che11 has quit [Remote host closed the connection]
xes has quit [Quit: Going offline...]
wazrus has joined #neo900
wazrus has quit [Quit: Lost terminal]
nox- has quit [Quit: Leaving]
Kabouik has quit [Ping timeout: 250 seconds]
nicksydney has quit [Quit: No Ping reply in 180 seconds.]
nicksydney has joined #neo900
<Oksana>
Good night
ShadowJK has quit [Ping timeout: 255 seconds]
ShadowJK has joined #neo900
nicksydney has quit [Quit: No Ping reply in 180 seconds.]
nicksydney has joined #neo900
sparetire_ has quit [Quit: sparetire_]
nicksydney has quit [Read error: Connection reset by peer]
kolp has joined #neo900
Pali has joined #neo900
Kabouik has joined #neo900
wazrus has joined #neo900
wazrus has quit [Quit: Lost terminal]
che1 has joined #neo900
che1 has quit [Ping timeout: 255 seconds]
che1 has joined #neo900
chainsawbike has quit [Ping timeout: 255 seconds]
unclouded has quit [Ping timeout: 272 seconds]
unclouded has joined #neo900
chainsawbike has joined #neo900
modem has quit [Remote host closed the connection]
modem has joined #neo900
jonwil has quit [Quit: ChatZilla 0.9.91 [SeaMonkey 2.30/20141013232806]]
modem has quit [Read error: Connection reset by peer]
modem has joined #neo900
che1 has quit [Ping timeout: 250 seconds]
<Pali>
DocScrutinizer05: I found buffer overflow in NOLO which is installed on that mozilla's n900
<Pali>
:D:D
<DocScrutinizer05>
LOL
<DocScrutinizer05>
Pali: we might need the hw memory controller init details of N9
<Pali>
nolo does not clear previous buffer and if new buffer is not NULL terminated and previous was larger then new...
<DocScrutinizer05>
we're about to source a first few parts of the 1GB RAM chip, but the datasheet is pretty much complete unobtainium, even the chip is near impossible to source. So for setting up the timing of memory controller we might need to copy N9/HARMATTAN
<Pali>
then I see invalid data on usb
<DocScrutinizer05>
Pali: haha great
<DocScrutinizer05>
0xFFFF ?
<Pali>
both flasher-3.5 and 0xFFFF :D
<Pali>
first string is "prod_code.......RX-51" (which returns device identifier) and second is "version:sw-release-51" which should return sw version
<Pali>
but sw version looks like is cleared in CAL
<Pali>
and nolo combined input buffer "version:sw-release" with previous output buffer "prod_code.......RX-51"
<Pali>
so it is: "version:sw-release-51"
<Pali>
funny :D:D
<DocScrutinizer05>
well, prolly an old NOLO version
<DocScrutinizer05>
NOLO got updated several times
<Pali>
so NOLO is at last version for sure?
<DocScrutinizer05>
no, I don't even know _which_ NOLO version is on those devices
<DocScrutinizer05>
missed to check it when I reflashed one device with my convenient flasher script
<Pali>
NOLO via USB told me: NOLO version 1.4.14
<Pali>
did you flashed full combined image?
<DocScrutinizer05>
yes
<Pali>
then if flashing process did not failed, there should be last version
<DocScrutinizer05>
sure, now there is
<DocScrutinizer05>
and it's "NOLO version 1.4.14"
che1 has joined #neo900
arcean has joined #neo900
sparetire_ has joined #neo900
<Pali>
now I fixed all known problems in 0xFFFF
<DocScrutinizer05>
WOW!
<Pali>
it can flash full RX-51_2009SE_21.2011.38-1_PR_COMBINED_MR0_ARM.bin image without any problem
<Pali>
what 0xFFFF still does not support is flashing MMC image
<Pali>
so for this last piece is flasher-3.5 needed
che1 has joined #neo900
<DocScrutinizer05>
hmm, flashing eMMC sounds like a weird concept anyway, since it's basically just like "start linux, execute a script to format the storage volume, then do some sort of ftp or whatever to copy some data via USB to it"
<DocScrutinizer05>
the above is exactly what NOLO and rootfs and flasher-3.5 do, and the reason why you can't do further MOLO commands in flasher after flashing eMMC. You need a -R aka "reboot N900" to enter NOLO flasher mode again
useretail has joined #neo900
<DocScrutinizer05>
the above is also the reason why flashing eMMC fails when your target device rootfs is borked
<useretail>
hey guys, how's neo900 development going?
<DocScrutinizer05>
busy
<DocScrutinizer05>
we're just about to source a few of the much wanted and hard-to-find 1GB RAM PoP chips
<DocScrutinizer05>
and such sourcing of risk parts is eating up our funds
<useretail>
cool. so how would you evaluate your work on neo900 in percentage?
<DocScrutinizer05>
hardly at all, it's pretty impossible to do that
<useretail>
approximate
<DocScrutinizer05>
we hope that we can start series production aka "build the sales devices" after prototype_v3, we are just inmidst of building prototype_v2
<DocScrutinizer05>
needed parts vanish from market and some of them reappear later on
<DocScrutinizer05>
for others we find a source after months of searching
<DocScrutinizer05>
yet for others we find a better alternative
<DocScrutinizer05>
and during all that our funds diminish
<DocScrutinizer05>
we need to source a 6..8k RGB LEDs now, after they reappeared unexpectedly
che1 has quit [Ping timeout: 264 seconds]
<DocScrutinizer05>
we need to secure as many of the 1GB RAM PoP chips as we can get, while making sure that they actually work in our device (have some chip tested on top of a DM3730 CPU in a working device like beagleboard) and also we don't want to buy fake or defect chips
<DocScrutinizer05>
we have a constant expense of ~5k EUR per month
<DocScrutinizer05>
and developing prototypes costs quite some money as well. So eventually we will run short on funds, and probably we will not be ready to start mass production and open up sales at that point in time
<DocScrutinizer05>
other risk parts we're about to secure: camera switch, power switch, volume switches
<DocScrutinizer05>
besides all that we are ordering small volumes of the mech parts (slider mech, LCD, touchscreen, flex cable, etc pp) to evaluate if they really are usable or are some fake. This also costs some money and quite a bit of time
<DocScrutinizer05>
useretail: anyway if you force me to give you a percentage, I'd say "50%"
<DocScrutinizer05>
with an error margin of +10/-30%
<DocScrutinizer05>
we're roundabout at our 1st aniversary, and we lost ~6 months for reorg, so the expected shipping date is mid of next year
<DocScrutinizer05>
useretail: thanks, we know all these :-)
<useretail>
will you have additional devices for sale?
<DocScrutinizer05>
useretail: try "KCE00E00CA"
<DocScrutinizer05>
useretail: nope, we cannot order excess material upfront, no funds for doing so
<DocScrutinizer05>
see above remarks about securing risk parts
<DocScrutinizer05>
we try to get excess qty of risk parts, but our funds are very limited to do so. So we may get a 1000 of each of the switches, but for sure not a 1000 of the RAM chips, when we right now expect to need max 500
<DocScrutinizer05>
so, given our current number of "preorders" (~380), plus excess devices we need for warranty swap, there might be a maybe 100 devices available to order in the end, for non-donors aka late-joiners. Maybe less
che1 has joined #neo900
<bencoh>
when do you think you'll close preorders ?
<DocScrutinizer05>
I simply don't know.
<DocScrutinizer05>
at the moment when we see "ok this part will never again get available and for our design it is needed" and we got a certain amount of that part and preorders reach that number
<DocScrutinizer05>
for now it seems we might be able to source ~50 of the 1GB RAM chips. We're still searching for sources though. We *might* end up with a situation where we can only offer a certain number (like 50) of devices with 1GB RAM, and all those who joned late can only get a 512MB RAM which are sufficiently easy to source
<DocScrutinizer05>
and mech parts situation is yet another horror story
<DocScrutinizer05>
bencoh: the plan always been to keep "preorder" campaign running until real sales start
<DocScrutinizer05>
I don't hope we run into a situation where we have to declare that we definitely cannot satisfy any further orders
<DocScrutinizer05>
actually I would love when we could secure all our needed resources for arbitrary high amounts at a lead time of < 8 weeks, so we could take arbitrary high amounts of real orders when sales start
<astr>
DocScrutinizer05, thanks for the info. might it be possible that to get 5*1* 1gb ram chips so one more device can be 1gb ram if someone did say a partial pre-"order" to pay for the extra one 1gb ram chip so they have more time to perhaps raise the rest of the money needed but if they don't, for the 1 extra 1gb ram chip to sold on so they get there partial pre"order" money back?
Kabouik has quit [Ping timeout: 255 seconds]
<astr>
i'm just being hope-full and daring :P . hehe
<DocScrutinizer05>
astr: for new the amount of ram chips we can source is mainly limited by availability of such component
<astr>
....no I don't see it working when you can't get more then 50 and anyway thoughs those that pre-ordered early get priority, arr well sillyly hopefull hehe
<DocScrutinizer05>
if we can find an "unlimited" supply then it's a question of available funds. A average of 100plus-a-few bucks per "preorder" doesn't allow sourcing/securing of multiple components with a pricetag of $$ each
<DocScrutinizer05>
mind you, we have monthly expense and R&D expense as well, and those 100 bucks per device hardly suffice for that
<astr>
oh so it really needs to be a more generous preorder :)
<astr>
I do wonder how your keeping it going
<DocScrutinizer05>
now let's assume the chips will cost 20 bucks, we would need to secure a at least 450 of them, based on our current number of preorders. That would be
<DocScrutinizer05>
~450*20
<infobot>
9000
<DocScrutinizer05>
bucks
<astr>
~9000/0.4
<infobot>
22500
<astr>
£ roughly
<astr>
opps
<astr>
~9000*0.4
<infobot>
3600
<astr>
~9000*0.
<astr>
~9000*0.5
<infobot>
4500
<astr>
~9000*0.6
<infobot>
5400
<astr>
looks more like it in £
<astr>
~450*100
<infobot>
45000
<DocScrutinizer05>
we got a monthly expense of ~5000 (there are a few people who work 24/7 for this project, who need to eat and have a flat to store their bed and PC in), and we're facing an expense of ~20k for the proto_V2. Our kitty holds around 35k right now
<DocScrutinizer05>
(a few people) plus a real company paying taxes and all
<astr>
hmm with xmass comming up, reminds me of these "I want on of those" catalogues... maybe a "I want on of those libre's" catalogue featuring neo900 generous pre-order today! hehe
Kabouik has joined #neo900
<astr>
hope I didn't cause concern with the 450*100 calculation, it was to put help puy into perspective
<astr>
puy=put
<DocScrutinizer05>
eventually we will need more preorders (though that's somewhat limited by component supply), or ask for more donations from those who already donated/preordered. Or we need to open real sales, estimating a sales pricetag we cannot honestly tell right now and telling people that they either only paid part of the whole bill or get something back when device's final price tag gets fixed
<DocScrutinizer05>
or we open a new business where people buy components for their Neo900 from our company, and those components stay at our factory and will get used when building the devices and to reduce the sales pricetag of device for the particular customer who bought the components
<bencoh>
?
<bencoh>
not sure how the latter would work
<DocScrutinizer05>
would look like this: "your order: 1 Neo900, sales price 0.999 GPL; rebate for: RAM chip already paid by you: 0.050 GPL; final invoice: 0.949 GPL"
<DocScrutinizer05>
GPL == Gold Pressed Latinum
<DocScrutinizer05>
you simply buy a chip that's your property then despite it for now stays in a safe place at factory waiting to get used to build a N900
<Humpelstilzchen>
less complicate = better
* freemangordon
hates gcc inline asm syntax :(
<DocScrutinizer05>
if you decide you rather want to hold that chip inyour hands, we'll of course send it to you, however please note that single chips cannot get fed to the production component placer engine, so whatever you do with your chip at home, it won't end in a Neo900 unless you solder it there by yourself
<DocScrutinizer05>
we need a (iirc) 8 RBG side looking LEDs per device, which like a miracle appeared out of nowhere again, after 6 months of unobtainium period. each such component costs roundabout half a buck. So that's another 4€ * 500 to spend on securing risk parts
<DocScrutinizer05>
4 switches which are EOL: another maybe 3 bucks per device
<DocScrutinizer05>
there's a lot more, e.g. main camera: ~15bucks, domesheet: maybe 20 bucks, etc pp
che1 has quit [Ping timeout: 245 seconds]
nox- has joined #neo900
<DocScrutinizer05>
for the RGB LEDs I might be able to source enough for 1000 devices, taking the risk to have a 4000 useless components on shelf in the end, with a net loss of 2000 bucks burnt on them
<DocScrutinizer05>
but it's quite obvious that Neo900 UG cannot proceed to secure components like that
xes has joined #neo900
<DocScrutinizer05>
then OTOH for ram the alternative is to go for 512MB when the pre-allocated amount of 1GB components is used up, and for LEDs we may go for monochromatic when the RGB got used up, leaving the late Neo900 with white kbd backlight and no/monochromatic AUX LEDs. Production planning and particularly running a crowd funded project isn't always mere fun :-/
che1 has joined #neo900
* Humpelstilzchen
won't care about led colors..but RAM is important
<DocScrutinizer05>
Humpelstilzchen: well, the problem about LEDs starts like "we either get kbd backlight light we planned it, or we fail tosource one of the components and there will be no kbd backlight at all" - the fallback alternative to use white LEDs is not confirmed to be feasible either, but it has much higher likelihood. Nevertheless we would need to find matching white side-looking LEDs that are <0.7mm high and when we want to switch from RGB to
<DocScrutinizer05>
white later, we also need them to have a footprint that would fit on the one of a RGB LED
<DocScrutinizer05>
s/ light / like /
<DocScrutinizer05>
possible future rant in fab: "OK, we used up that reel of RGB side looking LEDs. There's nothing else available that we can put there. So from now on the devices ship without kbd backlight"
<ds2>
polished your crystal ball yet? :D
<DocScrutinizer05>
just preparing for next campaign "secure your Neo900 risk components"
<ds2>
time to teach people the phrase "Life Time Buy"
<freemangordon>
jusa_: any idea how to test xprot? safe and sane that is
<freemangordon>
so far the outputs from stock and REed match
<freemangordon>
but my concern is that some code paths are only taken under circumstances I am not sure I can make
<jusa_>
don't know really, sorry
<freemangordon>
ok
<DocScrutinizer05>
freemangordon: emulate temperature >60°C
<freemangordon>
it is not that simple
<DocScrutinizer05>
I know, testframes often are more complex tan the tested module
<freemangordon>
the actions taken depend not only to bme temp, but on some prediction for coil temp
<DocScrutinizer05>
quite probably, yes
<freemangordon>
unfortunately I don;t know when the limiter should start behaving
<DocScrutinizer05>
you should see when you compare XPROT input to output ;-)
<freemangordon>
I compared stock vs REed function by function, with the same test raw file as foe -music and -record
<freemangordon>
*for
<DocScrutinizer05>
it might simply not be loud enough to make XPROT interfere
<freemangordon>
and that is my problem - what is "loud enough"?
<freemangordon>
lots of low freqs?
<DocScrutinizer05>
and I wonder where the heck the developers got the whole parameters from, since... who knows how much thermal capacity a Knowles TINY coil has, how hot it mey get, how's the thermal coupling factor to environment that cools coil down, etc pp
<freemangordon>
that one too :)
<freemangordon>
I guess they measured it
<freemangordon>
though... it is not exactly simple to measure a coil temp of a speaker
<DocScrutinizer05>
I think a pink noise bursts with 80% on time, -3dB, and some 1 to 5 bursts per second should give a very interesting output signal after XPROT
<DocScrutinizer05>
sox should be capable of producing such data file for you, easily
<DocScrutinizer05>
use at least 30s of audio for the test
<DocScrutinizer05>
it may take a while until XPROT thinks the transducer heated up
<DocScrutinizer05>
analyzing amplitude and spectrum of the output will tell you a lot about what XPROT actually does
<DocScrutinizer05>
make sure you note the ambient temperature XPROT gets from vme!
<DocScrutinizer05>
bme*
<freemangordon>
I am using test program which can feed whatever ambient temp I want
<DocScrutinizer05>
cool
<DocScrutinizer05>
that's the way :-)
<freemangordon>
will push it to gitorious shortly
<freemangordon>
I know
<freemangordon>
:)
<DocScrutinizer05>
seems you're not exactly a noob, eh? ;-)
<DocScrutinizer05>
Oksana: yes, but seekic is usually fake
<DocScrutinizer05>
they quote completely made up prices and availabilities of at least 10k/week, while when you ask they tell you "sorry we don't have this component on stock" and sometimes offer to try and source it for you
<Oksana>
:-) Will research further. It turns out that the wires that arrived are AC-5, not AC-8. Not much difference between 800mA and 890mA, but EBay has to be warned about such deliberate mis-advertising.
<DocScrutinizer05>
Oksana: as a rule of thumb, consider us being semi-professional when searching for component sources. There's hardly anything you can find in google which we didn't check yet. So unless you know some special thing about some provider that only offers parts to registered users, odds are you won't tell us sth new
<DocScrutinizer05>
of course you're free to try stuff like seekic and maybe you're lucky, we usually didn't even try those sources
<DocScrutinizer05>
I could post an example convo with a "source" like seekic, where first thing was some Lilian Wang or whomever asked me to share details about our project, then asked me if there maybe are other components as well that we need for such project, and then explained they cannot provide the PoP RAM but sent 20 sequel mails asking if I would like to source OMAP or Y or Z from them
<DocScrutinizer05>
took me 4 weeks to brush that noise out of my daily mail again, until she finally stopped spamming me
che1 has joined #neo900
<Oksana>
How reasonable-realistic is this description? Model: KCE00E00CA-A506 Brand: Samsung Parameters: 4Gb (256M x16) Mux OneNAND Flash + 8Gb QDP (128M x32 + 128M x32) 2 / CS, 2CKE Mobile DDR SDRAM
<DocScrutinizer05>
100% accurate
<DocScrutinizer05>
even has the fact that OMAP3 can address >512 MB RAM only by using 2nd CS
<Oksana>
Yes. Where else would they be made? Even Nokia's battery-charging-dock has "made in China" on the box.
<DocScrutinizer05>
sure, they are made by Samsung, in whatever fab they use for this
<DocScrutinizer05>
doesn't mean the website of the shop selling them has to be in chinese though ;-)
<DocScrutinizer05>
generally sourcing components in Asia is very risky business
<DocScrutinizer05>
see recent FTDI mess
<DocScrutinizer05>
and that's only a rare case where the chips were fake/counterfeited and still did work. More common are thinks like selling 1GB uSD as 32GB, with a simple cheap relabeling done on some b-rating 1GB noname uSDs
<DocScrutinizer05>
or simply using a similar looking completely different chip and relabel it. Possobly a case with no chip at all inside, etc
<DocScrutinizer05>
b-grade stuff sold as "totally new excellent quality A-grade!!!" is almost as good as it gets
<DocScrutinizer05>
when you're lucky
<DocScrutinizer05>
of course there are some 50% .. 99% non-scam sellers in asia too. But it's absolutely impossible to tell apart for us westerners who don't live there and may have a phonecall to a friend living same district as the shop selling the devices, so such friend could give that shop a visit and check if they are rogue or serious
<Oksana>
Ok, will ask them about available quantity. Why C8, if the model number is KCE00E00CA-A506 ?
<DocScrutinizer05>
huh?
<Oksana>
Last link from me: Quantity Available: 160000(contact us to confirm the available quantity) | Are you saying that there should be C8 suffix attached somewhere to KCE00E00CA-A506 ?
<DocScrutinizer05>
that CA in KCE00E00CA-A506 is not related to the general explanation of meaning of "package suffix" I'd guess
<DocScrutinizer05>
111MHz bus speed is ridiculous
<DocScrutinizer05>
well, or it's 111MHz but double bus width, making for a 222MHz netto
<DocScrutinizer05>
Oksana: we have no datasheet for that component, all we know is that it's used on top of a OMAP36xx processor in Nokia N9
<DocScrutinizer05>
so maybe rather hunt for datasheet first? :-)
<Oksana>
Google detects the language as Haitian Creole :-/ Have to point out to Google, manually, that the language is Chinese. | Trying to hunt for semblance of datasheet, yes. :-)
<DocScrutinizer05>
actually that page you have there looks pretty good, they offer a datasheet, though under "contact us"
<DocScrutinizer05>
then OTOH "Contact: winwin-ke@hotmail" WUT? HOTMAIL?
<nox->
i wonder what email services the gread chinese firewall blocks...
<nox->
-d+t
<nox->
maybe it was the least of evel or so
<nox->
evil even
<DocScrutinizer05>
evidently *@hotmail is no valid email at all, lacks top level domain
<DocScrutinizer05>
"FOB Price: US $1 - 30 / Piece; Min.Order Quantity: 20 Piece/Pieces; Supply Ability: 160000 Pack/Packs per Day" is exactly the usual scam found on alibaba when somebody offering components that they at best have seen a datasheet once, or just found it offered on some competitor's website, and that competitor again never sold those chips but rather... see a)
jonwil has joined #neo900
<Oksana>
The company is a middle-man between manufacturers in China and wider world. They do not have up-to-date information aggregated about their sources, that's why they say: contact us to confirm.
<DocScrutinizer05>
those sellers effectively killed any business with china, since it's impossible to find the 2 or 3 sellers who actually have the component on stock
<DocScrutinizer05>
Oksana: please see my above story about my contact to such "middleman"
<Oksana>
Yep. They can be quite persistent :-/
<DocScrutinizer05>
those "shops" offer verything, even components that never existed, just to make contact to prospect customers
<DocScrutinizer05>
everything*
<DocScrutinizer05>
and I guess they have absolutely zilch on stock
<DocScrutinizer05>
that's one of the most valuable know how of big companies like Nokia that their component sourcing department knows the 5 or 10 companies that are 'real'
<DocScrutinizer05>
I bet when you do sufficient amount of sophisitcated searching, you can find original sources of every single photo on that website
<DocScrutinizer05>
and none of them photos showing the winwin company
<Oksana>
Maybe. They probably do not see the chips they sell. They are organising the buyer-manufacturer link, the postage, and such. Relying on manufacturer's stock, not having their own stock. | I am hypothesising on the basis that Alibaba did somehow verify the existence and reliability of the company.
<DocScrutinizer05>
"she" will ask you for your mail address and then offer you a lot of stuff you never asked for, and when you're lucky then after a week they found out that they cannot source KCE00E00CA-A506
<Oksana>
What is NFC, Norwegian Forest cat or Near field communication? :-) | Thanks. Wish Samsung had datasheet available... I cannot even find KCE series. Considered obsolete? Then what are the compatible descendants?
<DocScrutinizer05>
~wtf nfc
<infobot>
NFC: no fucking clue
<DocScrutinizer05>
what is | ?
<DocScrutinizer05>
KCE00E00CA has a quote strange numbering that might indicate that it is a custom made product. also it seems Samsung doesn't build any FBGA168 MCP any more
<DocScrutinizer05>
so there simply is no descendant/successor
<Oksana>
"Break line" between short messages when they can be fit into one line, but are logically unrelated to each other. My IRC client does not understand meaning of "compact" aka "fit as much info onto the screen as possible". | For a custom product, it is strangely well-known.
<DocScrutinizer05>
OMAP3 (FBGA168) is obsolete
<DocScrutinizer05>
seems there been Samsung KA series, but no KC series