as a conductor. the resistance is typically of a few dozen kOhms, so you may not notice it unless you work with low currents. but when you notice it, it will be very irritating :)13:40
wpwrakbesides, a board full of flux is just dirty :)13:41
wpwraki clean the board also to be able to inspect things. if all is full of flux you see nothing.13:42
apeletewpwrak: damn, flux seems to be a field of its own, and I wasn't even planning on getting any initially (didn't know the little amount of flux in the solder wire wasn't enough)13:57
apeleteit seems there was a *lot* of things I was doing wrong when I soldered the serial line on the nanonote (tried many times before having a good joint back then)13:59
whitequarkI've always had moderate to good success with lots of RMA14:00
whitequarkand a truly epic failure with a water-soluble flux called TT14:00
whitequarkclaimed to be non-conductive14:01
whitequarkit is actually neither water-soluble nor non-conductive, which provided for some funny results and weeks of headscratching14:01
whitequarkseriously, you get a head-start just by virtue of not having access to solders and fluxes made in russia14:02
wpwraksometimes, less can indeed be more :)14:12
whitequarksolders, fluxes, ICs, connectors (they STILL sell about 500 SKUs of bakelite connectors), anything made of plastic, etc14:17
whitequarkold tools however can be seriously great. my father's caliper was amazing14:18
wpwrakheh :) old tools are often much better than new ones14:18
whitequarkiirc cost him ~40 roubles in USSR times. a book would cost, say, 1.514:18
whitequarkengineer's wage would be ~15014:18
whitequarka scope I had (C1-68) was however not so amazing14:21
whitequarkits weight in kgs was actually higher than bandwidth in mhz14:21
whitequark... about 10kg/mhz actually14:23
wpwraka 10 MHz scope ? nice ! ;-)14:23
whitequarkif only! 1mhz14:24
whitequarkwas quite a feat to transport it across moscow with my bare hands and public transport14:24
whitequarkbecause it was quite huge and also had a CRT. had to take care not to shatter that.14:26
wpwrakah, crt not firmly attached ? :)14:27
wpwrakor was it difficult because of the ice and snow ?14:27
wpwrakand the erratic movements of the horse carts ...14:27
whitequarkfirmly, actually I thought it was too rigidly attached -> easy to shatter with sudden movements14:28
whitequarkyeah, the whole ice and snow deal, too14:28
wpwrak(shatter) hmm, so no case, just the tube ?14:38
whitequarkno, it has a case14:42
wpwrakprobably made of hardened russian undestructium14:43
Action: dos1 received his NanoNote today :)18:06
wpwrakwelcome to the club ! :)18:17
DocScrutinizer05I'd love to learn how sharism(?) managed to build the cases and mech parts for NN20:04
DocScrutinizer05then otoh checking out status of sharism and NN today, I maybe don't need to know20:05
DocScrutinizer05except for learning what to NOT do20:05
DocScrutinizer05it all feels like trying to sell milk from customer's "own" known named cow to customer in *supermarket*, packed into a bottle of customer's design. Of course at a price competitive to standard milk20:08
dos1DocScrutinizer05: wasn't NN based on some chinese notebook/calendar/mp3 player, reusing its case and overall design?20:10
wpwrakDocScrutinizer05: you're about right about the price calculation :)20:12
Action: dos1 started dgClock and feels like a newbie who just launched vim20:12
wpwrakand yes, the design is licensed. it's a modification of a dictionary. they made a number of changes, but the basic design already existed.20:12
dos1damn it20:14
wpwrakdos1: broke it already ?20:15
dos1I was sure that I already tried every key and lots of different key combinations in order to find that one which quits that app with no success20:15
dos1so I downloaded the sources of dgClock and looked how to quit20:15
dos1turned out it was escape button *facepalm*20:15
dos1I was like "no way", presses esc, and then like "no way" again :D20:16
dos1actually it's the second time I overlooked esc button when trying to quit from some app... it's in a quite strange place20:17
dos1anyway, I thought it was cool even with stock rootfs20:19
DocScrutinizer05even worse: while it originally been invented for exactly this purpose ( dict "escape"), it is not even existing on e.g. N900. And generally not used that way in DOS and Linux20:19
dos1just reflashed... and boy, so many awesome stuff there20:20
dos1and it even (mostly) works! :D20:20
DocScrutinizer05originally there been other useful keys like "help" that now is "F1"...20:20
DocScrutinizer05"prtscreen" which is usually used for dunnowhat but not for printing a screen copy20:21
DocScrutinizer05"break" which is generally ignored nowadays, and if it's not then it's a lame hotkey for ^C20:22
DocScrutinizer05same with "pause" and ^S20:22
wpwrak"can you please F1 me with this ?"20:30
wpwraki think Pause is pretty much meant to be like ^S20:33
wpwrakand what's more worrying are all the "multimedia" keys they're putting on keyboards nowadays. as if they weren't big enough already.20:33
wpwrakand even small keyboards get an obligatory row of dedicated function keys. even the ben has that madness !20:34
DocScrutinizer05right. sorry I'm supposed to appreciate that I'm not well and shouldn't touch any IT. Of course "Pause" is ^S20:43
DocScrutinizer05funny enough it's sometimes also ^Q20:43
wpwraki think it's a toggle20:45
wpwrakgot a flu ?20:45
DocScrutinizer05yeah, that's what it's technically meant to be, emulated by sending ^Q when ^S-mode active, and ^S when no ^S-mode active20:46
DocScrutinizer05no, severe burnout all levels20:46
wpwrakuh uh. then you should perhaps have a beer and take a nap.20:47
DocScrutinizer05saw it coming, for like 6 months now20:47
DocScrutinizer05I'm afraid one beer and one nap will not do anything20:47
wpwrakwell, it'll be a start :)20:48
DocScrutinizer05:nod: pending since errr Thursday20:48
DocScrutinizer05the beer, not the nap20:48
DocScrutinizer05actually i had 11h of nap last night, feels like being sick20:49
wpwrakbeer FOUR days overdue ?!? oh dear. someone please call an ambulance !20:50
DocScrutinizer05actually not only the beer but also the shower and leaving flat20:50
DocScrutinizer05I prolly shouldn't elaborate on all that20:51
DocScrutinizer05I should shut down that friggin PC and take care to restore some normality to about every other aspect of my life. Internet is the only domain that actually doesn't need such care yet20:53
wpwrakyeah, we'll try to keep it running while you have your shower and your beer :)20:54
DocScrutinizer05funny side effect: you can't catch a flu when you live in quarantine20:54
wpwraknaw, it somehow finds you anyway20:57
wpwrakplus, your immune system gets weaker that way20:57
apeletelarsc: are you there ?21:21
apeleteI've been trying to figure out what was wrong with the dma code in jz4740_mmc, to no avail21:23
apeletelarsc: last log I got was http://paste.debian.net/78701/21:24
larscdo you get any data via dma?21:25
apeletedon't think so, the sd card does not mount anymore21:26
larscbut it looks as if the transfer is completed21:26
larscyou might need to invalidate the dcache after a transfer21:27
larscafter a read transfer21:27
larscand flush it before a write transfer21:27
apeletewhat's the dcache and how do I do that ?21:27
apeletelarsc: ah I see flush_dcache_page(miter->page); in the code21:28
larscthat looks good21:30
larscthe dcache is the data cache of the cpu21:30
larscyou need to flush it to the real memory since the DMA will read from real memory21:31
apeletelarsc: ok, right now in the code for pio mode flush_dcache_page is only use in read_data()21:31
apeleteso fpor dma I use it after read and before write operations ?21:32
larscflush before write, invalidate after read21:34
larscinvalidate means you tell the cpu that the data in the cache is outdated21:35
apeletelarsc: okay, I see that flush_dcache_page() does both writeback and invalidate operations (here: http://lxr.free-electrons.com/source/arch/mips/include/asm/cacheflush.h?a=mips#L24)21:43
apeleteso I guess I can use it to flush before write and invalidate after read as you said21:45
apeletenow, let's see how I get a struct page pointer from struct mmc_data21:48
apeletelarsc: same "error -22 whilst initialising SD card22:44
apeletegit diff -> http://paste.debian.net/78725/22:44
apeletelarsc: is there a way to check the data read by the dma ?22:45
DocScrutinizer05http://uk.farnell.com/texas-instruments/cc1111f32rsp/transceiver-rf-with-mcu-36qfn/dp/1573890  http://www.farnell.com/datasheets/1734581.pdf  300/433/868/900MHz transceicver CPU22:48
DocScrutinizer05Low-power SoC wireless application soperating in the 315/433/868/915 MH zISM/SRD bands22:49
DocScrutinizer05((<larsc> invalidate means you tell the cpu that the data in the cache is outdated)) something some CPUs do on a hw level by snooping on RAM addr/ctrl-lines to detect writes to addresses that are in dcache23:41
DocScrutinizer05flush is particularly important when you do mem-mapped IO23:42
DocScrutinizer05...or are working on a multicore platform23:43
DocScrutinizer05...for data that's supposed to get transferred from one CPU to the other, when both CPUs don't use same dcache23:44
DocScrutinizer05ARM has a special "mailbox" IP for that, basically a hw FIFO that creates some datalink between any arbitrary 2 cores on SoC23:45
DocScrutinizer05PL320 PrimeCell23:46
DocScrutinizer05on some modem+AP SoC the mailbox is used to transfer addr of a RAM buffer from sender to receiver CPU. In that case you need to make sure the buffer got flushed from dcache before you hand it over to the desination CPU23:48
DocScrutinizer05similar considerations apply to USB and HSI and generally any high speed DMA based IO interfaces23:50
wpwrakDocScrutinizer05: TI have quite a lot of CCxxxx chips with similar characteristics. the sub-GHz bands have long range and are usually not as crowded as 2.4 GHz but they have the disadvantage of being more restricted and having limited geographical reach23:55
DocScrutinizer05yeah. Anyway that chip looks quite nice for the requirement spec "general purpose 433/868MHz radio, for all sorts of home automation, garage door openers, stereo headphones, etc"23:57
DocScrutinizer05X10 is 433MHz?23:58
apeletewpwrak: couldn't find 63/37 solder wire in EU, it seems they only sell 60/40 around here23:58
apeletewpwrak: does this look good to you ? http://www.ebay.fr/itm/Solder-Reel-Leaded-Omega-II-250g-22SWG-Sn63-Pb-37-2-flux-/291053556827?pt=UK_Home_Garden_PowerTools_SM&hash=item43c424d45b23:58
wpwraklooks great23:59
--- Tue Jan 28 201400:00

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