| rjo_ | after the decade the arduino epidemic, the decade of "fpgauinos" has apparently dawned: http://www.xess.com/shop/product/xula2-lx25/ | 07:02 |
|---|---|---|
| rjo_ | s/the decade the/the decade of the/ | 07:02 |
| rjo | lekernel: thx for committing. i forgot to also add ngdbuild_opt to the non-edif case in _run_ise. its a one-worder. do you want a patch? | 08:41 |
| ysionneau | wpwrak: dunno :( 3 persons in CC were interested in the first place | 08:43 |
| ysionneau | if nobody is interested, let's say it's more free time for me | 08:43 |
| ysionneau | after all I have my M1, I don't care that much | 08:44 |
| ysionneau | now I'm interested if someone finds who the hell is desactivating DTLB (bit 0x40 in PSW) in my last email :D | 08:45 |
| ysionneau | I don't even write to PSW in the dtlb miss handler ... | 08:46 |
| ysionneau | all the other handlers are looping to themselves... and should freeze the M1 | 08:46 |
| ysionneau | maybe a strange interaction between IE and PSW (because they are somehow related because of PSW[2:0]) | 08:46 |
| ysionneau | I don't get it | 08:46 |
| lekernel | rjo, yw. yes, please | 08:47 |
| lekernel | the mixxeo boards just arrived here btw | 08:47 |
| ysionneau | \o/ | 08:48 |
| larsc | so they'll hopefully make it right on time for christmas | 08:50 |
| lekernel | should be fine, Germany domestic parcels aren't that slow | 08:51 |
| wpwrak | rjo: those kids grow up so fast ;-) | 10:01 |
| lekernel | ysionneau, "interest" dies quickly when you start asking for things like money :) | 13:02 |
| ysionneau | yes I think the price is not very attractive :( | 13:07 |
| lekernel | the hacker/maker market is full of low ballers who think hacking cheap Chinese gadgets is the best thing in the world | 13:09 |
| lekernel | I'm out of it now | 13:09 |
| ysionneau | I'm a little bit depressed about the fact that as soon as you want to make a real PCB with a few layers, a few connectors a some interesting components: then the price skyrockets | 13:10 |
| ysionneau | like if you don't want to spend 700 EUR, you're stuck with arduino like projects | 13:11 |
| ysionneau | it sucks | 13:11 |
| ysionneau | with 150 EUR you can buy nowadays pretty crazy electronic devices, or consumer stuff | 13:12 |
| ysionneau | but when you want to make something ... | 13:12 |
| ysionneau | you get nothing for 150 eur :( | 13:12 |
| lekernel | on the other hand, if you can deal with such PCBs, you can probably get a job that makes 700E very affordable | 13:12 |
| ysionneau | yes indeed | 13:13 |
| ysionneau | but it seems to exceed the price that even people who can afford it tend to spend for "toys" | 13:14 |
| ysionneau | I mean, for gadgets or hacking projets etc | 13:14 |
| ysionneau | I hope we'll find some manufacturing techniques which enable the price to go down sometime | 13:16 |
| lekernel | do those people do anything really interesting with those "toys"? | 13:16 |
| lekernel | have you seen an impressive raspberry pi project lately? | 13:16 |
| ysionneau | I've never searched fo raspberry stuff but I bet that I can find a few interesting projects if I browse a little bit | 13:17 |
| wpwrak | (2048 cycles stall) ouch ! :) | 13:17 |
| ysionneau | well, for instance the robotic club at my school used one to control their robot | 13:17 |
| ysionneau | wpwrak: yeah read-compare-write, for 1024 entries | 13:18 |
| ysionneau | they used arduinos to count ticks of "coding wheels" | 13:18 |
| ysionneau | before raspberry existed they used an eeepc stuck to the robot to control it | 13:19 |
| ysionneau | now they use a raspberry | 13:19 |
| lekernel | well, the rpi/arduino is hardly an essential component of that robot, right? you could replace it with a pc | 13:19 |
| lekernel | yes :) | 13:19 |
| ysionneau | the rpi is just the control stuf and contains the algorithm (path finding, avoiding collisions etc) | 13:20 |
| ysionneau | arduino counts ticks , report to rpi, and forward the rpi orders like "make the wheel go faster" | 13:20 |
| wpwrak | (700 EUR) well, milkymist is a complex device. you can make simple things cheaply, e.g., atben/atusb weren't all that expensive to make. to avoid the smell of arduino, atusb could nowadays also use some little ARM instead of an AVR. it would actually get cheaper :) | 13:21 |
| ysionneau | look at the neo900, same issue | 13:23 |
| ysionneau | it's very expensive | 13:23 |
| ysionneau | but indeed it's complex, and lot of components | 13:23 |
| wpwrak | yes. smartphones are about the worst you can have | 13:25 |
| wpwrak | plus, people have very strongly biased price expectations | 13:25 |
| ysionneau | hum looking at the M1 RC3 Bom it's $170 with PCB and smt | 13:25 |
| ysionneau | so what makes it so expensive? | 13:25 |
| wpwrak | at least for milkymist, there's nothing you could compare it with directly | 13:25 |
| wpwrak | PCBA price is USD 170 ? then you should be able to see it for ... about USD 500. the usual formula to apply is x 3. to cover yield, any taxes on your side, logistics, returns/conflict handling, and all that | 13:27 |
| wpwrak | you could go a little lower if you do everything direct and avoid expensive things like certifications | 13:28 |
| ysionneau | well I mean, not the price you would sell it if you have a company to run | 13:29 |
| ysionneau | lawyers, secretary, ingineers etc | 13:29 |
| ysionneau | engineers* | 13:29 |
| wpwrak | but then, you'd also want to make some money on it. have a basis for future projects and also have a reserve for when you hit an unexpected issue. e.g., some patent troll attack | 13:29 |
| ysionneau | I mean the price I would need to pay if I call a chinese factory and ask for 1 unit | 13:29 |
| ysionneau | if I want to make one for myself for instance | 13:29 |
| wpwrak | are you sure it's so low ? production run 1 -> USD 170 ? | 13:30 |
| ysionneau | I don't know : http://en.qi-hardware.com/wiki/Milkymist_One_RC3_BOM | 13:30 |
| wpwrak | already setting up the PCB-making process should be > USD 500 | 13:30 |
| ysionneau | it says $170 at the bottom of the page | 13:30 |
| wpwrak | that's ~USD 2000 for 95 PCBs | 13:31 |
| wpwrak | and does not include the cost of SMT | 13:31 |
| wpwrak | that's basically 100 bags of components | 13:32 |
| wpwrak | also the component number should be ~100. so that's considerably cheaper than buying in single-unit quantities. | 13:32 |
| ysionneau | ah I did not see the cell if empty for the "SMT" line | 13:32 |
| ysionneau | so to avoid all this expense | 13:34 |
| wpwrak | e.g., the MIC2550AYTS, to pick a random example, costs at digi-key only 70% per unit at 100 pieces than at 1 | 13:34 |
| ysionneau | I would need to do the PCB myself, so forget about the 1337 layers | 13:34 |
| ysionneau | and do something like top-bottom | 13:34 |
| ysionneau | forget about BGA | 13:34 |
| ysionneau | and only use QFP chips | 13:34 |
| ysionneau | which usually are the smaller ones :( | 13:34 |
| wpwrak | so for a single unit you get ... about USD 150/0.7 = USD 215, plus USD 500 PCB making setup, plus USD 15 for the one PCB | 13:35 |
| wpwrak | so, about USD 730. add shipping and you're around USD 810-850 | 13:35 |
| ysionneau | :/ | 13:35 |
| ysionneau | how realistic would it be to make a clone of the M1 with just top-bottom pcb? | 13:36 |
| wpwrak | and that's still just the bag of components. SMT will have to be done manually. | 13:36 |
| ysionneau | and then manufacture it yourself | 13:36 |
| wpwrak | dunno if you can handle the FPGA with two-layer | 13:36 |
| wpwrak | but if you pick a QFP, then yes, it may be feasible | 13:37 |
| ysionneau | ah yes, QFP | 13:37 |
| ysionneau | but then no LX45 | 13:37 |
| ysionneau | oly LX4 LX9 | 13:38 |
| ysionneau | only* | 13:38 |
| wpwrak | quite a piece of work, though. since there still are LOTS of things on the board | 13:38 |
| wpwrak | yup. and you have parallel NOR (get rid of it, use SPI), SDRAM, video in/out, etc. | 13:38 |
| wpwrak | for my projects, i usually get at least 10 components. if the component is cheap and reasonably generic, a lot more. | 13:40 |
| ysionneau | ok | 13:41 |
| wpwrak | that way, i can make several boards, can also make partial boards for testing some subsystem (e.g., the board i made for anelok that has wheel and OLED but no MCU) | 13:42 |
| wpwrak | and i may also have parts for later derivative projects. you never want to run out of components :) | 13:42 |
| ysionneau | yes sourcing can become difficult for instance | 13:42 |
| wpwrak | and you never want to be soldering some nasty chip, knowing that it's the last one you have :) | 13:42 |
| ysionneau | ahah | 13:42 |
| ysionneau | it's even more true when you are a soldering beginner | 13:43 |
| ysionneau | I mean, a 150 EUR M1 would be very sexy | 13:43 |
| ysionneau | even without say dmx/midi video in | 13:44 |
| ysionneau | M1 or Mixxeo, whatever | 13:44 |
| ysionneau | a lot more people would be willing to spend 150 EUR | 13:44 |
| ysionneau | it's even more true when they don't have yet a very precise idea about what they will do with it | 13:45 |
| ysionneau | if it's very expensive and you don't know yet what you will do with it: you don't buy | 13:45 |
| ysionneau | if it's not so much expensive: you buy, and you think you will figure out later | 13:45 |
| ysionneau | but I think I'm dreaming, such kind of pcb/components for 150 EUR will never happen I guess | 13:46 |
| lekernel | since there are apparently no process-dependent hardware bug, I'm probably not going to need all 6 Mixxeo boards I took for me | 13:47 |
| lekernel | I can ship a couple immediately... | 13:47 |
| wpwrak | ysionneau: there's your opportunity ;-) | 13:48 |
| ysionneau | wpwrak: what do you mean? | 13:49 |
| wpwrak | ysionneau: what lekernel just wrote :) | 13:50 |
| wpwrak | ysionneau: your path to an "inexpensive" mixxeo. well, you may have ordered one already. | 13:51 |
| ysionneau | I didn't order one | 13:52 |
| ysionneau | because honestly I don't think I have any use for it | 13:52 |
| wpwrak | ysionneau: (cheap process) you need to go over BOM and process with a fine comb and throw out everything that gives you trouble. e.g., M1 would be easier without the fancy connectors. | 13:52 |
| ysionneau | I already have a very shiny M1 board which can do so much things | 13:52 |
| ysionneau | if it was cheaper I would have order one, just to support the initiative | 13:52 |
| ysionneau | but it's really not cheap | 13:52 |
| wpwrak | then you have the wide busses for analog video, especially video out | 13:54 |
| wpwrak | also ethernet isn't exactly a small bus | 13:54 |
| wpwrak | granted, a bus is most of the time just wide. but still. | 13:54 |
| wpwrak | (not cheap) ah, how much is it ? | 13:55 |
| ysionneau | well it's something like 700 EUR, isn't it? | 13:56 |
| wpwrak | doesn't sound too bad to me for such a small run | 13:57 |
| ysionneau | well not too bad I guess | 13:58 |
| ysionneau | but for a specialized device I won't know what to do with, it's expensive :p | 13:59 |
| ysionneau | I mean I already have the M1 for doing FPGA/embedded-soft-dev playing | 13:59 |
| ysionneau | I'm not so much into VJing and mixing | 13:59 |
| ysionneau | my main interest is learning embedded stuff | 14:00 |
| ysionneau | at pcb-pool I can make a 100x100 pcb , 6 layers, for 134 EUR, at 1 unit | 14:03 |
| ysionneau | altera Cyclone 3 EP3C40 comes with PQFP package | 14:04 |
| ysionneau | aouch it's more expensive than spartan 6 LX45 | 14:05 |
| ysionneau | USD 80 for the EP3C40 | 14:06 |
| ysionneau | anyway, seems too hard a goal to make a cheaper M1 | 14:07 |
| ysionneau | and I don't have the time for it | 14:07 |
| lekernel | shipped! :) | 15:27 |
| larsc | thanks | 15:27 |
| ysionneau | gn8 | 21:05 |
| --- Sat Dec 7 2013 | 00:00 | |
Generated by irclog2html.py 2.9.2 by Marius Gedminas - find it at mg.pov.lt!