#milkymist IRC log for Friday, 2011-11-18

wpwrakno, the VBUS switch we consider adding00:01
wolfspraulI'm also for enabling software switching00:22
wolfspraulof course it's another risk, but so what00:22
wpwrakvery good :)00:29
kristianpaulsoftware switching reads like power saving feature for me :)00:42
kristianpaulbtw is anyone using FEL16 here? i guess/hope at least sebastien..00:52
aw_wpwrak, seems that you prefer to use an unpopulated 0402 footprint in series with them to s/w enable usb switch. Smart it is! so this won't affect if real know FPGA's pins in after reset as either Z or pull-up. Does it the idea go on? http://en.qi-hardware.com/mmlogs/milkymist_2011-11-17.log.html#t23:1302:33
wpwrakyes, it's independent. of course, it will default to on, which may not be ideal. but at least we can easily get control. and once this has worked for a while and sebastien stops worrying, we can simplify it :)02:40
wolfspraulaw_: I think we should make this work right away02:40
wolfspraulmake it really work or just don't do it at all :-)02:40
wolfspraulthe idea of replacing something that we worry about with something that only does the worrisome thing half-way sounds like a foolish attempt to battle with Murphy's law to me02:41
wolfsprauleither break through, or don't do it at all. I support either way.02:41
wpwraknaw, it doesn't do the worrying thing. but we can easily make it do it. so no risk but all the fun :)02:42
wolfspraulfoolish attempt to battle with Murphy02:42
wolfspraulthat's what that is02:42
wpwrakwell ... the things we do for peace :)02:42
wolfsprauland then some problem will show up with the half-way approach, guaranteed02:42
wolfsprauleither we make this work, or we don't do it at all02:43
wpwraklet's first see if any glitches pop up02:43
wolfspraulif it's too risky, then let's put it aside02:43
wolfspraulok that's design verification...02:43
wpwraki don't know yet whether to take sebastien's premonitions serious or not :)02:44
wpwrakand it may be a while until we really exercise this02:44
wolfspraulI just want to say we should focus on valuable working things, not on adding insurance everywhere.02:44
wolfspraulit seems there is some merit to the idea, with mwalle also supporting it02:45
wpwrakyeah, remember GTA02 and all the 0R/NC ? ;-))02:45
wolfspraulthe idea is that we can enable/disable power to USB ?02:45
wpwrakroh: during reconfiguration/partial reconfiguration02:45
aw_wolfspraul, i know your words. me also wanted this feature to be implemented actually. as management your thinking is good, but as executive side we can try this. from my views on routing, I don't see how it big risk on it.02:45
rohso what. what should these glicthes do? enable usb power? nothing bad should happen then02:45
wolfspraulaw_: ok. but like I said - I either want to work towards something that actually creates value, or not work on it at all.02:46
wolfspraulI don't want to work on virtual features, then we loose all focus and concentration.02:46
wpwrakroh: he thinks they could cut power. of course, there's a big silo cap right after the switch, so ...02:46
rohits not a power-amp for a man-sized metal mill which is contolled, its if a usbport has power within spec or not (and if it draws too much it will limit by hw, right?02:46
wpwrakwhen the switch os there, yes02:46
aw_wpwrak, btw, so your idea is to use only two routes to connect ENs of usb-switch. so how about FLGs or indicator, it seems that you wanted to use them for led.02:47
rohwpwrak: when? reconfig doesnt happen while using, right?02:47
wpwraksebastien's approach would be "always on". so the only thing he can worry about is accidental off02:47
wpwrakroh: not yet ...02:47
rohso if it happens once on boot, who cares? the usb will be rescanned after that anyhow02:47
aw_wolfspraul, true. understood. we can prepare parts first then design verification with Werner later, how so you think?02:48
rohif the chip gliches on generic io while reconfig we can not use that while the user has it02:48
wolfsprauldefinitely, get the parts, try, verify it's stable02:48
wpwrakaw_: ah, the FLG wold be nice, too. you can never have too much information :)02:48
roh'in production use' anyhow02:48
wolfspraulbut I need to understand which real goal we are chasing, and in this case it seems to be the ability to enable/disable usb power in software (from the fpga)02:48
wpwrakroh: as i said, i don't know what he's worried about and whether it makes sense02:48
rohwolfspraul: ack. not a bad idea. also wire the overcurrent-fail line also then (feedback from the switch to the fpga)02:49
wolfspraulthat sounds like a nice little feature to me02:49
wolfspraulroh: what is that? the FLG aw_ just mentioned?02:49
wpwrakyup. gain flexibility for whatever we may want to do with it. one use could be not powering USB when starting, so M1 would have a lower total current consumption around reset time02:50
rohwpwrak: 'normal' usb host ports on the pc disable/enable power to hard reset the bus as kind of last resort to get a port useable again after real bad issues (usually these come from bad/broken usb cables)02:50
wolfspraulsoftware control over power is good02:50
wpwrakroh: yes, that's another use case :)02:50
aw_wpwrak, if adding usb-switch in rc4, i can see i can just wire two wires from J21 to usb-switch while design verification stage. and connected +5V and VBUS power rail lines.02:50
rohwolfspraul: yes.. its sometimes called a flag. its simply a line which signals the status of the current limiter inside such usb-vbus switch/limit chips.02:51
rohfor lowering consumtion you need to also disable the usbphy also... but that would be another todo ;)02:52
aw_the usb-switch chip has a good two pins for two usb channel to indicate usb host current (charging status, etc) so my question is werner will you want to use FLGs to as led for two ports?02:52
rohe.g. http://people.openmoko.org/joerg/schematics/debug_board/OpenMoKo_Debug_Board_V3_MP.pdf02:54
roh ;)02:54
wpwrakaw_: no, i'd want FLG1 and FLG2 at the FPGA. and separate lines to the LEDs02:54
rohwpwrak: ?02:54
wpwrakaw_: no hard-wired power status logic, cf. ben charging led ;-)02:54
wpwrakbut now .. dinner :)02:55
aw_wpwrak, so this desire needs total four routes under fpga then connect to usb-switch.02:55
rohah. OC is called FLG there02:55
wpwrakaw_: yup02:56
aw_wpwrak, got it02:56
rohnice.. the chip you found even has seperate flags for each port? nice.02:56
aw_so i think we can pick active-low with AP2142A02:56
rohhm. and we need 2 pullups on flg1/2 then02:57
rohfrom my pov, if we add the possibility to hardwire the en_ pins we have basically zero added risk02:58
aw_wpwrak, i think later we can use J21's four pins  to implement this usb-switch while verifying.02:59
rohthe chip selected is built for what we want to use it, and with the possibility to place resistors to hardware it to on, its even a zero sw change needed.02:59
wolfspraulaw_: yes, I agree. do not connect leds to things that are automatically triggered by hardware/electrically.03:01
wolfspraulthat way we loose control over our board. Remember we are all software people in the end :-)03:01
wolfspraulthe report wires go back to the fpga, and the fpga controls the leds...03:01
rohhm. is the rc3 schem still altium? search is broken and one cannot click on signals *sigh*03:03
rohone question.. do we really need extra leds for that?03:04
aw_rc3 sche is broken? checking...03:04
rohwell.. not broken. just less featured03:05
rohseems pixels not vectors/text.03:06
aw_roh, oah..okay03:07
kristianpaulroh: yeah is a shame, but good side you develop patience and visual skills :)03:08
rohkristianpaul: i was just wondering. i thought we are on kicad now, not altium anymore03:09
wolfspraultoo many battles03:09
kristianpaulhe, how you get to that conclusion roh :)03:09
wolfspraulmy next kicad task is to uptick the keyboard patches (and work on 1-2 todo items regarding those from Werner) - haven't done that IN A YEAR03:10
wolfspraulwe could transfer the schematics into kicad as a first step03:10
wolfspraulbecause they are fairly stable03:10
wolfspraulthat way they would start to show up in schhist with proper version history etc.03:10
wolfsprauland we could boomify the bom, but that would most likely require more work in boom03:10
wolfspraulwork everywhere03:10
wolfspraulcan someone drop 500 people from somewhere? :-)03:11
roheasy... 500people working without pay would be impressive03:11
wolfspraulafter the schematics and bom, we could do the layout, but after we move the layout to kicad we loose access to the layout house we are using now, which is sad since they can share their experience and reduce risk03:12
wolfspraulbut at least on my todo list, that's many items away from the top :-)03:12
wolfspraulit starts with upticking keyboard patches and trying to upstream them into kicad03:12
wolfsprauldormant since January (!)03:12
wolfspraulroh: do you think you can find someone who is sourcing once in a while and could help develop boom?03:14
wolfspraulI want to use boom for m1 sourcing but I have so many tasks that realistically it just won't happen soon03:14
rohgood question. i dont have any idea about a specific person yet.03:27
rohbut i'll keep my eyes open03:27
rohbtw. we now have a extra plastic printer in the works03:29
wolfspraulyou mean a 3D printer?03:29
rohthe kit arrived yesterday... we completed it today. now its up to testing and calibration03:30
wolfspraulwhich model?03:30
wolfspraulgreat, please keep posting how it goes03:30
wolfspraulperformance, cost, etc.03:30
rohthe wiki is on http://wiki.shapercube.com (documentation etc)03:30
kristianpaulbtw shapercube said (i asked time ago) they dont have problems with hihg volumes (<500 units)03:31
wolfspraulI'm not one of the believers of the "3d printers will change the world" faction, but they are a great tool among many other tools, so that's good03:31
rohwe basically bought it because its more massive and bigger than a cupcake and not so much more expensive in europe03:31
kristianpaulof course if just a _said_03:31
wolfsprauland also there are many different variants and processes (as always in mechanical)03:31
kristianpaulah sorry sorry03:31
kristianpauli read bad and understood shapeways ,)03:32
rohreprap-based machines are still pioneer tools. and as such highly 'custom' over time.03:32
rohthere is a lot of variants one could test in the details and basically thats crowdsourced... best results are that what lands in the next revision of a machine03:33
wolfspraullike I said. I think a lot of the people currently joining the 3d printing hype seem a little naive in thinking that earth was created when the first reprap appeared out of dark matter, or so.03:33
wpwrak(keyboard patches) you mean command-line :)03:33
rohi think its nice to have, nice to hack, nice to use to hack, but not really good for mass-prduction/selling stuff03:33
wolfspraulin reality there are thousands of tools with all sorts of optimizations, in speed, materials, cost, etc.03:33
wolfspraulour friend marcan from OpenLase seems to be doing a little detour to a laser sintering 3D printer first, if I understood him correctly03:34
wpwrak(3D printing) the good ones seem to produce pretty respectable results. but you have to put money on the table. if you flinch when you hear about 5-digit figures at the point of going 6-digit, then that technology is not for you.03:36
wpwrak(yet :)03:36
rohnice idea. havent seen any supply of materials there (sinterable stuff etc)03:36
kristianpaulthere are actually roh , lase sintering looks very promising03:37
wolfsprauloh I think there is lots of stuff available, used, for this or that03:38
wolfspraulindustry equipment, from cheap to expensive, optimized for this or that parameter03:38
rohkristianpaul: depending on what your goals are. if its 'availability of raw materials/parts' should be good everywhere, you are a bit limited03:38
rohcurrent reprap variants can be built basically everywhere from local sourceable materials. (that is/was a design goal in most projects afaik)03:39
kristianpauland accurate, more natural at least can swap dark matter to chemical soup :)03:39
rohwe currently have white(natural), black and red ABS, and natual colored PLA (transluscent to transparent)03:40
kristianpauland now image this laser tech to micro umeter scale like azonenberg work :)03:41
kristianpaulroh: what color M1 next printed case will be? :)03:41
rohnice idea. different goal.03:41
rohkristianpaul: whatever wolfgang orders ;=)03:41
rohnext we want to upgrade our laser.03:41
rohmaybe we could do sintering experiments on that then03:42
azonenbergkristianpaul: if someone gets a device for under $2K that can do 365-405nm direct write at <50um resolution, i'm interested03:42
kristianpaulyou just put that stl somwhere i can copy it :)03:42
azonenbergi dont need a lot of power, this is for photolithography and not ablation03:42
rohgiven that we find a source for sinterable dust03:42
rohazonenberg: i dont really know your setup.. is it documented somewhere?03:43
azonenberghomecmos.googlecode.com has some info03:43
azonenbergmainly i'm interested in mask fab03:43
azonenbergevaporate or sputter metal over a piece of glass03:43
azonenbergspin coat in photoresist03:43
azonenbergthen draw my pattern at 1:1 scale into the photoresist with a laser03:43
azonenbergdevelop and etch the metal, then strip resist03:44
wpwrakazonenberg: we still haven't explored that direct bluray writer idea :)03:44
azonenbergwpwrak: i know03:45
azonenbergbluray has the power and then some03:45
azonenbergit's just focusing and aiming03:45
rohi know of people using a blueray diode on a old pen-plotter to cut paper03:46
wpwrakroh: wow03:47
wpwraki didn't imagine it to be that powerful03:47
azonenbergits a lot03:48
azonenbergi'd be firing it in short pulses03:48
azonenberglike 10-100ns03:48
azonenbergthen moving the beam and firing another shot03:48
wolfspraulazonenberg: do you know marcan and OpenLase?03:48
wolfspraulok I'll point him to this chat snippet03:49
wpwrakdifferent scale :)03:49
wolfspraulnah, but he is doing a 3D printer now and the galvanometer stuff is the same I believe03:49
azonenbergWell, what I *really* want to do is skip the laser entirely and do e-beam :p03:49
azonenbergBut thats a little... smaller03:49
wolfsprauldon't know whether there is overlap or not, but there might be some, at least in knowledge/experience03:49
wpwrakmaybe they're converging ;-)03:49
azonenbergor EUV :P03:50
azonenbergi say EUV because even the real semiconductor industry cant get it working :p03:50
wpwraksee. there's you chance to be waay ahead of the pack :)03:50
azonenbergI need a SEM first03:51
wpwraksemi-explainable matter, vulgo "dark matter" ?03:54
azonenbergscanning electron microscope03:54
wpwrak(sem) the price tag of your mad scientist's lab keeps on increasing. already the bribes for the environmental agency and all your chemicals must be a fortune :)03:55
azonenbergwpwrak: no bribes, everything is above board03:56
rohazonenberg: wish for a FIB workstation eh?03:56
azonenbergroh: A SEM is actually on the feasible list03:57
rohah. yes03:57
azonenbergwhile i'd love one of these, i dont expect to be able to get it any time soon http://www.semtechsolutions.com/node/138/zeiss-neon-40-esb-crossbeam03:57
azonenbergMy school has one in the cleanroom that cost $1.3M new03:58
azonenbergconsensus among a few friends is that used in good condition it'd probably sell for $500K ish03:58
azonenbergOn the other hand the JSM-840 they sell on the same site is probably in the low five figures03:59
azonenbergabout the price of a decent car03:59
azonenbergWhich is entirely affordable for someone with a PhD making >$100K a year, as i hope to be by 2016ish03:59
azonenbergBut i'm not going to spend five years' pay on one piece of equipment :p04:00
wpwraklet's talk about that in five years :)04:04
rohwell... give it time and a lot of hacking.. maybe it comes down to 'hackspace-affordable'04:31
rohi wouldnt have dreamed about being able to play with cnc equipment 5 years ago, in my own time in my own space04:32
rohand with industries and businesses crubling... keep your eyes open for surplus/used equipment :)04:35
azonenbergroh: yeah05:16
azonenbergI am hoping to bring IC fab to that level05:16
azonenbergthough its still a ways out05:17
wpwrakroh: spoken like a true geek. the world is coming to an end => TOYS !!10:12
rohwpwrak: :)10:16
lars_wpwrak: more importantly: _free_ toys13:25
lars_or at least highly subsidized13:25
lars_for the hamburg hacker space we recently managent to get hold of ~8 barely used cnc machines which were intended to go to the junkyard13:28
wpwrakwhat kind of equipment ?13:30
lars_and http://www.youtube.com/watch?v=vOpaRhF2RVY13:32
wpwrakholy shit13:35
wpwrakwhat materials are they good for ?13:36
lars_depends on the tools used13:36
wpwrakthat too :)13:37
wpwrakthe aluminium results seem to need a bit more work13:37
lars_i think we wrote a mail to the manufactur of the drill and they responded with which speeds we should use to get a proper result13:38
wpwrakah, good. i see that the fixation also seems to be an issue13:39
lars_that one was build using the cnd machine13:40
wpwrakthe speed is amazing. very fast.13:42
wpwrakwolfspraul: now we know where to get the aluminium M1 case from :)13:42
wpwrakthe quadcopter frame looks quite tidy indeed13:43
lars_unfortunately i'm no longer in hamburg13:43
wolfspraulyeah well. ownership of tools is only one small part of the problem, I think.13:44
wolfspraulin my experience13:44
wpwrakthey got these mills to lure you back ;-)13:44
wolfspraulsometimes the people sitting on the most expensive stuff are getting the least done13:44
wpwraktrue :)13:44
wolfspraulthe next step is how to translate the power of such tools into tangible results13:44
wolfspraulI have come to find out that it's best to just use the services of specialized small service providers/shops.13:45
wolfspraulthey are all over, and they will squeeze the last penny of value/revenue out of the machines they own (or have leased/etc).13:45
wolfspraulI'm not against hackerspace, academic 'shared/open' labs etc at all, just from practical experience. tedious opening hours, tedious communication channels, tedious access/waiting lists. stuff is 'unaccessible' for all sorts of minor maintenance problems, and so on. infinite list.13:46
wolfspraulmeanwhile whoever is running the hacker/shared/open space still has enough to take pictures and write blog posts :-)13:46
wpwrakyeah, you must be able to afford low duty cycles13:48
wpwrake.g., my mill is ready whenever i am. but it's idle 99.9% of the time. of course, the transaction cost of doing things at a shop would be very large, in comparison13:49
wolfsprauldon't understand13:50
wpwrakso i think for development, this sort of equipment is good to have13:50
wolfsprauloh sure, theoretically13:50
wolfspraula shop has an efficient system of processes, communication and people built around investment in tools13:50
wolfspraulis that happening at the hamburg hackerspace?13:50
wolfspraulthey push the inefficiencies outward because they have no incentive to make access to those tools easier13:51
wolfspraullet's say "could", because I don't know what exactly will happen13:51
wolfspraullet's find out in 1-2 years :-)13:51
wolfspraulthe difficult part is not in getting tools/machines, I think the difficult part is to efficiently keep them running13:51
wpwrakwell, if i did my milling to a shop, each cycle would take at last a day, probably several. and i may need a number of cycles to get the results i want.13:51
wolfspraulI have seen so much expensive stuff collecting dust by now, I don't want to see it anymore.13:52
wolfspraulI stand firm here making the point that the more expensive tools someone has, the less they get done. :-)13:52
wpwrakmy volume is low, so there's little benefit in the shop then being able to do larger runs (which would be more efficient)13:52
wpwraknaw, not true13:52
wolfspraulhe :-) it wasn't entirely serious...13:53
wolfspraulbut let's say that happens sometimes, right? :-)13:53
wpwraki see it a lot of times that i'm spending huge amounts of time and effort to work around equipment limitations13:53
wolfspraulwasn't there a guy at MIT saying he has this super great laser scanner and would scan the NanoNote parts in 30 seconds?13:53
wpwraki think it depends on how you decide what to buy13:53
wolfspraulI think we are still waiting for anything but that mail from him...13:54
wpwrakif it's "this could be useful", then there's a high risk of it being unused13:54
wolfspraulthat's not a lonely exception, in fact that is my expectation right when I read such a mail :-)13:54
wpwrakif it's "it would have helped with this task in the past, and there will be similar ones in the future", then that risk is low13:54
wpwrakheh. those case scans are indeed an interesting area :)13:55
wpwraki think one of the expensive tools issues are also ritualized purchases. at a lot of places, you can't just go and buy things when you need them. you have to wait for the right moment. and then you suddenly have a ton of cash and need to quickly grab everything you'll need for the next year(s)13:57
wpwrakand then you have managers making engineering purchase decisions instead of engineers. then it can become a matter of prestige to buy the shiniest, whatever it does13:58
wpwrakyou weren't around yet at openmoko when they shut down fiwin. you would have loved it ;-)13:59
wpwrake.g., that stack of GSM protocol analyzers (each the price of a moderately luxurious car) that was just gathering dust in a corner of a room has "buying frenzy" written all over it14:03
wolfspraulyeah, that's why I say. my experience: expensive stuff collects dust, cheap stuff is used to produce results.14:08
wolfspraulit's not the whole truth, but there is some truth in it.14:08
wolfspraulhas to do with human psychology, I guess :-)14:08
wolfspraullet's see what happens to the Hamburg machines, I hope they will be used and are easily accessible to as many worthy projects as possible14:08
wolfspraulbut just right there the nasty little details start, right? :-)14:08
wolfspraulyou wouldn't want some 'unqualified' (?) person break it? :-)14:09
wpwrakyeah, public access is always tricky14:12
wpwraki prefer the ownership model - i own my tools and nobody else gets to touch them. keeps things simple ;-)14:14
lars_well, you are allowed to use them if you have undergone some basic training14:14
rohlars_: well.. if you need access to a mill in berlin.. holler ;)14:14
lars_roh: i'm neiter in berlin :/14:15
rohah. right.. you went to muc?14:18
wpwraknow that he's in muc, he can't muck with the toys in hamburg :)14:25
rohbut if you are at it.. we still have space for more cnc machines here at raumfahrtagentur. and the cccmuc would surely also gladly adopt one... and hh would have more free space to build more mills with mills14:44
wpwrakthey're probably busy to turn them into a single monster with 8*3 = 24 degrees of freedom. then they can experimentally prove string theory :)14:50
--- Sat Nov 19 201100:00

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