00:38.35 | *** join/#neo900 infobot (ibot@rikers.org) |
00:38.35 | *** topic/#neo900 is http://neo900.org | conversations are logged to http://infobot.rikers.org/%23neo900/ | 2013-11-04 - the day our fundraiser reached its goal, 25k EUR | 2013-12-02 - 200 devices reached! | 12-14 50035EUR, 232 units | 01-17 60kEUR, 300 units | 02-28 333 units, 70k⬠| 03-28 350 units, 400 donations, 73555⬠| 0501 360 410 75k | 0712 183 ~30k | 0810 300 ~49k | 0914 346 ~56k |
00:38.35 | *** mode/#neo900 [+v infobot] by ChanServ |
00:38.40 | DocScrutinizer05 | ~wb |
00:38.40 | infobot | It's great to be back! |
01:22.33 | wpwrak | hmm. looking at the RF side of GPS, some doubts becreep me. what we have is the modem powering the LNA (VGNSS) and the CPU watching that signal and detecting any wrongfulness |
01:23.31 | wpwrak | with the addition of the GPS kill switch, that one gets controlled by the same net as well |
01:24.43 | wpwrak | so gps kill is under the control of the modem, though the CPU can see if anything untoward happens there |
01:26.52 | wpwrak | we could give the CPU more authority, though: since we use VGNSS for control only, not for power, we could add a series resistor on the modem side, allowing the CPU to command an a shutdown the modem can't veto |
01:27.38 | DocScrutinizer05 | that's a interim design |
01:27.56 | DocScrutinizer05 | the final design is meant to be exactly what you sketched |
01:28.39 | DocScrutinizer05 | or even use a gate instead of a resistor |
01:28.41 | wpwrak | an even more paranoid choice would be to have two lines: one on the kill/lna side (to command a shutdown) and the other on the modem side (to see if the modem attempts to do anything, even if we've already prevented it from succeeding) |
01:28.53 | DocScrutinizer05 | yep, exactly |
01:29.10 | wpwrak | so, 2 signals ? |
01:29.23 | DocScrutinizer05 | prolly, when we're not short on GPIO |
01:29.44 | DocScrutinizer05 | (btw same applies to SIM-VDD) |
01:30.04 | DocScrutinizer05 | and there even RFID goes in between as well |
01:30.30 | wpwrak | RFID = ? |
01:31.01 | DocScrutinizer05 | NFC, http://www.ic-on-line.cn/download.php?id=1698663&pdfid=95F70B32A8CE7F4EDBE287FEA56A3511&file=0350%5Cpn544_2767746.pdf |
01:31.16 | wpwrak | SIM-VDD: switch _and_ sense ? or just sense ? |
01:31.43 | wpwrak | but we dom't have NFC ? |
01:33.05 | DocScrutinizer05 | 9.3Supply of a secure element |
01:33.07 | DocScrutinizer05 | PN544 provides integrated PMU to ensure supply voltage for both secure elements |
01:33.08 | DocScrutinizer05 | connected via SWP or NFC WI interface. |
01:33.10 | DocScrutinizer05 | 9.3.1Supply of SIM with SWP interface |
01:33.29 | DocScrutinizer05 | we're supposed to have NFC, if any possible |
01:33.38 | wpwrak | oh dear ... |
01:33.42 | DocScrutinizer05 | yeah |
01:34.20 | wpwrak | what other surprises do you have in store ? it seems that every time we approach a point where the design seems to settle, you pull a new rabbit out of your hat :-( |
01:34.42 | DocScrutinizer05 | we might get problems with the antenna, so I like to think of antenna as a user-retrofit-option |
01:34.58 | wpwrak | project sisyphus ... |
01:35.15 | DocScrutinizer05 | sorry |
01:37.03 | DocScrutinizer05 | http://projects.goldelico.com/p/neo900/issues/546/ http://projects.goldelico.com/p/neo900/issues/539/ |
01:39.50 | wpwrak | so the PN532 goes into the design as well ? |
01:41.54 | wpwrak | kewl, has an 8051 :) |
01:42.33 | DocScrutinizer05 | hehe, good you love it |
01:43.22 | DocScrutinizer05 | please see the 539 ticket. It's still unclear what Nik supports to go in there |
01:43.44 | wpwrak | is that v2 item ? |
01:44.05 | DocScrutinizer05 | Harald recommended PN532 iirc (been almost a year since I chatted with him about which chip to get) |
01:44.43 | wpwrak | "chip cost (1Q and 1kQ)" as if anyone ever bought chips at 1Q ;-) well, maybe some house-prices FPGA |
01:44.57 | DocScrutinizer05 | (v2 item) not mandatorily, since we have no CPU and no final antenna design, and even hackerbus isn't specified yet |
01:46.10 | wpwrak | as i understand nik, he wants to stabilize the component placement on LOWER. so if NFC is a LOWER item, we should at least mention it |
01:46.29 | DocScrutinizer05 | I left it out from discussion so far since it's not THAT importaant in V2 yet, and we maybe first want to see how stuff looks like when we got all the mandatory subsystems, before we try to cram in stuff that's unclear if it even works |
01:46.43 | DocScrutinizer05 | good point, indeed |
01:47.34 | DocScrutinizer05 | assuming we use my suggestion how to impleent uSD-daugtherboard and hackerbus, it would likely go to UPPEr though |
01:48.59 | DocScrutinizer05 | and from UPPER straight to the two antenna pads next to uSD on daughterboard, via a B2B conn passing through a hole in LOWER |
01:49.40 | wpwrak | B2B RF ? urgh |
01:49.55 | DocScrutinizer05 | meh, that's not "RF" |
01:50.04 | DocScrutinizer05 | that's a cheesy 14MHz |
01:50.09 | wpwrak | ;-) |
01:50.38 | wpwrak | so you'd just route it unshieldedly along the digital signals ? |
01:51.38 | DocScrutinizer05 | of course not |
01:51.54 | wpwrak | else, you need to match a regular b2b connector with some shielded critter, for which i wish thee luck ;-) (remember the usb board disaster of milkymist ? well, i think they've been clumsy there, but still ...) |
01:52.15 | DocScrutinizer05 | a) it's balanced iirc, b) you of course need a GND between it and the digital signals |
01:52.35 | wpwrak | okay, but the same unshielded connector ? |
01:52.56 | DocScrutinizer05 | completely unspecified so far |
01:53.21 | DocScrutinizer05 | at end of this connector, it diectly enters the antenna anyway |
01:53.51 | DocScrutinizer05 | so a 5mm more of "antenna" won't hurt that much, eh? |
01:54.04 | wpwrak | you said the antenna may be a customer option anyway, so perhaps we could just have a connector at UPPER/LOWER and leave it at that ? |
01:54.28 | DocScrutinizer05 | please elaborate |
01:54.45 | wpwrak | "we might get problems with the antenna, so I like to think of antenna as a user-retrofit-option" |
01:55.08 | DocScrutinizer05 | I know what I wrote |
01:55.19 | wpwrak | so why not let the user connect to where the signal originates ? |
01:55.46 | DocScrutinizer05 | because that would be somewhere on the PCB wher nobody can reach it |
01:56.28 | wpwrak | alright, at the edge of the pcb where the signal originates ? |
01:56.42 | DocScrutinizer05 | how's that any difference? |
01:57.13 | DocScrutinizer05 | the edge of the PCB is civered by the case (or the spacer frame, for UPPER) |
01:57.19 | DocScrutinizer05 | coverd |
01:58.00 | DocScrutinizer05 | alternatively I considered having it on the N900 battery bay testpoint area |
01:58.46 | DocScrutinizer05 | but that would make it pretty hard to come up with a feasible and somethat decent solution how to connect an antenna there |
01:58.59 | wpwrak | is anything antenna-sized that attaches to them safe when the board is removed from the bottom shell ? |
01:59.21 | DocScrutinizer05 | sorry, please rephrase |
01:59.45 | wpwrak | i mean: if the antenna attaches there, would it be at risk of breaking off when disassembling the device ? |
01:59.56 | DocScrutinizer05 | where? |
02:00.03 | wpwrak | in the battery bay |
02:00.18 | DocScrutinizer05 | it would be at risk of breaking when somebody changes the battery |
02:00.42 | wpwrak | that doesn't sound nice |
02:01.13 | wpwrak | that is, unless we call it the iNeo900. then it's cool ;-) |
02:01.56 | DocScrutinizer05 | I think http://projects.goldelico.com/p/neo900/issues/539/#ic1561 is to the point |
02:01.58 | wpwrak | connectivity to the CPU would be I2C, i assume ? |
02:03.41 | wpwrak | (ant on lid) yeah, if there's enough vertical space that sounds reasonable. kinda like nokia connect to the antennas, too. just that they have [access to] simpler contacts. |
02:03.45 | DocScrutinizer05 | I2C if no conflicts, which I'd not expect. Otherwise we could use SPI or UART |
02:04.35 | wpwrak | "which I'd not expect" = you expect conflicts ? |
02:04.57 | DocScrutinizer05 | no, I expect not to see any conflicts |
02:05.35 | wpwrak | good. neither do i. with to may buses to choose from :) |
02:08.54 | DocScrutinizer05 | another reason why I didn't mention NFC yet: we *might* provide the needed signals on hackerbus and place the PN544 on the flexPCB together with the antenna |
02:08.54 | wpwrak | any other major subsystem that's not in the block diagram ? |
02:09.30 | DocScrutinizer05 | right now I'm not aware of any that we would need in V2 |
02:09.55 | wpwrak | and for v3 ? |
02:12.26 | DocScrutinizer05 | well, CPU, RAM, eMMC... |
02:12.43 | wpwrak | that's already in the block diagram :) |
02:13.04 | DocScrutinizer05 | I haven't checked the blocj diagram versus V3 |
02:14.17 | DocScrutinizer05 | I'm not going to make a legally binding statement here and now. It's sunday morning 4AM |
02:15.12 | wpwrak | sure, take your time. but i think we better make sure the major things that shall be present in the end are on the table (= in the block diagram) at v2 |
02:17.16 | DocScrutinizer05 | check N900 schematics, GTA02A schematics and TRM, http://projects.goldelico.com/p/neo900/issues/ and neo900 feasibility study and particularly http://neo900.org/specs |
02:18.49 | wpwrak | i did all that already. but for example i thought nfc had been dropped. |
02:19.42 | wpwrak | and there are also things i never saw coming, like the complex IR subsystem. so it's really you who has to tell me what's missing |
02:21.42 | DocScrutinizer05 | complex IR? |
02:22.18 | DocScrutinizer05 | it seems to me we have a total BOM of less than 2 dozen components for IR |
02:22.37 | wpwrak | btw, the specs have an "Ambient light" sensor. is that also missing ? |
02:23.12 | DocScrutinizer05 | err, we had it at one time |
02:23.40 | DocScrutinizer05 | on your block diagram it's definitely not missing since it's in display half |
02:23.50 | DocScrutinizer05 | good night! |
02:24.14 | wpwrak | aah, an N900 component. nice ;) |
02:58.46 | wpwrak | hmm. newark.com are down ... but they do have a pretty error page |
03:08.20 | DocScrutinizer05 | sth is buggy today |
03:09.54 | DocScrutinizer05 | dos1's internet down, freenode IRC got hacked, freenode webserver been down, now newark (whatever that is) |
03:10.23 | DocScrutinizer05 | and funny enough just the warning LED of my home automation lits up |
03:10.26 | wpwrak | newark.com is the US side of farnell (element14) |
03:10.40 | *** join/#neo900 Kabouik__ (~quassel@252.133.136.88.rev.sfr.net) |
03:10.40 | *** join/#neo900 Kabouik_ (~quassel@252.133.136.88.rev.sfr.net) |
03:10.57 | wpwrak | "warning: general, unspecified threat of unknown magnitude" :) |
03:11.34 | DocScrutinizer05 | http://wstaw.org/m/2014/09/14/plasma-desktopkq1987.png |
03:12.02 | DocScrutinizer05 | o/ |
03:12.55 | DocScrutinizer05 | btw of course Conrad had no decent photodiodes |
03:14.34 | wpwrak | mail-order is your friend ;) and that way, you can get the real part already, without the need to find something sufficiently similar |
03:16.04 | DocScrutinizer05 | no use for "the real part", it would look too weird when I solder wires to it |
03:16.54 | wpwrak | doesn't strike me as unusual :) |
03:37.10 | DocScrutinizer05 | btw which shop is you're using to order 3 LEDs and 3 photdiodes, and 5 opamps? |
03:37.24 | DocScrutinizer05 | has it free shipping? ;-) |
03:40.45 | wpwrak | hmm, i'd search locally. and probably just go there :) |
03:41.17 | wpwrak | else, digi-key. most convenient, most reliable. |
03:43.34 | wpwrak | that is, with credit card. you'd want to find one that accepts paypal or that's not too messy for wiring money |
03:45.49 | DocScrutinizer05 | opamp: 28ct. photodiode: 69ct, LED: 52ct |
03:47.03 | DocScrutinizer05 | digikey and farnell bot minimum order 30EUR iirc |
03:49.03 | wpwrak | maybe there are some other parts / things you'll need ? |
04:04.34 | *** join/#neo900 b1101 (~b@fsf/member/b1101) |
07:21.48 | DocScrutinizer05 | err nope |
07:23.29 | DocScrutinizer05 | funy detail: somehow the red LED I used "just for fun" works better as light detector than the photodiode I got (the only one they had) |
07:33.21 | kerio | <infobot/#maemo> [02:38:27] DocScrutinizer: infobot joined! |
07:33.24 | kerio | ;_; |
07:43.28 | DocScrutinizer05 | sorry? |
07:45.19 | DocScrutinizer05 | that's hardly a recent quote. I'm not a user of that channel anymore |
07:45.44 | DocScrutinizer05 | and the chanserv command actually been an autoresponder I've set up |
07:46.45 | DocScrutinizer05 | oooh, funny, the autoresponder been mine, it's not channel specific ;-P |
07:50.21 | DocScrutinizer05 | http://wstaw.org/m/2014/09/14/plasma-desktopnw1987.png I already wondered how the heck you managed it to post three lines without my irc client prepending them with your name |
08:05.21 | wpwrak | possible IR diodes placement: http://neo900.org/stuff/werner/tmp/ir-box.pdf |
09:02.52 | DocScrutinizer05 | wpwrak: looks good |
09:04.38 | *** join/#neo900 paulk-aldrin (~paulk@armstrong.paulk.fr) |
09:04.58 | DocScrutinizer05 | actually we can allow RX diode to go to an even more convenient place if needed. Sensitivity is high enough for the intended usecase, even when the diode would get mounted completely behind the TX diode. And while I'm at it: my experiments as reported above made me wonder if we actually *need* an RX diode ;-D |
09:05.33 | DocScrutinizer05 | if we would allow half-duplex only, we could abuse the TX diode ;-D |
09:07.40 | DocScrutinizer05 | additional funny detail: the red LED was next to completely insensitive to IR, while the photdiode received a signal sufficiently strong to show flicker on a yellow LED after a darlington, with a distance between CIR remote control and diode of 50cm |
09:08.17 | DocScrutinizer05 | for white (torch) LED light the effects been exactly the inverse |
09:08.56 | DocScrutinizer05 | well, not exactly, the photodiode worked better with white than the LED with IR |
09:09.33 | DocScrutinizer05 | anyway clearly the LED with white beats all other pairs hands down |
09:10.09 | enyc | wonders if neo900's GPS is likely to be lower power =) |
09:10.28 | enyc | I observed GPS eating N900 battery like mad or something |
09:10.44 | kerio | how's a LED actually acting as a sensor? |
09:11.23 | enyc | kerio: seemingly that is sonraml they have this bidirectional properyty, a photodiode is like an LED but simple tued for the purpose, or something like that ... |
09:12.20 | DocScrutinizer05 | I've seen pictures of 2N3055 decanned and emitting red light ;-D |
09:13.17 | DocScrutinizer05 | 2N3055 is a TO3 transistor and not meant to emit any light :-D |
09:14.11 | enyc | So.... |
09:14.12 | DocScrutinizer05 | enyc: I'm quite sure I've seen power rating about GPS as well, in PHS8 TRM/datasheet |
09:14.32 | enyc | DocScrutinizer05: oh? awht waht did you learn? |
09:24.40 | DocScrutinizer05 | http://wstaw.org/m/2014/09/14/plasma-desktopVI1987.png http://wstaw.org/m/2014/09/14/plasma-desktopbF1987.png |
09:24.58 | DocScrutinizer05 | so, an additional ~30mA for GPS |
09:25.41 | DocScrutinizer05 | on top of the basic 46mA for WCDMA (or GSM) operation and active USB |
09:26.40 | DocScrutinizer05 | even the total of 75mA is still good for a 15h continuous GPS usage, if you don't need other power hogs, like LCD backlight |
09:27.01 | enyc | aah i see and this is the chip used for all wireless et.c in the Neo900 |
09:27.14 | DocScrutinizer05 | no, this is the modem |
09:27.34 | enyc | i mmean iwiress lmoed,m , sorry being confusing |
09:27.48 | enyc | WLAN etc. is indeed separate! |
09:27.53 | enyc | lme looking over diagram now |
09:27.59 | DocScrutinizer05 | "all wireless" would be wl1273 for wlan, bt, fm |
09:28.48 | enyc | looking at diagram the PHS8/PLS8 is either USB connected, or USB provider |
09:29.56 | DocScrutinizer05 | the modem is peripheral, CPU/SoC is host |
09:30.06 | enyc | kk |
09:30.21 | enyc | and USB is 'safe' delineated interface from the pov of not being able to access host memory directly |
09:30.32 | DocScrutinizer05 | exactly |
09:30.36 | enyc | but not against becoming a 'rouge' 'pretentd to be something else' usb device |
09:30.53 | DocScrutinizer05 | err, funny idea |
09:31.04 | enyc | well the whole 'badusb' thing ... |
09:31.14 | enyc | been made a fuss about lately ;p |
09:31.19 | DocScrutinizer05 | anyway nope, the USB host is exclusively locked to the modem handler |
09:31.27 | enyc | right =) |
09:32.26 | DocScrutinizer05 | at least we will recommend that, it's a OS sw thing how you implement that |
09:32.54 | enyc | ''trust'' wider issue very interesinng complex, no simple answers, ... |
09:33.06 | DocScrutinizer05 | you are free to try and mount a mass storage device on that USB host |
09:33.16 | enyc | =) |
09:33.18 | DocScrutinizer05 | ;-) |
09:33.39 | DocScrutinizer05 | but it's completely under your (resp OS') control |
09:33.53 | enyc | is there a similar diagram for how N900 connected up ? |
09:34.03 | DocScrutinizer05 | err, yep |
09:34.12 | DocScrutinizer05 | in service manual L3_4 iirc |
09:34.26 | DocScrutinizer05 | less detailled though |
09:35.11 | DocScrutinizer05 | I might be mistaken, didn't look into L3_4 for more than a year |
09:36.08 | enyc | coo I didn't know that existed =) |
09:36.34 | kerio | hold on, usb host? |
09:36.41 | kerio | ah, the internal one |
09:40.00 | DocScrutinizer05 | http://wstaw.org/m/2014/09/14/plasma-desktopga1987.png |
09:40.26 | enyc | I'm not sure I 'evyr got N900 USB host working well |
09:40.52 | DocScrutinizer05 | that took a while to spot that on page 165 |
09:40.52 | enyc | DocScrutinizer05: oh indeed i just got to that page was already looking at that diagram =) |
09:43.02 | enyc | wondels what SSI / LoSSI serial interface is / means |
09:43.19 | kerio | RIP n900 usb otg |
09:43.28 | DocScrutinizer05 | wpwrak: http://wstaw.org/m/2014/09/14/plasma-desktopga1987.png "production test pads <--> UART" ;-) |
09:43.34 | kerio | we miss you so much :c |
09:44.12 | DocScrutinizer05 | SSI is slow HSI |
09:44.44 | enyc | Hrrm that document has a Glossary that defines neither SSI nor HSI =) |
09:47.47 | DocScrutinizer05 | ~wiki mipi-hsi |
09:49.30 | DocScrutinizer05 | http://mipi.org/specifications/high-speed-synchronous-serial-interface-hsi |
09:50.08 | enyc | interseting this serial link was supposedly 'just' for that purpose modem -- application |
09:50.13 | enyc | 'but can now be used for more stuff |
09:50.21 | enyc | I guess thats' how things get built, by necessity |
09:50.38 | DocScrutinizer05 | http://www.omappedia.com/wiki/HSI |
10:08.44 | *** join/#neo900 che1 (~che@g238.tum.vpn.lrz.de) |
10:10.36 | *** join/#neo900 kolp (~quassel@55d4593f.access.ecotel.net) |
11:04.16 | enyc | nodsnods |
11:31.19 | *** join/#neo900 modem (~modem@fsf/member/modem) |
13:11.36 | wpwrak | (uart to tp) it's out there ... somewhere ;-) |
13:12.33 | wpwrak | (ir placement) the box in the picture is the plastic structure in the n900 case. so there aren't all that many placement options. that is, without cutting into the box |
13:13.27 | wpwrak | btw, clearances are in the order of 0.3 mm. so yes, if you can find a way to make do with just a single diode, that would be a major improvement :) |
13:46.15 | enyc | thinks ... plastic structures |
13:46.30 | enyc | wpwrak: could you find osme integrated I/R tx/rx blob ? |
13:51.53 | wpwrak | all the integrated ones i found were quite huge. with a nice distance between TX and RX, but unfortunately far too big even for the IR window in N900, let alone the box |
13:52.22 | wpwrak | (back then i didn't know about the box so i only worked with the window size. the box adds more constraints.) |
13:52.45 | enyc | hum |
13:52.49 | wpwrak | so going from the original single diode to two diodes will be a very tight fit. |
13:52.50 | enyc | silly question but how does the N900 do it? |
13:53.14 | wpwrak | easy: it has one diode, not two :) |
13:53.23 | enyc | it managed half-duplex ? |
13:53.45 | enyc | I've never tried IR on n900 infact wasn't really waare it even existed =) |
13:54.14 | wpwrak | no, N900 is TX only. for CIR. no IrDA or such. |
13:54.49 | wpwrak | yeah, i suspect that the fancy IR capabilities of neo900 will go greatly underappreciated in the field |
13:55.01 | kerio | to be completely fair, TX is probably what you want |
13:55.14 | kerio | although it's nice to be able to read a remote instead of having to select it from a list |
14:03.51 | enyc | is the tight-fit IR separate diodes doable?? |
14:04.15 | enyc | I mean we know this was all going to be a squeeze anyway.... |
14:10.10 | kerio | how's battery hotswapping supposed to happen? |
14:15.08 | *** join/#neo900 M13 (~MirandaLS@2a02:17d0:d5:2300:d59a:12d4:8480:e2e7) |
14:26.45 | wpwrak | maybe we should pick a smaller diode. VEMD10940F looks friendlier. and has better availability. less sensitive and has a very wide angle, but all that's probably more feature than bug. of course, it'll doom full-duplex even more :) |
14:35.39 | wpwrak | so that would be the plan B: http://neo900.org/stuff/werner/tmp/ir-box2.pdf |
14:37.17 | wpwrak | minimum clearances increase to 0.5 mm and the height at the rear (may become important when rotating the pcb into place) goes down from 2.3 mm to 1.0 mm |
14:39.37 | wpwrak | as a small undeserved bonus, the part is even orderable at digi-key: http://www.digikey.com/product-detail/en/VEMD10940F/VEMD10940FCT-ND/3915201 |
14:41.25 | *** join/#neo900 modem (~modem@fsf/member/modem) |
14:43.17 | *** join/#neo900 Nokiabot (~Nokiabot@223.176.22.88) |
15:00.48 | Nokiabot | Vivrator is an awsome invention thats cratering our varied needs daily from pleasure to reinforcing concrete to notifications on communication devices to dancing 6310 so who knows what the heck a dozen pair of vivrators on neo900 could be used for.. :) |
15:03.38 | Nokiabot | Only limited by ones imagination if there are a couple of vivrators be available on neo900 to spare :p |
15:05.10 | jake42 | vivrator O.o |
15:05.41 | wpwrak | Nokiabot: here's some industrial design homework for you: enhance the stylus such that it a) is suitably inflatable, and b) has a built-in supercap-powered vibration device |
15:06.05 | jake42 | interessting how prices for VEMD10940F differs between digikey and mouser.. |
15:06.09 | Nokiabot | Jake42:you seem to get my point :) |
15:07.28 | Nokiabot | Wpwark:keep your nefarious intentions to yourself :p |
15:08.22 | jake42 | oh.. me beeing stupid.. mouser is â¬, digikey is $ |
15:09.39 | Nokiabot | Honestly what harm a couple of vivrators would do provided if there is enough space for one more atleast :) |
15:13.47 | Nokiabot | Wpwark:btw what is that supercap vivrator thingy :p |
15:15.17 | Nokiabot | ~Botsnack |
15:15.17 | infobot | :), Nokiabot |
15:16.50 | Nokiabot | Oo so this is highlighting was wondering on the changed fonts :p |
15:36.31 | Nokiabot | Wpwark:i see most chineese craps have torch with a hardware switch which works even when the phone is off a very useful function indeed so if possible feisable can same be done with flash leds using any of the switches like vol button maybe :) |
15:37.13 | *** join/#neo900 freemangordon (~freemango@46.249.74.23) |
15:37.25 | drathir | mornin... |
15:38.34 | Nokiabot | Night... |
15:39.38 | *** join/#neo900 Sicelo (Sicelo@gets.high.on.ircspeed.net) |
15:48.55 | Nokiabot | dinner over 18 chappatis in cant move now :p |
15:49.53 | *** join/#neo900 xe4l (~unknown@162.243.112.233) |
16:02.51 | wpwrak | ah, here :) |
16:03.17 | wpwrak | yes, i think the flash controller has a "torch" mode |
16:03.33 | wpwrak | (with reduced current, so the led doesn't burn up) |
16:04.22 | Nokiabot | Err read it again i stated something diffrent |
16:07.11 | Nokiabot | Wpwark:no you didnt get it |
16:13.58 | Nokiabot | Oo |
16:14.31 | Nokiabot | wpwrak:... |
16:28.34 | drathir | wpwrak: if good remember depend which kind of flash light used some of them dont like continious long flashing... |
16:50.14 | wpwrak | drathir: yes, LEDs have a certain maximum pulse current (for a very short time) and a lower maximum continuous current (for as long as you want) |
16:51.22 | wpwrak | continuous current is basically how fast the package can dissipate the heat generated by the junction |
16:52.12 | wpwrak | maximum pulse is the thermal capacity of the junction. at the end of the pulse, it would be at its maximum allowed temperature. then you have to turn off to let it cool down again |
16:52.14 | *** join/#neo900 xes (~xes@unaffiliated/xes) |
16:53.37 | wpwrak | we've actually mentioned this in http://neo900.org/stuff/papers/ir.pdf (page 10, last paragraph) |
16:53.48 | wpwrak | (mentioned) very very briefly, though |
17:04.50 | drathir | the project with all days amazing me more and more... |
17:05.29 | drathir | wpwrak: possitive way ofc... |
17:42.41 | wpwrak | heh, thanks ! :) |
17:44.27 | drathir | wpwrak: its np... |
18:05.49 | enyc | wpwrak: supposedly IrDA is halfduplex light anyway anyway |
18:07.44 | DocScrutinizer05 | yep |
18:08.02 | DocScrutinizer05 | exactly what I recall as well |
18:09.07 | DocScrutinizer05 | it's just about IR-UART which wouldn't understand ^S/^Q anymore |
18:09.18 | DocScrutinizer05 | ;-) |
18:10.12 | DocScrutinizer05 | but I don't see why duplex gets affected by placement or type of RX semiconductor |
18:10.37 | enyc | DocScrutinizer05: in as much as the transmitter overwhelms the local receiver |
18:10.41 | DocScrutinizer05 | sensitivity yes, duplex however...? |
18:10.59 | DocScrutinizer05 | no, we compensate for that |
18:11.15 | enyc | how?? |
18:11.37 | DocScrutinizer05 | and our S/N is pretty good for the planned usecase, no matter what the receiver placement |
18:12.02 | DocScrutinizer05 | how? simple feedback |
18:13.01 | DocScrutinizer05 | when I know my TX crosstalks with gain G to RX, then I simply have a direct electrical feedback with -G |
18:14.02 | DocScrutinizer05 | that's really EE 101 |
18:16.04 | DocScrutinizer05 | (actually the term "feedback" is slightly misleading, it's a forward compensation) |
20:14.23 | jake42 | DocScrutinizer05: hi, while I was composing a email to the list of our fablab, I realized I still have a hard time imagining how spacer frame and the two pcbs are supposed to fit together |
20:15.35 | jake42 | is the frame supposed to go on top or in between both pcbs? |
20:27.21 | wpwrak | jake42: there are two spacing issues: one part that provides a bit of sidewall and integrates the keyboard frame. this part is what we call the "spacer frame" |
20:28.53 | wpwrak | there other is for distributing pressure from the top board downward. this one is less clearly defined. DocScrutinizer05 wants to add metal spacers to the upper board. i think a plastic structure would be both easier to make and more versatile. |
20:29.49 | jake42 | so there will be a visible "external" frame and a invisible "internal" one |
20:30.49 | jake42 | part of the external one should be shaped like the original kbd-frame |
20:32.45 | wpwrak | yes, exactly |
20:33.28 | DocScrutinizer05 | no, I never said I want to use "metal spacers" |
20:33.45 | DocScrutinizer05 | you completely missed that |
20:34.28 | DocScrutinizer05 | I already elaborated there is no use in any plastic filling the (non-existent) empty space between UPPEr and LOWER |
20:36.00 | wpwrak | "metal force propagators" then ? |
20:36.01 | DocScrutinizer05 | what I said is to have rigid steel posts (like nails or contact posts) soldered to UPPER and reaching through holes in LOWER to finally sit on bottom of case |
20:36.18 | wpwrak | yes, that's what i mean. post = spacer |
20:36.55 | DocScrutinizer05 | jake42: see http://neo900.org/stuff/joerg/spacerframe/sketch.pdf |
20:36.57 | wpwrak | and i still think that this is more easily done with plastic. exact shape to be defined. |
20:37.07 | jake42 | so UPPER will part lie partially on modem? |
20:37.13 | DocScrutinizer05 | NO! |
20:37.25 | DocScrutinizer05 | that's a sure way to break stuff |
20:37.46 | jake42 | what is filling the empty space? |
20:37.56 | DocScrutinizer05 | nothing |
20:38.06 | wpwrak | i kinda wonder what sort of forces you expect on that keyboard ... |
20:38.07 | DocScrutinizer05 | a) there's not much of empty space anyway |
20:38.14 | jake42 | how is it non-existant? |
20:38.24 | DocScrutinizer05 | b) we need that to *avoid* any force to modem |
20:39.20 | DocScrutinizer05 | UPPER and LOWER are a PCB sandwich with component spread |
20:39.28 | jake42 | but there will be ~2mm headroom between both pcbs (expet where there are components) |
20:39.31 | jake42 | right? |
20:39.47 | DocScrutinizer05 | exactly, *expet where there are components* |
20:39.58 | jake42 | good |
20:40.17 | DocScrutinizer05 | there will be components virtually everywhere |
20:40.24 | jake42 | so the idea with those metal posts seems reasonable to me |
20:40.40 | DocScrutinizer05 | [2014-09-14 Sun 22:34:28] <DocScrutinizer05> I already elaborated there is no use in any plastic filling the (non-existent) empty space between UPPEr and LOWER |
20:41.28 | wpwrak | sigh. not filling, being there, to distribute the load. what would be the point of "filling' ? |
20:41.58 | DocScrutinizer05 | we must not "distribute the load" |
20:42.18 | jake42 | so if you type on the kbd one usually squeezes the button on top and the back of the device together |
20:42.19 | DocScrutinizer05 | LOWER must not get mechanically coupled to UPPER |
20:43.11 | DocScrutinizer05 | any "load" has to get supported from case bottom, not from LOWER |
20:43.15 | wpwrak | why would distributing the load imply coupling with LOWER ? you can still have all the holes and posts you want. |
20:43.25 | wpwrak | i think you're mixing up concept |
20:43.26 | wpwrak | s |
20:43.39 | wpwrak | a) take load from UPPER -> distribute it downward |
20:43.44 | DocScrutinizer05 | I'm pointing out that your concept is not exactly clear |
20:43.58 | wpwrak | b) see what works best down where |
20:45.40 | wpwrak | anyway, so hacklab will be in charge of ME ? |
20:46.02 | wpwrak | because then i won't look into making a spacer frame |
20:46.14 | jake42 | DocScrutinizer05: so you think it will be enough support for the kbg from the 4 posts |
20:46.24 | DocScrutinizer05 | any plastic "nails" will have to be much thicker in diameter than steel, and you cannot solder them but need some relatively huge structure on top side of UPPER to fix them there |
20:46.49 | DocScrutinizer05 | jake42: did I say this? |
20:47.12 | jake42 | wpwrak: don't know yet, have to inquire first. hacklab per se doesn't do anything. you have to do it yourself. but maybe someone is willing to do it |
20:47.22 | jake42 | DocScrutinizer05: sry, this was ment as a question |
20:47.26 | DocScrutinizer05 | I don't even know what are "THE 4 posts" |
20:48.25 | DocScrutinizer05 | [2014-09-14 Sun 22:36:01] <DocScrutinizer05> what I said is to have rigid steel posts [edit: several, like 6 or 8] (like nails or contact posts) soldered to UPPER and reaching through holes in LOWER to finally sit on bottom of case |
20:48.38 | jake42 | so sorry again. I imaginged that the posts you mentoned will be where the 4 screws go |
20:49.46 | jake42 | DocScrutinizer05: thanks, know the image in my head should be more syncronizied with your idea :-) |
20:50.04 | DocScrutinizer05 | :-D |
20:50.10 | jake42 | s/know/now |
20:50.42 | jake42 | to an extend nothing goes over discussing ideas |
20:50.49 | DocScrutinizer05 | most of them posts will end with their tip on inside of the battery bay steel |
20:53.33 | jake42 | will the 4 existing posts in the display half also be "made" longer? |
20:53.43 | DocScrutinizer05 | we anyway can place them relatively arbitrarily to wherever they don't conflict with modem or large chips (we don't want to drill holes into those components) |
20:53.48 | wpwrak | this limiting factor for the minimum thickness of such posts is likely the amount of lateral force they experience (in relation to their ability to distribute this to whatever they attach to), not compressive strength. so a high-strength material alone isn't enough - it also has to be prevented from titling |
20:54.00 | DocScrutinizer05 | jake42: those get spacer washers |
20:54.11 | jake42 | DocScrutinizer05: ok, good |
20:55.45 | DocScrutinizer05 | it seems that in N900 original design those bolts sit on top of PCB (which in N900 is LOWER) and thus fix it when you tighten the screws |
20:56.37 | DocScrutinizer05 | in N900 they will sit on top of UPPER then, pressing it down when tightening screws, against the washers and LOWER |
20:58.03 | DocScrutinizer05 | s/N900/Neo900/; the thickness of UPPER PCB plus washers will define how much higher the display is in Neo900 than in N900 |
20:58.41 | DocScrutinizer05 | this is 2mm for "empty space" (where components live) plus 0.8mm for PCB |
20:58.53 | DocScrutinizer05 | --> 2.8mm |
20:59.06 | DocScrutinizer05 | which our spacer frame needs to compensate for |
21:00.09 | jake42 | from your sketch I looks to me like the spacerframe will also add to the vertical thickness |
21:00.35 | DocScrutinizer05 | that's sloppy sketch then |
21:00.40 | DocScrutinizer05 | it shouldn't |
21:01.37 | DocScrutinizer05 | that sketch is just illustrative to explain the idea. It doesn't contain and design details that are useful to use them as specs |
21:01.44 | jake42 | as the posts where the screws go in will go through the sketched holes |
21:02.01 | DocScrutinizer05 | wpwrak: nope, fablab is not in charge to make a spacerframe yet |
21:02.43 | DocScrutinizer05 | however I thought you were not fond of the idea of doing it |
21:03.19 | DocScrutinizer05 | didn't we discuss 3D CAD and that we both don't really like to mess with that sort of application? |
21:04.18 | DocScrutinizer05 | if you feel like doing the 3D modeling, you're more than welcome |
21:04.24 | wpwrak | jake42: i'd say that the spacer frame largely defines the thickness (since it has to close "nicely"). but of course, the parameters for the desired addition come from elsewhere |
21:05.42 | wpwrak | DocScrutinizer05: yes, my capabilities here would be limited to a rough shape: machined from only one side (2.5 D), and with a rather crude CAD side |
21:05.55 | DocScrutinizer05 | no, the needed clearance for components plus the thickness of PCB defines the spacerframe |
21:06.40 | wpwrak | basically at the level of what i've done for anelok. but at the hacklab you should be able to find better facilities. |
21:06.46 | DocScrutinizer05 | we're talking 3D CAD modeling. Not CNC milling or 3D printing |
21:06.54 | jake42 | wpwrak: of course, I meant that sketch has holes in the frame which might even make sense for stability, though also might add to thickness |
21:07.26 | DocScrutinizer05 | jake42: yes, I actually made a mistake there maybe |
21:08.19 | DocScrutinizer05 | the holes need to be large enough to let the complete display bolts pass through, but probably should be precise enough to stabilize the spacerframe |
21:08.49 | wpwrak | (3d model) fped here. not what it's meant to be used for, but all the alternatives just seem to suck even more ;-) |
21:08.50 | jake42 | now we're talking :-) |
21:09.37 | jake42 | openSCAD looks fun ;-) |
21:10.05 | DocScrutinizer05 | I pondered whether we could make the spacerframe go in between the sandwitch to replace the washers in the "hole area", but that's pretty complicated |
21:10.15 | wpwrak | jake42: it looks nice but it never seems to be willing to do quite what i want ... |
21:11.14 | wpwrak | DocScrutinizer05: so you'd insert the upper PCB horizontally into the spacer frame ? |
21:11.33 | DocScrutinizer05 | that's why I say "that's complicated" |
21:11.41 | wpwrak | yeah :) |
21:12.14 | DocScrutinizer05 | the spacer frame concept isn't 100% finalized yet |
21:12.35 | wpwrak | 3D printer or a fancy mill or fancy tools. not impossible but considerably harder |
21:12.42 | DocScrutinizer05 | I'm hoping for the ME guys to know their business |
21:13.10 | jake42 | DocScrutinizer05: good to know |
21:13.16 | wpwrak | is there local expertise at the hacklab you can recruit ? |
21:13.33 | jake42 | it's at the university |
21:13.46 | wpwrak | aah ! :) |
21:13.46 | DocScrutinizer05 | *maybe* we actually could slide in UPPER and then plug LOWER to it, but that would basically result in a block of stuff that's impossible to mount to the case then |
21:13.49 | jake42 | run by students |
21:14.36 | DocScrutinizer05 | I think it's not feasible to plug an UPPEr to a already mounted LOWER |
21:14.52 | jake42 | if anyone wonders: https://fablab.fau.de/ |
21:15.24 | jake42 | (germany only, sry) |
21:15.39 | DocScrutinizer05 | so the most likely procedure to assemble is: plug UPPEr onto LOWER, mount the sandwich to case, then mount spacerframe to that |
21:16.26 | DocScrutinizer05 | which would rule out any idea of having spacerframe go between UPPER and LOWER and act as "washers" there, aiui |
21:20.56 | jake42 | hm.. I'm not yet satisfied with this discussion but some other time |
21:20.57 | DocScrutinizer05 | probably we should just use washers large enough in diameter so they will support the spacerframe against getting pushed in from side |
21:21.49 | jake42 | good night.. gotta catch some sleep before the craftsman show up in the morning |
21:21.51 | DocScrutinizer05 | note that UPPER needs some clearance to spacerframe, or any tiny impact will break the B2B connectors between UPPER and LOWER |
21:22.15 | jake42 | good point! |
21:22.16 | DocScrutinizer05 | jake42: good night! :-) |
21:22.39 | jake42 | cya |
21:23.19 | DocScrutinizer05 | only thing sturdy enough to take such lateral impact are the 6 screws |
21:24.26 | DocScrutinizer05 | thus have washers sitting between UPPEr and LOWER and fixed by such screws which are large enough in diameter to support the outer structure which will take the impact: spacerframe |
21:25.46 | DocScrutinizer05 | this should work for 3 of the corners, lower right corner is a tad off and thus needs some more care to design this detail right there |
21:27.40 | DocScrutinizer05 | I'd love to discuss all those little design details with the guy creating the spacer frame 3D-CAD model |
21:34.46 | DocScrutinizer05 | wpwrak: maybe those "washers" could actually be made of metal or plastic in a CNC, to optimally adapt to the sandwich shape and maybe even have tine bolts/noses that latch into 1mm holes in upper and lower to keep the "washer" thingies in place during assembly |
21:35.47 | DocScrutinizer05 | s/tine/tiny/. |
21:37.31 | wpwrak | careful: if it's upper AND lower, you need two-sided machining again. (or a good 3D printer) |
21:37.49 | DocScrutinizer05 | yeah, already thinking about that |
21:38.10 | DocScrutinizer05 | prolly "either or" is enough |
21:38.21 | wpwrak | one problem with 3D printers is that you don't have a lot of choices for the material. if you're lucky, what they use just happens to work for you. else, you lose. it's pretty much all or nothing. |
21:38.41 | DocScrutinizer05 | spacerframe also needs some way to fix it to case at upper end next to the buttons |
21:39.36 | wpwrak | (or) yes, should do. if you want to get fancy, you could add holes for screws to do the other side. through-drilling is quite doable. but then, this weakens the material and requires a tool change. so something to not consider lightly. |
21:39.43 | DocScrutinizer05 | not that it could move too much in Z-axis up since there's display which stops it, but we don't want a dangling flimsy plastic thingie there |
21:40.41 | wpwrak | you mean the thin "bridge" that in n900 has extensions that slider under cans and some vaguely Z-shaped metal structure ? |
21:41.02 | DocScrutinizer05 | (screws) thought about that, do somehow fix UPPEr and LOWER together, even without getting squeezed between case shell and display bolts |
21:41.43 | DocScrutinizer05 | no I mean the parallel structure to that, which is part of the outer case wall next to the buttons |
21:41.48 | DocScrutinizer05 | switches |
21:41.55 | DocScrutinizer05 | power and volume |
21:42.32 | wpwrak | ah yes, that one will be quite fragile, too |
21:43.13 | DocScrutinizer05 | idly wonders if we shouldn't use a 2-pats design consisting of a mere spacer frame and a separate (original) kbd frame, and the *glue* the spacerframe to case |
21:43.15 | wpwrak | well, so-so. it's almost 3 mm thick. that's a fair bit of material |
21:43.29 | DocScrutinizer05 | parts* |
21:43.51 | DocScrutinizer05 | and then* |
21:43.59 | wpwrak | that seems to be both more complex and less stable |
21:44.05 | DocScrutinizer05 | yes |
21:44.28 | DocScrutinizer05 | unless the glue is strong |
21:45.13 | wpwrak | that way, you probably also miss the opportunity to stabilize the spacer frame in the middle |
21:45.47 | wpwrak | or rather, "an opportunity" |
21:45.56 | wpwrak | in general, the simpler, the more stable :) |
21:46.08 | DocScrutinizer05 | the whole opportunities are in glue with such design |
21:46.21 | DocScrutinizer05 | but I also think it's not feasible |
21:50.19 | DocScrutinizer05 | maybe several of those "problems" could get mitigated if we use alu or even steel for the spacerframe |
21:51.08 | DocScrutinizer05 | RF design should not see any impact from that, since all antennas are on opposite side of the PCB |
21:52.40 | DocScrutinizer05 | I'd hope dural or steel would be less flimsy and allow for fewer fixing points e.g along the upper buttons "bridge" |
21:52.57 | wpwrak | but also a lot harder to machine |
21:53.32 | wpwrak | -> $$$ |
21:54.14 | DocScrutinizer05 | yep, I know |
21:55.38 | DocScrutinizer05 | anyway steel would probably even allow for a 2 0.1mm thick 2ears" with holes in them, at upper right and left side, to actually fix the frame *under* the display bolts |
21:56.00 | DocScrutinizer05 | we could reduce PCB thickness by 0.1mm accordingly at that area |
21:56.17 | DocScrutinizer05 | HMMMM |
21:57.07 | wpwrak | we we going to left-shift the price of Neo900 a bit ? :) |
21:57.07 | DocScrutinizer05 | who says UPPER needs tho get fixed by the display bolts at all? maybe some latch on spacerframe coult fix it instead |
21:58.01 | DocScrutinizer05 | wpwrak: we cannot design stuff from a "we got xx bicks, now let's see what we can do for that amount" perspective |
21:58.11 | DocScrutinizer05 | bucks even |
21:59.16 | DocScrutinizer05 | when spacer frame isn't working in acrylic, we need to investigate other alternatives, then after we found all solutions that may actualy work, we check which is the most economic one |
22:00.16 | DocScrutinizer05 | I had dural CNC stuff made already, been $$, not $$$ |
22:01.10 | DocScrutinizer05 | when spacer frame costs 15 bucks then we will survive that, even when it's not nice. We even will *survive* 50 bucks per frame |
22:01.48 | DocScrutinizer05 | spacer frame is a mission critical part. It has to just work |
22:02.55 | DocScrutinizer05 | we cannot do somthing that "somehow works" and gives us the benefit of only costing 2 bucks per frame but gives us all sorts of annoyances and problems |
22:03.46 | DocScrutinizer05 | spacer frame already is a botch by design. So let's at least try to make something that doesn't suck too much |
22:05.08 | DocScrutinizer05 | as a first simple rule, you must not be able to destroy the spacerframe by pressing at it with your fingernail |
22:08.37 | DocScrutinizer05 | zinc cast comes to mind |
22:08.58 | DocScrutinizer05 | need to check what's the options and cost with that one |
22:10.40 | DocScrutinizer05 | most likely way too expensive since aiui you need those damn molds again. Same situation like with case |
22:14.54 | wpwrak | regarding cost, plastic = $, aluminium = $$, steel = $$$. and yes, casts cost |
22:15.14 | DocScrutinizer05 | souds about right |
22:19.12 | DocScrutinizer05 | sill not off the table: make UPPER PCB so large that it creates the outer wall of case on that level |
22:20.22 | wpwrak | sounds like a bag of problems :) |
22:20.45 | wpwrak | and unless you gold-plate the sides of the pcb, it'll look shitty |
22:20.46 | DocScrutinizer05 | have a thin rubber sealing between UPPEr and the rim of case shell. Mount UPPER to case shell rigidly via the 6 screws we have. Mount LOWER only to UPPER then, giving it some clearance to case shell |
22:21.13 | DocScrutinizer05 | I planned to do exactly that gold plating, but Nik told me "can't get done" |
22:21.27 | wpwrak | oh, it can, but it has a price ... |
22:21.44 | wpwrak | also, the fancier the pcb spec, the fewer fabs that can make it |
22:21.54 | DocScrutinizer05 | is that price higher than a spacer frame of palstic, alu, steel, zinc? |
22:22.16 | wpwrak | but you still need the spacer :) |
22:22.27 | DocScrutinizer05 | huh? |
22:22.36 | wpwrak | well, or some piece of plastic. whatever you call it |
22:22.47 | wpwrak | you need a keyboard frame |
22:22.54 | DocScrutinizer05 | we need a kbd frame then, that's true |
22:23.15 | DocScrutinizer05 | that however would look pretty simple then |
22:23.56 | DocScrutinizer05 | could get mounted to UPPER with a few screws *before you marry the sandwich |
22:24.33 | DocScrutinizer05 | assembly would be dead easy |
22:24.55 | wpwrak | and your new keyboard frame also needs to provide a bit of wall all around the device |
22:25.09 | DocScrutinizer05 | lterally drop in the whole sandwich incl kbd frame into case shell, fix with the screws |
22:25.21 | wpwrak | granted, you can now glue all the plastic to the upper board, but ... |
22:25.45 | DocScrutinizer05 | not glue, screw |
22:26.14 | wpwrak | in any case, you would have exactly the same options also if the pcb is not part of the outside |
22:26.24 | DocScrutinizer05 | no |
22:26.48 | DocScrutinizer05 | impossible to assemble, way more complex frame, less rigid design |
22:27.09 | wpwrak | also, if pcb goes to the outside, you have a rigid mechanical connection. so anything that hits the device will go straight to the connectors and such |
22:27.22 | DocScrutinizer05 | you cannot fix the spacer frame as we have it now with screws to UPPER |
22:27.44 | DocScrutinizer05 | no, since I mounted LOWER to UPPER |
22:28.39 | DocScrutinizer05 | >>Mount LOWER only to UPPER then, giving it some clearance to case shell<< |
22:29.02 | wpwrak | well, think a bit about all the places where things mate and you'll see where you run into problems |
22:29.33 | wpwrak | you've basically divided one problems into two, each roughly of the same magnitude as the original |
22:29.34 | DocScrutinizer05 | I already did. I had that idea... some time ago |
22:30.35 | DocScrutinizer05 | why don't _you_ "think about all the places" so I could check what I missed? |
22:30.51 | wpwrak | how would it connect to the bottom shell ? just lie flat on top ? or try to have a rim ? |
22:31.10 | DocScrutinizer05 | flat, rim, we can do both |
22:31.26 | DocScrutinizer05 | some "latches" |
22:31.56 | DocScrutinizer05 | basically it gets fixed by the 6 screws |
22:32.34 | wpwrak | for them rim, you'd need a very fine structure. there you may actually have no choice by to use some metal |
22:32.53 | DocScrutinizer05 | why do we need such fine structure? |
22:32.58 | wpwrak | on the other side (rear), you have no choice but have the pcb rest flat on the case |
22:33.31 | DocScrutinizer05 | maybe I already missed what we're actually talking about |
22:33.41 | wpwrak | also, how would all this give you the 2 mm internal separation between upper and lower ? |
22:34.35 | wpwrak | actually, lemme check this one |
22:36.13 | *** join/#neo900 MonkeyofDoom (~~~~~~~~Mo@71-14-188-191.dhcp.stls.mo.charter.com) |
22:37.49 | DocScrutinizer05 | http://projects.goldelico.com/p/neo900/issues/518/#ic1581 |
22:37.54 | wpwrak | hmm, distance between top of upper and the edge of the bottom shell is about 1.9 mm |
22:38.35 | DocScrutinizer05 | and bottom of the one before |
22:39.12 | DocScrutinizer05 | till http://projects.goldelico.com/p/neo900/issues/518/#ic1585 |
22:42.50 | DocScrutinizer05 | the distance between upper side of LOWER and the rim of case shell is < 2mm |
22:43.04 | wpwrak | so you a) need a structure on top of UPPER-XL that provides keyboard frame and the part of the wall that normally comes from the bottom case shell |
22:43.16 | DocScrutinizer05 | distance betwee upper side of LOWER and lower side of UPPEr is 2mm |
22:43.31 | DocScrutinizer05 | a) yes, exactly |
22:43.57 | wpwrak | and b) if you want to have a rim at the bottom you need some thin U-shaped structure on the pcb there as well |
22:44.05 | wpwrak | lemme check the dimensions of that |
22:44.57 | DocScrutinizer05 | elastic and ~0.2mm thick. See http://projects.goldelico.com/p/neo900/issues/518/#ic1585 |
22:46.47 | DocScrutinizer05 | we however should have some non-elastic distance blocks in there, so forces to UPPEr like key pressing (or impact from fropping device) will not give all load only to those distance nails I mentioned above |
22:47.03 | DocScrutinizer05 | dropping* |
22:47.46 | wpwrak | that U-structure would have to be 1.2 mm - tolerance tall. width from 0.5 mm to as much as you want |
22:48.41 | DocScrutinizer05 | the new (now only) kbd frame would get mounted to UPPER by a few screws you drive in from bottom side of UPPER to hold frame in place |
22:48.56 | DocScrutinizer05 | sorry? |
22:49.21 | DocScrutinizer05 | where from are those 1.2mm? |
22:49.31 | wpwrak | U-structure to connect to the rim of the bottom shell. making sure board and shell don't bend away from each other too much |
22:49.39 | wpwrak | that's the depth of the rim |
22:50.12 | DocScrutinizer05 | oooh that one, we don't care about that stair in there anymore |
22:50.51 | DocScrutinizer05 | well, we could |
22:51.19 | wpwrak | would make the thing less wobbly |
22:51.30 | DocScrutinizer05 | but we also can go "deeper" and support the case where it has its full wall thickness |
22:51.47 | wpwrak | yes, that's what you would have to do on the rear side |
22:51.56 | DocScrutinizer05 | exactly |
22:54.06 | DocScrutinizer05 | maybe use slightly triangle shaped noses that are from kbd frame and go through holes in PCB to form the support structure? |
22:54.14 | wpwrak | adding rubber sheets to the stack is likely to be messy if they go all the way to the edge: a) you get even more parts that need to be finished with very high precision (and rubber !), b) they're likely fray and to trap dirt, so that edge will probably look messy |
22:54.38 | DocScrutinizer05 | good point |
22:54.48 | wpwrak | (triangles) that would be an option, yes |
22:56.18 | DocScrutinizer05 | we don't want foam or rubber or sth like that there |
22:56.32 | wpwrak | noises would also improve xy registration between the top plastic and the upper board |
22:56.47 | DocScrutinizer05 | yes, that been the idea |
22:58.12 | DocScrutinizer05 | Nik will kill us both for binning the complete ME-testing done in V0 when we go this path |
22:58.39 | wpwrak | i thought most of the ME was on LOWER ? |
22:58.47 | wpwrak | ME testing i mean |
22:59.03 | DocScrutinizer05 | yes, exactly, and LOWER must not touch the case in this stacking |
22:59.33 | wpwrak | well, it can touch it. just mustn't snap into it |
22:59.53 | DocScrutinizer05 | in this stacking it's UPPER that needs to fit to the case, and LOWER needs the clearance we now have on UPPER (tentatively) |
23:01.02 | DocScrutinizer05 | we need that clearnace to allow deformation and thermal/other shape distortion will not create shearing forces to the B2B connectors |
23:01.35 | DocScrutinizer05 | I actually hate the fact that those B2B have so few x/y "elasticity" |
23:02.03 | wpwrak | if you screw lower to upper then the spacers you use for that will take these forces |
23:02.08 | DocScrutinizer05 | Nik already searched for some that have maximum x/y clearance available |
23:02.33 | wpwrak | (spacer taking forces) which of course means that, as you tighten them, they may break your b2b if you're not careful |
23:02.46 | wpwrak | well, plan B would be FPC (-:C |
23:03.10 | DocScrutinizer05 | that would be great, from an ME / EE POV |
23:03.39 | DocScrutinizer05 | not though from a components sourcing and selection POV |
23:04.14 | wpwrak | do it nokia-style ? ;-) |
23:05.18 | DocScrutinizer05 | err yes, technically that would actually be feasible |
23:05.45 | DocScrutinizer05 | though wasting a ~ 3cm^2 PCB real estate |
23:06.43 | DocScrutinizer05 | but then we have 8-layer on both PCB, and that embedded FPC stuff is expensive and extremely nasty |
23:07.08 | wpwrak | i imagine it to be |
23:09.27 | DocScrutinizer05 | FPC is great, but only with a connector on both ends |
23:09.42 | wpwrak | i still suspect that in this kind of stack, things can easily shift out of alignment. so you may have the pcb stick out some 0.2 mm at some place, suffer abrasion, etc. |
23:09.45 | DocScrutinizer05 | kinda like GTA02 debug cable |
23:09.59 | wpwrak | and we all know just how much that one sucked :) |
23:10.36 | DocScrutinizer05 | well, it's not meant to withstand abuse by user |
23:10.39 | wpwrak | (gta01/02 debug) well, you've been spared the worst i think |
23:10.58 | DocScrutinizer05 | ? |
23:11.03 | wpwrak | oh, it didn't need abusive users. we've all been very very careful ;-) |
23:11.24 | DocScrutinizer05 | FPC are for inside a device |
23:11.44 | DocScrutinizer05 | there they work just fine |
23:12.19 | wpwrak | we had basically everything possible go wrong with these cables. and it was a pita to even get it manufactured. and the junk they sent us was mostly garbage. |
23:12.59 | DocScrutinizer05 | :-/ |
23:13.26 | wpwrak | gta02 side too wide (so you had to trim it with a cutter), board side too wide/small/off-center, so you had to wiggle it around until you had enough signals connect (no point in hoping to get all at the same time) |
23:13.41 | DocScrutinizer05 | doesn't mean this is system immanent |
23:14.18 | wpwrak | then material fatigue in the connector on the board. and the copper in the fpc broke very easily, just by regular bending. no abuse needed. |
23:14.56 | wpwrak | i mean i made some support structures that held the whole mess in place to avoid any load whatsoever on the precious fpc ... |
23:16.03 | wpwrak | yes, i think that one was the worst fpc experience possible. it can only get better from there ;-) |
23:16.37 | DocScrutinizer05 | the wuestion is: could we get a FPC (possibly even some off-the-shelf stuff) that sucks less re quality, and can we get some connectors for that too? |
23:22.09 | *** join/#neo900 freemangordon (~freemango@46.249.74.23) |
23:23.03 | wpwrak | there may be some short parallel ones |
23:23.11 | *** join/#neo900 nicksydney (~quassel@197.139-50-210.dynamic.dsl.syd.iprimus.net.au) |
23:23.11 | wpwrak | how many signals do we need ? |
23:23.48 | wpwrak | (digi-key: Product Index > Cable Assemblies > Flat Flex, Ribbon Jumper Cables) |
23:24.20 | wpwrak | there should also be solderable such critters. but digi-key doesn't seem to have that kind |
23:25.52 | wpwrak | (solderable) like on this critter: http://www.buydisplay.com/media/catalog/product/cache/1/image/c577d1203d4a53d9f98182eb6081b1d6/s/e/serial_spi_1.3_inch_128x64_oled_display_module_ssd1306_white_on_black_1.jpg |
23:25.54 | DocScrutinizer05 | err, dunno, what have we got so far? 2 pcs 60pin |
23:26.42 | DocScrutinizer05 | yeah, soldering one end is a reasonable option |
23:27.46 | DocScrutinizer05 | I'm pretty sure those rigid 60pin B2B-conn will eventually give us headache |
23:28.10 | wpwrak | 2 x60 ? hmm. that's a lot. |
23:28.26 | wpwrak | here's one with 40. 0.5 mm pitch, 17 mm long: http://www.digikey.com/product-detail/en/050R40-17BO/RE109-010-ND/2263535 |
23:28.33 | *** join/#neo900 Neros_ (~quassel@128-79-158-242.hfc.dyn.abo.bbox.fr) |
23:29.02 | wpwrak | and i'm sure china has some cheaper ones ;-) |
23:29.16 | DocScrutinizer05 | well, 3 of them will replace 2 pcs 60pin B2B |
23:29.38 | DocScrutinizer05 | WTF? 20 bucks? |
23:29.47 | wpwrak | you found it ;-) |
23:30.04 | wpwrak | that's just digi-key's hopelsss patriotism |
23:30.24 | DocScrutinizer05 | awesome datasheet¡ |
23:31.39 | wpwrak | the L variants may be interesting: "lasered". that could be the soldering holes |
23:32.02 | DocScrutinizer05 | :nod: |
23:32.30 | DocScrutinizer05 | idly mused about that Z-axis connecting sticky tape |
23:32.54 | DocScrutinizer05 | but that's not as reliable as we would like it to be, I guess |
23:33.38 | DocScrutinizer05 | though industry seems to actually use it for exactly such stuff like attaching FPC to glass of LCD |
23:37.30 | wpwrak | hmm. 4pcb.com have "gold fingers" which is on the "edge". but only options are 0-2 sides. |
23:37.52 | DocScrutinizer05 | hm? |
23:38.07 | DocScrutinizer05 | what is 2gold fingers"? |
23:38.42 | wpwrak | ah, plated edges is if what we'd want. alas, "(Pricing Subject To File Review) " |
23:38.54 | wpwrak | (2) gold fingers on 2 sides |
23:39.01 | DocScrutinizer05 | aaah, PCB gold rim |
23:39.48 | DocScrutinizer05 | I dunno, the PCB edge looks pretty clean |
23:41.27 | DocScrutinizer05 | but gold edges (actually segmented, for contacts) would be great |
23:42.00 | wpwrak | let's see ... 8 layer, didn't specify anything fancy. with gold fingers on 2 sides, 2 week: USD 529 tooling, then USD 49.68 @ 50 |
23:42.30 | DocScrutinizer05 | for the plating alone? |
23:42.32 | wpwrak | without gold fingers: same NRE, USD 38.60 @ 50 |
23:42.36 | DocScrutinizer05 | aaah |
23:42.54 | DocScrutinizer05 | now that doesn't sound bad at all |
23:43.03 | wpwrak | whole pcb. but under-specified. we probably need some fancy stuff that would make it non-quoteable |
23:43.25 | wpwrak | that's just fingers. full edges may be harder |
23:43.32 | wpwrak | but at least it's a start :) |
23:43.39 | DocScrutinizer05 | expected as much. wouldn't see why via plating is cheap and edge plating cost more than gold (literally) |
23:43.57 | wpwrak | usual process -> $$$ :) |
23:44.24 | wpwrak | by and large it's always the same |
23:45.34 | DocScrutinizer05 | >>And when we already have gold there, why not make this a few cotact pads for e.g. charger-cradle pogopins to contact there. << |
23:46.04 | DocScrutinizer05 | http://projects.goldelico.com/p/neo900/issues/518/#ic1581 |
23:46.39 | DocScrutinizer05 | with FPCs we could even keep LOWER ME as is |
23:46.42 | DocScrutinizer05 | \o/ |
23:47.21 | DocScrutinizer05 | neither orientation nor distance between UPPER and LOWER would have to match to the 1/10 |
23:48.16 | DocScrutinizer05 | sou we would mount UPPEr on top of LOWEr only loosely |
23:48.54 | DocScrutinizer05 | weird how my typng always fails to keep the shift key pressed on ending "r" |
23:50.19 | DocScrutinizer05 | mucle memory needs retraining |
23:50.19 | wpwrak | the fpc idea is also orthogonal to what happens with the spacer frame |
23:50.28 | Oksana | DocScrutinizer05how the heck Oksana was able to tell we use pelican? I didn't recall the name |
23:50.30 | Oksana | See file "pelicanconf.py"? http://neo900.org/git/?p=www;a=tree |
23:50.45 | DocScrutinizer05 | :-D |
23:52.07 | DocScrutinizer05 | wpwrak: please explaon "orthogonal" |
23:52.23 | DocScrutinizer05 | in context |
23:52.42 | DocScrutinizer05 | or simply, please rephrase |
23:55.18 | wpwrak | independent from |
23:55.27 | DocScrutinizer05 | wpwrak: aah then I disagree |
23:56.01 | DocScrutinizer05 | with FPCs we could even keep LOWER ME as is |
23:56.44 | wpwrak | i was more thinking of fpc also being an option with the old design that keeps the pcbs inside |
23:57.07 | DocScrutinizer05 | aaah |
23:57.13 | DocScrutinizer05 | yes, of course |
23:57.31 | DocScrutinizer05 | and I meant that B2B conns will give us headache either way |
23:57.48 | DocScrutinizer05 | I don't like them for the stacking we got right now either |
23:57.51 | wpwrak | yes, it looks as if they had a good chance of doing so |
23:58.47 | DocScrutinizer05 | if we could go s/B2B/FPC/ we *should* do this |
23:58.55 | wpwrak | you'd basically want a rigid package: e.g., upper screwed (with spacers) to lower, then b2b in the middle. so all stress goes to the spacers |
23:59.19 | wpwrak | or fpc to connector board, like nokia did the display |
23:59.27 | DocScrutinizer05 | :nod: |
23:59.58 | wpwrak | b2b as mechanical support works great in the arduino world, but ... :) |