pagurus has quit [Remote host closed the connection]
pagurus has joined #neo900
pagurus has quit [Remote host closed the connection]
arossdotme has joined #neo900
<wpwrak>
naw, it says 2016 :)
<DocScrutinizer05>
typo, sorry
ravelo has joined #neo900
<DocScrutinizer05>
I need a 21€ item from mouser, do we need anything else? to reach free shipping
liteIRC_ has joined #neo900
ravelo has quit [Disconnected by services]
liteIRC_ is now known as liteIRC__
liteIRC__ is now known as ravelo
<DocScrutinizer05>
I highly doubt the 02-25 shipping date
rjeffries has joined #neo900
<wpwrak>
pogos ?
arossdotme has quit [Ping timeout: 240 seconds]
<DocScrutinizer05>
hmm, which ones?
<DocScrutinizer05>
I actually feel tempted to get some low profile ones (~2mm high max) to mount to a battery-shaped PCB
<DocScrutinizer05>
I prolly also could use higer ones and then use a shim of sorts to make the PCB 'hover' several millimeters above bottom of battery bay and still sit flush and parallel to bottom
arossdotme has joined #neo900
ravelo has quit [Read error: Connection reset by peer]
ravelo has joined #neo900
<DocScrutinizer05>
hmmm, or fix the pogopins to testpoints with a blob of wax, than use a slaughtered BL-5J where cell removed, insert it to the bay and fill with epoxy
<DocScrutinizer05>
well, maybe a smart idea to also check minimal possible pitch and compare the the needed pitch
arossdotme-planb has joined #neo900
<DocScrutinizer05>
*very* tight
<DocScrutinizer05>
might need sime milling to take off from the 2mm dia 'dish' of the pogos
<DocScrutinizer05>
some, even
arossdotme has quit [Ping timeout: 240 seconds]
<wpwrak>
test points have a pitch of about 1.6 mm, no ?
liteIRC_ has joined #neo900
ravelo has quit [Disconnected by services]
liteIRC_ is now known as liteIRC__
liteIRC__ is now known as ravelo
<DocScrutinizer05>
~2mm
<DocScrutinizer05>
1.9 maybe
<DocScrutinizer05>
I had a caliper opened to 6mm and I had a 3.3 testpoints in there
pagurus has joined #neo900
<DocScrutinizer05>
since this is a one-shot for freemangordon, I get away with dremel and a 'oblong' milled into the 2mm dia dish of the pogo
<wpwrak>
maybe 1.8
<DocScrutinizer05>
well above 1.5
<DocScrutinizer05>
which is the pogo core dia
<wpwrak>
so the pogos shouldn't be wider than ~1.6 mm, better 1.5 mm (it's very hard to place them in a strictly vertical orientation)
pagurus has quit [Remote host closed the connection]
<DocScrutinizer05>
strictly verical isn't as much of a problem. Worst case I ram some isolator plastic in between
<freemangordon>
DocScrutinizer05: you know what? isn;t it better to connect that device to europa, the same way as bb-xm?
<wpwrak>
phew :)
pagurus has joined #neo900
<freemangordon>
that way more devs will have chance to use it
<DocScrutinizer05>
I still need to connect to serial, no?
<freemangordon>
sure
<DocScrutinizer05>
so I prefer sending it to you rather than setting up Europe again
<freemangordon>
ok
<DocScrutinizer05>
and I guess I will simply solder the testpoints with thin wires
<freemangordon>
what is wrong with Europa?
<DocScrutinizer05>
ideally I find wires thin enough so a battery still fits in
<freemangordon>
DocScrutinizer05: do as you wish, but think about ^^^ once again before sending :)
<DocScrutinizer05>
Europa is fine, just collecting dust and i'm happy about that since I don't need to manage all the VLAN and port forwards and stuff
<freemangordon>
for example Pali will have a chance to get serial output too
<DocScrutinizer05>
I need a better caliper, something for blind people
<DocScrutinizer05>
well, how would you restart and flash such a remote device?
<freemangordon>
:D
<wpwrak>
(solder instead of pogo) yes, i think that's a tough one with single pogos. and pogo blocks may be very hard to find (or custom-made)
<freemangordon>
DocScrutinizer05: anyway you solder stuff, you can put the battery away fro mthe device and connect it to a managed switch
<wpwrak>
(restart) power-cycle with relays ?
<DocScrutinizer05>
flashing might still work, via USB, but actually power cycling is... nasty
<freemangordon>
flashing is not a problem
<freemangordon>
if device is in r-d mode, it is flashable on boot without 'U' pressed
<DocScrutinizer05>
and I don't like a N900 under full remote control in my living room ;-P
<freemangordon>
:D
<freemangordon>
ok, now I am convinced ;)
<wpwrak>
it's a N900, not a Transformer ;-)
<freemangordon>
DocScrutinizer05: but, cameras still does not work in mainline kernel
<DocScrutinizer05>
it's a linux device with camera and microphone
<freemangordon>
yeah, correct
<freemangordon>
ok, send it then
* freemangordon
is afk
<DocScrutinizer05>
nah, kernel debugging is extremely inconvenient to do remote. When Pali needs a console-enabled device as well, I can provide one to him too
<freemangordon>
well, what I am afraid of, is that he will ask me to do tha debugging :D
<DocScrutinizer05>
hehehe
<wpwrak>
black tape for the camera and we won't see you walk naked through your living room. microphone ... hmm, listening to joerg grumble ...
<DocScrutinizer05>
no, my parties are strictly off limits ;-)
arossdotme-planb has quit [Ping timeout: 240 seconds]
<wpwrak>
or just short the mic ?
<freemangordon>
DocScrutinizer05: "remote kernel debugging" - come on, I did remote xloader debugging, kernel should be a piece of cake compared :)
<freemangordon>
anyway, bbl
<DocScrutinizer05>
indeed, when you put it like that
<bencoh>
:]
* DocScrutinizer05
checks mouser for decent calipers
arossdotme-planb has quit [Ping timeout: 272 seconds]
<Kero>
do I want a russion kbd? found out that I don't want a british kbd...
<DocScrutinizer05>
swapping keymat is pretty simple
<MonkeyofDoom>
4000+ is a pretty number
<MonkeyofDoom>
:)
hashcore has quit [Ping timeout: 250 seconds]
<DocScrutinizer05>
well, there are several shops quoting up to ~2000, I guess it's as usual, they all refer to same original source
<MonkeyofDoom>
ah
paulk-collins has joined #neo900
<wpwrak>
DocScrutinizer05: N950 hasn't arrived yet. dunno where it is exactly - the tracking is either broken or they really screwed up the user interface :-( in any case, the next two days are holidays here, so the earliest possible delivery would be wednesday
<wpwrak>
DocScrutinizer05: (hb) yeah, meaning to ask you to confirm or change the current ratings i suggested. once i have them i can calculate the connectors, which in turn (may) affect the exact geometry. so that would then hopefully close that whole mess of dependencies.
<DocScrutinizer05>
please assume current limits as they derive from the connectors used
<DocScrutinizer05>
there's no design requirement for current ratings
<DocScrutinizer05>
as good as it gets
<DocScrutinizer05>
a 500mA for VBATT would be nice
<wpwrak>
VBAT_RAW or VBAT_SWITCHED ?
<DocScrutinizer05>
1A would be great
<DocScrutinizer05>
VBAT_SWITCHED is low power
<DocScrutinizer05>
max 300mA
<DocScrutinizer05>
rather less
<wpwrak>
i have 1.5 A (charging) for VBAT_RAW, 100 mA for VBAT_SWITCHED, 0.5 A for VBUS_OTG, and 100 mA for GPIO_d/3V3 (or whatever voltage we end up putting there)
<wpwrak>
okay, 300 mA is a good number :)
<wpwrak>
i think it's the lowest per-pin current rating i came across for connectors that looked suitable for this scenario
<DocScrutinizer05>
ok :)
<wpwrak>
so anything <= 300 mA is okay for a single pin. then i need to bundle. and hopefully i'll end up with enough ground to match the positive rails ;-)
<DocScrutinizer05>
we have 4 GND pins iirc
<wpwrak>
it's not just HB :) we also need to connect to the memory card and the flash
<DocScrutinizer05>
yep
<DocScrutinizer05>
pogopins go up to 1A at least ;-)
<DocScrutinizer05>
if everything else fails
<wpwrak>
pogos sounds like a good plan B. "just hold BOB in place while taking a picture" ;-)
<DocScrutinizer05>
err nope, BOB is fixed by case anyway
<DocScrutinizer05>
at least in this direction
hashcore has joined #neo900
<wpwrak>
in Z ? yes. which would make it fairly entertaining to shift it around in XY until most of pogos connect to something that at least won't cause trouble :)
arossdotme-planb has joined #neo900
<DocScrutinizer05>
shifting in XY can't happen due to the B2B-conns
<DocScrutinizer05>
and 2 B2B should have enough retraction force to withstand one pogo
<wpwrak>
yes, but we're talking about those b2b conns. so if you use pogo for them, shifting does happen :)
<wpwrak>
ah, selective pogo. heh :)
arossdotme has quit [Ping timeout: 250 seconds]
<wpwrak>
naw, let's not go there. mixing connectors is sending a nice invitation card, printed on expensive paper, with rsvp, to trouble
<wpwrak>
(i'm still amazed how thoroughly that backfired on milkymist. and that was MUCH simpler than what we're doing with BOB)
<DocScrutinizer05>
trouble with connectors?
<wpwrak>
it has two different types of connectors for the little board that had an FTDI chip for the serial console. they weren't matched carefully ... and all sort of problems developed :)
<wpwrak>
(and no, i wasn't involved in that design - but i learned from it nevertheless :)
rjeffries has quit [Ping timeout: 245 seconds]
<DocScrutinizer05>
yes, poor alignment is one of my largest concerns with all those B2B conns in the sandwich and to BOB
<DocScrutinizer05>
that's why I suggested to revisit that detail and find a solution that literally offers more flexibility
<DocScrutinizer05>
I wouldn't like to find we need to pair LOWER to UPPEr and rework 70% of the boards
<wpwrak>
guess PCB and SMT will just have to be very precise ...
<wpwrak>
reworking the big connectors between LOWER and UPPER would certainly be a treat
<DocScrutinizer05>
we'll only know after PV, if not after we picked a 30% yield from fab line
<DocScrutinizer05>
I wouldn't want to manually rework a 400 boards
<wpwrak>
well, at least we know that sufficient precision for this sort of sandwich is possible. USRP had this kind of arrangement, apparently without undue problems
<DocScrutinizer05>
mhm, nice to know
<wpwrak>
bigger connectors, though. but then the whole thing is bigger. and they had "real" RF in there for fun, too
<DocScrutinizer05>
bigger connectors is the key word here
<DocScrutinizer05>
size of the whole thing doesn't matter as much as the size (and wiggling room) of the connectors themselves versus the precision of placing and *soldering* the components
<DocScrutinizer05>
components tend to drift away during soldering
<DocScrutinizer05>
if you need precise placement of connectors, you use a type with notches that go through fitting holes in the PCB
<DocScrutinizer05>
Nikolaus promised there will be zilch problems. however he's not here to keep his promise
<DocScrutinizer05>
I'm still pretty worried and flirting with stuff like zebra strips
arossdotme has joined #neo900
<DocScrutinizer05>
for the sandwich at least
arossdotme-planb has quit [Ping timeout: 272 seconds]
<DocScrutinizer05>
actually we could use a short flex PCB and solder it to the edge of both PCBs to connect
<DocScrutinizer05>
or solder to one PCB's edge while using a B2B con on the other end
arossdotme-planb has joined #neo900
qwazix has joined #neo900
<DocScrutinizer05>
or, when we use those B2B conns, make a bay in otuer layer prepreg so the connector fits into the bay with no more error than what we can tolerate for that type of connector
<DocScrutinizer05>
outer, even
arossdotme has quit [Ping timeout: 264 seconds]
jurov has quit [*.net *.split]
jurov has joined #neo900
<DocScrutinizer05>
maybe we can glue the connectors into place, prior to soldering. Given the placers are working accurate enough
arossdotme-planb has quit [Ping timeout: 256 seconds]
<DocScrutinizer05>
we *might* even join the two conns with a mounting guide that holds them in a proper relation to each other, until soldered
<DocScrutinizer05>
only needs a minimal 'PCB' structure with two B2B receptacles *glued* into proper location
<DocScrutinizer05>
then plug to connector components to that thing and place the whole object for soldering
<DocScrutinizer05>
s/ to / two /
modem_ has quit [Ping timeout: 240 seconds]
<DocScrutinizer05>
however that idea doesn't fly for automated assembly, if only since we'd need multiple of those helper thingies which just replicates the problem to placement of the jacks on the helper PCB
<DocScrutinizer05>
could we use LOWER as guide for placing and glueing the conns on UPPER? would actually create pairing, but at least it would guarantee stuff fits
<DocScrutinizer05>
dunno how to place solder paste then
xes has quit [Quit: bye]
<saper>
what is your opinion on using cheap mediatek soc's to build phones?
arossdotme-planb has joined #neo900
<DocScrutinizer05>
no opinion here, since it's not applicable for our project
<DocScrutinizer05>
one of the major requeirements is compatibility to N900 - mtk chips don't fit in there
<saper>
thank you
<DocScrutinizer05>
onother aspect is compatibility to GTA04 which also uses an OMAP3
<DocScrutinizer05>
IOW a cheap SoC is not necessarily a good SoC, it needs to be available and open docs too
<DocScrutinizer05>
at least when you care about openness and are not Nokia or Samsung or Sony etc, to dictate what the chip makers offer to you
arossdotme-planb has quit [Ping timeout: 248 seconds]
<DocScrutinizer05>
>>ED: Next stop: MediaTek. Asian company, so probably easier to get than from the huge global players. However, MediaTek only offers Android drivers, and they do not plan to support Linux – well, not for our small quantities anyways. So they decided to not sell them to us as well.<<
<DocScrutinizer05>
note **they decided to not sell**
<MonkeyofDoom>
such a shame how Android chokes out Linux proper in situations like that
<MonkeyofDoom>
Google has enough resources they could afford to rebase their universe onto a modern kernel :(
jonsger has joined #neo900
<MonkeyofDoom>
not that the kernel is the whole story, of course, since the meaty bits of video drivers and lots else are pretty much all in userspace on those systems
<DocScrutinizer05>
or closed blobs no matter where they actually are
<MonkeyofDoom>
I think lots of users are less upset by closed blobs (e.g. for nvidia video cards) than they are by vendor support not tracking upstream kernels
<MonkeyofDoom>
of course the latter kind of falls out of the former eventually when it becomes no-longer-profitable to support your old hardware
<DocScrutinizer05>
the two topics are thoroughly entangled though
<MonkeyofDoom>
yeah
<DocScrutinizer05>
anyway we build and sell an *open* hardware, no blobs from chip manufs, definitely not on the SoC
<MonkeyofDoom>
yeah
<DocScrutinizer05>
if the system can't work without blobs, we missed our design goal
<DocScrutinizer05>
the idea exactly being independance from manfacturer support
<DocScrutinizer05>
so you could upgrade to kernel 14.876 in 2040
<DocScrutinizer05>
we also don't want any blobs on APE for mere security reasons
<DocScrutinizer05>
audit of a blob for vulnerabilities and backdoors isn't impossible but terribly hard
<MonkeyofDoom>
right
<MonkeyofDoom>
what's APE?
<DocScrutinizer05>
APplication Environment
<MonkeyofDoom>
oh, ok
<DocScrutinizer05>
or even app Processor Env
<DocScrutinizer05>
iow the linux CPU
<DocScrutinizer05>
basically the domain FSF should limit their RYF to ;-)
* DocScrutinizer05
just had an idea for the rating in UYL (our RYF evolution): device classes: *) full "RYF" compliant *) theoretically possible to reboot into a fully operational fully RYF compliant system from a system that does sorts of init which are not RYF compliant (this would be Neo900 then) *) RYF compliance only for a system stripped down in functionality since the needed blobs defeat some system functions when removed
<DocScrutinizer05>
the latter would be the case for example with WLAN that needs a blob for softMAC and doesn't work without such blob
<DocScrutinizer05>
or simply put: closed linux drivers for some of the subsystems
<DocScrutinizer05>
for Neo900 that applies for the 3D parts of PowerVR, however the system is fully functional without them
jonwil has joined #neo900
<DocScrutinizer05>
well, Neo900 doesn't even comply to RYF on the reboot level definition since we offer you can update the modem firmware whenever Gemalto provides such update blob
<MonkeyofDoom>
I'm still waiting for someone to write an open-source driver from the leaked PVR stuff, legality be damned
<DocScrutinizer05>
nah, actually when imagination is hiring FOSS developers to further their linux support, then why not start with 'legalizing' what's already leaked anyway? Must be lawyer mumbojumbo
<DocScrutinizer05>
we didn't see Nokia doing such thing either. Supposedly because they didn't 'own' the code
<DocScrutinizer05>
they should have talked to colabora and agreed upon releasing everything
ashneo76 has joined #neo900
<DocScrutinizer05>
I don't know enough about imagination and the PVR stuff to even start to speculate why they didn't release stuff
<jonwil>
More than likely the newer PowerVR parts are more friendly to being opened up than the older SGX parts (i.e. less "secret sauce" in there that needs to be dealt with), also the leaked stuff likely contains things imagination cant share (e.g. code imagination doesn't own the rights to or things they cant share because of patent agreements)
arossdotme-planb has joined #neo900
arossdotme has quit [Ping timeout: 248 seconds]
<DocScrutinizer05>
:nod:
<jonwil>
I strongly suspect the older PowerVR chips (e.g. SGX) are the chips least likely to end up with FOSS GPU drivers