#milkymist IRC log for Tuesday, 2012-03-06

wolfspraulHodapp: that's great that you found Milkymist via github!00:14
wolfsprauland I do agree the social aspect of such sites is very important, and we are probably not the smartest in using the 'social web'00:15
wolfspraulI was surprised once how many new people showed up for the Ben NanoNote after a simple version uptick in freshmeat00:15
wolfspraulthe at some point xiangfu tried to get a milkymist soc or milkymist firmware project registered at freshmeat but he was turned out because it seems they didn't understand how the different milkymist sub-projects relate to each other00:16
wolfspraulturned down00:16
wpwrakfreshmeat used to reach a lot of people. i saw this with some of my more popular linux projects. nowadays, also secondary media may rebroadcast things from freshmeat.00:19
wpwrakperhaps we could register FN ?00:19
wolfspraulstrange. analog's official distributor says the ADV7441 is what they are pushing now? but ADI's webpage says that one is 'obsolete' http://www.analog.com/en/analog-to-digital-converters/video-decoders/adv7441/products/product.html00:26
wolfsprauland the 7441 doesn't look good at octopart either, oh well00:28
wolfspraulconfusion all around00:28
wolfsprauland the analog.com website manages to say 'obsolete' but links to replacement parts only lead to general search engines, urgh. as if they couldn't list some likely replacement candidates if they are already able to say it's 'obsolete'00:31
wolfspraulthat makes me feel better about the Qi wiki :-)00:32
wolfspraulreplacement is 7441A :-)00:34
wpwrakat least they have one ;-)00:34
wolfspraulok, but that looks like a huge and expensive chip00:35
juliusbhey guys, quick question - how many resources of a spartan 6 does the lm32 consume in the config you use it in in the m1?00:35
wolfspraul30 USD at digikey, 20x20mm, includes hdmi receiver, etc.00:36
juliusbhow man LUT and FF, basically, i'm curious about00:36
juliusbs/man/many/00:36
wpwrakdid you want digital video in ? :)00:36
wolfspraulin R4?00:36
wolfspraul:-)00:36
wpwrakdetails ;-)00:36
wpwrakdual hdmi receiver. hah.00:37
wolfspraulok, it would help me if someone could look at the 7441A datasheet and let me know how hard/difficult/wanted/unwanted it would be on m100:37
wolfspraulor just wait a few days, I try to discover more on the sourcing ground00:37
wolfspraulthis stuff is tedious00:37
wolfspraultoo many chips, too many roadmaps, too small margins, too many technical details, too many direction and marketing changes, too too too :-)00:38
wpwrakand you can get it without HDCP. very thoughtful :)00:39
wolfspraulthe less chips we have on m1 in the future, and the more we do in 1 high-performance fpga, the better00:39
wolfspraulI rather source external cables with built-in dacs and adcs00:39
wpwrakhow many M1r4 do you expect to make ? (estimate within 1-2 orders of magnitude)00:40
wolfspraulwe have to make a 8 pc verification run first00:40
wpwraki wouldn't source too fancy cables :)00:40
wolfspraulI hope adam has most of the parts for that00:40
wpwrakbut agreed on trying to do things in the fpga. there, we have nearly total control.00:40
wolfspraulsince boom looks like a little delayed00:40
wolfspraulit's tough all around, but all good work00:41
wolfspraulI need to secure larger orders, I have no strong feeling on the actual r4 run size right now00:41
wolfspraulwe have to keep m1 in stock00:41
wolfspraulthat is for sure00:41
wolfspraulso how long until r5? when do we reach the point that word-of-mouth or other sales & marketing activities start to really gain traction?00:41
wolfspraulall very difficult to predict00:42
wolfspraulfor now I go day by day00:42
wolfsprauljust speed up00:42
wpwrakyeah, boom will only really make sense with kicad anyway. otherwise, you just have to manually copy things around and nobody will be particularly motivated to check the accuracy. pretty pointless.00:42
wolfspraulI want the 8pc R4 verification run asap00:42
wolfspraulwe have to kick the design up00:42
wolfspraulthe 7181 sourcing is a good reminder00:42
wolfspraulin the long run we should not depend on such chips00:42
wolfspraultoo specialized, too expensive, only used in fringe use cases00:43
wolfsprauleverybody will invest to get away from needing them :-)00:43
wpwrakhmm. seems that i won't get my answer :) my question would have been: if you had to buy all the codec you need for M1r4 today, would you expect to be able to find them ? (with decent quality and at a not overly exaggerated price)00:43
wolfsprauloh sorry, wait00:43
wolfspraulI really try to answer a question, not avoid or dismiss it00:44
wolfspraulsec00:44
wolfspraulhad to buy?00:44
wolfspraulI want to buy00:44
wolfspraulnow00:44
wolfspraultoday00:44
wolfspraul:-)00:44
wolfspraulsince we don't have boom, we have no good visibility into which chips exactly are not easily sourcable at digikey00:45
wolfspraulthe slx45 we source in China, and for now that looks good00:45
wpwrakhmm, tracking sourcing with boom seems like an interesting extra feature00:45
wolfspraulother key troublemakers could be the nor chip and adv718100:45
wolfspraulnot 'tracking', but it can quickly reduce the number of things you look at00:46
wolfspraulsourcing is all about economics too00:46
wolfspraulor only00:46
wolfspraulok, back to 7181 - let's assume for now we don't want the 744100:46
wolfspraulI will just get that 7181 somewhere, b or c00:46
wolfspraul:-)00:46
wolfspraulincluding shenzhen backyards00:47
wpwrakone problem we'll have to solve is the database. right now, it's locally built. if we were to redistribute it, we'd reduce the setup time, load on the distributor's servers, but also duplicate information they may consider falls under their copyright00:47
wpwraktricky00:47
wolfspraulthe dark alleys where only real men go00:47
wpwrak(7181 from the dark underbelly of china) okay, problem solved :-)00:47
wolfspraulI cannot imagine they cause problems, this is about sales00:47
wolfspraulin fact we should approach their marketing dept and tell them about boom00:48
wpwraki guess that depends how badly their legal department feels it needs to justify its existence00:48
wolfspraulthey may or may not like it. they may not like the fact that we make the different ones so easily comparable00:48
wpwrakyes, that would be a possibility00:48
wolfsprauloh, always directly talk and embed00:48
wpwrakpro: if they're reasonable, this can make things a lot easier for everyone involved00:48
wolfspraulwe don't need to think about everything and everybody00:48
wpwrakcontra: if they're not, they may get angry if we use them anyway00:48
wolfspraullet's just make a good job, and be 100% open and transparent and inviting about it00:49
wolfspraulthen we stop using them00:49
wolfspraulI would pick the option that makes boom actually *work* asap00:49
wolfspraulI need this tool00:49
wpwrakregarding tracking, if we keep historical data around, you could call up the stock and price history of a part (and possible alternatives, if boom knows of them)00:50
cladamwwpwrak, thanks for adequately express my lack summary. ;-)00:51
wolfspraulcladamw: good morning! :-)00:51
wpwrakthe quick way to making boom work is what i do now: crawl their component search00:51
wolfspraulyes00:51
cladamwgood morning !00:51
wolfspraullet's strictly follow our practical needs, my #1 wish00:51
wolfspraulwe want to make this thing, and make it cheap, and high-performance, and make it function today00:52
wpwrak:)00:52
wolfspraulcladamw: did you hear back regarding the ADV7181C ?00:52
wolfspraularrow shanghai said it's not in stock there (and also element14), and they mentioned the ADV7441A instead, urgh :-)00:52
cladamwwolfspraul, no, will directly call again.00:52
wolfspraulyes00:52
wolfspraullet's zoom in a little.00:52
wolfspraulforget about the D00:53
wolfsprauland let's just buy now, no more questions, just cash on the table and where are the chips? :-)00:53
wolfspraulif needed, buy a whole tray00:53
wpwrakhmm. why don't we have TVS on MIDI TX ?00:53
wolfspraulcladamw: do you agree? just buy whole tray if that accelerates the case.00:54
wolfspraulI want to accelerate. otherwise we waste a lot of money with our time, and could work on better things.00:54
wolfspraulalso I want to 100% rule out that M1 goes out of stock, I have to keep it in stock.00:54
cladamwwolfspraul, i'd rather order C in AVNET directly.00:54
wolfsprauldo they have it?00:55
wolfspraulhow many do you want now?00:55
wpwraki think one feature i'll add as soon as we're on kicad with the schematics is automatic suppression of everything DNP ;-)00:55
wpwrakwe00:55
wpwrak're approaching openmoko-grade badness levels ;-)00:56
wolfspraulwhere?00:56
wolfsprauldon't make the nightmares come back00:56
wpwrakwith DNP components. kinda everwhere :)00:56
wolfspraulafter Om I know how good my life can be00:56
wolfspraulis it bad? should we remove some?00:57
cladamwAVNET has 105pcs in stock.00:57
wpwrakyeah. with kicad, we could just filter them from the .sch file before printing00:57
wolfspraulat Om, I seriously had fever every 6 weeks or so, from all the stress. and since that nightmare is over, maybe once in 3 years. call it quality of life - priceless.00:57
wpwrakmost of them are there for a reason. so for now, i don't seem much chance for a purhe00:57
wolfspraulcladamw: then why don't we order at avnet?00:57
wpwrakpurGe00:57
wpwrakyeah, om was tough00:58
wolfsprauldatasheet urls also here http://en.qi-hardware.com/wiki/Milkymist_One_RC3_BOM01:01
wolfsprauland yes, I like the middle way of dsv01:01
wolfspraulelegant and professional and still real-life enough to leave some backdoors :-)01:01
cladamwwolfspraul, i was not planned to order ADV7181C in last order batch, since gathered parts on Digi-key, Mouser for sample parts, not for a real R4 run order. The ADV7181C was highlighted while checked in octopart, not the thing why we don't order at avnet. ;-)01:01
wolfsprauldon't understand. what is your point?01:02
wolfspraulwe are planning a 8pc r4 verification run01:02
wolfspraulhow many adv7181c do you have in stock now? how many do you need to buy?01:02
wolfspraulwhat is the moq at avnet? if you can buy from avnet, let's just do it!01:02
wolfspraulwe can also reserve an extra 5 adv7181c for upgrades maybe?01:03
cladamwwpwrak, i've not edit R4 wiki bom page with datasheet link.01:03
wpwrak(links) thanks !01:03
wpwrakah !01:03
wolfspraulwpwrak: can we detect the difference between adv7181b and c at sw runtime?01:04
wpwrakwell, i think i'll make a BOOKSHELF file (for dsv) anyway. too tedious to look up things otherwise. just want to make sure i'm not duplicating a lot of existing research.01:04
wpwrakyes01:04
wpwrakand even without asking the user to hold the camera above it :)01:05
wolfspraulwell, you will like the links on the rc3 bom page :-)01:05
wolfsprauladam spent a lot of time looking them all up, in preparation for a nicer solution like dsv later01:05
wpwrakyes the rc3 links are great01:06
wpwrakwow. found one FB that didn't just say "FB" ;-)01:06
wpwrakactually, all the USB ones are like that. nice01:07
cladamwwolfspraul, will order 10 ~ 12 pcs 7181C in AVNET for R4 & if upgrade in case if need. i didn't include it in my last sample batch of orders.01:09
wpwrakhmm ... http://en.wikipedia.org/wiki/Circuit_diagram  "Schematic wire junctions" box. 2a = evil. 3a, right side = good01:10
wolfspra1lsorry got disconnected01:11
wpwrakat least usb has a few 2a-style junctions01:11
wolfspra1lcladamw: I didn't understand you earlier. can you repeat?01:11
wolfspra1lcladamw: ok, read backlog :-)01:13
wolfspra1lsounds great! so the adv7181c can be bought from avnet taipei?01:14
wolfspra1lwe will try to find out some more information from shanghai and shenzhen, it may be helpful in the future01:14
wolfspra1lbut if you can buy a few chips for R4 now, great!01:14
cladamwwolfspra1l, last week my samples order was to get them from digikey or Mouser for R4 few parts not for whole 8 sets of run. but just discovered 7181C was only sell in avnet web no else, so I didn't order ADV7181C in avnet.01:14
cladamwmy next works: two -01:15
cladamw1. write notes in Chinese to house layout 2. prepare R4 wiki Bom page (which sorry Werner just asked.) ;-)01:16
wolfspra1lok01:16
wolfspra1lbut now it seems you can buy the adv7181c from avnet in taipei?01:16
wolfspra1lyou did excellent work on the rc3 bom page, that's a good starting point for Werner01:16
cladamwnot from taipei, from usa avnet web directly, i remembered C came from US.01:17
wpwrakabove i mentioned usb (FB and junctions). that should have been power. got confused :)01:18
cladamwi don't trust taipei especially the 'time' for waiting, since i am too 'small'. last time I ordered 7181C from avnet: http://avnetexpress.avnet.com/store/em/EMController?action=products&catalogId=500201&storeId=500201&N=0&langId=-1&slnk=e&term=ADV7181CBSTZ&mfr=ANA&hrf=http%3A%2F%2Foctopart.com%2Fadv7181cbstz-analog%2Bdevices-10480349&CMP=KNC-Octopart_VSE01:18
wolfspra1ljust order, then we see01:20
wolfspra1lwe find out more data from china in parallel01:20
wpwrakwow. "top seller" :)01:20
wolfspra1lbut you definitely please go ahead and order01:20
wpwrakis that us stock ? if yes, why does it say "Not available in Americas" ?01:21
wolfspra1lI wanted to ask about the parts for the r4 design verification mini-run anyway01:21
wolfspra1lif we want to make 8pc r4 - how many parts are missing?01:21
wolfspra1ldo you have leftover parts from the rc3 run?01:21
wolfspra1lactually we can learn from this - if we do a run, we source 10-20% too much in case the next one needs a small design-verification first01:22
cladamwwpwrak, i meant last time not this current page.01:22
cladamw(leftover parts) wolfspra1l, i can't answer you now until i edit R4 done then that time to start order. but if worry 7181C then  we order it firstly with others.01:23
cladamwfor me, it's quite too earlier to order. only design-verification samples.01:24
wolfspra1lI think we can order adv7181c now, just in case01:25
wolfspra1lwpwrak: I think in general you can forget about the availability info you find on the web01:26
wolfspra1lespecially for expensive and specialized ICs like that one01:26
wolfspra1lwhatever they say, they may just order them from somewhere else, etc. it's a connected world :-)01:27
wolfspra1lwpwrak: what did you mean with "FB"?01:29
wpwrak(availability) dunno. i think at digi-key, it's pretty firm. dunno about avnet.01:30
wolfspra1lyes digikey is one of the very few where the availability/stock data still has some quality01:31
wpwrak(FB) beads. many of them are just labeled "FB"01:31
wolfspra1lbut even at digikey I think they start to play a little with "can be ordered" etc.01:31
wpwrak(digi-key) the other extreme would be the gazillion chinese resellers. they all have everything. but you have to call/mail :)01:31
wolfspra1lthey cannot stock all sorts of specialized expensive ics, that are also changing all the time01:32
wpwrakyeah, the non-stock items. usually with huge MOQs.01:32
wolfspra1llook how much trouble we have even with the few chips on our m1 board that several people pour over for years01:32
wolfspra1land now imagine thousands and thousands of chips and revisions, and obsolete, replacement, company ownership change, management change, renaming, new process, marketing change, and and01:33
wolfspra1land then we also like digikey because of their competitive pricing01:33
wolfspra1lor do we like it because of the quality of their availability data, or how much they have in stock in their own warehouse?01:33
wpwrakall of the above ;)01:33
wolfspra1lanyway we are on a good path overall, moving to more general chips01:33
wpwraksourcing from a "cheap" but unreliable place can get expensive very quickly :)01:34
cladamw(FB) hmm...we have two kinds of FB only , one is 1206, the other is 0402, mmm...i should have edited and enabled their text for 0402.01:35
wpwraksorry, power trouble01:39
wpwrakthere's a bug in my apartment's wiring that trips the circuit breaker if there's a lot of rain. then i need to wait until wherever the trouble is dries again. thought it was already safe, but no such luck.01:42
wolfspra1lsee01:43
wolfspra1lChina is in fact leading on *everything*01:43
wolfspra1lthe rest of the world is behind01:44
wolfspra1lI have stories of that type 3 times a day01:44
wolfspra1lfeel so advanced now01:44
wpwrak;-)01:54
wpwrakyeah, that electricity issue is a tiny nuisance. they never noticed before because they didn't have this kind of circuit breaker. not sure what it is. could be water dripping into a cable pipe.01:58
Thihiwolfspra1l, http://www.atom.com/funny_videos/haha_america/ familiar with this then? ;)01:59
wpwrakand of course, it's in the original electrical installation. so it's somewhere in the walls. meanwhile, the new one in had added works flawlessly. i just need to cut the old, reset the circuit breaker, and, if the rain is very persistent, plug the fridge into the "good" net.02:00
wpwrakso the annoyance level is just slightly below my "do something about it" threshold :)02:01
kristianpauljuliusb: few K Luts, 2 - 2.5K if remenber02:03
wolfspra1lThihi: no didn't know, looks like crap though :-) I focus on the positive, all the time.02:11
Thihiwolfspra1l, well, I remember laughing my ass off when I first saw that in 2006. I guess it is negative, in the sense that they are making fun of americans, but I found it pretty funny ;P02:13
ThihiWhat you said about hearing that message many times a day just brought that back into mind.02:13
ThihiI'm respectful towards all the nations of course, but somehow making fun of the world police has always appealed to my humor. ;P02:15
ThihiWell, as respectful as one can be after realizing that a nation can accurately be portrayed as a violent thug, being held back by full submission to it's demands ;P02:18
ThihiAny nation that is.02:18
wpwrakthat video looks kinda unfriendly to both countries. probably some finns with their wry sense of humor made it ;-)02:18
Thihi:D02:18
ThihiAnyway, let's not get into politics. :)02:19
ThihiWe're all individuals here as far as I can tell, so no reason not to look everyone in the eye and smile ;P02:19
Thihieyes, sorry.02:19
ThihiI'm the only cyclops .)02:19
ThihiAlthough I have to say, that the wry (or deadpan) humor is prolly the best quality of the finns ;P02:21
wpwrakthey're also fun to watch when going home from the disco in a winter night. nobody is walking along. because they couldn't ;-)02:23
Thihi:)02:24
ThihiAnyway, sorry for disrupting your conversation. Do carry on. :)02:25
wolfspra1lwpwrak: oh btw, in preparing for better communication with Lattice, I read up a little on the company02:47
wolfspra1lso it seems they withdrew from the 'top-end general-purpose fpga' market02:47
wolfspra1lleaving only xilinx and altera to compete with each other there02:47
wolfspra1land instead lattice focuses on mid-range, and various specialized things like power and clock management, mixed-signal, etc.02:48
wpwrakmid-range would be us, too. right ?02:48
wolfspra1lthey claim that their mid-range fpgas are cost-competitive with the altera/xilinx offerings, and try to justify why they are several generations behind on process nodes (I think they are still at 90nm), but at least for my quick glancing over this I did not find anything even closely attractive to m102:48
wolfspra1lif it wouldn't be slower and also more expensive, yeah, maybe :-)02:49
wolfspra1lI think they simply cannot keep investing at the speed and amount that xilinx and altera can, and dropped out there02:49
wolfspra1lnow they are focusing on smaller more profitable niches02:49
wolfspra1land that does seem to work for them02:49
wpwrakah, i see02:50
wolfspra1lthat's my current understanding02:50
wolfspra1lit's not entirely convincing to me, especially I don't see how chips that are interesting for us could come out there02:50
wolfspra1lit looks like they will invest more and more in specialized (and of course proprietary) IP, mixed-signal fpgas etc.02:51
wpwrakwith an open core to tie it all together02:51
wolfspra1lI don't even think that open core is an important part of their thinking now02:52
wolfspra1lother than it's something that is still 'theirs' and they try to monetize02:52
kristianpauli doubt about tie, i mean i dont see how strong is that lm32 for then..02:52
kristianpaulyeah !02:53
kristianpaulafortunatly its open core is _very_ usefull for us :)02:53
wolfspra1lI'm sure they can find a profitable niche for themselves, it's a big big world. but I don't see what could be in it for us, which chips.02:53
wolfspra1lso maybe we just take the open core and that's it. and I don't expect lattice to do any meaningful investments into this, other than lip service which we already see :-)02:54
kristianpaulchips, well yeah... they claim a lot for low power stuff02:55
kristianpaulof course i just had read their marketing :)02:55
wolfspra1lyes, they have all sorts of specialized chips and ip/service too02:55
wolfspra1lthat was their 'turnaround strategy', after my reading02:55
wolfspra1ldrop out of the high-end race, leave it to xilinx/altera, and focus on more profitable smaller niches where they can reuse the tech they have, and carefully invest for their own future growth02:56
kristianpauland the built-in flash well, could be and advantage could not.. i need evaluate that when get a osmo-sdr hope soon! :)02:56
wolfspra1land it did work, after years and years of losses the company seems stabilized now02:56
wolfspra1losmo-sdr uses a lattice fpga?02:57
kristianpaulyes02:57
wolfspra1linteresting. which one?02:57
kristianpaulXFP202:57
kristianpaulXP2*02:57
kristianpaullet me confirm02:58
wolfspra1lyeah, just looking at octopart quickly02:58
kristianpaulyes that one02:58
wolfspra1lI find it expensive for the small 'raw performance' it has02:58
wolfspra1lbut that is their strategy02:58
wolfspra1lthey are adding stuff that makes it more interesting for specialized cases02:59
kristianpaulthe plus was actually built-in flash and reasy to source (dont aske me details)02:59
kristianpaulalso i think they used to use it :)02:59
wolfspra1lif you compare raw performance, you will pay way way more than with high-end xilinx or altera chips02:59
wolfspra1lbut that is their strategy, and it seems to work02:59
wolfspra1llattice was able to stop the bleeding02:59
kristianpaulGPS!!! range (ilinx or altera )03:00
kristianpaulGBPS*03:00
kristianpauloops03:00
kristianpaul:)03:00
wolfspra1ldon't understand. what is your point?03:00
kristianpauli mean i always read from both xilinx and altera, that advertice hihg speed bus transfers in their chips03:01
kristianpaultranceivers. etc..03:01
kristianpauli dont know if i explain correctly03:01
kristianpauli just undertood that like high-end03:01
kristianpauls/like/by03:02
kristianpaulergh03:02
wolfspra1llattice and xilinx both invest aggressively toward what is typically understood as 'latst and greatest performance'03:03
wolfspra1lthe spartan-6 is made in a 45nm process, and the upcoming -7 series in a 28nm process03:03
wolfspra1lbut lattice explains why you can do all the good things with chips made in 90nm and rather invest the money differently into mixed-signal, power, clocks, etc.03:03
wolfspra1lI think both strategies may work, maybe it was smart for lattice to drop out of the race, surely they had to or they would have simply gone out of business.03:04
wpwraksometimes, a little analog circuit can go a very long way. their main problem should be that engineers who can do that are dying out03:07
kristianpaulmid-range also involve mobile?03:08
kristianpauli remenber lattice adquired recently a company related to mobile solutions and i guess pld/fpga etc..03:08
wolfspra1lI cannot explain or comment on lattice too much, just reporting my findings03:08
kristianpaulsure :)03:09
wolfspra1lyes they seem to focus on cplds too03:09
wolfspra1lall sorts of things away from the typical xilinx/altera comparison03:09
kristianpaulmore than usual it seems to me03:09
kristianpaulindeed03:09
wolfspra1lhow many luts? max mhz? how many usd/lut? etc.03:09
wolfspra1lthat they dropped out of, after loosing for years and years :-)03:10
kristianpaulhe03:10
wolfspra1land if you compare the chips they sell today along those lines, they are indeed totally uncompetitive03:10
wolfspra1lbut since they seem to be selling their chips, the additional value/special features must work03:10
wolfspra1lhow long this works I don't know03:10
wolfspra1lthey will not reenter the high-end race thouhg, I think after dropping out the capital needed would be so huge - they would never find an investor for that03:11
wolfspra1lso that is down to xilinx & altera now03:11
wolfspra1lI definitely don't expect more open source stuff from them03:12
wolfspra1l(from lattice)03:12
wolfspra1land I don't expect high-performance fpgas that can compete (on raw performance) with anything from altera or xilinx either03:12
wpwrakwhen you talk like this, then usually the opposite happens ;-)03:13
wolfspra1lthat leaves little :-) unless some of their specialized chips suddenly stands out...03:13
wolfspra1lfine fine, it's all recorded. we see.03:13
wolfspra1lI can just do my homework and read.03:13
wolfspra1lit sounds as if the company is stable and profitable now. that's a start.03:14
wolfspra1lI think they held a record for most consecutive quarters with losses in Oregon State, or something like that :-)03:14
wpwrak;-)03:14
wolfspra1lthat does make me like them a bit03:14
wpwrakpersistence :)03:14
Action: kristianpaul wonder what happended with the lm32 ASIC once commented on the ML03:15
Action: xiangfu just tested tl 703n with as m1 wifi works just fine.04:37
xiangfuwith OpenWrt images.04:37
xiangfuthe only problem is you have to setup the WIFI connect by using your pc.04:39
xiangfubecause there is no browser in m1.04:39
xiangfu703N image files is here: http://downloads.qi-hardware.com/people/xiangfu/icarus/openwrt-ar71xx-generic-trunk-29936/openwrt-ar71xx-generic-tl-wr703n-v1-squashfs-factory.bin04:42
xiangfuI have tested the latest openwrt trunk build.04:42
xiangfuthere is no web server enable in latest openwrt trunk build.04:42
wolfspraulxiangfu: we have to find an easy way for someone to enter ESSID and password on the m1, and then transfer it to the 703n05:16
wolfspraulor even scan networks, pick a essid, enter password05:16
wolfspraulor, if it's too difficult, maybe you have to connect to the wifi router with another computer and set it up then?05:16
wolfspraulwe need to think throuhg how this becomes realy fool-proof and working, yet safe05:16
wolfspraulshould be possible without another computer...05:17
xiangfuout for lunch. bbl05:32
lekernelwpwrak: manufacturers do not care; if they do I'll take it down08:44
wolfspraulyes but you open your entire project up for easy dismissals "built on illegal basis, just look here"08:45
wolfspraulnot that I care much either, but werner's dsv is a nice middle way08:45
wolfspraulit allows for easy downloading, easy viewing, easy archival & backup in case something goes missing...08:45
lekernelwpwrak: and it's all public files, it's no worse than google cache or archive.org08:46
wolfsprauland no, the manufacturers we work with are not know (to me) to aggressively enforce these things08:46
wolfspraulwe don't even consider chips of that type...08:46
wolfspraullekernel: for most of these documents, there will be no redistribution rights granted to you08:47
wolfsprauland without those, your uploading is considered illegal in major western jurisdictions08:47
wolfspraulnot that they would do much, see the various datasheet aggregator sites08:47
wpwrakthe issue is mainly that they may feel they have to do something or it'll weaken their position in cases that actually matter (and there's probably good legal reasoning behind it)08:52
lekernelI'm not saying this is not a copyright violation, I'm saying it's unimportant08:52
wpwrakprobably not their lawyers, whose task is specifically to avoid weakening of their IP's legal strength08:53
lekerneland regarding the "built on illegal basis" comment - no one noticed this repository before, it's a bit far fetched, and only uninteresting trolls would say that08:53
lekernelso it acts as a natural filter08:54
wpwrakbut now you announced it ... :)08:54
wpwraklegal reasoning is often marked by a mix of fears. basically anything that happens with and around your product is perceived as a threat. this yields typical psychotic control freak behaviour. if you do anything they didn't explicitly approve, that's an attack. even if the outcome of you actions is irrelevant or even helps them.08:57
wpwrakno, some companies are worse than others. but i don't think it's particularly interesting to find out where exactly each stands.08:57
lekernelphew08:59
lekernelthis stuff is copied all over the place08:59
lekernelgoogle cache08:59
lekernelarchive.org08:59
lekernelvarious datasheet aggregators sites08:59
lekerneluniversity websites08:59
lekernelpersonal websites08:59
lekerneland more. and no one cares.08:59
wpwrakof course is it. and some of these companies are spending millions on lobbying for things like ACTA because they hate that so much.09:00
lekerneldo they? it's just free publicity for the chips they sell09:00
lekerneland I don't remember semiconductor companies lobbying for ACTA09:01
wpwrakthat's not how their legal departments thinks. their task is to make sure the company is not perceived as "weak", that none of their precious IP loses value (unless approved), and if that really important document should leak and they want to do something, that nobody could claim that this was just usual practice.09:02
wpwraksometimes, you can get a company to give you redistribution rights. then everyone is happy: you can legally redistribute and they didn't open a hole in their defense09:04
wpwrakbut it's hard to get to a point where they'll listen to you. and it still depends on how this resonates with internal policies. if they're close-minded and already hate it that they can't just NDA all their regular data sheets, you'll have a very hard time beating such a culture.09:05
wpwrake.g., in Openmoko, we had such success with NXP. but that was after months of very good contacts with them. plus, many people there understand open culture.09:06
wpwrakthere's also a subtle line between plausible deniability. e.g., a company may be very well aware that you're doing something they shouldn't let you do. but they realize that it's not worth stopping you (and they may actually like what you're doing)09:10
wpwrakbut they wouldn't necessarily want others to do the same.09:10
wpwrakso as long as you don't make noise about it, they'll just look the other way. but if something happens that makes it difficult for them to claim they hadn't noticed, then they have to act.09:11
wpwrake.g., if you redistribute, say, some xilinx data sheets. and then fpga-acme.cn does the same, then xilinx may want to fight them. and attacking that redistribution would be one thing they could do. but if there's already established precedent of them not caring about people redistributing without permission,09:13
wpwrakthey'd lose the ability to get a preliminary injunction. this is a powerful legal instrument but you have to be able to justify that there is urgency.09:14
wpwraksometimes, also the things that legal cares about are seemingly trivial details. again, that is to have a point they can attack in bigger cases. for example, philips had troubles with asian companies cloning their high-end shaving machines. it would have been hard to fight them on the whole design.09:16
wpwrakwhat they used instead was copyright on the pattern of non-functional holes in the blade. that apparently gave them a strong enough case that they could fight that competition for a while.09:18
wpwrakhmm, i wonder if any english-speaking person would guess what adam meant here :) http://en.qi-hardware.com/wiki/Milkymist_One_RC3_Known_Issues#darn_of_never_booted_up09:23
wpwrakwolfspraul: btw, do you know if adam is making that r4 BOM with data sheets now ? if yes, i'll just wait and generate my BOOKSHELF from the wiki page. if not, i'll start writing my BOOKSHELF from scratch.09:26
wolfspraulfrom scratch?10:22
wolfspraulyou can definitely use the rc3 wiki page, that's what I'm sure adam will do10:22
wolfspraulI think there is an underlying .ods file that he will update10:22
wolfsprauland I think for the time being Adam should still do this10:22
wolfspraulI think the solution Sebastien chose at the beginning is a smart and acceptable compromise. Sebastien alone cannot work on thousands of details.10:23
wolfspraulthe datasheets are in a separate git repo, nicely kept away from the rest10:23
wolfspraulhe would definitely reply to takedown notices, which are unlikely to be filed given the nature of those datasheets and companies10:24
wolfsprauland in the long run we can work on what I believe are even better approaches such as dsv. all easy... :-)10:24
wolfsprauljust my 2c10:24
wolfspraulwpwrak: this is the bom source file http://downloads.qi-hardware.com/hardware/milkymist_one/bom/rc3/milkymist_one_bom_rc3.ods10:25
wolfsprauladam will update that no matter what for r4, I am sure10:26
xiangfuwpwrak, just report a bug on your patch: https://www.rtems.org/bugzilla/show_bug.cgi?id=203611:32
xiangfuhttp://www.milkymist.org/wiki/index.php?title=Pending_RTEMS_PRs updated11:32
xiangfuwpwrak, 'm1/patches/rtems/' also updated.11:33
wpwrakwolfspraul: yeah, just tried to explain to sebastien why you're likely to run into problems with an aggressive approach even though it seems at first sight to be in everyone's best interest.12:45
wolfspraullet's just step it up one level now with dsv etc then in the long run the datasheet repo will become unnecessary anyway12:46
wpwrakwolfspraul: (.ods) aah, now i see the workflow. thanks !12:46
wolfspraulthen we still have (theoretically) our 'deep level linking' friends, but I think even for you and me that's too much to think about :-)12:46
wpwraki think we're also in a much better negotiating position when is comes to deep linking12:47
wolfspraulI'm just saying we all have a line where it's enough and we want to get back to real work12:48
wpwrakafter all, we have a perfectly valid need for keeping those data sheets well accessible12:48
GitHub89[migen] sbourdeauducq pushed 2 new commits to master: https://github.com/milkymist/migen/compare/aac9752...2c375e913:05
GitHub89[migen/master] fhdl: add simulation functions in fragment - Sebastien Bourdeauducq13:05
GitHub89[migen/master] sim: remove default sockaddr - Sebastien Bourdeauducq13:05
Fallenou /win 4013:05
GitHub123[migen] sbourdeauducq pushed 1 new commit to master: https://github.com/milkymist/migen/commit/7230508e7cbcc783c302de1f0a05405d20487a2913:24
GitHub123[migen/master] fhdl/namer: do not reference objects with __del__ methods to avoid uncollectable cycles - Sebastien Bourdeauducq13:24
GitHub179[migen] sbourdeauducq pushed 1 new commit to master: https://github.com/milkymist/migen/commit/06de17b16cf43059d6262195d9ff37c290f9866613:26
GitHub179[migen/master] sim: remove temporary files and socket - Sebastien Bourdeauducq13:26
GitHub50[migen] sbourdeauducq pushed 1 new commit to master: https://github.com/milkymist/migen/commit/22b3c11b93436d9f15459ce3380aec82776a105314:06
GitHub50[migen/master] sim: clean startup/shutdown - Sebastien Bourdeauducq14:06
[florian]lekernel: hey14:07
GitHub174[migen] sbourdeauducq pushed 1 new commit to master: https://github.com/milkymist/migen/commit/9da512dbf5effc90a0c2a867f3a1e387250f8fa414:32
GitHub174[migen/master] sim: VCD generation - Sebastien Bourdeauducq14:32
GitHub23[migen] sbourdeauducq pushed 1 new commit to master: https://github.com/milkymist/migen/commit/0a23cadd38c831eff5e89718f2a7b15f20b5334d14:57
GitHub23[migen/master] vpi: install target - Sebastien Bourdeauducq14:57
GitHub28[migen] sbourdeauducq pushed 2 new commits to master: https://github.com/milkymist/migen/compare/0a23cad...6f829c715:52
GitHub28[migen/master] fhdl/verilog: fix signed constant conversion - Sebastien Bourdeauducq15:52
GitHub28[migen/master] sim: support for signed numbers - Sebastien Bourdeauducq15:52
GitHub172[migen] sbourdeauducq pushed 1 new commit to master: https://github.com/milkymist/migen/commit/db8f8bf2e375ddee9f5b3fdb376a1c2545e0a62d17:40
GitHub172[migen/master] fhdl: register memory objects with namespace - Sebastien Bourdeauducq17:40
wpwraksometimes, a click counter can be rather embarrassing ;-) http://www.chipcera.com.cn/en/articleinfo/detail_12_59_226.html18:03
lekernelwhat do you mean? no one uses their stuff?18:08
rohwpwrak: hrhr.. you propably doubled that counter just by pasting here18:08
lekernelwhat's a "Gao Rong pin"?18:11
wpwraklekernel: that's what they use on people who ask this kind of question ;-)18:11
roh'its just a pin, how bad can it be' *ducks*18:12
wpwraknice. taiyo yuden write english text with japanese fonts obscure enough to leave xpdf's output blank and to hang adobe's acrobat reader.18:28
wpwraksolution: install package xpdf-japanese (apparently "non-free")18:28
GitHub88[migen] sbourdeauducq pushed 1 new commit to master: https://github.com/milkymist/migen/commit/8160ced2e9ce96b4fba3c109a3ca16c7734ff83f18:36
GitHub88[migen/master] sim: memory access - Sebastien Bourdeauducq18:36
GitHub66[migen] sbourdeauducq pushed 1 new commit to master: https://github.com/milkymist/migen/commit/98e96b39522d92588cbfa8f5c48fa69df845f35218:50
GitHub66[migen/master] sim: make initialization cycle optional (selectable by function attribute) - Sebastien Bourdeauducq18:50
LmAthi, voidcoder.  Welcome.  Would you like to sit down and have some tea?  Tell us, what's on your mind?18:59
voidcoderwow, are you a bot LmAt?18:59
LmAtI'm not a bot.18:59
voidcoderyou greeted me so quickly, it kind of scared me into thinking you were one19:00
voidcoderthanks for inviting me, anyway19:00
voidcoderand I surely wouldn't mind some tea19:01
LmAtI'm not a bot.19:02
Action: LmAt gives voidcoder some tea.19:02
Action: voidcoder thanks LmAt for the tea and sits down in a comfortable chair19:03
LmAtI'm not a bot.19:03
wpwraknow .. how to convert an .odf spreadsheet to .csv from the command line ? jodconverter doesn't seem to work. complains about not being able to connect to ooffice -headless. and indeed, i don't see it listening on any port19:10
wpwrak(and the whole concept of firing [los]ffice as a kind of server is a bit scary anyway19:12
voidcoderI think there should just be a BSD licensed library for these kinds of things19:13
voidcoderI've done the whole UNO thing and it's just too complicated for most purposes19:13
wpwrakjust a nice little library and a command-line tool to transmogrify from one variant to the other would be wonderful, yes19:17
lekernelanyone succeeded in simulating lm32 with a recent icarus-verilog? now that we have pretty good simulation support in migen, we can easily do funny things, like connecting to the simulated soc with gtkterm :-)19:19
wpwrak;-))19:21
wpwrakjust make sure the simulation can't render in real time. or we'll have a marketing problem ;-)19:22
FallenouI think last time I tried with iverilog it failed to "compile" the source code19:29
Fallenouof lm3219:29
Fallenouso you advised me to use isim19:29
wolfspraulwpwrak: yesterday we briefly touched on cables with built-in data converters, and you more or less dismissed them as exotic23:19
wolfspraulbut I wanted to bring this back again for discussion since I do believe there may be a point there for Milkymist One or future products23:20
wolfspraulvideo in and out are quickly moving to digital23:20
wolfspraulI am paying attention to this all the time now and that's definitely something I see23:20
wolfspraulvga is disappearing faster than rca23:20
wolfspraulI already found cables for 5-10 USD that include a DAC/converter to go from DisplayPort to VGA23:21
wolfspraulI think what will happen is that the manufacturers of new devices will increasingly remove the analog video-in and out from their new products23:22
rohmost devices dont have analog video in anymore23:22
wolfsprauland for the large installed base of devices that want or emit RCA or VGA, there will then be a growing market for cables with built-in converters23:22
wolfspraulyes, just spelling it out23:22
wolfspraulso the market for video cables that include a DAC or ADC will grow, such cables will become more common23:23
wolfspraulimho23:23
rohi havent seen those around here23:23
wolfspraulI'm not saying we should dump our adc and dac now, and we couldn't even because it works today and is our only option.23:23
rohsure. no worries.23:24
wolfspraulbut one possible path forward could be to add digital video support, in and out, then look for cables, then one day remove the built-in adc and dac23:24
rohnothing wrong thinking about the actual new revision +123:24
wolfspraulif we are successful, I would bet that's what's going to happen over the next 1-2 years23:24
wolfspraulso we are doing the right thing. add digital video, first out, then see about in. make it work well (lots of work in the SoC). then start thinking about removing the analog path.23:25
wpwrakwell, if and when such cables are ubiquitous, you can consider such a change. similar to usb-midi.23:27
wolfspraulI'm not so much describing what I want to change, but what I believe are trends.23:27
wpwrakand you want digitial video support anyway, even if you keep analog around23:27
wolfspraulsure, that's the first step23:28
wolfspraulbut your point yesterday was that such cables are ... forgot what you said 'exotic' maybe?23:28
wolfspraulbut I don't think so23:28
rohyou will have enough to do with explaining the resolutions anyhow ;)23:28
wolfspraulnew devices will *remove* analog23:28
wolfspraulbut there's a large installed base of devices that only have analog23:28
roh'why doesnt it do 720p/1080i' ;)23:28
wolfspraulmeaning that the market for this type of cable will grow23:29
wpwrak(trends) maybe. could also be that products fork - all-digital for places that have a high replacement rate (with very expensive converters), and mixed or even all-analog for places where equipment only gets replaced after the 100th failed repair attempt23:29
rohdigital connections usually have the tendency to be copy protected and or not intended to be used as inputs23:29
wolfspraulroh: I think we are completely out of that type of comparison now23:29
rohe.g. i know of only very few hdmi-input cards (framegrabbers)23:29
wpwrakbut if these trends go as you describe, they don't require us planning much ahead anyway. we can just opportunistically pick technology that becomes available :)23:30
rohan nothing which does display-port in23:30
wolfspraulwpwrak: yes, absolutely23:30
wolfspraulour path is 100% correct23:30
wpwrakroh: sebastien one mentioned the possiblity of adding a pixel doubler :)23:30
wolfspraulI just wanted to add a snippet I realized about video cables with built-in converters23:30
wolfspraulwhich I will not even go search now, because we don't need them23:30
rohwpwrak: ack. anyhow.. lets add the dvi and get digital out working first. maybe we can abuse it to do digital in also... (for prototyping)23:31
wpwrak(DVI-in) ah yes, why not :)23:31
wpwrakand before we know, we'll want a means to power down DCC5V ((((-:C23:31
wolfspraulalright! :-) we think those are related?23:33
wolfspraulcan we (ab)use the dvi-i connector for digital video-in prototyping?23:33
wolfspraulI'll get a coffee first :-)23:33
wpwraknot sure if DVI-in would work electronically. Adam has looked at it quite a bit. maybe he has an idea.23:39
wpwrakif you can really just plug one into the other with the usual cable, you'll probably need a means to set HPD. DCC5V would be a minor issue, too23:42
wpwrak(not sure if anything horrible would happen if two 5 V supplies would work against each other. particular considering that ours if pretty well isolated already)23:43
rohno need to change it now. we can mod a cable for testing and nobody would use it as in because it would block the out besides testing23:43
wpwrakyeah. we'll always have patch wires :)23:47
wpwraklekernel: there's more to your gao rong pin than meets the eye. step 1: try to extract technical characteristics from this page: http://www.chipcera.com.cn/en/productinfo/detail_12_25_93.html23:52
wpwrakstep 2: consider the placement of the numbers 1-423:52
wolfspraulwhat is gao rong pin and what is this product?23:56
wolfspraulare we using those caps on m1?23:56
wolfspraulany chinese supplier naturally makes me nervous :-)23:58
wolfspraulthat one seems to be Taiwanese maybe? hard to tell23:58
wolfspraulin general I would advise against components from a Chinese supplier unless you personally visited and qualified the supplier, and have some China experience23:59
wolfsprauland those visits will kick out >95% of candidates you will visit...23:59
wpwraknaw, just kidding. they're a bit of untranslated chinese in the middle of a so-so translated page. and no, we don't use them :)23:59
wolfspraulphew, good23:59
wolfspraul:-)23:59
--- Wed Mar 7 201200:00

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