| 00:15.41 | *** join/#edev prpplague (~prpplague@107-206-64-184.lightspeed.rcsntx.sbcglobal.net) |
| 03:03.10 | *** join/#edev JoeLlama (~snork@unaffiliated/joellama) |
| 03:22.13 | *** join/#edev MrDark (4a8b2098@gateway/web/freenode/ip.74.139.32.152) |
| 03:22.34 | MrDark | anyone know what i should look in to, to solder embedded cpu chip to a board ? |
| 03:22.52 | SpeedEvil | a soldering iron |
| 03:23.02 | MrDark | i been looking at arm cortex |
| 03:23.09 | MrDark | very very tiny pins |
| 03:23.56 | MrDark | i had someone design the board layout |
| 03:24.31 | SpeedEvil | what package? BGAs? qfn? |
| 03:24.47 | SpeedEvil | also, ##electronics would be more appropriate |
| 03:24.57 | MrDark | i thought this was for embedded ? |
| 03:25.35 | MrDark | and it qfn |
| 03:26.10 | MrDark | im more of a hardcore software designer, i been wanting to learn a lot more about hardware |
| 03:27.56 | MrDark | i seen stuff about surface mounting, anyone done that before ? |
| 03:31.58 | MrDark | my project based off of pandaboard if that helps |
| 03:33.06 | SpeedEvil | this is embedded, but that's a 'pure' electronics question |
| 03:33.19 | SpeedEvil | and you're more likely to get a response there |
| 03:34.41 | MrDark | alright thank you, what all do you talk about here ? |
| 03:35.05 | MrDark | i been working with the embedded devlopment kit (Pandaboard) for around 3 months |
| 03:36.26 | SpeedEvil | 'based off' - you mean attached to? |
| 03:36.54 | SpeedEvil | I don't t think any of the panda board class CPUs come in qfn |
| 03:38.29 | MrDark | i might been confused about the cpu sorry speedevil and yes attached to, i written a custom kernel and interface for it |
| 03:40.50 | MrDark | i would not say custom but i modified a lot of the code and written a lot of new functions in it |
| 03:41.47 | prpplague | hehe, no OMAP44xx series is not available in QFN |
| 03:42.18 | prpplague | there are some nice arm9 and arm11 soc's available in QFP but you'll find most modern SoCs are going to be BGA |
| 03:42.32 | MrDark | i was just wanting to move the cpu and parts i needed to a smaller board |
| 03:42.58 | MrDark | only issue i was having was trying to figure out how to attach the cpu to the board |
| 03:42.58 | SpeedEvil | is this a Linux kernel? |
| 03:43.05 | MrDark | yes SpeedEvil |
| 03:43.12 | prpplague | MrDark: http://www.phytec.com/products/som/Cortex-A9/phyCORE-OMAP4460-OMAP4430.html |
| 03:43.21 | prpplague | MrDark: essential the panda without the connectors |
| 03:43.37 | SpeedEvil | if you are asking that, you're at the wrong place in your design. |
| 03:43.56 | prpplague | hehe |
| 03:44.01 | SpeedEvil | you work out what capabilities you have for manufacture, then pick the parts |
| 03:44.52 | MrDark | is there a place or factory that would make a prototype ? |
| 03:44.56 | SpeedEvil | if you have to ask, you are not going to properly solder the first one. even with the equipment |
| 03:44.59 | SpeedEvil | yes |
| 03:45.21 | SpeedEvil | http://www.mauve.plus.com/opensourcehw.txt |
| 03:45.33 | SpeedEvil | is a canned rant of mine on the topic |
| 03:46.22 | *** join/#edev MonMotha (~monmotha@2001:470:c5f2:1:224:7eff:fedc:893e) |
| 03:47.49 | prpplague | MrDark: what is it you building to interface to the pandaboard? |
| 03:48.32 | MrDark | prpplangu i had a board designed small enough to fit in my project |
| 03:49.17 | MrDark | i was just asking around if people know how to solder or mount a cpu chip, but ill probably end up sending it to a company to produce it |
| 03:50.00 | prpplague | hehe |
| 03:50.24 | MrDark | there a lot of parts i dont need with pandaboard |
| 03:50.31 | prpplague | didn't actually answer the question |
| 03:50.45 | MrDark | you mean interface ? |
| 03:50.58 | MrDark | like microcontrollers or input to it ? |
| 03:51.37 | prpplague | MrDark: if you are going to use the OMAP4 which the pandaboard is based on, only two companies are authorized to do prototype designs using OMAP4 |
| 03:51.41 | prpplague | svtronics and circuitco |
| 03:52.07 | prpplague | MrDark: like i said, you probably should look at a SoM from phytec |
| 03:53.18 | MrDark | what do you think would be the charge for a single board? |
| 03:53.27 | MrDark | kinda running low on budget |
| 03:53.38 | MrDark | i got the board designs and list of all the parts |
| 03:53.49 | MrDark | there around $11,000 left |
| 03:57.34 | MrDark | i guess i can can stick to pandaboard but i liked the idea of having my own board |
| 03:57.38 | MonMotha | MrDark: the OMAP4 line, if you can even get info on them let alone physical parts, is a pain in the butt to physically assemble as it's PoP |
| 03:58.19 | MonMotha | there are some other ARM Cortex-A parts on the market that are much more low-volume friendly (though that still often means fine-ish pitch BGA which will cost you a premium on assembly in low quantity) |
| 03:58.38 | prpplague | MrDark: no clue as to what you design is, but i just did a run of 10 custom OMAP44xx boards, just 10 boards, and the batch cost about $120,000 USD |
| 03:58.52 | MonMotha | the system-on-module cards available from various places are much more amenable to low volume/one-off designs |
| 03:59.06 | MonMotha | prpplague: curious, was that bare PCBs or full assemblies? |
| 03:59.17 | prpplague | MonMotha: full assemblies |
| 03:59.30 | MonMotha | prpplague: hum...not terrible, but still salty of course |
| 03:59.55 | prpplague | MonMotha: yea, it was high than expected due to the stackup |
| 04:00.38 | MonMotha | prpplague: the PCB rules needed to bring those things out are kinda insane. I can easily see $5000+ PCB in that quantity. |
| 04:01.01 | MonMotha | 0.5mm pitch, right? |
| 04:01.07 | prpplague | 0.4mm |
| 04:01.15 | MonMotha | oh, even better! |
| 04:01.22 | MonMotha | so...4/4 via in pad? |
| 04:01.39 | prpplague | yea with burried and blind vias |
| 04:01.41 | MonMotha | not knowing the pinout, I'm guessing at least 8 layers |
| 04:01.45 | prpplague | MonMotha: 112 |
| 04:01.47 | prpplague | oops |
| 04:01.48 | prpplague | 12 |
| 04:02.01 | MonMotha | heh, I'm sure some fabricator would quote a 112 layer design just for grins :) |
| 04:02.11 | prpplague | with 2 layers dedicated for thermal and ground items |
| 04:02.28 | MonMotha | wow...I'm used to multiple ground layers but a dedicated thermal layer is pretty intense |
| 04:02.39 | prpplague | MonMotha: ended up with .072 thickness |
| 04:03.04 | MonMotha | heh, just close enough to be 0.062 compatible, I guess? |
| 04:03.23 | prpplague | MonMotha: yea a copper flood bringing it out to screw hole points to connect to metal housing |
| 04:03.27 | prpplague | yea |
| 04:03.31 | MrDark | i been overclocking the arm cortex |
| 04:03.36 | MonMotha | in some ways, I miss getting to do these kinds of designs. OTOH, there's a lot of pressure to get it right. |
| 04:03.46 | prpplague | MonMotha: yea |
| 04:03.53 | MonMotha | prpplague: woah, what was driving that much thermal power? |
| 04:04.08 | prpplague | MonMotha: omap4470 running at 1.8GHz |
| 04:04.14 | MonMotha | what's the dissipation on that? |
| 04:04.28 | MrDark | liquid nitrogen burns the chip out in about 3 days |
| 04:04.52 | prpplague | MonMotha: i'd have to look at the numbers, but 4470 mounted on the panda, will hit thermal shutoff at 120C in under 3 minutes |
| 04:05.22 | prpplague | MonMotha: i normally test it with a sandybridge heatsink and cooling fan |
| 04:05.24 | MonMotha | MrDark: LN2 seems a bit excessive in terms of trying to overclock a Cortex-A. Most of them are at most just a few watts. You just need a big enough heatsink and good thermal conductivity from the CPU |
| 04:05.45 | MonMotha | prpplague: hum, that sounds like maybe a couple watts. The package is pretty darned small. |
| 04:05.55 | MrDark | prpplague but liquid nitrogen or water cooling will avoid thaty |
| 04:06.17 | MrDark | but burns the chips insides up quick |
| 04:06.19 | prpplague | MrDark: not for a commercial product |
| 04:06.29 | MonMotha | then again, "a couple watts" in a typical tight handheld or similar can be hard to get rid of |
| 04:06.57 | prpplague | MonMotha: yea |
| 04:07.06 | prpplague | MonMotha: the extra thermal layer does wonders |
| 04:07.27 | MonMotha | prpplague: makes sense |
| 04:07.31 | prpplague | MonMotha: we wanted to get one in for the panda, but the management wanted it in as few layers as possible to give people a reference design |
| 04:07.44 | MonMotha | prpplague: I usually can just use the system ground plane for that, but that may not work at 1.8GHz |
| 04:11.25 | MonMotha | prpplague: do you know how many drill pairs you ended up with on that board? |
| 04:12.39 | prpplague | MonMotha: not off the top of my head, the design is a little foggy since it was done before the holidays and before the RiF |
| 04:13.19 | MonMotha | no problem |
| 04:13.21 | MonMotha | I was just curious |
| 04:13.45 | MonMotha | every time I've considered doing blind/buried vias, I've backed away after checking the cost and accepted the larger board |
| 04:15.14 | prpplague | MonMotha: yea avoid it unless you have a darn good reason and you got a customer with cash |
| 04:15.43 | MonMotha | no kidding |
| 04:16.19 | MonMotha | I'm just sometimes amused at how much PCB space I burn with vias, and that's even being very careful with the fanouts and routing |
| 04:55.00 | *** join/#edev [7] (~quassel@rockbox/developer/TheSeven) |
| 05:27.14 | *** join/#edev mrcan (~mrcan@unaffiliated/mrcan) |
| 06:06.19 | _av500_ | prpplague: your PCB prototype costs are insane |
| 06:12.14 | SpeedEvil | prpplague: $10k/board??? |
| 06:12.22 | SpeedEvil | is this multiple spins? |
| 06:21.46 | *** join/#edev clio (~andrej@85.159.109.222) |
| 06:34.47 | _av500_ | we move all proto stuff to china |
| 06:51.15 | MonMotha | SpeedEvil: that's quite reasonable for a board of that complexity including assembly |
| 06:51.17 | MonMotha | (in that quantity) |
| 06:51.40 | MonMotha | the set up costs are insane, and I wouldn't be surprised if the PCB shop threw half the boards away |
| 06:52.13 | SpeedEvil | I see the above rant may have been optimistic |
| 06:52.24 | MonMotha | the OMAP CPU is a nasty beast physically - 0.4mm pitch BGA with IIRC no cutout PoP with another part of similar density |
| 06:53.16 | MonMotha | if it were usable on more standard design rules, yeah that would be a bit insane for anything other than very rapid turn |
| 06:53.42 | SpeedEvil | oh, right |
| 06:54.00 | SpeedEvil | I guess that requires microvias and other nastiness. |
| 06:54.03 | MonMotha | it's pretty much exclusively targeted toward mobile/handheld |
| 06:54.25 | MonMotha | 4/4 mil trace/space, via in pad, and probably at least 5 blind/buried drill pairs |
| 06:54.30 | MonMotha | and apparently 12 layers |
| 06:55.02 | MonMotha | in qty 10, I could easily see paying $5000/pcb for the bare board without any expedite costs |
| 06:56.10 | MonMotha | I'm sure the board had a bunch of other fun mechanical quirks (weird shape, counterbores, etc.), but at that point those aren't a big deal |
| 06:58.18 | SpeedEvil | thinks there has to be a better way. |
| 07:12.46 | *** join/#edev gia (~fire@unaffiliated/security) |
| 07:35.55 | *** join/#edev JoeLlama (~snork@unaffiliated/joellama) |
| 07:58.02 | *** join/#edev megha (~fire@unaffiliated/security) |
| 08:28.39 | *** join/#edev florian_kc (~fuchs@port-217-146-132-69.static.qsc.de) |
| 08:28.39 | *** join/#edev florian_kc (~fuchs@Maemo/community/contributor/florian) |
| 08:35.17 | *** join/#edev megha (~fire@unaffiliated/security) |
| 09:11.00 | *** join/#edev jackmitchell (~Thunderbi@195.171.99.130) |
| 10:08.34 | *** join/#edev megha (~fire@unaffiliated/security) |
| 10:57.26 | *** join/#edev SpeedEvil (~quassel@tor/regular/SpeedEvil) |
| 11:09.10 | *** join/#edev jackmitchell (~Thunderbi@195.171.99.130) |
| 11:18.51 | *** join/#edev gustavoz (~gustavoz@host4.190-138-192.telecom.net.ar) |
| 12:06.19 | *** join/#edev g1powermac (~g1powerma@rrcs-74-87-150-6.west.biz.rr.com) |
| 12:06.19 | *** join/#edev g1powermac (~g1powerma@unaffiliated/g1powermac) |
| 12:48.01 | *** join/#edev calculus (~calculus@gentoo/user/calculus) |
| 12:58.09 | *** join/#edev megha (~fire@unaffiliated/security) |
| 14:47.17 | *** join/#edev megha (~fire@unaffiliated/security) |
| 15:33.27 | *** join/#edev prpplague (~danders@192.94.92.14) |
| 15:36.23 | *** join/#edev keesj (~keesj@dellpc132.few.vu.nl) |
| 15:47.28 | *** join/#edev megha (~fire@unaffiliated/security) |
| 16:07.18 | *** join/#edev GPSFan (~kenm@64.92.145.112) |
| 16:07.19 | *** mode/#edev [+o GPSFan] by ChanServ |
| 16:24.33 | *** join/#edev sndcrb (~sndcrb@dsl-hkibrasgw2-50def8-126.dhcp.inet.fi) |
| 16:34.46 | *** join/#edev florian (~fuchs@Maemo/community/contributor/florian) |
| 16:53.41 | *** join/#edev NightMonkey (~NightrMon@pdpc/supporter/professional/nightmonkey) |
| 16:57.00 | *** join/#edev megha (~fire@unaffiliated/security) |
| 17:05.43 | *** join/#edev JoeLlama (~snork@unaffiliated/joellama) |
| 17:11.19 | *** join/#edev megha (~fire@unaffiliated/security) |
| 17:41.19 | *** join/#edev megha (~fire@unaffiliated/security) |
| 18:26.16 | *** join/#edev sndcrb (~sndcrb@dsl-hkibrasgw2-50def8-126.dhcp.inet.fi) |
| 18:51.50 | *** join/#edev megha (~fire@unaffiliated/security) |
| 19:03.42 | *** join/#edev NightMonkey (~NightrMon@pdpc/supporter/professional/nightmonkey) |
| 19:24.54 | iospace | yay segfaults -_-;; |
| 19:26.51 | *** join/#edev megha (~fire@unaffiliated/security) |
| 20:20.42 | *** join/#edev NightMonkey (~NightrMon@pdpc/supporter/professional/nightmonkey) |
| 20:29.05 | *** join/#edev aarias (~aarias@48.Red-80-59-144.staticIP.rima-tde.net) |
| 21:50.18 | *** join/#edev solar (~solar@smtp.gentoo.org) |
| 22:09.23 | *** join/#edev NightMonkey_ (~NightrMon@pdpc/supporter/professional/nightmonkey) |