2010-08-23 00:00 (kicad) phew. lots of questions. i need some food first :) 2010-08-23 00:08 [kicad] sure, take your time. Most of these things are not technical anyway, rather process oriented and how multiple people and even vendors work together. 2010-08-23 00:08 that's as much a communication challenge as it is a technical challenge 2010-08-23 00:08 but we need to start/continue :-) 2010-08-23 00:11 also we need to have a process that is practical enough for the real people and vendors we are working with, not something that only works in theory 2010-08-23 00:22 heh. i would gladly be a tester for that. havend had time for working myself into kicad 2010-08-23 00:23 mostly because of every time i needed some real schem or pcb, some eagly guy was around who 'just did it' 2010-08-23 00:25 roh: just jump in. asking questions is the first step, that's basically what I do all day long :-) 2010-08-23 00:25 for kicad, my personal opinion is not everything needs to be stored in revision control 2010-08-23 00:26 it makes sense to have the schedule in the wiki, and to work with wiki as well in some other areas, like collecting errata, test reports, layout notes, etc. 2010-08-23 00:27 I would probably export generated files such as gerber/ai/stencil as well, and upload into the wiki. it's too hard to generate them from the real 'source' data, and depends on too many things that can break too easily, even with kicad version upgrades. 2010-08-23 00:27 wolfspraul i think the fist one is: users dont compile sw themselves. we need proper packages. 2010-08-23 00:27 but let's see, it's a big discussion and we try these things out step by step rather than just announcing the perfect process that actually is not perfect at all and nobody follows anyway :-) 2010-08-23 00:28 wolfspraul i got like 10 computers i would need kicad to work on, and i am not really in the mood for being the one doing packages for them soon 2010-08-23 00:28 why 10? and why do you need kicad on all 10? 2010-08-23 00:29 wolfspraul doing and using schematics is a team effort. and in our hackspace we got compuers for all and some for special purposes. and everybody got his own also. so i would need to install it on atleast 4 or 5 stationary computers and the rest is personal notebooks. all with a bit different set-ups 2010-08-23 00:29 luckily most of them ubuntu 2010-08-23 00:30 well that's why I think our projects server could auto-plot schematics into ps/pdf/png files on every commit. 2010-08-23 00:30 it doesn't work yet but that's one thing I think I already firmly have on my todo list 2010-08-23 00:30 then we see who uses it and how 2010-08-23 00:31 also where those files should be stored, whether it's important to keep old revisions around or not 2010-08-23 00:31 another problem is that kicad is not very scriptable, so this won't help us when it comes to the gerber files 2010-08-23 00:32 true. would want the same too, but its hard. kicad's pro arguments are a)text files instead of binaries in the svn and b) opensource tool. when it comes to 'usability' (which includes not needing to compile som svn snapshot) i think it still needs to go a long way. 2010-08-23 00:32 also I am looking for practical little steps, not some grand plan that means no real production for years :-) 2010-08-23 00:32 anyhow.. ubuntu has packages from 2007 or 2008.., i guess those are incompatible to what you guys do now? 2010-08-23 00:32 next run is Milkymist One RC2, and it will be Altium-based again 2010-08-23 00:32 can't wait for the KiCad process 2010-08-23 00:33 I don't know, different people use different KiCad versions, and upstream is not very open-minded about some patches either 2010-08-23 00:33 even Werner's --plot parameter to be able to auto-plot schematics was not accepted upstream :-) 2010-08-23 00:34 I think the whole GUI should be scriptable... 2010-08-23 00:34 wolfspraul my only point is: the people who can help you most with making process, are not the ones which are in the mood manually installing stuff in undefined versions and keeping that stuff up to date. thats what packages are for. in the end i would be happiily using some ppa if it includes packages for x86 and x86-64, for ubuntu 9.10 and 10.04 2010-08-23 00:35 yes I agree. That's what a lot of my mail is about, no? 2010-08-23 00:35 i am just the wrong guy to generate and care for that ppa. thats somebody who got a clue about kicad needs to do (package maintainer) 2010-08-23 00:35 the barriers of entry are too high right now 2010-08-23 00:36 how does gEDA compare to KiCad? 2010-08-23 00:36 ack. i dont know any cad sw which doesnt support scripting nearly everything 2010-08-23 00:36 gEDA was basically much too simplistic. 2010-08-23 00:36 even for hobbyists. atleast as i looked last time. 2010-08-23 00:37 and segfaulting enough that it annoyed me away. 2010-08-23 00:37 i think its a few years of work away from eagle and kicad when it comes to details. 2010-08-23 00:38 I've not seen it segfault ( installed from the Debian Lenny repos) 2010-08-23 00:38 geda got a nice thing tho: symbols are the same as schems. 2010-08-23 00:38 so to define a 'part' you can use the same tools and workflow as for using these 'parts' in bigger assemblys. 2010-08-23 00:38 kinda recursive, but very nice. 2010-08-23 00:39 unclouded i think lenny wasnt released when i last tried to use it really. 2010-08-23 00:39 does geda have converters for part libs now? 2010-08-23 00:39 shame it's not tough enough for professional work yet then.Â Â it's got everything I need but then maybe I'm not even a very advanced hobbyist ;) 2010-08-23 00:42 roh: sorry, I don't know. I create my symbols and footprints by hand with digital calipers 2010-08-23 00:43 probably not yet judging by this page: http://geda.seul.org/wiki/geda:format_translation 2010-08-23 01:02 ah.. there is a ppa https://launchpad.net/~paxer/+archive/ppa 2010-08-23 01:02 but it doesnt have werners patches i guess 2010-08-23 01:48 hello, my nanonote down when I update flash. 2010-08-23 01:48 The screen say In: serial Out: lcd Err: lcd 2010-08-23 01:48 Is there anybody can help me ? 2010-08-23 09:53 ah, interesting. stefan schmidt is working with the cc2420, the predecessor of the cc2520. it's a small world :) 2010-08-23 10:06 wow, didn't know 2010-08-23 10:06 wpwrak: some feedback from me regarding your brick&mortar vs. blueprint thoughts 2010-08-23 10:06 in general I think 'in the industry', there is a lot more reviewing and sharing of knowledge going on than you realize or acknowledge 2010-08-23 10:07 every good IC vendor offers real solid FAE tech support 2010-08-23 10:07 every module maker, even makers of passive components! 2010-08-23 10:08 so by going to a KiCad process (which we will), I just want to be very clear where we make it harder for others to look over our shoulder, or how we can avoid such complications in advance, and provide them with the files they need to be able to help 2010-08-23 10:08 the typical thing as in sharing full EDA files won't work if we use KiCad, because nobody will be able to deal with them 2010-08-23 10:09 but if we know this in advance, and can share PDF schematics, or PDF/PNG/GERBER/whatever other general format layouts, I think we can still achieve 90% of the benefits without sacrificing our tool of choice 2010-08-23 10:10 I'm somehow too lazy to post this to the list, I wait until we have some hard facts and specific cases, then we can discuss a specific solution 2010-08-23 10:10 rather than trading opinions forever 2010-08-23 10:11 and yes - PCB makers (depending on who it is and what the nature of the project is) may have excellent people on staff, I won't even start with the SMT places 2010-08-23 10:12 if we cut ourselves off from all these experience people - not good :-)000 2010-08-23 10:15 (reviews) okay, FAEs and such, yes. i just didn't fine the pcb/smt example very convincing, since these are production shops, not really engineering shops. 2010-08-23 10:15 ah, you just wrote about it :) 2010-08-23 10:15 (catching up) 2010-08-23 10:15 not true 2010-08-23 10:15 it's your choice of course 2010-08-23 10:16 you can work with a shop and keep them blind as in "hey I know everything, just run this file through your machine and send me the results" 2010-08-23 10:16 people that work like that often will then start into an endles rant about 'incompetencies' 'stupid mistakes' etc. etc. 2010-08-23 10:17 the people that I know work in actual real production (=high volume) all don't work like that 2010-08-23 10:17 this is not how the industry functions, really 2010-08-23 10:17 the modus operandi i know of are those mail order pcb shops. you send them the files, they somehow toss them into their machines, out come the pcbs. even with electrical testing and such. they really don't seem to care what you send them, as long as it's a gerber or they can generate one from it. 2010-08-23 10:17 it's the opposite 2010-08-23 10:17 oh sure 2010-08-23 10:17 and of the x billion devices made each year, how many do you think are made in this way? 2010-08-23 10:17 no idea :) 2010-08-23 10:17 I don't know either, but I'd say 0.01% or so 2010-08-23 10:18 there is a lot of 'experience' traded all the time, you really need to hook into tha 2010-08-23 10:18 that 2010-08-23 10:18 and it is offered, constantly 2010-08-23 10:18 because everybody knows without it you cannot be successful anyway :-) 2010-08-23 10:18 but what feedback do you get from the pcb shop ? any process limitations really ought to be in your design parameters 2010-08-23 10:19 we need to look into a specific case, definitely me personally I can't give you a convincing example now 2010-08-23 10:19 of course it also depends on the complexity of the board, the riskiness etc. 2010-08-23 10:20 my point was just that by using a non-standard EDA tool, we need to take extra precautions to not cut ourselves off from valuable help 2010-08-23 10:20 I think as long as we do that, we can minimize the damage. PDFs and other commonly used file formats will probably do as well, if we only have them. 2010-08-23 10:21 i can't really imagine things much beyond "here's what we want to do. we know it's tricky. what do you recommend ?". but you wouldn't need access to the design files for that. maybe a screenshot. 2010-08-23 10:21 yes, but don't underestimate the power of habits 2010-08-23 10:21 ;-) 2010-08-23 10:21 they are used to getting the PADS files 2010-08-23 10:21 checking here, checking there 2010-08-23 10:22 everybody has licensed or pirate copies of all standard EDAs on their notebooks 2010-08-23 10:22 i can see a benefit in having them as a means for making sure people really sent them all the files and everything in a consistent state, yes 2010-08-23 10:22 you are just putting yourself out of that universe 2010-08-23 10:22 my point is more about cutting yourself off from feedback 2010-08-23 10:22 and the dangerous thing is - you will never know if there would have been feedback :-) 2010-08-23 10:22 the problem with those big names EDA tools is that they create a huge barrier for cooperation inside our projects 2010-08-23 10:22 yes sure, no worries 2010-08-23 10:23 no need to discuss with me 2010-08-23 10:23 I just wanted to alert you to the maybe unimaginable level of personal support that is typically happening. 2010-08-23 10:23 really 2010-08-23 10:23 maybe they make it easier to talk to the outside, but you're probably trading a small evil for a big evil 2010-08-23 10:23 it's irrational sometimes 2010-08-23 10:23 I was in endless such meetings. 2010-08-23 10:23 hi can someone tell me what is intel i5 750 equivalent amd processor? 2010-08-23 10:23 wpwrak: give you one example - Murata 2010-08-23 10:24 in the phone of our former employer, the Murata components added up to 10 cents or so, we had almost none 2010-08-23 10:24 johnny6: wrong channel 2010-08-23 10:24 yet we were offered access to their FAE, at any time, access to their lab, lab equipment, etc. 2010-08-23 10:25 we seem to attract a lot of "pc shop" type of questions. i wonder if people think there is a place called "qi hw" at the mall ;-) 2010-08-23 10:25 I directly asked them how they could ever hope to make this back with our bom and volume, but they said they don't work like that. 2010-08-23 10:25 wpwrak: I think they think qi somehow means questions ;) 2010-08-23 10:25 wolfspraul: sounds great. true idealists even :) 2010-08-23 10:25 their assumption is that without this type of inside-out support, their customers won't be successful anyway 2010-08-23 10:25 no 2010-08-23 10:25 this is how the industry works 2010-08-23 10:26 you can complain about mistakes in datasheets the rest of your life 2010-08-23 10:26 ever wondered why people can still produce and sell working stuff today? 2010-08-23 10:26 wolfspraul: with time, you acquire the skills to interpret the data sheet bugs :) 2010-08-23 10:26 because they work in their basements? ;) 2010-08-23 10:26 you think from the perspective of a strong individual with vast knowledge 2010-08-23 10:27 but most companies are not run like this, it's more like lots of people throwing all they know together to maybe make a working product 2010-08-23 10:27 but lo and behold, they are actually quite successful that way 2010-08-23 10:27 so anyway, my style would be to always keep the possibility open to get support 2010-08-23 10:28 welcome all offers, make sure we have what they need to look over our shoulders 2010-08-23 10:28 i don't think we're creating much of an obstacle there. for most things, you should be able to explain them with pencil and paper. 2010-08-23 10:29 well whether it's an obstacle or not I would like to hear from them 2010-08-23 10:29 KiCad definitely is an obstacle 2010-08-23 10:29 there are areas where things get trickier. but then you're probably beyond the limits of the "standard" EDA tools already. (emi/rf simulation and such) 2010-08-23 10:29 nobody, literally nobody, will be able to 'quickly' open that file and give you some feedback 2010-08-23 10:30 kicad is unfamiliar and they're not using it. so i think you'll be fine as long as you don't make them think you want them to invest resources in having kicad too. 2010-08-23 10:30 oh of course not 2010-08-23 10:30 but you want them to be able to look over your shoulder as I said 2010-08-23 10:30 so if KiCad doesn't work, it needs to be in other common file formats 2010-08-23 10:31 gerber, pdf, ... 2010-08-23 10:31 yes 2010-08-23 10:32 it probably does take more effort to prepare the questions that way. you can't just dump them the design files and say "who does it not work ?" 2010-08-23 10:32 s/who/why/ 2010-08-23 10:32 exactly 2010-08-23 10:32 stupid high-level questions 2010-08-23 10:32 "which of your 25 ICs should I use here?" 2010-08-23 10:32 but then, this degree of effort is probably already necessary to run open projects in a sane way 2010-08-23 10:34 some of these questions may simply be a means to avoid giving an answer based on insufficient information by making the question somebody else's problem ;-) 2010-08-23 10:34 if you pass it around long enough, someone will just shrug and give an answer, even if they don't quite know either :) 2010-08-23 10:34 it's about finding someone with experience to share their snippet of experience with you 2010-08-23 10:35 as opposed to assuming over time you will know everything anyway and everybody else just needs to press buttons 2010-08-23 10:35 yes. most of the time, what you really want to ask is "what do all the others use here ?" :) 2010-08-23 10:35 the first approach will bring you to an actual production of sellable goods soonder, the second approach much much later, even though you are definitely wiser by then :-) 2010-08-23 10:35 yes, that also 2010-08-23 10:35 there are many things really 2010-08-23 10:35 everybody tries to help, share what they know 2010-08-23 10:36 but if the process and tools are uncommon, that gets a lot harder 2010-08-23 10:36 that was my only point 2010-08-23 10:36 it's managable I think 2010-08-23 10:36 well, i prefer to have that knowledge in the open part of the project. that way, it can be reused. 2010-08-23 10:36 in fact we need to create those 'easy viewable' files ourselves, because even for many of our contributions the KiCad barrier is too high 2010-08-23 10:36 yup 2010-08-23 10:37 yes true, of course, nobody is against that. but at some point I assume you also want to manufacture something. 2010-08-23 10:37 nobody is against documentation, it's just that things are constantly moving forward, so many people don't take the time 2010-08-23 10:37 you need to pull in every bit of help you can get, that's my experience 2010-08-23 10:37 as you know I have worked at some real big disaster projects :-) 2010-08-23 10:39 like I said, I think if we create those easily vieable/reviewable file formats for ourselves, the same stuff should work for outside vendors who want to help 2010-08-23 10:39 but I need to double check in specific cases that we don't lock someone out by using KiCad 2010-08-23 10:39 well, documentation in a very broad sense. probably nobody will sit down and write a book. but if you need to solve X and you see that there's something similar to that X already in another project, then it's nice if you can find out about the reasons why that X was done in this or that way. 2010-08-23 10:40 sure 2010-08-23 10:40 we spent quite a bit of time on those layout notes, even you think it's too much? :-) 2010-08-23 10:40 (viewable files) yes, i fully agree. like the things i generated in gta02-core. 2010-08-23 10:43 with the layout, where did things actually fail ? did anyone actually try to make a layout yet ? 2010-08-23 10:43 (manually, with the non-Free push router, whatever) 2010-08-23 10:43 don't know, too many projects and too long ago 2010-08-23 10:44 ;-) 2010-08-23 10:44 the next one is Xue, which is also getting a bit chaotic already. 2010-08-23 10:44 well open process is hard 2010-08-23 10:44 but we try, definitely some good pieces now 2010-08-23 10:44 i think people just need to get their hands dirty. then we can discuss concrete problems. right now, all the layout discussion is a bit too abstract for my taste. 2010-08-23 10:44 I will take a closer look at how you created 'easily viewable' files in 02-core 2010-08-23 10:45 chaos is normal :) 2010-08-23 10:46 so my point about industry help got across? 2010-08-23 10:46 you don't feel it that much in Buenos Aires probably, digi-key, pcbcart.com, etc. 2010-08-23 10:46 one small gotcha: the easily viewable files are just schematics. no layout there so far. 2010-08-23 10:46 but that's not how things work once you are in Taipei, Shenzhen, Shanghai, etc. personally 2010-08-23 10:46 there are FAEs everywhere, 'open labs' 2010-08-23 10:47 they _WANT_ you to use their lab to test their stuff 2010-08-23 10:47 they have their people sitting there trying to help you 2010-08-23 10:47 you just call a bit in advance, a few days, like with a doctor, so they reserve time for you 2010-08-23 10:47 you go there with your stuff, and of course their lab is perfectly equipped around whatever ICs they are selling 2010-08-23 10:47 (industry help) yup, got it. but i don;t really think we're creating a huge barrier. of course, we shouldn't be arrogant and, say, demand that Murata review our KiCad files. 2010-08-23 10:48 yes exactly, that will never happen :-) 2010-08-23 10:48 maybe some day, if it gets popular enough. but we're not quite there yet :) 2010-08-23 10:50 (easily-viewable files) not sure what's best for the layout. very often, you actually do need the pcb program to navigate the layout in a meaningful way 2010-08-23 10:51 e.g., if you just have a stack of gerbers or - worse - pdfs, it's incredibly painful to trace tracks across vias 2010-08-23 10:51 so that's for reviewing the routing and such 2010-08-23 10:52 if you have things like ground clearances, emi in general, or specific patterns (footprints, thermal, etc.), gerber/pdf are fine 2010-08-23 10:53 pdfs are also quite handy for spotting inefficient routes in the x/y plane. sometimes you don't see them on the screen, but they pop right out on a print. 2010-08-23 10:55 yes but that also makes a strong point for people sharing in file formats they know, specifically the (proprietary) EDA formats 2010-08-23 10:55 what definitely should be there is the component placement. maybe also with a map for going from component reference to position 2010-08-23 10:55 because they will already know how to work with these tools in the way that will make it easiest for them to spot what they wanted to go after 2010-08-23 10:56 but if you give them flat/umodifiable PDF/PNG/whatever, that may be a lot harder, or at least unusual 2010-08-23 10:56 and don't ask me for a specific example, I don't have one, I am just saying this 'in general' 2010-08-23 10:56 if they have to debug our routing, yes :) 2010-08-23 10:57 because the normal thing you would share are just PADS/OrCad/Altium files 2010-08-23 10:57 it could be all sorts of things, down to footprint issues, mechanical placement, heating, EMI, RF, whatnot 2010-08-23 10:57 you don't know in advance, that's my point 2010-08-23 10:58 if I cannot share a PADS file with someone, I also don't know what that person might have been able to help me with. But he says "oh, if you had a PADS file I could do all this wonderful stuff for you" 2010-08-23 10:58 well yes. why get out of bed. you don't know what might happen ;-)) 2010-08-23 10:58 "everybody else has that, you don't" 2010-08-23 10:59 I am careful to dismiss such things as "nothing there anyway". 2010-08-23 10:59 and they can probably do even more wonderful stuff if you use windows :) 2010-08-23 10:59 and you wouldn't believe all the incredible things they can do with those binary-only drivers 2010-08-23 10:59 and if you sign just that NDA, then .... 2010-08-23 10:59 like I said, from my experience, a number of years now, I get more and more careful to dismiss it. 2010-08-23 10:59 it's just not compatible with an open process 2010-08-23 11:00 no the NDA in most cases I know is totally separate 2010-08-23 11:00 all people working on real tech issues hate NDAs 2010-08-23 11:00 binary-only drivers, same, no. we are talking about the ee side here. 2010-08-23 11:01 I am not talking about salesmen who try to sell you some strange usb dongle. they will give you the 'windows driver' story :-) 2010-08-23 11:01 you're basically trying to organize your process from the outside in. so you maintain compatibility with .. 2010-08-23 11:01 mompls 2010-08-23 11:02 no, I just want to be careful to not exclude valuable input. 2010-08-23 11:03 but we are aware of it already, I will continue to dig and see what we can do to get all valuable things out... 2010-08-23 11:03 and if there is nothing, then there is nothing... also good. 2010-08-23 11:09 wpwrak: if Adam has a bit of time, I will ask him whether we got anything back in the Milkymist One RC1 run we had. That one used Altium and only 3 or 4 vendors helped I think. 2010-08-23 11:10 Xilinx FAE, SMT, PCB, layout house 2010-08-23 11:10 maybe he still remembers whether anyone provided valuable input or not, and if so whether they would have also come in had we used KiCad 2010-08-23 11:14 back. had to get a delivery 2010-08-23 11:14 ah yes, that would be a good case study 2010-08-23 11:15 wpwrak: last one for today - you will like it's it a bit subversive... 2010-08-23 11:15 even the component distributors (!) act as hubs for full EDA files 2010-08-23 11:15 how so? 2010-08-23 11:15 wow ;-) because people can't find the "BOM" button ? :) 2010-08-23 11:15 because they are under pressure from their customers (who are buying the components) to give them some copy/paste starting points 2010-08-23 11:15 no no 2010-08-23 11:16 starting points ? 2010-08-23 11:16 so you get the component, and you also get the EDA files of other products using that component :-) 2010-08-23 11:16 hihi ;-) 2010-08-23 11:16 yeah, welcome to the real world 2010-08-23 11:16 well, it's not too different from what i'd hope would happen with projects centered around open tools :) 2010-08-23 11:16 at least those guys are not claiming they 'review' anything 2010-08-23 11:16 oh, in China it's all open already 2010-08-23 11:17 also all software is free software already 2010-08-23 11:17 binaries are preferred over source, more practical (doesn't need the compile step) 2010-08-23 11:17 (everything "free") rms' dream come true :) 2010-08-23 11:17 he wouldn't like the binaries, though 2010-08-23 11:18 except free speech ;) 2010-08-23 11:18 but anyway, yes, seriously, the distributors also play an important role here, they can give quite a bit of guidance as to common/uncommon components, and also 'reference' implementations 2010-08-23 11:18 but that's not such a big concern for me on the KiCad side 2010-08-23 11:19 I'm more worried about FAEs 2010-08-23 11:19 common/uncommon is the one i tend to worry most about. luckily, price is often a good indicator. if it's common, it's usually cheap. if it's expensive, it won't become common. nice self-reinforcing feedback :) 2010-08-23 11:19 I call it a day, cu tomorrow 2010-08-23 11:20 if you interact with the faes directly, if all else fails, you can just show them on your laptop :) 2010-08-23 11:20 price is not a good indicator, ICs have different marketing strategies too 2010-08-23 11:20 that's because many IC companies also act as a sort of 'peacemaker' among their customers 2010-08-23 11:21 (price) not even for the same functionality ? (as in "almost the same") 2010-08-23 11:21 some ICs are sold in a tightly controlled way, others are just dumped openly to anyone who pays 2010-08-23 11:21 okay, the former aren't so good to have anyway 2010-08-23 11:22 only with very high volume ICs, say those in phones, you are really sure they cannot control distribution anymore, it's just overwhelming 2010-08-23 11:22 likely come with non-public documentation and all that 2010-08-23 11:23 i'll read up tomorrow... n8 2010-08-23 11:23 cya ! 2010-08-23 11:47 is there a way to dualboot on the nn yet? 2010-08-23 11:50 because if I can I might try jlime, but if I can't.. openwrt is still my primary dev target 2010-08-23 12:04 bartbes, you can install jlime to your SD-card 2010-08-23 12:05 bartbes, then boot from the SD-card by holding M while powering up the NN 2010-08-23 12:05 I briefly saw something about partitioning it, that seems a bit over-the-top though 2010-08-23 12:05 Yeah, you need a swap partition 2010-08-23 12:05 otherwise it sounds great 2010-08-23 12:05 though you might be able to have the swap as a file on the main partition 2010-08-23 12:05 i think that requires a kernel recompile though 2010-08-23 12:05 :( 2010-08-23 12:05 ask in #jlime 2010-08-23 12:06 maybe rafa will be a nice guy and hook you up :P 2010-08-23 12:39 [commit] Werner Almesberger: Renamed fw/boot/ to fw/atspi-boot/. Fixed build problem in fw/atspi-boot/. http://qi-hw.com/p/ben-wpan/d6b045c 2010-08-23 12:39 [commit] Werner Almesberger: Moved fw/ and tools/ into atrf/ as well. http://qi-hw.com/p/ben-wpan/c2dd238 2010-08-23 13:27 [commit] Werner Almesberger: Registers and register values are now CPU-specific. http://qi-hw.com/p/f32xbase/1c7150c 2010-08-23 13:32 [commit] Werner Almesberger: - fw/common/regs-f320.h: include mcs51/C8051F320.h, not mcs51/C8051F326.h http://qi-hw.com/p/f32xbase/0f1c0ea 2010-08-23 13:34 [commit] Werner Almesberger: Basic framework for CNTR firmware. http://qi-hw.com/p/ben-wpan/5109e0b 2010-08-23 13:37 bartbes: hey you can use the jlime without swap. The swap is needed to install more stuff. But if you want to use the current applications there you can try without swap. 2010-08-23 13:37 swap as permanent storage? 2010-08-23 13:38 ? 2010-08-23 13:38 "The swap is needed to install more stuff." 2010-08-23 13:38 anyway, that sounds better 2010-08-23 13:38 does it still need to be ext2 though? 2010-08-23 13:39 bartbes: you can set swap just when you need to install more stuff. But like the release mail says.. we need to fix that, because still with swap opkg takes years until it install something. You can enable/disable swap, it is not needed to be permanent. 2010-08-23 13:39 oh 2010-08-23 13:39 opkg needs the swap 2010-08-23 13:40 ah 2010-08-23 13:40 bartbes: if you want to try a swap file on sd try with (for example and just an example): cp /boot/uImage /swapfile; mkswap /swapfile ; swapon /swapfile 2010-08-23 13:40 if that works, whatever FS you have on sd, then you can do a bigger swapfile 2010-08-23 13:41 cool 2010-08-23 13:41 and i don't need a modified image either? 2010-08-23 13:42 modified image?.. for what? .. no .. you are enabling swap manually, or I am not following you 2010-08-23 13:42 to boot from 2010-08-23 13:42 but I guess that's a no 2010-08-23 13:42 no.. you just use or ubi or tar.gz for nand 2010-08-23 13:42 the ones on jlime.com 2010-08-23 13:42 for nand? 2010-08-23 13:43 bartbes: you need to install jlime. You can install jlime on nand or on sd 2010-08-23 13:43 yes, but I want it on sd 2010-08-23 13:43 bartbes: have you already installed? 2010-08-23 13:44 the goal of this all is *not* removing openwrt 2010-08-23 13:44 bartbes: there are two images: a jlime ubi image to install on nand, and a jlime.tar.gz image to install on sd. You choose 2010-08-23 13:44 bartbes: then you need to use jlime tar.gz to install on microSD 2010-08-23 13:45 just do the ext2 partition on sd and ungzip untar the tar.gz image inside. And you need do that as root from your pc or it will fail to boot 2010-08-23 13:45 and you will have jlime on microsd without modify openwrt 2010-08-23 13:46 or at least chown 2010-08-23 13:46 yeah, that makes sense 2010-08-23 13:46 no chown 2010-08-23 13:46 that doesn't work? 2010-08-23 13:47 if you ungzip/untar as NO ROOT your user will not be able to create devices files 2010-08-23 13:47 ah 2010-08-23 13:47 gives up on fat32 and formats a small bit of his sd card with ext2 2010-08-23 13:50 bartbes: qi uboot just read first ext2 partition IIRC. So if your microSD is 1) vfat 2) ext2 perhaps uboot will not find the system 2010-08-23 13:50 I'll try 2010-08-23 13:50 windows does the same thing I believe 2010-08-23 13:50 and I guess most devices do 2010-08-23 13:50 you need 1) ext2 2) fat I guess 2010-08-23 13:50 yeah, but windows only reads the first partition 2010-08-23 13:50 most devices do 2010-08-23 13:51 haha.. but uboot could read other partitions.. just that it needs some work perhaps in the source code. 2010-08-23 14:03 rafa: it indeed needs the first partition to be ext2 2010-08-23 14:04 bartbes: you can get a 256MB microsd for $10 or less I guess 2010-08-23 14:04 you can't get those here 2010-08-23 14:05 it's hard to find a 2gb one here 2010-08-23 14:05 so I got a 4gb one 2010-08-23 14:05 or install on nand 2010-08-23 14:05 ;) 2010-08-23 14:05 yeah, but I *need* openwrt 2010-08-23 14:05 to survive? 2010-08-23 14:05 I was hoping I could 'try' jlime 2010-08-23 14:05 it's my dev target 2010-08-23 14:11 ah well, it was worth a try 2010-08-23 17:56 [commit] Bas Wijnen: working boot loader http://qi-hw.com/p/iris/aef8331 2010-08-23 18:56 [commit] Werner Almesberger: Added timer register values. http://qi-hw.com/p/f32xbase/d1b3966 2010-08-23 18:56 [commit] Werner Almesberger: Completed the basic counter firmware. Now it just has to work ... http://qi-hw.com/p/ben-wpan/5514c33 2010-08-23 20:28 [commit] Werner Almesberger: Minor comment cleanup. http://qi-hw.com/p/f32xbase/d254ca8 2010-08-23 20:49 [commit] Werner Almesberger: atspi-trim: utility to set the crystal trimming capacitors. http://qi-hw.com/p/ben-wpan/b770051 2010-08-23 21:15 wolfspraul: so i still have to "sell" idbg to you ;-) 2010-08-23 21:15 already did 2010-08-23 21:16 have you used it ? 2010-08-23 21:17 (sell) i mean for Ben II/Ya/... . doesn't make the "likely features" lists yet :) 2010-08-23 21:18 need to see how expensive it really is later, and what the value is. 2010-08-23 21:18 I cannot tell now, the whole 'Ya package' is not clear to me. 2010-08-23 21:19 but it's very good taht we have it, and we should think about whether we can make some as 'Ben hacker giveaway' or something like that. 2010-08-23 21:19 c8051f326 plus a usb connector (micro or mini, whatever you prefer) 2010-08-23 21:19 that's the first thing I think about 2010-08-23 21:20 yes, it's not very easy to install, but sufficiently doable that many people who currently solder other kinds of serial consoles might do it 2010-08-23 21:21 wpwrak how much are these c8051f326? (in Eur) 2010-08-23 21:21 (not very easy) unless you're use to do this sort of stuff. then it's nothing special. 2010-08-23 21:21 +~ 2010-08-23 21:21 roh: hmm, where's a distributor with prices in EUR ? :) 2010-08-23 21:22 btw.. my avr-soft-usb is working 2010-08-23 21:22 roh: digi-key want USD 2.851 a piece if you buy 1000 units 2010-08-23 21:23 roh: how good is it doing usb to serial conversion at 57600 bps ? 2010-08-23 21:24 dunno. will need to test 2010-08-23 21:24 but i got a free uart, so i can use it for that 2010-08-23 21:25 roh: did you know that there is an entire AVR soft-ic in Milkymist now, to implement the USB OHCI protocol? 2010-08-23 21:25 called Navre 2010-08-23 21:26 hihi 2010-08-23 21:26 well.. its a really nice to 'frickel' platform 2010-08-23 21:26 gives fast result 2010-08-23 21:26 makes people enthusiastic and do more 2010-08-23 21:28 silab's direct orders list USD 2.81 for 1000 units 2010-08-23 21:29 roh: what is the frickel platform? 2010-08-23 21:29 you mean Milkymist? 2010-08-23 21:29 roh: avr is nice. and the peripherals are bit less quirky than those of many competitors 2010-08-23 21:29 it seems to be tested and working with 38400bps with a 12mhz crystal 2010-08-23 21:29 roh: needing a crystal isn't nice 2010-08-23 21:30 wolfspraul do you know the german term 'frickeln'? - to tinker ;) 2010-08-23 21:30 wpwrak high baud always need crystals or arent really stable/exact 2010-08-23 21:30 roh: not if you can sync your clock from USB ;-) 2010-08-23 21:30 had bad experiences in generic with rc and or even slightly (3%) detuned bauds 2010-08-23 21:31 well.. thats a feat. true. doesnt make me like 8051 ;) 2010-08-23 21:31 yes sure I know 2010-08-23 21:31 but what is the frickel platform? 2010-08-23 21:31 avr? milkymist? 2010-08-23 21:31 wpwrak avr in generic. arduino or not 2010-08-23 21:31 roh: at the end of the day, it's all C ;-) 2010-08-23 21:31 most people use 'frickel platform' as a negative term, but I just thought one could just market it like that :-) 2010-08-23 21:32 redirects roh's last message to wolfgang :) 2010-08-23 21:32 wpwrak yes. but peripherals are quirky.. always ;) 2010-08-23 21:33 wpwrak i think thats the biggest gainer arduino does. abstracting the peripherals. make you not need to worry about them. call stuff. 2010-08-23 21:33 roh: agreed. the 8051 peripherals suck more than avr's. currently wresting with the c8051f320, a bigger brother of the c8051f326. still have to at least get it to light a LED ... 2010-08-23 21:34 well, you could have a cover-up library on any architecture. i like it that the silabs usb mcus don't need a crystal. makes it so much easier to use them. 2010-08-23 21:35 wpwrak: you were right. In the last Milkymist One RC1 run, neither the PCB nor SMT factory requested the complete Altium files, or would have provided any valuable feedback based on those files. 2010-08-23 21:36 wolfspraul: good. i would have found that a little scary :) 2010-08-23 21:36 it was all based on standard files being traded, gerber, ai, bom, parts etc. 2010-08-23 21:37 wolfspraul: FAEs and the like, there it makes sense. but factories, not really. 2010-08-23 21:37 I will still be careful that our free tools don't unnecessarily exclude us from standard industry practices. But no worries, no fear tax etc... 2010-08-23 21:37 heh ;-) 2010-08-23 21:38 we did work with a layout outsourcing house last time, and that would have been impossible outright with KiCad 2010-08-23 21:38 but that's another story, that part of the work we want to get under control ourselves anyway 2010-08-23 21:41 yup. having to rely on outsourcing for this isn't good. 2010-08-23 21:47 would the SIE be a suitable platform for that Uwe Dippler ? he's looking for something to tinker with 2010-08-23 21:47 or is it too FPGA-centric ? 2010-08-23 21:52 I doubt it. 2010-08-23 21:52