#qi-hardware IRC log for Sunday, 2011-02-20

wpwrakkristianpaul: i think parametric is the way to go. (stable version) hopefully soon. well, over the last week, i already thought three times i had the final version ... one even progressed to the point of having boards cut, etched, and via-ed. and then i realized it still had some features i'd rather not have there :-(00:04
kristianpaulwpwrak: development hell trap? :p00:05
kristianpaulargg i give up, no more tries of printing the buttons..00:07
kristianpauli'll make work what i alredy printed00:07
wpwrakkristianpaul: well, #1 was me realizing that i could make quite a dramatic improvement to the layout. #2 was me realizing that some things in the layout where too tight (also discovering some new traps in how kicad handles zones). #3 was me overlooking one flaw. now, i allowed me one extra day to review #4. let's hope this was time well spent, and that it works now ;-)00:13
kristianpaulphew, finally it fits and i be pressed00:20
wolfspraulbartbes: hi00:45
bartbeswoah00:46
bartbesdelayed response00:46
wolfspraulkristianpaul: this is absolutely awesome!!!00:46
wolfspraulcan you do a blog post, or mail, or even just a few lines here describing the steps it took you to make the case?00:47
wolfspraullike how did you find your local acrylic supplier, how did you find the laser shop?00:47
kristianpaulsure sure00:47
wolfspraulhow long did it take you, what did it cost, which difficulties did you run into (tolerances, etc)00:47
wolfspraulhow about screws and spacers, other small parts?00:47
kristianpaulgood point00:48
kristianpaulI had printed my own buttons, is not the _perfect_ thing but it works00:48
kristianpaul(find your local acrylic supplier, how did you find the laser shop) is same shop00:49
kristianpaulhttp://dataprint.webs.com/00:49
kristianpaulI just send case, pay, choose color and thats it00:49
wolfspraulcan you upload some of those pics (or maybe just all) to the wiki? it should be a news item in our community news00:49
kristianpaulsure00:50
kristianpauldifficulties so far, are missing small parts, mostly the spacers that came with the mm1 are not compaible with the case, so i have a slight un-aligned00:51
kristianpaulI do not have plans of buy screws00:51
kristianpaulI already made my own hole for let jtag-serial cable to go inside case (i used it most of the time)00:52
kristianpaulI'll blog later, an upload pics to wiki, righ now i'm wiring sige board to mm1, and i need to see how handle the ribbon cable there..00:53
kristianpaulcost: 10usd material and cut process + 5 usd shipping to my home00:54
wpwrakkristianpaul: soon you'll need another case - a presentable one, without all the extra holes ;-)00:54
kristianpaulwpwrak: hmm, may be, also i can try green color ;-)00:54
wpwrakthe "smoke" looks very nice00:55
kristianpaulwolfspraul: i just send cad file as it is on repo plus a pic from a already made case then i got the quote00:56
kristianpaulI got the link to this company as a guy have a reprap machine told me about then00:57
kristianpaulI think tolerances in general will be good with the right small parts..00:57
kristianpaulbuttons are really tricky00:57
kristianpaulwpwrak: smoke is nice indeed,  i'm planning also put may be some smalls blue leds in two sides to get a nice effect, (for future events)00:59
kristianpaulwolfspraul: thats  it !00:59
kristianpaulat least now mm1 is easy to handle :-)00:59
wolfspraulkristianpaul: ok, maybe write a little 'how I made my own m1 case' story somewhere01:02
wolfsprauland you know how it is, it's either done fast or never. now you still remember the details :-) If you don't get to it, I understand. So many things. But it would be a great story I think.01:02
kristianpaulwolfspraul:  (story) sure01:03
wolfspraulrejon: did you see this? kristianpaul made his own m1 case, totally independently in Colombia!01:05
wolfspraulcc alive, in hardware :-)01:05
rejonthat is she shit01:06
rejonkristianpaul i want to come to columbia01:06
rejonyou got a pic?01:06
kristianpaulrejon: http://kristianpaul.org/gallery/mm1%20case/index.html ?01:06
wpwrakkristianpaul: soon, people will do pilgrimages to buga just to see your case ;-)01:07
rejonyeah01:07
rejoni want to come to argentina and colombia01:07
rejonthat is killer!01:08
rejoni talked with roh about that01:08
rejonhow we can make some future products but with even more customized bits01:08
rejonlike how can we make some products that are special to the owner01:09
rejonhe was skeptical :)01:09
rejonhahah01:09
rejonok dinner...bbiab01:10
wolfspraulkristianpaul: are they licensed under cc-by? pd? I will just do the wiki upload if you don't mind...01:26
wolfspraul(the pics I mean)01:26
kristianpaulwolfspraul: yes sure, and thanks !01:27
kristianpaulhttp://kristianpaul.org/ <- botton BY-SA 3.001:27
wolfspraulgreat, will do01:29
wolfspraulkristianpaul: where are those dataprint guys located? how did they ship to you (how much did shipping cost and how long did it take?)01:33
wolfspraulah I found it - in Bogota01:35
kristianpaulwolfspraul: http://kristianpaul.org/comoblog/Local_fabbed_case_for_the_Milkymist_One__Full_log_.html01:40
kristianpaulhope all answers are there..01:40
Action: kristianpaul dinner01:40
zrafawpwrak: if rejon is coming some day we should invite kristianpaul as well, so with his mikymist one we can do the qi party here :)01:48
kristianpaulzrafa: :D01:49
zrafakristianpaul: wpwrak has a nice terraza for asados :)01:49
zrafakristianpaul: in the middle of Buenos Aires01:50
wolfspraulkristianpaul: wow yes, fantastic, this post answers it all.01:51
zrafakristianpaul: do they know your case and M1 is open hardware? (the guys who did the case)01:51
kristianpaulzrafa: nope.. i may send the blog post.. what do you think?01:52
zrafakristianpaul: I think that you should explain them about the machine and case. They could get interested and be part of qi community some day01:52
kristianpaulokay i can give a try :-)01:52
wolfspraulkristianpaul: I uploaded your pics. btw, we have a MultiUpload feature now http://en.qi-hardware.com/wiki/Special:MultipleUpload01:57
wolfspraulit's a bit buggy with strange warnings after upload, but the stuff gets uploaded still.01:57
kristianpaul(multiupload) great01:57
wolfspraulsometimes Adam has series of scope pictures, or x-ray pictures, so he needed that feature02:01
wolfspraulbtw, it's depressing to see that our friends from linuxencaja are already buried under spam and I guess gave up on removing it http://wiki.linuxencaja.net/wiki/Special:RecentChanges02:04
wolfspraulthey need to disable anonymous edits asap02:06
kristianpaulor add catcha..02:07
wolfspraulfirst disable anonymous edits, immediately02:08
wolfspraulthe spam bots are ripping their beautiful wiki apart right now02:08
wolfspraulfrom the last 500 edits it looks like 50% or more are spam02:09
wpwrakstill better than mail :)02:22
kristianpaulhmm, 10Mhz, 3V3, 15pf load per pin, i should not extend this so much..02:33
kristianpaulwolfspraul: What you think about ARM clones? i mean use it on real products02:36
wolfsprauldon't understand your question02:39
kristianpaulI mean arm is getting popular, soon or later it may get implemented on fpga and described with some HDL language and use it somwhere02:44
kristianpaulNon couting the already linux adn gcc  support it make interest for developers02:44
wolfspraulyou need to ask Sebastien, I would think ARM will vigorously protect their IP02:45
kristianpauloh sure02:45
kristianpaulbut clones will appear one next to other02:45
wolfspraulso if you just ask me, I can see arm cores inside ASICs, and they can be great for us02:46
wolfspraulwhen it comes to an integrated IC design, I believe in the Milkymist/LM32 path, which we can only implement in fpgas right now02:46
wolfspraulwhat do you mean with 'clones will appear'?02:46
kristianpaulimplentarios of arm-like cores compatible with gcc/linux02:47
wpwrakwolfspraul: how about the pre-FPGA nanonote line ? would a ya with, say, a Marvell PXA be something you'd consider ?02:48
wolfspraulI don't see that at all right now for myself, so the question is very hypothetical.02:48
kristianpaulsure ;-)02:49
wolfspraulputting that aside, what's the difference between a Marvell PXA and an Ingenic chip? None I think.02:49
wolfsprauljust comes down to features02:49
wolfspraulbut given all the things we did to bring the NanoNote alive, I would try to protect our software investments above everything else first.02:49
wolfspraulsame with Milkymist, as we improve the tools around it02:50
wpwrakwolfspraul: yeah, i mean if the feature set would be more attractive02:50
wolfspraulbottom line: someone else needs to make that Marvell NanoNote :-)02:50
wolfspraulyes, and if the NanoNote would not have been started, then yes, sure02:50
wolfspraulthe Marvell chip may be more expensive and harder to source02:51
wolfsprauland Marvell would be far less flexible than Ingenic02:51
wpwrakusing the same core has of course the advantage of a more-or-less binary-compatible user space02:51
wolfspraulno hopes of ever installing a server behind the Marvell firewall that will do a nightly rsync to shuffle some sources out :-)02:51
wpwrak;-)))02:51
wolfspraulI like working with Ingenic, even if it's dormant RIGHT NOW.02:51
wolfspraula large Ingenic customer contacted us recently because they saw what we did on the NanoNote.02:52
wolfspraulI don't expect any big business opportunity there, but that's how things grow if you do solid work in some place, rather than jumping around and building random boards with random chips, and nothing ever works well.02:53
wolfspraulkristianpaul: my main worry about ARM 'clones' would be patents, and the way ARM will defend their IP.02:53
kristianpaulwolfspraul: indeed02:53
wolfspraulyou need to ask Sebastien about this, I trust and follow his judgment. Milkymist is just starting...02:53
kristianpaulsure02:54
wpwrakwolfspraul: (jumping around) yeah, agreed. continuity is good.02:54
kristianpaulhey andres-calderon02:54
wolfspraulbut I am not 'against' ARM cores, that would be so stupid I think. we are trying to build great copyleft hw, fast, that really works, and has great features/specs.02:54
larscthere exist some ARM-compatible desgins.02:54
larscbut mostly for early ARM versions02:55
kristianpaulandres-calderon: noticed the spam edit around linuxencaja? http://wiki.linuxencaja.net/wiki/Special:RecentChanges02:55
wpwrakwolfspraul: just wanted to know your general opinion with respect to ARM, apart from OMAP overengineering02:55
wolfspraulsure but there can be several 'back stories' to this02:55
wolfspraul1) they exist because they operate and sell in places where ARM does not have judicial reach02:55
wolfspraul2) they exist because they are small, and ARM is better off letting them grow first, then apply their tax02:56
wolfspraul3) they exist because ARM is actually not vigorously patenting and defending their IP, maybe (theoretically) because they even support such clones?02:56
wolfspraul4) they exist because they already have a licensing agreement with ARM02:56
wolfspraulthose are the 4 options that come to my mind right now02:56
larsci meant opensource designs02:57
wolfspraulyou mean a gpl licensed hdl source code?02:57
larscyes02:57
wolfspraulif an implementation infringes ARM patents, that changes nothing02:58
wolfspraul5) they exist because it is an open source software design that no company would ever dare to manufacture or sell because of ARM patent infringement02:58
wolfspraulI added that one :-)02:59
kristianpaulwolfspraul: You care about US export laws about lattice mico32?03:00
larsci'm not arguing about that. just wanted to point out that there _are_ ARM compatible designs, because kristianpaul said there _will_ be.03:00
kristianpaullarsc: yep03:00
wpwrakup to ARM6 may be free from patents or close to it (~1-2 years), as far as the 20 years limit applies03:00
wolfspraulthat concept never worked for MIPS03:01
wpwrakyeah, but who knows why ...03:01
wolfspraulkristianpaul: I care, but that's an entirely different subject and typically outside of scope for free projects. There are a lot of embargoes...03:02
wpwrakmaybe MIPS paid them to stop claiming it was free from patents ;-)03:02
wolfspraulwpwrak: oh I know why. because MIPS is doing very aggressive lobbying and very aggressive legal actions.03:02
wolfspraulI spoke in person with the people running it on both sides.03:03
wpwrakwolfspraul: ah, so it's the sue-to-lose-but-kill-the-enemy-anyway approach03:03
wolfspraulif you think you can make a MIPS-clone under the 'patents are expired' idea, you can go try, and MIPS Inc. will be your enemy03:03
wolfspraulthe Chinese government tried, and failed!03:03
wolfspraulif that is not enough proof for you, MIPS will gladly take on Werner Inc. next03:04
Fusinwb /me ;)03:04
wpwrakwolfspraul: lemme first rig some laws my way :)03:04
wolfspraulARM is in the UK, they could use less leverage via their government. But not that much less, UK is still someone...03:05
Fusinmy ben still doesn't boot :(03:05
wolfspraulI think it's simple. If you believe in ARM clones, get a written statement from ARM Inc. that your theory is OK and they acknowledge that what you do is legal.03:06
wolfspraulonce you have that, we can move03:06
kristianpaulFusin: you we're flashing it? or tryin boot?03:06
FusinI reflashed03:06
Fusinbut still doesnt boot03:06
kristianpaulormris: Did you get usbboot to work finally?03:06
wolfspraulyou ran reflash_ben.sh ?03:07
Fusinyep03:07
kristianpaulFusin: what it said on screen?03:07
Fusinscreen is black03:07
wolfspraulafter reflashing, unplug the USB cable, take out the battery, wait 20 seconds03:07
kristianpaullsusb show something?03:07
wolfspraulFusin: then battery back in, then press power-on button for a solid 5 seconds03:08
FusinBus 001 Device 009: ID 0525:a4a1 Netchip Technology, Inc. Linux-USB Ethernet Gadget03:08
wolfspraullet's try that first. maybe the reflash was successful and you just didnt' reset/reboot yet.03:08
Fusinok i try power off03:08
larscthat looks as if the kernel is actually booted03:08
kristianpaulmay it be lcm bus unpluged?03:08
kristianpaulFusin: it just happen? it was working well before?03:08
wolfspraulthat would be bad, but not very likely that it happens in conjunction with a reflash03:09
Fusinno did not work before03:09
kristianpaulhmm..03:09
Fusini wait 20 secs ;)03:09
wolfspraulFusin: when did you get your Nano? did you ever see it boot/turn on the screen?03:09
Fusinnever saw it running03:09
Fusinnew buy, got it yesterday03:09
Fusinconnected to laptop for charging03:10
Fusinwaited until led light goes off03:10
larsclooks like your nanonote screen is broken/not connected03:10
Fusinthen tried boot03:10
wolfspraultoo bad, could be that your LCM cable/connection is bad and you need to return/replace the device03:10
Fusinthen read lotsa in the wiki03:10
wolfspraulcan you ping the nano?03:11
Fusinhow without ipconfig on nano?03:12
kristianpaulit have default ip03:12
wpwrak192.168.254.10103:12
Fusin--- 192.168.254.101 ping statistics --- 9 packets transmitted, 0 received, +9 errors, 100% packet loss, time 8010ms03:13
Fusinno pongs :(03:13
wpwrakfirst you have to  ifconfig usb0 192.168.254.100 up03:13
wolfsprauldo you have usb0 setup for 192.168.254.1/2403:13
Fusinerror in usb03:15
kristianpaulwhat it said?03:15
Fusinusb0: FEHLER beim Auslesen der Schnittstellenmerker: Kein passendes Gerät gefunden03:15
Fusindoesnt found device on usb003:15
kristianpaultry dmesg03:15
kristianpaultoo see if is thre03:16
kristianpaulthere*03:16
wolfsprauldoes lsusb show 0525:a4a1 as you posted before?03:16
Fusin--- 192.168.254.101 ping statistics --- 6 packets transmitted, 6 received, 0% packet loss, time 5001ms rtt min/avg/max/mdev = 0.360/0.875/2.828/0.875 ms03:16
Fusinnow got pongs03:16
wolfspraulhe, let's try one more thing03:17
wolfspraulon the Ben, press 'Ctrl-Alt-F2', then 'enter'03:17
Fusinbut auto disconnected as linux mint told me03:17
wolfspraul(I know the screen is blank)03:17
wolfsprauldisconnect? try 'telnet 192.168.254.101 22' - see whether you get a ssh response03:18
Fusinscreen is still black03:18
wolfspraulyes, leave it like that for now03:18
wolfspraulcan you try 'telnet 192.168.254.101 22'?03:18
kristianpaul22?03:18
wolfspraulssh, yes03:19
kristianpaulwhy not ssh..03:19
wolfspraulI think it will not work without setting a password on ben first.03:19
wpwrakdefinitely looks like a display connection issue03:19
kristianpaulah yes..03:19
wolfsprauland because he said something about 'disconnect', I first want to see whether he get the ssh response03:19
wolfspraulFusin: to set the ssh password on the Ben, try this (even with blank screen)03:20
Fusinconnection refused :(03:20
wolfspraul1. Ctrl-Alt-F2Enter03:20
wolfspraulsorry I meant03:20
wolfspraul1. Ctrl-Alt-F203:20
wolfspraul2. Enter03:20
Fusinpasswd03:20
wolfspraul3. passwd<Enter>03:20
Fusinetc pp03:20
FusinI'm used to linux :D03:20
wolfspraulok great03:20
wolfspraultype it twice03:20
wolfspraulthen ssh root@192.168.254.101 from your host03:21
wolfspraulmaybe you can login :-)03:21
wolfspraulbut you need to do Ctrl-Alt-F2, and <Enter> first, otherwise passwd will not run03:21
Fusinstill connection refused...03:21
Fusinshitty situation :(03:22
wolfspraulok the ssh connection would have just been for fun anyway03:22
wolfspraulwhen you did telnet 192.168.254.101 22, did you see the SSH response?03:22
Fusinfusin@limint-t43 ~/Downloads $ ssh 192.168.254.101 fusin@192.168.254.101's password:  Permission denied, please try again.03:24
Fusinshit shit03:24
Fusinseems it doesnt changed root passwd03:24
wolfspraulyou did Ctrl-Alt-F2? and then <enter>?03:24
kristianpaulfusin@?03:24
wolfsprauljust try again with Ctrl-Alt-F3 :-)03:24
kristianpaulroot@..03:24
Fusinoops03:24
wolfspraulah good point!03:25
Fusinfusin is my username on mint..03:25
Fusinredo as root03:25
kristianpaulis not on ben :-)03:25
Fusinroot@192.168.254.101's password:    BusyBox v1.15.3 (2010-12-14 14:36:17 CET) built-in shell (ash) Enter 'help' for a list of built-in commands.  root@BenNanoNote:~# i'm in :D03:26
FusinI'm in the ben03:26
kristianpaulwolfspraul: do you have a extra set of "small parts" s for a mm1 perhaps?03:26
Fusinbut screen is still black03:26
Fusinso i guess ribbon is faulty, eh?03:26
kristianpauljust in case.. gmenu2x03:26
kristianpaulFusin: ribbon may be, but that part need to be checked visually03:27
kristianpaulat lest linux can diagnostice that  part, larsc ?03:27
Fusintop said gmenu is running03:27
kristianpaulah ok :-)03:28
wolfspraulFusin: yes, I am really sorry but you need to return/replace the device, that's the easiest way.03:28
wolfspraulwe've had 5-10 cases like this so far, in about 1100 nanos.03:28
kristianpaulwolfspraul: all fo the just disconnected lcm bus?03:29
kristianpauls/fo/of03:29
Fusinthat's bad for me :(03:30
Fusinreturn nano and wait new one takes time...03:30
larscFusin: where are you from?03:30
wolfspraulwhere did you buy it?03:30
FusinI usualy work approx 200 - 500km from home03:31
Fusinso it take weeks before I'm back home...03:31
Fusin<- Nürnberg Germany03:31
wolfspraulwhere did you buy it? maybe the replacement can be sent there?03:31
Fusinoh, i will03:31
Fusinit's from Pulster03:31
Fusinalso sold me my Freerunner ;)03:32
wolfspraulhe. welcome to the NanoNote club then, even though you are having a hard first time...03:32
Fusinreminds me my first Linux :D03:32
wolfspraulkristianpaul: there seems to be a variety of causes 'around' the LCM cable03:32
FusinSoftlanding System 0.99pl6 :D03:33
wolfspraulthe connector on the mainboard side, the connector on the LCM board side, the cable itself03:33
Fusintooks days to install on an 386 :D03:33
Fusintook also one week to get X11 working in greyscale03:34
kristianpaulwolfspraul: oh, really?.. getting fun with hardware then :-)03:34
Fusinbut i survived :P03:34
wolfspraulFusin: one sec let's try another thing, just for fun (you can still replace)03:34
Action: Fusin listens03:35
wolfspraulthe LCM cable goes through the hinge on the right side03:35
wolfspraulwhere the 'mic' hole is03:35
wolfspraulpress your thumb on that area (keep the Ben running), see whether the LCM comes on03:36
wolfspraulyou can try the same on the other side (the LCM), roughly where 'qi-hardware' is printed03:36
Fusinhmm, how do I quit gmu?? *g*03:36
Fusinwith key commands?03:36
kristianpaulalt + enter03:37
wolfspraulor just kill it in ssh :-)03:37
Fusinthx03:37
wolfspraulFusin: when you press on those places, does the LCM turn on?03:40
Fusinnada :(03:40
Fusinmeans : no03:40
wolfspraulyou can press quite hard, just not so much as to break the plastic03:40
wolfspraulok, was a long shot anyway03:41
wolfspraulI think there were 1-2 people who said pressing there helped03:41
wolfspraul(of course the devices were exchanged still)03:41
kristianpaulor try open screeen to max then close again.. and see03:41
Fusinok, thx for help03:42
Fusini goto bed now, it's 4 o'clock in the morning :)03:43
wolfspraulsorry again for the LCM problem03:43
wolfspraulkeep us posted how your exchange is going03:43
Fusinok03:43
Fusinn8 guys03:43
kristianpaulnite03:44
kristianpaulwolfspraul: You do stress test on nanonote lcm before leave?03:45
kristianpaulI mean let said, 50 open and close, or vibration tests..03:45
kristianpaulhehe, just dreaming :-)03:45
wpwrakkristianpaul: i guess soon he will :)03:46
wpwrakwolfspraul: i think if the connection is really hosed, the LCM will miss the initialization sequence, so pushing later wouldn't do much03:46
kristianpaulI remenber when my laptop failed, lemote said we tested the power suply 50 times to make sure it will came back OK03:46
wolfspraulkristianpaul: not every device is stress-tested, and I'm not sure that would help.03:47
wpwrakwolfspraul: i have one problem LCM, too (developed after a while - didn't examine it yet), and there the contact was intermittent. so the LCM came up, but then the contact failed. in this case, it could be brought back with pressure once.03:47
wolfspraulthe LCM connection is definitely the weakest spot in the hw design03:48
kristianpaulagree03:49
wpwrakyup. very fragile.03:50
kristianpaulwell i damaged my self lcm connector too03:50
kristianpaulis kinday weak... i think, i dint aplied too much force i remenber03:50
wpwraki suspect the cable can just develop hairline cracks anywhere. i don't think the receptacle is nearly as likely to fail.03:54
wpwrakflexible cables are just one of these troubled technologies. i'm sure wolfgang has vivid memories of that from his openmoko days, too :)03:56
kristianpaulbut freerunner dint have mobile parts..03:56
wpwrakthe cable between mainboard and (fixed) lcm worked quite well. i never heard of problems there.03:57
wpwrakbut then it also had the debug board. that connected with a long FPC to the main unit. endless fun has been had with this one ...03:58
wpwraka particularly nice moment was when we took an OEM project under our wings. that one was loosely based on the freerunner technology. they had a horrible contraption for their debug board, though. so we suggested to replace this with the openmoko debug board.04:00
wpwrakso we made the design changes. then, the plan was to ship something like ten or twenty debug boards with cables to our customer (in the US). so i got the items and did a quick test to make sure everything was alright ... much later that night, after beginning to seriously doubt my sanity, i knew that about 10% of the whole batch really worked.04:02
wpwrakthe rest had all sorts of issues. also, many worked sometimes and then not. turned out that our previous FPC supplier had vanished and we had to order cables from a new source. that new source had not made them with the necessary accuracy. so basically the whole production run had to go back (well, or into the trash. don't know where they ended up.)04:04
kristianpaul:-/04:04
wpwrakone of the issues was that many cables were not wide enough. so you could shift them a little. if you *just* hit the right spot, they would work. if not, some pins wouldn't connect. so you may get serial but no jtag. during one of those nights, i figured out the failure pattern and learned how to position the critters.04:06
wpwrakbasically, insert it, then gently move to the left until you hit the border. then to the right, until you hit that border. then move back half the distance (a few dozen micron). then lock the thing and test. if it doesn't pass, repeat.04:07
wpwrakthe next day, we verified the scratch marks on the contacts with a microscope. and yes, they were all over the place.04:08
wpwrakanother batch of these cables was too wide. so you could 1) try to force them in anyway, or 2) try to cut them down to the right size. 1) would sometimes work, sometimes not. 2) was a game of chance.04:09
wpwrakand then these cables developed hairline fissures quite easily. so when you bent it one way, it may give contact. bend it the other way, and it's gone. and again, this usually just affected one signal out of more than a dozen. so you never quite knew if a cable didn't have some defect that you just hadn't looked for yet.04:10
wpwrakoh yes, i learned to love those FPCs ;-)04:11
wpwrakeventually, i was sufficiently fed up with that mess that i made idbg. that solved the problem for good - at least for me :)04:12
kristianpaulfdti was not so popular in that time?04:13
wpwrakftdi is ancient :) we had an ftdi on the debug board. i picked the silabs chip for idbg because it's more flexible and needs less space than the ftdi chips (they didn't have a crystal-less solution back then)04:17
wpwrakftdi documentation also sucks. lots of secrecy that's just plain annoying.04:17
wolfspraulftdi is surprisingly unpopular among people who really tried to use it, and push to the limits04:19
wpwrakand what's documented still leaves lots of things unclear04:19
wpwraki'm not surprised at all ;-)04:19
wolfspraultheir brand is strong though, so I guess it gets selected by 'managerial decision' into many places still, whether the engineers are complaining or not04:19
wpwrakof course, i'm one of them. if "try anything but dumb serial" already counts as "push to the limits"  :)04:19
kristianpaulso what you recomend instead of fdti?04:20
wpwraka microcontroller04:20
kristianpaul:-)04:20
wolfspraulwell but you are demanding on the details, you demand clear documentation, etc.04:20
wpwrakwolfspraul: is it demanding when you want to know which bits in their EEPROM you have to set to enable bit-banging on the "C" bus ? the manual just tells you that this needs to be configured, but doesn't tell you how.04:21
wolfspraulthe ft2232hq on the m1 jtag-serial board seems to fulfill its purpose well though04:21
wpwrakwolfspraul: also, due to the poor documentation, there are all sorts of bugs in the drivers04:21
wolfspraulmaybe the problem is just that people try to use the ftdi chip it all sorts of diverse environments, and then it fails04:22
wpwraknaw, the issue is that they somehow feel they shouldn't tell you what you need to know. this is very different from the usual incompleteness you find in documentation.04:23
wpwrakthere are also obscure bugs that manuals don't warn you about. like data being lost apparently at the byte level.04:24
Action: kristianpaul stick on his trusty or well know microchip for me pic1804:25
wpwrakthe ft2232 has a special engine for things like jtag. the simpler chips don't have that engine. but their documentation claims capabilities that would make it perfectly feasible to accomplish the same, only perhaps less efficiently04:25
wpwrakbut eventually you realize that you probably won't even get something like SPI to work reliably04:26
kristianpaulthats shame04:26
kristianpaulanyway for me fdti was just usb2ttl and now jtag with mm104:27
wpwrakand amoong the two or three possibilities for implementing it the manual suggests exist, you'll find that at least one is not accessible because of incomplete documentation04:27
wolfspraulok but I think for the very specific purpose that the ft2232hq is hard-wired on the m1 jtag-serial board, it's a good choice and the problem is solved04:28
wolfspraulthat's not to say that I don't hear the type of feedback we hear from you now many times already04:28
wpwrakand so on. i tried to make one of the ftdi critters work for programming the c8051f32x chips, but eventually gave up after trying for one or two weeks.04:28
wolfspraulI would be very careful to see the ftdi chips as some sort of debug-board-swiss-army-knife. for sure they are not.04:28
wpwrakyes, if you have a use case that someone else already validated and you don't to anything else but exactly that, then you're probably fine04:29
wpwrakit's kinda like the stories you tell of OMAP :-)04:30
wpwrakwell, one good thing came out of all this - in the end, i used the ben for programming my c8051f32x chips, with a predecessor of ubb :-)04:32
wpwrakbtw, also the ft2232 has its bugs. if you use its regular big-banging mode, it can also "swallow" commands. i saw when making a remote reset tool for the openmoko debug board. hadn't realized the context back then, though.04:34
wolfspraulcan we run into those bugs with the m1 jtag-serial board?04:41
wpwrakif all you do are jtag and serial, then i think you should be fine04:41
wolfspraulgood04:42
wolfspraul:-)04:42
wpwrakif you control anything else, e.g., a led or so, you may have to send the command several times before it "takes"04:42
wpwrakhttp://svn.openmoko.org/developers/werner/dbgrst/dbgrst.c (set_high) and http://svn.openmoko.org/trunk/src/host/norwp/norwp.c (main)04:45
wpwraki don't know if these are really reliable, though. dbgrst should be pretty good, because i used it in scripted tests and would have noticed a failure rate > 10% or so. norwp never saw such intense testing.04:46
qwebirc39061.06:29
kyakmirko: hey, are you there?09:13
kyakmirko: the top Makefile is also not quite OK. eg. it has staging_dir/toolchain-mipsel_gcc-4.3.3+cs_uClibc-0.9.30.1/usr/include/QtGui in INCPATH. But this is not correct, must be staging_dir/target-mipsel_uClibc-0.9.30.1/usr/include/QtGui09:19
kyakmirko: but then you are right that further Makefiles have even more problems (for example, empty CC)09:20
kyakmirko: i'm trying it with "Tile" application, which is quite simple and has only one Makefile09:20
wpwrakkyak: btw, speaking of the openwrt build: after the discussion on build time, i tried to see how long it would take to run it on my pc. one thing i noticed is that the "jikes" dependency seems to be an anachronism that cannot be satisfied with any recent ubuntu. (the build failed subsequently somewhere in gcc, which is when i dropped the idea)11:51
kyakwpwrak: $ rpmquery -a |grep jikes returns nothing. Why do you have this dependency?11:53
wpwrakkyak: it's listed on xiangfu's page about the build process11:54
kyakbtw, at takes ~2.5 hours to build from scratch here. I don't build all the packages, maybe only half of them (my rootfs size was 256 Mb)11:54
kyakso i guess full build would take 5-6 hours11:54
kyakwpwrak: you should ask xiangfu then11:55
kyaki don\t have jikes installed on my build11:55
kyakwpwrak: i want to disassemble my Ben :)11:58
kyakneed to clean behind the LCD glass11:58
kyakthere is some dust and particles these11:58
kyakhow i can do this? is the top case disassemlable?11:59
wpwrakthe plastic is easy to remove. but you may scratch it in the process.11:59
wpwrakthe plastic sheet is just glued. you can lift it off with a knife.11:59
kyakoh. then i would need to prepare a glue to glue it back?12:00
wolfspraulkyak: I very much recommend you not to try this.12:01
wolfspraulI'd say chances are 90% or more that after you are finished, it will look worse.12:01
wolfspraulthe piece above the LCM module is called 'PC sheet', it's plastic12:02
kyakthat's bad.. this dust behind the LCD is bothering me12:02
wolfspraulit has black paint on the back side, if you try to lift it off (it is glued), you have to be extremely careful to not scratch off that black paint from behind12:02
wolfspraulyou mean behind the PC sheet? on top of the LCM module probably12:02
wolfspraulwell you can try, but here are 2 things you really need to watch big time12:02
wolfspraulfirst of all - the PC sheet is glued and you need some thing razor-like instrument to lift it off on the sides12:03
wolfsprauland when doing that, you have to be super careful to not exert pressure from below onto the PC sheet, because you will scratch off the black paint otherwise12:03
kyakit seems that these particles are on both PC sheet and LCM module12:04
wolfspraulassuming you are able to lift off the PC sheet without damaging it, you then need to be able to clean both sides (the top side of the LCM module and the bottom side of the PC sheet) so that it is cleaner than right now12:04
wolfspraulmost likely that will only work with some sort of air gun12:04
kyakwolfspraul: ok, so i won't do that :)12:05
kyaklooks too hard12:05
kyakand the result is unpredictable12:05
wolfspraulhttp://en.qi-hardware.com/wiki/File:Cleaning_the_PC_Sheet.jpg12:06
wolfspraulyou can see the PC sheet clearly in this picture, before it is being glued onto/around the LCM module12:06
wolfspraulit's not unpredictable, but the problem is that you have only 1 unit, so your margin for error is 0 :-)12:07
wolfspraulif you have 10 units, I'm sure you would get 7 or more of them super clean :-)12:07
kyakoh yes, this is the PC sheet12:07
kyaki wonder how the dust even got inside there12:07
wolfspraulone issue is that on the bottom side of the PC sheet, there is no protective film over the black paint12:07
kyakbetween the PC sheet and LCM12:07
wolfspraulso the black paint is really easily scratched off once you are underneath, which is where you have to go if you want to lift it off12:08
wolfsprauland even if you can lift it off without such scratches, how do you plan to clean both sides?12:09
wolfspraulthe smallest particle you leave will be visible in that spot :-) I doubt you can clean it with any type of cloth.12:10
wolfsprauldo you have some pressurized air?12:10
kyakno, i have nothing12:12
wolfspraulhttp://en.qi-hardware.com/wiki/File:Disassemble_LCD_protect.JPG12:12
wolfspraulhttp://en.qi-hardware.com/wiki/File:Disassemble_LCD_protect_detail.JPG12:12
wolfspraulhttp://en.qi-hardware.com/wiki/File:Disassemble_LCD_protect_glue.JPG12:12
wolfspraulsome pictures that may help12:12
wpwrakmaybe alcohol would work, too. "pressurized" air (in cans) is actually not air. it's some other inert gas.12:12
wolfspraulyes it may work, but I have never tried and from similar experiments I am careful to not overestimate my cleaning ability.12:13
wolfspraulit's amazing what small particle you see once it is pressed right in front of the bright backlight coming out of the LCM module12:14
wpwrak;-) yeah, i wouldn't recommend cleaning attempts if it's just for appearance. if the dirt is so thick that you can't see though it anymore, that would be different.12:14
wolfspraulcorrect, fully agree12:15
kyakwolfspraul: ok, i give up. Seems there is no easy way to do that. From my previous experience, this could end badly :)12:15
wolfspraulyou would need a few Bens to play with before you are really good at this12:16
wolfspraulI could send you a replacement PC sheet, I think I have some flying around somewhere, but that may not even fix your (much bigger) problems anymore once you have too much dirt there and no way to clean it properly.12:16
wolfspraulI suggest you wait until it gets really bad and you get into the needed 'what the heck' mood.12:17
kyakhehe, ok :)12:17
wolfspraulsuch as 'worst case, I continue to use my Ben without the PC sheet, then I can always easily swipe directly over the LCM module'12:17
wolfspraulonce you are there, I'd say you are ready12:18
wolfspraulxiangfu's personal Ben was like that for a while. plastic on top of the keyboard missing, keyboard held down with tape. pc sheet missing.12:18
wolfspraulcouple other parts missing or 'hacked'12:18
kyakthis must look cool12:19
mirkokyak: i think i've a solution :)12:27
mirkokyak: the issue is: the actual qmake command gets the (more or less) correct environment variables passed12:27
mirkohowever the created Makefile again calls qmake to create further Makefiles - this call isn't prefixed with proper variables..12:28
kyakhm, i see12:29
mirkokyak: http://pastebin.com/VhDwrEua12:29
mirkothis should help here12:29
kyaki will try it now12:30
mirkobe aware, qt4 gets cleaned and recompiled12:31
mirkoto avoid this, apply the patch to the file inside your staging_dir12:32
mirkoas "staging_dir/target-mipsel_uClibc-0.9.30.1/mk/qmake.mk"12:32
mirkootherwise the modification will be noticed and qt4 will be rebuilt12:32
kyakdamn, too late :)12:33
kyakok. shouldn't take very long12:39
mirko~1-2hours12:42
kyakmirko: NanoMap just built fine13:31
kyakwithout setting those values in Build/Prepare13:31
kyakmirko: Tile, however, won't build13:32
kyakstaging_dir/target-mipsel_uClibc-0.9.30.1/usr/include/QtGui (and similar, if needed) are not in INCPATH13:33
kyakthere is toolchain dir, which is not correct...13:33
kyakmirko: qstardict, qball also won't build without manually setting of those variables13:43
kyakmirko: same with gottet13:58
mirkookay, will take a look at those soon14:18
kyakok, thanks!14:20
qwebirc39061 /quit14:36
kristianpaulkyak: and the end you'll let the nanonote with no screws, so you can disamble it by quikc :-)15:09
kristianpaulhttp://en.qi-hardware.com/wiki/File:AVT2_small_run_expenses.jpg15:21
kristianpaulhttp://en.qi-hardware.com/wiki/File:AVT2_layout_boot_switch.jpg15:22
kristianpaulhttp://en.qi-hardware.com/wiki/File:AVT2_mechanical_boot_switch.jpg15:22
wpwrakreally good prices15:24
kristianpauli dint get what was wrong with avt2...15:25
kristianpauli mean to get produced..15:25
kristianpaulah ok http://en.qi-hardware.com/wiki/AVT2_RC1_Reference_Board#KNOWN_ISSUES15:27
kristianpaulwpwrak: from you experience whats the max long to extend a bus with logic signals around 10Mhz and 15pf of load per pin15:28
kristianpaulI'm about cut ribbon cable for mm1 <-> sige comunication but it should be a bit longer as the exp connector is not in the edge of the board.15:29
kristianpaulI can test signal anyway with scope (yay!)15:29
kristianpaulbut may be you already had experience to share abou this topic :-)15:31
wpwrakhmm, 10 MHz ... very very short :)15:31
wpwrakwill you have ground between signals ?15:31
kristianpaulyes15:32
kristianpaulsige is powered from mm115:32
kristianpaul(very very short) damn i know :/ i think i'll end fiting sige inside the mm115:32
kristianpaulmay be i can find a place for it next to the jtag-serial board..15:33
wpwrak(ground) i mean: are signal traces separated by ground traces ? or will you have crosstalk between signals ? also, do you have termination ?15:36
wpwrakinside the box would definitely be an advantage. a properly designed cable can go for maybe 1 m. a not so properly designed one probably just a few cm.15:37
kristianpaulwpwrak: separated by ground traces, yes15:37
kristianpaulwpwrak: termination, prone to interference..15:39
wpwrak(termination) so is this a yes or a no ? :)15:40
kristianpaulyes15:40
wpwrakokay, then it's probably safe to have a few tens of centimeters15:43
wpwrakif you consider IDE PATA cables, they also go quite fast and they're relatively long15:44
kristianpaulhmm, good point15:44
kristianpaulI already bougt ribon cable, but i'll take better my bus from a IDE PATA one15:45
wpwrakthe cable material is probably very similar15:49
kristianpaulIDE wires look thin than the colored-ribbon i have15:50
kristianpaulah, but connector differ so i'll use the colored-ribbon15:55
r-wosHi everybody! Quick question: has anyone successfully compiled ScummVM for the NanoNote yet?16:13
r-wosI tried to compile it directly on the NanoNote (mad, I know) - and it works up to the final link, which dies with a signal 10, bus error...16:15
wpwrakwow, bus error is a misalignment. these have gotten rare in recent years16:21
mirkowpwrak: you didn't use macosx recently did you? ;)16:21
r-wosno, but I did a quick search for "bus error", too ;-)16:22
wpwrakmirko: where's the crucifix and the garlic ?16:23
kyakr-wos: compiling it on Ben might be a bad idea, better set up the openwrt build environment and take over the broken scummvm packages from qi-openwrt-feeds16:30
r-woskyak: yeah, that's probably better... Hm, I thought I could get away without this cross-compile-juggling...16:34
mirkowpwrak: wp16:43
wpwrakmirko: 0x007f 0x007f ... ? what's that ? :)16:48
kyakmirko: pretty weird. I changed the order of directories here: http://pastebin.mandriva.com/21972 and it all works now16:55
kyaksorry, here: http://pastebin.mandriva.com/2197316:55
kyakmirko: it seems that qt4 is appending QtCore/QtGui/etc (whatever is needed) to the last directory in TARGET_INCDIRS16:56
kyakso it didn't work when toolchain dir was at the end.16:57
kyakmirko: it seems that QMAKE_INCDIR_QT is not designed to hold multiple directories17:38
kyakby default, it is $$[QT_INSTALL_HEADERS] which is a single directory17:38
kyakmirko: maybe it is correct to have $(STAGING_DIR)/usr/lib only in TARGET_INCDIRS (which goes into QMAKE_INCDIR_QT and then gets appended by QtGui/QtCore etc)17:45
kyakand then use QMAKE_INCLUDEPATH to add other directories that currently live in TARGET_INCDIRS17:46
kyak(btw, i think those other dirs are not really necessary)17:47
kristianpaulsteve|m: hey19:24
kristianpaulsteve|m: Do you know if the osl support vcd output?19:25
kristianpaulah, wait19:26
kristianpaulneat it does :-)19:29
kristianpaulamazing i can record ~2ms of data at 10Mhz19:38
kristianpaulbut nothing trusty, 50mhz should be ok19:48
steve|mhttp://hackaday.com/2011/02/20/rf-control-from-just-about-any-device/22:29
kristianpaul:-)22:30
kristianpaulmirko: !22:30
mirkothey misspelled my name --.--22:39
--- Mon Feb 21 201100:00

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