#milkymist IRC log for Thursday, 2012-05-10

cladamw(AD m1r4 latest placement) http://downloads.qi-hardware.com/hardware/milkymist_one/pcb/r4/051012/MILKYMISTONE.pdf07:04
cladamw(AD m1r4 design files) http://downloads.qi-hardware.com/hardware/milkymist_one/pcb/r4/051012/r4_design_20120510.tar.bz207:04
cladamw(AD m1r4 DWG file) http://downloads.qi-hardware.com/hardware/milkymist_one/pcb/r4/051012/MILKYMISTONE.DWG07:05
cladamwwpwrak, you could see that pdf and dwg for placement. Meanwhile I'm also checking.07:06
wpwrakhmm, DWG is proprietary. let's see how things go with the other files ..09:12
Fallenoudwg, it reminds me microstation format09:12
Fallenouoh no autocad09:13
wpwrakDXF's evil twin09:13
wpwrakaha, L3 is not yet placed09:13
wpwrakhmm. now. what is it ...09:15
wpwrakah. there is no L3. that's why :)09:16
wpwrak.SchDoc are useless without AD09:18
wpwrakcladamw: can you generate gerbers from all this ?09:19
wpwrakyeah. not enough information in the PDF. we need gerbers.09:33
cladamw(L3, L19) you can ignore them now, since they must be there now for importing and deal with copper filling later.09:36
wpwrakah, technical meta-components :)09:37
cladamw(dwg) hmm... good reminders.09:38
cladamwdeleted09:38
cladamwso you can use gerber to generate dwg file ?09:39
wpwrakmaybe someone could. not me :)09:39
wpwrakbut i can do other things with gerber :)09:39
wpwrakmany other things. fun things :)09:39
cladamwsince ask to request for generating gerber files, to house means to pay first. I have to deal with them first. i can't generate gerber files in AD.09:41
wpwrakargh :-(09:41
cladamwsince we need to make 'placement' done first then routing, not to generate gerber files firstly,09:42
wpwrakah, so it's not fully routed yet. good. i was wondering how they had done DVI :)09:43
cladamwno, it's not routed yet though,09:43
cladamwso now you and me to review that pdf files firstly to those all placements, if we agree current placement, then we tell house GREEN to start to route, or ask few to modify09:45
wpwrakhmm, could they try to move C178 "east" of DMX TX ? that way, it wouldn't be under the expansion board09:49
wpwrakalso, if they could shift J24 a bit to the south (1-2 mm), that would be good. it's low-priority, though. if they can't we can work around it on the adapter board09:51
cladamw(C178) yes, i also noticed this capacitor, it can be moved to the place upper area of U2709:53
wpwraki wonder if we should move SW2 ~1 mm "south", to increase clearance to J1609:57
wpwrakyeah. that's all i see at the moment. what did you find ?10:02
wpwrakand did they have any comments ? e.g., regarding the overall difficulty or some problem areas ?10:06
cladamw(C178) is for usb A/B, it should be close to U26 as possible. (C286) should not be under that supposed usb dongle area. and can be moved to east of DMXTX con.10:08
cladamw(J24) i suggest not to move it south a bit now.10:09
wpwraki just noticed that C178 used to be in the upper right in M1rc3. so it's traveling south, then north again :)10:10
wpwrak(U26 & C178) ah, i see. hmm. tricky10:11
cladamw(C178) wpwrak, yes, but their function are different to target now.10:11
cladamwbut still for decoupling and surge goal10:11
wpwrakC286 is not a big issue. but if we can move it, better (gives us more flexibility for the adapter board)10:12
cladamwso a C178/220uF should be still there to close U26.10:12
cladamw(C286) yes, I'll ask to move to east of DMX TX.10:13
wpwrakcan we send U26, C259, and C279 to the bottom of the PCB ?10:13
wpwrakthen C178 could be behind SW210:14
wpwraks/C279/C280/10:15
cladamw(bottom side idea) i was thought this, but the 220uF is quite high  a bit, if we don't care the newer case design, YES, we can let them to be bottom side, then everything are happy.10:15
cladamw(U26) i don't know if put bottom is good or not, I'll ask for chance.10:18
wpwraks/C280/C260/   hard to read :)10:18
wpwrak(C178) i mean moving some the parts that are behind SW2 to the bottom. so that would be C258, C259, U26, etc. then C178 could be on top, behind SW2.10:19
wpwrak(U26) ok. great.10:19
cladamw(C280) now is sit south side of U30 (bottom side) , i can read it from design files, no worry on it.10:21
wpwraki mis-read. i wrote C280 but i meant C26010:22
wpwrakand then i confused it with C258. so C280 should really mean C258 :)10:22
cladamw(C260) current it's on top side, but they are group parts for usb part a/b : U26, C178, C[256..260], so I can ask them for if placing at bottom side.10:23
wpwrakor maybe i'm still confused :) it's very hard to tell which component is which there10:23
wpwrakokay, you got it :)10:23
cladamwsure , since you are read them from pdf, it's quite hard to read a bit. :)10:24
cladamwwe go through a bit then i noted them in wiki.10:24
cladamw(J16/J20) last time they are not in right placement, now they are correct to be far away from board edge a bit, so i asked  them.10:27
cladamw(led cut off area) http://en.qi-hardware.com/wiki/File:M1r4_apa1606surck_cutoff_layer.png10:28
cladamwthis is the new, but i pdf you can not see it wider.10:29
cladamws/i/in10:29
cladamw(8 leds) between C178 and U8, i think they are okay.10:31
cladamw(U23) for 4V3 audio supply parts are moved to J21, should be also okay.10:32
wpwrak(J16/J20) yes, looks good to me10:33
wpwrak(8 leds) yup. i like them :)10:34
wpwrakspeaking of LEDs. in the kicad schematics, they still need to be labeled10:35
wpwrak(LED cut out area) looks good. the 1 mm on both sides are clearances to other components ?10:35
wpwrak(LED in PDF) i don't see that area at all :-(10:36
cladamw(J5, J6) they are now surely moved south a bit. but not a big improvement for solving problems on plugging right angle use cable.10:36
wpwrak(J5/J6) we should see if we can make a case for a straight cable there :)10:37
cladamwyes, 1mm for expending cut off area only10:38
wpwraknot sure if the laser could make an "almost through" cut. such that one could manually break out the piece10:38
wpwrakroh: what do you think ? is this something laser-cutting can do ? or would it be a mess ?10:38
cladamwso those 8 leds between C178 and SW2, they are connected closely without clearance. but for example if you see led for U8(IR), the cut off area would be bigger.10:40
wpwrak(leds) hmm. not sure what you mean. all the leds seem to have room for the cut-out area and for ~1 mm lateral clearance10:41
wpwrakof course some leds have a component on top of the cut-out area. but that's life :)10:42
cladamw(J27 mechanical mounting hole) it's close to J21, i like its position now, how do you think ?10:43
cladamw(leds) i'm capturing pictures ...10:43
wpwrak(J27) you have good eyes ;-) it looks decent. may be better a little bit (~ 3 mm) more to the south, though.10:45
cladamwhttp://downloads.qi-hardware.com/people/adam/m1/tmp/m1r4/m1r4_led_bigger_cutoffarea.png10:47
wpwrak(cut-out) you mean you want to move the LEDs away from the board edge ?10:48
cladamw(J27) i was tried to think it should be close to fpga chip as well. :-)10:48
cladamw(cut-out) no, not away from board edge.10:49
cladamwwider for more light radiate as clear as possible through pcb.10:50
cladamw(J27) so you want it really a bit more south ~ 3 mm ?10:51
wpwrak(J27) i was thinking of having it closer to the center of J21, so that it can absorb the force more directly10:51
wpwrak(J27) i _think_ it would be a little bit better. maybe it doesn't make a difference. maybe i'm wrong :)10:52
cladamw(J27) ha, i don't think so, also maybe i am wrong, since we all don't have mechanical simulation s/w to express how force is and how bend it will. :-)10:53
cladamwbut I think the solderaibility on through hole headers pins are much more stronger than fpga. :-)10:54
wpwrak(LEDs) ah, i see. it probably doesn't make much of a difference, given that the LEDs are already very close to the board edge. but why not.10:54
wpwrak(header stronger than FPGA) yes, agreed on that ;-)10:55
cladamwokay, so let's still move south J27 to 1.5 mm or ... hehe   :-)10:55
wpwrak;-)10:56
cladamwelse being discovered ?10:57
wpwraknope. i didn't notice anything else10:58
cladamw(ugly reference texts) there're many ugly reference text around each part, you can ignore them now, since house use chinese version for it, and things came out.10:58
wpwrakokay, chinese would be even uglier for me ;-))10:58
cladamwhouse will clean them up once in the end when we ask to generate gerber files.10:59
cladamw(J7, J8) wpwrak, could you check our KiCad symbol and current AD layout gerber (in rc3) once you item by item to check both ?11:00
cladamwsince I perticularly added shield ground pins, you can see their difference in symbols. (schematics of KiCad and AD), but no rush for now...11:02
wpwraki see a ground pin on J8. none on J7. what do you want me to check ?11:03
rohwpwrak: do what?11:05
wpwrak... which looks good, given that J7 has 1 and GND combined, according to the data sheet11:05
cladamwJ8's added pin in KiCad, i am thinking if that we've done wrong on previous design ?11:06
wpwrakroh: make a partial cut that would allow a small rectangular area (for a mini-USB cable) to be broken out of the sidewall (for that JTAG board)11:06
rohwpwrak: very difficult to make reliable on different batches of material11:07
rohalso looks shitty on the breaking point (unlinear breaks)11:07
wpwrakcladamw: ah, i see. hmm yes, might be good to do that in AD as well.11:07
rohor bigger parts will break off.. acryllic is not very flexible.11:08
cladamwthat pin in all previous version are not connected to even system ground or anti-emi shielding ground. so far I didn't know it's purpose. i think Neutrik designed it with purpose but we didn't use it.11:08
wpwrakhmm yes, for clean breaks, we would have to cut from both sides11:08
rohwpwrak: thats not possible. it breaks apart when taking it out11:12
cladamwi go for dinner first, cu11:12
rohor.. different way.. it would be much cheaper to add another different, wallpart with the hole than to make one which is breakable properly11:13
wpwrakroh: oh, the part you remove can be destroyed. just the rest of the sidewall would have to remain intact11:13
wpwrakyes, that's plan B :)11:13
rohwpwrak: acryllic is brittle. when it breaks is nearly impossible to make it only break where you want it. on linear cuts and with a v its possible. but not with something small with 3 sides11:14
rohit most likely would break along the line beyond the v-cut and destroy more than intented11:14
roheh intended11:15
wpwrakhmm. haven't tried braking acrylic yet11:15
wpwrakokay, so plan B then :)11:15
rohwhy... do you want to have the debugboard in all cases?11:16
rohor just package the sidewall with the debug board11:16
wpwraki think the jtag board will still go into every device. just in case.11:16
wpwrakif the jtag board would be a separate item, i'd say the plan B wall would be perfect. if it's preinstalled, then plan B wall is a bit trickier.11:18
rohwell.. lets see11:26
qi-botThe firmware build was successful, see images here: http://fidelio.qi-hardware.com/~xiangfu/build-milkymist/milkymist-firmware-20120510-1228/12:11
qi-botThe firmware (using branch) build was successful, checkout the VERSIONS for detail, see images here: http://fidelio.qi-hardware.com/~xiangfu/build-milkymist/milkymist-firmware-20120510-1411/14:15
FallenouISE 14.1 has been released19:57
Fallenouwill it be a good, or a bad surprise ? :)19:58
larscwe have some designs which don't work with 14.1 anymore :/19:58
Fallenou:/19:59
Fallenoudo they even test for regressions ?19:59
larscthe main feature of 14.1 is the new zynq support anyway19:59
Fallenousure19:59
Fallenouwhich we don't care about19:59
Fallenouand 4 Artix-7 removed from the tools :)20:01
Fallenoutoo bad for those who bought them20:01
wpwrakeven worse for those who designed for them20:02
kristianpauland used lots of xilinx libs..20:02
Fallenouand MicroBlaze updates20:02
larsci guess one of these updates breaks our designs ;)20:03
larscour = at work20:03
larscbtw. i've recently read that xilinx bought/is working on a new design suite which is said to be 10x faster at building a project20:05
larscand also supports partital rebuilds20:05
Fallenoumaybe a good news !20:06
kristianpaul:o20:08
kristianpaulvivado?20:09
larsccan't remember the name20:12
wpwraklarsc: have they decided to compile the bubble-sort with -O9 ? :)20:12
larschaha20:12
larsci suppose they've come up with better heuristics or something20:13
kristianpaulhope that doest impact resource utilization..20:14
larscbut i'm really looking forward to the partital rebuild feature. no more re-building everything, just because you changed a single line20:14
kristianpauls/doest/doesnt20:14
kristianpaullarsc: if this works for s6 will be cool, but i guess not..20:14
kristianpaulindeed, thats a lot of time saving20:14
larscand the suite is also said to have a intermediate layer for hdl code, so both verilog and vhdl get translated to that intermediate layer and optimizations are run on it20:16
larsckind of like llhdl was supposed to be20:16
GitHub162[milkymist] sbourdeauducq pushed 1 new commit to master: https://github.com/milkymist/milkymist/commit/3b7019b1641a9b8db969034c0091a0fd440857de23:21
GitHub162[milkymist/master] navre: fix missing NBA conversion - Michael Walle23:21
wolfspraullarsc: you start to believe in software announcements :-)23:27
wolfspraulFallenou: which artix-7 were removed from ise 14.1 ? (checking too...)23:28
wolfspraulok, I looked at the release notes http://www.xilinx.com/support/documentation/sw_manuals/xilinx14_1/irn.pdf23:50
wolfspraulyes, they removed support for xc7a8, a15, a30t and a50t. but that was already announced a while ago, and I think the main reason is that they want to focus on first getting some chips out, any artix-723:51
wolfspraulit seems they have thrown their initial support behind the -100 -200 and -300 variants23:51
wolfspraulI'm sure 'later' (2013 or later) there will be other variants, depending on sales opportunity and where to spend their resources (read: trouble with the -100 -200 and -300), etc.23:52
wolfspraulbut the good news in ise 14.1 is that it's now official that the artix -100 and -200 will be supported by webpack. -300 will not.23:52
wolfspraulthat makes the XC7A200T the most powerful low-cost chip supported by webpack23:54
wolfspraulnow we only need to wait another 6-12 months, then we have prices for the XC7A100T and XC7A200T and can think about whether this is an attractive option for Milkymist...23:55
--- Fri May 11 201200:00

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