#qi-hardware IRC log for Tuesday, 2012-05-29

wolfspraulgood morning everybody01:15
wolfspraulcladamw: hi! :-)01:15
wolfspraulI am reading over the m1 schematics and I am so impressed about their quality!01:15
rjeffriesThis project is new to me. seems very open, but I am not expert on licenses etc.: http://www.ethernut.de/01:15
cladamwgood morning 01:15
wolfspraulthat is really fun to read through and think about, one can feel it's not the typical sloppy junk with all sorts of loose ends and "don't cares"01:16
wolfspraulthank you (and Werner) so much for your excellent work on this01:16
cladamwthe quality derived from werner's suggestions. :)01:16
wolfspraulI have a few small questions here and there, clueless as I am about circuits01:16
wpwrakand adam's relentless work on applying them with great care :)01:17
wpwrakcladamw: i'm actually a bit puzzled how you avoided making a lot more mistakes :) i mean trivial things that you just overlook. after a while, i simply don't see my own mistakes anymore. but try as i might, i found almost nothing in your work.01:19
wpwrakwolfspraul: and thanks :)01:19
rjeffrieswolfspraul when do you plan to have thge new itereatuion of MM available for customers to oreder?01:19
qi-bot[commit] Adam Wang: ir.fpd: added front edge silkscreen (master) http://qi-hw.com/p/kicad-libs/b88a13301:19
wolfspraulok I start a little01:20
wolfspraulon the pdf side, we already know we could be better in technical things like searchable text, links, page titles, what not01:20
wolfspraulsome things I noticed on the power supply page01:21
wolfspraulU11 pin4 "Inhibit" ends in a little 'x'01:21
wolfsprauldoes that mean it's not connected/floating?01:21
wolfspraulI would think so but that may not be 100% clear01:21
wpwrakall those little "x" mean unconnected01:21
wolfspraulmost of my feedback is about 'understandability'01:22
cladamwwpwrak, hehe ... i eventually got used to your rules but still not convinced myself to good quality once I saw all explanations you mentioned. All reasonable !01:22
wpwrakit's something you put there manually, so that kicad warn you if you forgot to assign some pins01:22
wolfspraulsomething I noticed in the 5V bottom-left part - D15 says "5.6Vdc" but C142 is just "100n"01:23
wolfspraulso the cap doesn't have a unit, but the voltage does01:23
wpwrakyes, we use the compact unit-or-multiplier-as-decimal-point notation for the most common values01:23
wolfspraulonly occured to me because the two are right next to each other01:24
wpwrakwe only switch to regular SI style for the more exotic parts01:24
wolfspraulbut not for 5.6Vdc? :-)01:24
wolfspraulvoltage is not exotic01:24
wpwrakR and C are a lot more common than zeners :)01:24
wolfspraulI agree to leave the Vdc there btw, but it makes me feel sad about the missins 'farad' :-)01:24
wolfspraulsure sure, just saying because I saw those two back to back01:24
wpwrakbesides, zeners have multiple voltages. so the "dc" does convey information.01:24
wolfspraulit felt wrong01:24
wolfspraulnext one - I noticed those blocks "POWERED" on nets01:25
wpwrakdo you also feel bad about the missing Omega for resistors ?01:25
wolfspraulI'm wondering whether that is more helpful or more confusing01:25
wolfspraulI understand the system01:25
wolfspraulbut you have to admit seeing those "5.6Vdc" and "100n" next to each other feels unbalanced01:25
wpwrakso our instincts work in similar ways ;-)01:25
wolfspraulit feels like "one of them must be wrong"01:25
wolfspraulsmall detail01:26
wolfsprauljust my impressions01:26
wolfspraulbasically we want to clarify where the power lines are01:26
wpwrakmy instinctive response to 100n is also that it's a little odd. but i have a hard time coming up with a rule that would require an F for caps but no Omega for resistors.01:26
wolfspraula "strong and continuous" signal :-)01:26
wpwrak"POWERED" is to help kicad with error checking01:27
wolfspraulI find that "POWERED" confusing because I think there must be a wire/net "POWERED" somewhere01:27
kristianpaulhi wolfspraul :)01:27
wolfspraulis it?01:27
wolfsprauluntil I understand it's more like a comment/note01:27
wpwrakyou can have pins that are defined as "power input". you must connect them to a "power output"01:27
wolfsprauland even then it's quite fuzzy since "powered" is still about electrons and physics as everything else01:27
wpwrakthat can be a pin marked as such, or some other pin with the POWERED on top01:27
wolfspraulagain, just describing my confusion01:27
wolfspraulbut eventually I got it01:28
wpwrakit's a confusing concept :)01:28
wolfspraulall sorts of lines are marked "POWERED" :-)01:28
wolfspraulnot sure what to think of this01:28
wpwrakkicad makes it even more confusing by just having a mysterious diamond-shaped symbol01:28
wolfspraulsome devices can be powered using just the energy coming over signal wires...01:28
wpwrakat least we got rid of that one :)01:28
wolfspraulyou could argue this forever01:28
wolfspraulpower is flowing all over01:28
wpwrakthink of it as a declaration01:28
wolfspraulyeah but very strange one01:29
wpwraksort of a comment for kicad01:29
wolfsprauldon't know what the point should be01:29
wolfspraulyou could easily take this concept apart I think01:29
wolfspraulthinking about something as simple as a PWM circuit01:29
wolfspraulso first it's not a "powered" line, and then it is? after the lowpass filter?01:29
wpwraki consider it borderline useful, too. but that's what you have to do to get kicad to be happy with your design01:29
wolfspraulsure sure01:30
wolfsprauljust my feedback01:30
wolfspraulhere's another one...01:30
wolfspraulwe have a lot of ICs, on the power page for example "TPS76301"01:30
wolfspraula regulator? so I am wondering whether ICs should have 2 more fields in the schematics, just text: 1) type 2) maker01:30
wpwrakif you run an ERC on, say, atben, you'll find more weirdness. nasty things i didn't work around. such as "output" pins that connect to ground. (for RF hardening) of course, kicad complains. but it doesn't provide a means for modeling such oddball concepts.01:31
wolfspraulI think chip brands are important, and so might be a quick categorization like "regulator" or "ldo" or whatever01:31
wpwraki wouldn't be opposed to specifying the maker.01:31
wolfspraulhow about category/type ?01:32
wolfspraulspeeds up reading I think01:32
wpwrak(maker) because boom may be mislead, too. it does matching on vendor part numbers.01:32
wpwrakdunno. it's usually pretty clear what the things do. just a question of recognizing design patterns.01:33
wolfspraulwe continued the thread in 2 channels :-)01:33
wolfspraulsorry about that01:33
wolfspraul#milkymist then01:33
wolfspraulor here?01:33
wolfspraulI'm confused now01:33
wolfspraulso for ICs - add maker & type01:34
wolfspraulif possible, I would vote for keeping that "POWERED" box out altogether01:34
wolfspraulin the PDF01:34
wpwrakif you drop the POWERED, ERC gets messy01:35
wolfspraulI don't want to drop a kicad concept01:35
wolfspraulI only talk about the PDF I am reading01:35
wolfsprauland I describe the things that cause me a little mental bump01:35
wolfspraulwithout any subsequent aha moment01:35
wolfspraulmore like "urgh" moment01:35
wolfspraulI have some more01:36
wpwraki think it's just another concept you have to get used to. i think the "POWERED" is already much better than the diamond. e.g., you guessed right that it's just some sort of comment. you'd have had a much harder time figuring out what the diamond means.01:36
wolfspraulin a PWM you first have a pulse, then a DC current01:36
wpwrak(diamond) consider this: http://projects.qi-hardware.com/schhist/atben/pdf_atben.pdf01:36
wolfspraulwhere does "POWERED" start?01:36
wpwrakthe critters are in the lower right corner01:37
wolfsprauljust with my tiny little knowledge of electronics I already kinda swallow hard on this 'powered' thing01:37
wpwrakyou're over-analyzing :-)01:37
wolfspraulgood thing we don't document intricate regulator circuits01:37
wolfspraulnot over-analyzing01:37
wolfspraulI run into this, it wastes my brain energy01:38
wolfsprauluntil I mentally dump it into the trash bin01:38
wolfspraulbecause I see those "5V" marks clearly01:38
wolfspraulor "1V8"01:38
wolfspraulso what could that be?01:38
wolfspraula DC current?01:38
wolfspraulthen what is "POWERED"?01:38
wolfspraulnothing, just confusion01:38
wpwrakperhaps we should have left the diamonds. you may have found them easier to overlook :)01:39
wpwrakthe bikeshed problem :)01:39
wolfsprauland again, I do want to learn. in a PWM, where does 'powered' start?01:39
wolfspraulfirst it's a pulse, right?01:39
wolfspraulthen it becomes a DC current01:39
wolfspraulor did I misunderstand this01:39
wolfspraulseems very artificial to me where the "POWERED" starts01:39
wpwrakit starts on the net to which you connect to a "power in" pin and which doesn't already connect to a "power out" pin01:39
wolfsprauland your example of shunting to ground or whatever in atrf as well01:40
wolfspraulthis "powered" concept feels wrong to me01:40
wolfspraulbut the "1V8" are clear01:40
wpwrakthink of it as a directive to guide ERC01:40
wolfspraulit's in the mental trash bin now01:41
wolfspraulI don't think about it anymore :-)01:41
wolfspraulI can't see it01:41
wolfspraulit disappeared! :-)01:41
wolfspraulanother one01:41
wpwrakits presence does not affect what your circuit does. only what complains you get from ERC.01:41
wolfspraulwe have a nice little table with hw revision on one of the fpga sheets01:41
wolfspraulI understand [kicad/erc]01:41
wpwrakmaybe we should call it "FNORD" then :)01:41
wolfspraulbut I talk about readability/understandability of the pdf01:41
wolfsprauland I think I may have a point there01:42
wolfspraulcoming from that angle...01:42
wpwrakit's a quirky kicad concept01:42
wolfspraulso, that hw revision table01:42
wolfspraulI like those little tables01:42
wolfspraulon the power supply page, we have a bunch of voltages we are generating01:42
wolfspraulwouldn't it be nice to have a little table that lists who the different voltages are for?01:42
wolfsprauljust high-level, but I think it could help01:42
wolfspraulin fact, on the top left side we have a 4V3 = 4.3V? who is that for?01:43
wpwraksure. you commonly have power distribution diagrams. also things like current would be interesting.01:43
wpwrakdoesn't even have to be part of the schematics01:43
wolfspraulyes, current too01:43
wolfspraulalthough we have a few little "4A" etc here and there01:44
wolfspraulmy idea was just to start with a little table on the power supply page01:44
wolfspraulvery simple01:44
wolfspraulwho is the 4V3 for?01:44
wpwrakyes, that's the beads. since openmoko, i'm very picky about seeing them properly specified ;-)01:44
wpwraki think it's for audio01:45
wpwrakyes, audio01:45
wolfspraulah, ok01:45
wolfspraulwell then, "little overview table"01:45
wolfspraulI have some more01:45
wolfspraulon the fpga decoupling page01:45
wolfspraulthe larger caps are polarized, the smaller ones not?01:45
wpwrakthis touches on a broader problem: it would be nice if global labels indicated there else the signal appears01:45
wolfspraulit there an obvious reason/explanation for that that any EE would know, or should we add a little comment?01:45
wolfspraulsay for example C88 is polarized (4u7) but C98 (470n) is not01:46
wolfspraulmaybe this is obvious for more knowledgeable folks01:46
wpwraki'm not even sure if all of the caps drawn as polarized are really polarized. i'm almost certain they're not.01:46
wolfspraulwell ok01:46
wolfspraulthen I even may have a point01:46
wolfspraulsome more01:46
wolfspraulon the U22C page (btw I don't know which page is bank 0/1/2/3)01:47
wolfspraulI really hate those DNP/0R resistors without explanation01:47
wolfspraulfeels like some uncertainty/sloppiness/unfinishedness to me01:47
wpwrakthat's bank 201:47
wolfspraulR158 is 0R DNP, R159 is 0R01:48
wpwrak-> xilinx manual ;-)01:48
wolfspraulso it has to be like this?01:48
wolfspraulcan R159 also be DNP?01:48
wpwrakM0/M1 set some sort of boot mode. but the details are probably complicated.01:48
wolfspraulcan all those 4 be removed?01:48
wolfspraulok but you admit that a block of 2*2 resistors, all 0R, two DNP two not is confusing!01:49
wpwrakit's really a "go to the current xilinx docs if you need an answer" type of question01:49
wolfspraulsomething like this I mentally want to rip out entirely and understand what is really going on01:49
wpwraki mean, in the general sense01:49
wolfspraulor it's a leftover of unfinished/partial works01:49
wolfspraulI have some more :-)01:50
wolfspraulhope this is interesting for anyone...01:50
wolfspraulsimilar issue on the nor flash page01:50
wolfspraulR60/61, R18401:50
wpwraka 0R / DNP pair is usually clear: we want to be able to generate H or L. there's one choice that's the normal condition for our board. but someone may what to change it. therefore, we add a footprint that allows doing that01:50
wolfspraul60 and 61 are 10k dnp01:50
wolfspraul184 is 4k701:50
wpwrakthink USB_BOOT on the ben01:50
wolfspraulso those are all pull-up resistors?01:51
wolfspraulwhy is one 4k7 and two are 10k?01:51
wpwrakwhere are they ?01:51
wolfsprauland the 10k - is the pullup needed or not?01:51
wolfspraulthose are the ones we fiddled around with forever in the reset battles :-)01:51
wolfspraulnor flash page01:51
wolfspraulbut I see those 3, and it looks confusing/unfinished01:52
wolfspraulif r60/r61 are dnp now, can we remove them?01:52
wolfspraulno pullup needed?01:52
wpwrakon which page are they ?01:52
wolfspraulI may misunderstand it of course01:52
wolfspraulu9 nor flash01:52
wpwrakthanks ! ah yes, they're indeed a bit odd01:53
wolfspraulsome more01:53
wolfspraulC127 on that page, value is 1u, but not polarized01:53
wolfspraulthe bigger ones before were polarized, the smaller ones not01:53
wolfspraulwhat is the cutoff/reason?01:53
wpwraki think polarization is quite arbitrary01:54
wolfsprauljust saw it. C128 is 100n C127 is 1u01:54
wpwrakas i said, several caps drawn as polarized probably have components that aren't01:54
wolfspraulshould we remove the + then?01:54
wolfspraulin some use cases the difference can be quite dramatic though I would think01:54
wpwraki don't know what rules adam applies. in the end, i does't bother me too much, since they're nearly equivalent01:55
wolfspraulnext one first01:55
wpwrak(nearly equivalent) in the cases we have here. you're right that there are situations where things are less obvious.01:55
wolfspraulon the usb a/b page, some more wonderful DNP resistors01:56
wpwrakDNP can mean a number of things :)01:56
wolfsprauland even better, a note saying "full speed: mount R168" "low speed: mount R169"01:56
wolfspraulnow that is confusing01:56
wolfsprauldo we really mean a difference between full and low?01:56
wolfspraulor between host and slave?01:56
wolfspraulwhy is that comment/dnp resistors missing on the usb c/d and e/f pages?01:57
wpwrakdesign variants, things where we want to leave a plan b open, and then stuff we simply leave to the user to equip01:57
wpwrakthere's a lot of the latter in audio01:57
wolfspraullow and full?01:57
wolfspraulwith resistor values?01:57
wolfspraulwhat do we mount on R4 then?01:57
wolfspraulboth dnp?01:57
wolfspraulI don't get it01:57
wolfspraulit's just confusing01:57
wolfspraulso with both dnp, neither low nor full with work?01:57
wolfspraulsomething is not clear there01:57
kristianpauldammit, never close kicad even you already saved... for some reason things get messy...01:57
kristianpaulat least eeschema..01:58
wolfspraulthose are 1k5 (dnp) resistors on the D+ and D- lines? huh?01:58
wpwrakwolfspraul: it's for configuring the port for a full or low-speed device01:58
wpwraktotally unsupported by gateware and firmware, of course :)01:58
wolfspraulI don't get it01:58
kristianpaulclose by powering off computer*01:58
wolfspraulDNP means the circuit is open there?01:59
wolfspraulit connects to 3V3!01:59
wpwrakthe pull up/down configuration depends on whether you have a host, low-speed device, and full-speed device01:59
wolfspraulI don't get it, sorry01:59
wpwrakwe have only host ports, hence only the pull-downs but no pull-ups01:59
wpwrakthat's how USB works :)01:59
wpwrakwhat this circuit does is give you the choice to configure it for any of these modes, should you choose to do so02:00
wolfspraulmaybe we should write "full speed client" and "low speed client"?02:00
wolfspraulright now one could very easily think this somehow affects full and low speed in host mode02:00
wolfspraulbut I think it doesn't02:00
wpwrakwe could add a comment that it's for configurint the port as a device, yes02:00
wolfspraulit's just confusing, to me02:00
wolfsprauldevice, sorry02:01
wolfspraulhost and device02:01
wpwrakno, host mode is unaffected02:01
wolfspraulbut that's not obvious02:01
wpwrakit's a common source of confusion :)02:01
wpwraksuch is life in usb land :)02:01
wolfspraulso then let's write "FULL SPEED DEVICE" and "LOW SPEED DEVICE" instead02:01
wpwrakwhen i first saw it, i thought someone had misunderstood the USB spec02:01
wolfspraulnah come on, that one word can help clarify things. it's not impossible to explain...02:01
wpwrakonly then did i realize these were meant as configuration alternatives02:01
wolfspraulthe lack of clarifying "device" confused me02:02
wpwrakkristianpaul: if you do that, do at least a few  sync  first :)02:02
wolfspraulthat's it!02:02
wolfspraul1. 'powered' designator02:03
wolfspraul2. add maker and type to ics02:03
wolfspraul3. small table to clarify power users and current02:03
wolfspraul4. clarify decoupling caps polarization02:03
wpwraki wouldn't touch "powered". it's a cost of living with kicad02:03
wolfspraul5. R158-R161 could be documented better/cleaned up02:04
wpwrakbut you can try to make a patch that lets eeschema omit things like POWERED and submit it ;-)02:04
wolfspraul6. same for R60/R6102:04
wolfspraul7. usb device config resistors could be documented better02:05
wolfspraulthat's it02:05
wpwrakthough it may be tricky to get that right. there are other pieces of meta-data you don't want to suppress. and you'd also have to follow the wires, maybe convert junctions, too. etc. it's a bit of work.02:05
wolfspraulI feel most strongly about #2 and #3 in this02:05
wolfspraulgod no, I won't touch that [powered]02:05
wolfspraulthen I think #4, and #5/#6 should also be possible to improve slightly02:05
wolfspraul#7 is just adding one word02:05
wolfspraul#1 is just ranting02:05
wolfspraulbut overall the schematics are a joy to read!02:06
wolfspraulcladamw: can you help with a power diagram/table?02:06
wolfspraulthis concludes my feedback :-)02:06
wpwrakkewl. not too bad :)02:07
wolfspraulif even I can read that stuff, it's already written for dummies02:08
wpwrakhehe :)02:08
wolfspraulthe IC maker/type could help even more02:08
wolfspraulor, at least in the online PDF, some cool little links to wikipedia one can click on to find out about some background concepts02:09
wolfspraulbut those are just bells and whistles, plus the PDF has some serious technical shortcomings like text search02:09
wolfspraulability to follow wires by clicking on them02:10
wpwraksome concepts are difficult to convey because they depend on a lot of background. just like that usb pull business. it's a very common source of confusion everywhere. but yes, we can make it a little bit more accessible on our side.02:10
wpwrak(pdf) all things that really need support from kicad itself02:10
wolfspraulas I said I feel most strongly about my #2 and #302:10
wolfspraulthose are really valuble02:10
wolfspraulthe small cleanup is always nice too02:10
wolfspraulpolarity, dnp resistors, etc02:10
cladamwwpwrak, wolfspraul do you mean like this ? http://en.qi-hardware.com/wiki/Milkymist_One_Power_Tree02:10
wpwrakanother thing that would be nice to clear up would be to have chip pin numbers on all the bypass caps :) we have that now for the RAM02:11
wolfspraulyes, although that also doesn't include the mystery 4V3 user :-)02:12
wolfspraulbut yes, that's what I mean just integrated nicely as a little table or so in the "power supply" schematics page02:12
wolfspraulright now we have a set of regulators and circuits prepping this or that voltage02:12
wpwrakand it was a pretty good lesson - before these things weren't there, which led to a good amount of weirdness and miscommunication with layout. fortunately, harmless. this time.02:12
wolfspraulbut little info on where this goes, current, overview02:12
wolfsprauldefinitely [chip pin numbers for bypass caps]02:13
wpwrakmaybe the power tree needs updating. i see that it still has that ancient noisy codec02:14
wolfspraulmaybe we can move it into the schematics?02:14
wolfspraulremove from wiki?02:14
wpwrakthat should be possible, yes02:14
cladamwnot to remove it now, i think that i can use KiCad editor to edit a new power tree.02:15
cladamwa power tree i can equip them with power net name, chip reference, circuit block and etc.02:16
cladamwregards to chip pin numbers with bypass caps, i'll also clean them up.02:17
wpwraki would put the circuit block before the chip reference. less intimidating :)02:17
cladamwthe chip pin name don't need to be accurately same as current m1rc3 if later m1r[4..n] producing.02:18
cladamwfor examples two 100nF placed to pin 1 and pin 19, layout guy can place whatever 100nF to pin1 or pin19, 02:19
cladamwwe made unclear and unreadable in previous runs.02:20
kristianpaulwpwrak: no it not02:20
kristianpauli still deleted same components02:20
kristianpaulclose kicad02:20
kristianpaulno poweroff02:20
wpwrak(pin 1 vs. 19) well, it shouldn't cause much work for the layout guys either way :)02:20
kristianpauland got EESchema file text load error at line 78802:20
cladamw(circuit block) wpwrak  ?02:20
kristianpauland now there are lots of missing components on the  screen...02:21
wolfspraulcladamw: yes, good!02:22
wolfspraulof course, first make a new and clean version in kicad before removing the wiki one02:22
wolfspraulnow that we have kicad, we have a better place for this info than the wiki02:22
wpwrakcladamw: i mean "Audio  LM4550BVH"  looks nicer than "LM4550BVH Audio".02:22
wolfspraulbefore we needed the wiki to have any place outside of the dark and closed AD files :-)02:22
wpwrakkristianpaul: did you use a weird character in some name ?02:23
cladamwwpwrak, okay. :)02:23
kristianpaulwpwrak: let me see, but this wasnt happening before the component cutoff..02:23
kristianpauljust can see 3 components right now afaik,02:25
kristianpaulthe rest wires and nodes..02:25
wpwrakcan you paste line 788 and its context ?02:27
kristianpaulis empty..02:28
wpwrakand what is around it ?02:29
kristianpaulis very curious, after check the diff with tig, i noticed new text when i just removed compoments why?..02:33
qi-bot[commit] Adam Wang: tsop.fpd: added new TSOP module (master) http://qi-hw.com/p/kicad-libs/dfcaa9d02:35
kristianpaullooks like a bug02:35
kristianpauli just deleted one block now and the same error popup..02:36
wpwrakkristianpaul: it gets better: pick a text comment. double-click to edit the text. delete it. then press OK. you'll get "Empty text !". and then kicad segfaults ;-)02:36
wpwrak(at least here)02:36
kristianpauldont cheerme up that much ;)02:37
wpwrakperhaps you can recover it by putting some text in the empty line with a text editor02:37
kristianpauli'll try found root cause02:37
kristianpaulby trial and error :)02:37
cladamwwpwrak, would you mind if I outline frame to your header or he10 module ?02:37
cladamws/I/I add02:38
kristianpaulor try another versio  perhaps? this is 2011-05-2502:38
cladamwwpwrak, are you qi-bot ? hehe ... 02:38
wpwrak(mind) not at all. i've just been lazy :)02:40
kristianpaulputting some text in the empty line seems works, but how many empty lines left :)02:42
kristianpaulperhaps i need sed help02:42
cladamwwpwrak, also since a dedicated pin number 1, 2, or n, (n-1) there would be good. :)02:43
wpwrakyou mean to mark #1 ? the default kicad lib uses quare pins for #1, round for the rest02:44
kristianpaulargh, EESchema file undefined object at line02:44
kristianpaulbetter replace not remove empty spaces..02:44
cladamwwpwrak, i'd like a feature for Fped. i.e. place number (txt) in outline, can it be now ?02:44
wpwrakno, it can't do text02:44
cladamwCan KiCad do #n, #n-1 ?02:45
wpwraklemme see ...02:45
cladamwhmm .... i was thought your pin# can be shown inside solder paste in Fped, but can't add number text in outline in the future ?02:46
wpwrakyeah, the module editor has text02:46
cladamwhmm ... alright.02:47
wpwraki only show the pin number as non-physical text (i.e., something pcbnew shows but that doesn't go to the gerber)02:47
wpwrakthe problem with supporting text that goes to the gerber would be matching kicad's fonts and such. tricky.02:48
cladamwif yours(Fped) can do, that would be super ole.02:48
wpwraki'll put it on my to do list for 2013 :)02:49
kristianpauli just save it as another name (no mod) and same error02:49
cladamwwpwrak, tks a lots ! not bad. :-)02:49
qi-bot[commit] Werner Almesberger: reset reachable_pkg in the instantiation process and restore on inst_revert (master) http://qi-hw.com/p/fped/59b90b302:59
qi-bot[commit] Adam Wang: tsop.fpd: added 28 and 32 pins option (master) http://qi-hw.com/p/kicad-libs/d0f16e903:55
qi-bot[commit] Adam Wang: added to-252 link (master) http://qi-hw.com/p/kicad-libs/b8f4df303:59
qi-bot[commit] Werner Almesberger: modules/qfp-gen.fpd: QFP modules generated from IPC-7351 formulas (WIP) (master) http://qi-hw.com/p/kicad-libs/6a2427204:33
qi-bot[commit] Werner Almesberger: Merge branch 'master' of projects.qi-hardware.com:kicad-libs (master) http://qi-hw.com/p/kicad-libs/b977bc904:33
wolfspraulah, another small schematics comment04:59
wolfspraulthis one about atben - there is a crystal X1 16MHz 8pF 40ppm ESR=80R05:00
wolfspraulmaybe we could relax those specs a little so nobody gets stuck finding *exactly* such a crystal05:01
wolfspraulis that possible?05:01
wolfspraulfor example the ESR - probably anything above or below 80 should be ok, no?05:01
wolfspraulor the ppm, if it's 30 or 20, what would be the problem?05:01
wolfspraulif that info is missing then someone may not know what to do with a part they found with say ESR 100...05:02
wpwrakoh, that's a maximum value. like a resistor tolerance of 5% doens't mean we'd reject a 1% resistor05:02
wolfspraulok but we can be more helpful05:02
wpwrakthe capacitance must match C10 and C11, though05:02
wolfspraulthe ESR is a minimum or maximum?05:02
wolfspraul40ppm is the precision, right? if so, would <= 40 ppm be more helpful?05:03
wolfspraulI just want to convey good positive vibes to the reader, less worries :-)05:03
wpwrakmaximum. but i wonder where it comes from ...05:04
wpwraki think 40ppm is pretty unambiguous :)05:04
wolfspraulthe point is to make those specs more reader/user *friendly*05:04
wolfspraula higher precision crystal will work as well, so why not write "<= 40 ppm"?05:05
wolfspraulif the schematics should be an exercise in "think yourself", then I can also have a lot of ideas how to achieve that, of course :-)05:05
wpwrakah, the ESR is from the transceiver data sheet. so far so good. but they actually relaxed the spec. now it's <= 100 Ohm05:05
wolfspraula large part of the industry seems to believe in that model anyway05:06
wolfspraulhow good that they say <= :-)05:06
wpwrak(higher precision) because it's obvious ? :)    that's a bit like     foo++; /* increment foo */05:06
wolfspraulso I think we should do that too, obvious or not, if it helps people more quickly navigate around the specs and focus their time on where it matters to them05:06
wolfspraulyou could say "all crystal ppm are meant <=", but then it's better to just write <= there imho05:07
wpwraknaw, if you fill the schematics with trivial junk, people will not notice the places where you actually have something to say05:07
wolfspraultwo characters05:07
wolfspraulESR is <= 100R now?05:07
wpwrakwill be, once i finish dinner :)05:07
wolfspraulI agree about not repeating obvious/redundant information05:09
wolfspraulbut then you want to invite people and accelerate their first steps05:09
wolfspraulESR= is a good example of potential time waste05:09
wpwrakyes, the ESR is bad05:09
qi-bot[commit] Werner Almesberger: atben/atben.sch, atusb/atrf.sch: correct and clarify ESR of X1 (reported by Wolfgang Spraul) (master) http://qi-hw.com/p/ben-wpan/42f1a7805:50
Aylais there a way to tell GCC that a branch will probably be used more often than another one?10:36
qi-bot[commit] Adam Wang: to.fpd: added TO-252 package (master) http://qi-hw.com/p/kicad-libs/976203810:50
qi-bot[commit] Adam Wang: Merge branch 'master' of projects.qi-hardware.com:kicad-libs (master) http://qi-hw.com/p/kicad-libs/983ee3510:50
wpwrakAyla: there is a builting for it. linux uses it a lot: http://kerneltrap.org/node/470512:33
Aylayep, I knew it on linux12:33
Aylabut linux uses likely() / unlikely()12:33
Ayla(which are macros)12:34
wpwrakthe article also describes what's underneath12:35
mthAyla: this is what we use in openMSX:12:58
mth#if __GNUC__ > 212:58
mth#define likely(x)   __builtin_expect((x),1)12:58
mth#define unlikely(x) __builtin_expect((x),0)12:58
mth#define likely(x)   (x)12:58
mth#define unlikely(x) (x)12:58
AylaI used something like that12:58
Aylabut why __GNUC__ > 2?12:59
mthbecause GCC 2 doesn't have the feature and non-GCC compilers probably don't have it either (not with identical syntax, anyway)12:59
mthI hope no-one still uses GCC 2, but we've had this code for a long time13:00
mthI should check how it is expanded for clang13:01
mthbecause they do support many GCC extensions13:01
Aylawant my clang toolchain for mips? ;)13:01
mthI can check on OS X13:02
mthI do want clang for MIPS at some point, but let's not get distracted by that right now13:02
Aylawe have only 3 days left ;)13:03
mthwe should make a list of issues and priorities13:05
mthmaybe in the other channel would be more on-topic13:05
wolfsprauluwe_mobile__ uwe_ - congrats on your sigrok project btw! that's shaping up really nicely23:43
wolfspraulI recently bought a FX2-based CWAV AX clone, now trying out sigrok and the free firmware23:44
wolfspraulmaybe I also give the Hantek dso-2090 a shot, but I understand support for that is still kinda wip23:44
wolfspraulbut it's a great project, good luck!23:44
wolfspraulsigrok I mean :-)23:44
--- Wed May 30 201200:00

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