Topic for #qi-hardware is now Copyleft hardware - http://qi-hardware.com | hardware hackers join here to discuss Ben NanoNote, atben / atusb 802.15.4 wireless, and other community driven hw projects | public logging at http://en.qi-hardware.com/irclogs
<wolfspra1l>
good morning everybody :-)
<Ayla>
good morning
xiangfu has joined #qi-hardware
<wolfspra1l>
Ayla: still there? :-)
<wolfspra1l>
what's up lately? are you still supporting the Dingoo? any exciting news there?
<Ayla>
yes, still here
<Ayla>
and yes, there are exciting news
<Ayla>
like, all the time
<wolfspra1l>
good
<wolfspra1l>
shoot! ;-)
<Ayla>
my bootloader is now able to boot linux on Dingoo from the NAND
<Ayla>
(formatted with UBI)
<wolfspra1l>
great!
<wolfspra1l>
sounds to me like that was possible for a while on the NanoNote - were you able to look at or reuse some codes?
<Ayla>
actually I used a lot of code from booboo's hwinit
<kristianpaul>
morning
<kristianpaul>
or hi :)
<wolfspra1l>
Ayla: what it booboo's hwinit? url?
<Ayla>
it's a program which inits the hardware
<wolfspra1l>
kristianpaul: how's GPS? how about the switch to Maxim?
<wolfspra1l>
I would like to understand it - have you decided to move over to a Maxim chip?
<wolfspra1l>
if so - let's do it! let's make GPS real... :-)
<Ayla>
it probably works on NanoNote too, who knows
<kristianpaul>
Maxim, not swtiched yet but is on the plan
<Ayla>
the second improvement I made lately, is the ability to change the ILI93xx panel at runtime through sysfs
<wolfspra1l>
you mean disconnect and reconnect?
<wolfspra1l>
kristianpaul: does that mean the Maxim fully replaces need for the SiGe chip?
<wolfspra1l>
so we all just focus on Maxim going forward?
<kristianpaul>
yes it evetually replace sige chip
<Ayla>
wolfspra1l, in software, yes
<xiangfu>
Ayla, yes. u-boot support load kernel from UBI under nanonote. but by default still load kernel from raw nand data.
<kristianpaul>
and move forward, yes, comapre later
<kristianpaul>
compare*
<wolfspra1l>
don't understand. you mean compare the two?
<wolfspra1l>
but it seems you already like the Maxim better?
<Ayla>
xiangfu: is that recent? IIRC last time I looked uboot wasn't able to load a kernel from UBI
<Ayla>
only jffs2 or raw
<wolfspra1l>
I have no problem at all in switching, not much to 'switch' anyway, just want to support you and whatever you like better is what I will go for.
<kristianpaul>
yes sure, the comparison is not a priority
<xiangfu>
Ayla, from 'Image 2011-02-23'
<wolfspra1l>
I somehow feel the urge to make a working product though that includes GPS :-)
<wolfspra1l>
and it seems thanks to the work of kristianpaul artyom etc we can soon be crazy enough to try!? :-)
<xiangfu>
"F4 will load kernel inside the ubi rootfs of nand.(we need put kernel at /boot/uImage)" but it's very slow compare to raw nand data
<kristianpaul>
and i like maxim, yes, its avaliable documentation is better than the non-public one i got for sige
<wolfspra1l>
perfect
<wolfspra1l>
that's a big deal for us
<kristianpaul>
wolfspra1l: also the Takuji Ebinuma works
<Ayla>
xiangfu: great. Then I don't feel bad about creating a new bootloader :)
<wolfspra1l>
who is that?
<kristianpaul>
he ported namutu to its nios II soc ;)
<kristianpaul>
he was trying milkymist, dunno was happened..
<kristianpaul>
or may be i tought i was.. dont remenber that at all..
<wolfspra1l>
wow, so great we get the name from you! thanks!
<kristianpaul>
and he uses maxim too, i think we have too many confirmations already ;)
<wolfspra1l>
I think I want to make an extra effort to document better where stuff comes from, from whom we use what, etc.
<kristianpaul>
he yes currently bit messy
<wolfspra1l>
yes, the reason for sige at the beginning was simply that this 1 guy who promised to go open but since then seems to have disappeared said he liked sige better than maxim
<wolfspra1l>
but since he also seems to like closedness better than openess, let's all move to the happy open maxim camp and have fun
<wolfspra1l>
and done
<wolfspra1l>
:-)
<kristianpaul>
:-)
<wolfspra1l>
Takuji Ebinuma... he knows about Milkymist SoC already - good
<wolfspra1l>
maybe it's interesting for him, maybe not
<kristianpaul>
i cannot get he proper feedback for him
<kristianpaul>
i'll try later what happened
<wolfspra1l>
Ayla: I think it's safe to say that *nothing* can match the experience of doing your own thing, in this case a UBI bootloader
<wolfspra1l>
is he living in Japan?
<wolfspra1l>
how's his English?
<kristianpaul>
dunno
<Ayla>
yes, definitely
wej has joined #qi-hardware
<wolfspra1l>
sometimes I wonder whether it makes sense to write in English and include a copy/pasted Japanese snippet out of google translate under it :-)
<wolfspra1l>
Ayla: what do you use your Dingoo for mostly?
<kristianpaul>
his englihs looks good to me
<kristianpaul>
english*
<kristianpaul>
i had crossed several mails with Takuji Ebinuma
<Ayla>
wolfspra1l: coding stuff on it, almost exclusively
<kristianpaul>
he also talked about a source code published... still now news ;)
<wolfspra1l>
nice! anything we can use on the Ben as well?
<wolfspra1l>
yeah, I can imagine
<wolfspra1l>
I know little about Japan, but if it's anything close to China, there are probably massively other cultural misunderstandings than a simple 'translation' of text manages to overcome
<wolfspra1l>
we have to live with that and try to understand each other as much as that is possible
<Ayla>
wolfspra1l: you already do use them :)
<wolfspra1l>
and slowly it will get better :-)
<wolfspra1l>
Ayla: oh :-)
<Ayla>
qi-kernel, gmenu2x
<wolfspra1l>
well of course
<kristianpaul>
wolfspra1l: are you suscribed to the osmo-sdr list?
<wolfspra1l>
no I don't think so
<wolfspra1l>
lemme see :-)
<wolfspra1l>
if the traffic is not too high I will add it
<kristianpaul>
is not too hihg and when get mails very interesting topics
<wolfspra1l>
I can imagine, it's a wonderful project
<wolfspra1l>
subscribed :-) thanks!
<kristianpaul>
:)
<wolfspra1l>
do you have the Maxim chip already?
<kristianpaul>
yes
<wolfspra1l>
anything I can do to help?
<kristianpaul>
not yet :)
<wolfspra1l>
alright... standby then
<wolfspra1l>
we should do something mobile, or with a simple battery-powered solution
<wolfspra1l>
like one of those USB phone recharger thingies
<wolfspra1l>
I continue to be amazed at how little power my entire m1 setup needs
<wolfspra1l>
the m1 itself - 5W
<wolfspra1l>
my laser projector - 5W
<wolfspra1l>
camera - 1.2W
<wolfspra1l>
wifi dongle - 0.5W
<kristianpaul>
I like Takuji Ebinuma's maxin receiver board, will be nice manufacture some
<kristianpaul>
afaik not very public details about it, may be you can write him?
<kristianpaul>
is 5W battery-powered ready? i mean worth to battery
<wolfspra1l>
tell me more
<wolfspra1l>
url?
<wolfspra1l>
well, the laser projector ships with a battery ;-)
<kristianpaul>
oh
<kristianpaul>
how long last?
<wolfspra1l>
but of course battery will still run down quickly
<wolfspra1l>
1-1.5h maybe
<wolfspra1l>
but wait, that's not my point
<kristianpaul>
ah
<kristianpaul>
i'm lost
<wolfspra1l>
power consumption will continue to go down
<wolfspra1l>
so let's say I want to put all this stuff into a portable box, and put some batteries into that box as well
<wolfspra1l>
I can lug 10 AA batteries around, no problem :-)
<wolfspra1l>
right now we are a little unflexible on the power side
<wolfspra1l>
we ship m1 with that switching power supply and that's it
<wolfspra1l>
that is fine today, but how about tomorrow?
<wolfspra1l>
M1 needs 5V, that's good. can probably hook it up to one of those USB phone recharger battery packs - I should try
<wolfspra1l>
not for a real application, just for fun now :-)
<kristianpaul>
ah ok
<kristianpaul>
i tought was for real, my bad ;)
<wolfspra1l>
is that a blog you think we should add to the Qi planet?
<wolfspra1l>
checking...
<wolfspra1l>
sure for real, but as an experiment now
<wolfspra1l>
has to start somewhere...
<wolfspra1l>
he, I see a URL on the picture there :-)
<wolfspra1l>
oshugnss.blogspot.com
<wolfspra1l>
firing up my freedom network... :-)
<kristianpaul>
but blog dont load
<wolfspra1l>
what do you mean with "tell mroe"?
<kristianpaul>
anyway, my point is we should have a maxin board for gps receiver
<wolfspra1l>
yeah, indeed
<wolfspra1l>
how strange is that
<wolfspra1l>
well as I told you :-)
<wolfspra1l>
the culture...
<kristianpaul>
ok
<wolfspra1l>
of someone goes to the great length of silkscreening a url on a pcb, you would think they register (for free) the corresponding blog? :-)
<wolfspra1l>
s/of/if/
<qi-bot>
wolfspra1l meant: "if someone goes to the great length if silkscreening a url on a pcb, you would think they register (for free) the corresponding blog? :-)"
<wolfspra1l>
if of - I think it's clear
<wolfspra1l>
maybe a typo?
<wolfspra1l>
what are you looking for exactly?
<wolfspra1l>
electrical design / bom ?
<kristianpaul>
yes
<wolfspra1l>
looks very simple, can even tell half of it from the picture :-)
<kristianpaul>
we already have Artyom's board
<kristianpaul>
is kicad
<kristianpaul>
not sure bom
<kristianpaul>
but have extra components we dont need
<wolfspra1l>
sure sure, I am not trying to 'take' something, but rather make a contact with this guy and see what we can learn, understand his work, etc.
<kristianpaul>
oh sure !
<kristianpaul>
DocScrutinizer, said somethim worry me a lot as my ignorance on this field
<kristianpaul>
and is th interference generated from M1
<kristianpaul>
that can affect the receiver
<kristianpaul>
if we put the antenna near to it
<kristianpaul>
wich is not what happen right now..
<kristianpaul>
did i explain my selft correctly?
<kristianpaul>
xD
wolfspraul has joined #qi-hardware
<wolfspraul>
kristianpaul: sorry, ran my battery into the ground
<kristianpaul>
no problem
<wolfspraul>
can you tell me more about the Japanese guy - you emailed him?
<kristianpaul>
he first :)
<wolfspraul>
should we add his blog to the Qi planet?
<kristianpaul>
if you dont mind perhaps, but i feel it lacks a bit more sources
<kristianpaul>
it just tell, okay i'm doing this, or looks that..
<kristianpaul>
but..
<wolfspraul>
ok but maybe only some small encouragement is missing
<kristianpaul>
but blog is good
<kristianpaul>
i mean that as a improvement
<wolfspraul>
he may not realize that he has avid readers
<wolfspraul>
and that dead url on the silkscreen... maybe the stuff is somewhere and we just don't find it
<kristianpaul>
perhaps blog is more local, for universiry collegues? etc..
<kristianpaul>
you read my lines about my concern about M1 and interference ?
<wolfspraul>
no
<wolfspraul>
just now?
<kristianpaul>
DocScrutinizer, mentioned time ago
<wolfspraul>
interference are good problems to work on
<kristianpaul>
i undertood, if have gps atenna near M1 it will lead problems
<wolfspraul>
since we want to make stuff others can use, not a fake prototype/one-off thing with lots of real life bugs
<kristianpaul>
that dint happen right now
<wolfspraul>
oh I'm sure. lots of problems :-)
<wolfspraul>
but we have hope, we have Joerg? :-)
<kristianpaul>
:)
<wolfspraul>
just kidding - eventually we find out the missing pieces to make it work. we are trying to make the best of high-performance open hw, right?
<wolfspraul>
and we need to search the entire world for the very best pieces of wisdom, experience, design, contributors, feedback, what not
<wolfspraul>
it won't be easy
<wolfspraul>
but look how many other good people in that area you already found, I think we are not alone
<kristianpaul>
oh no
<wolfspraul>
that's the whole point of collaborative hw
<wolfspraul>
GPS signals are very weak, the antenna thing will be tough for sure
<wolfspraul>
I try to ignore it for now and put all my hope on finding people smarter than me to help
<kristianpaul>
Okay ;-)
<kristianpaul>
so thats all, in the mean time i try document and organice all i learn about this
<kristianpaul>
and make it work of course :)
<kristianpaul>
osgps rtems port ;)
<kristianpaul>
what remains for a fix
urandom__ has joined #qi-hardware
<wolfspraul>
cool
<wolfspraul>
still googling for Takuji Ebinuma's stuff
<wolfspraul>
his project is what? to make this board in the picture?
<wolfspraul>
does it have a name?
<kristianpaul>
dont know
<wolfspraul>
maybe osqzss?
<wolfspraul>
:-)
<wolfspraul>
or qzss. how do you read his blog? google translate?
<kristianpaul>
yes i translate by google
<wolfspraul>
it seems all in Japanese
<wolfspraul>
yeah
<wolfspraul>
for the Qi planet, we should have a feature to pull things through an auto-translater
<wolfspraul>
kristianpaul: well, thanks a lot for bringing Takuji's work up here
<kristianpaul>
the least i can do ;)
<wolfspraul>
it's perfect. we need to search around more for great open-minded projects to join.
<wolfspraul>
and you already ran into a whole bunch over time :-) cool!
<qi-bot>
Jon Phillips: Phones, Tablets, Computers are over already. Apple won. The show is over. @qihardware + @milkymistvj are post-apple. Let's run! ( 177650736413417472@rejon - 31s ago via Ping.fm )
<wpwrak>
but that didn't go beyond early experiments (i.e., until the point where i realized why it would get complicated :)
<viric>
ok
<wpwrak>
my latest idea is to use fped to produce a series of 2D drawings of "slices" of the 3D object, then construct a 3D object or directly a toolpath from them
<wpwrak>
but for this to be practical, fped needs the ability to reuse points in different frames. right now, points don't cross frames
<wpwrak>
this is still work in progress
<roh>
i think you are reinventing the wheel there. with a ui nobody besides you can use
<wpwrak>
(not too happy with either) openscad is a bit more powerful than cadmium but the resulting meshes were full of problems, particularly if i used subtractive models. cadmium got that right but only had the most primitive of primitives.
<roh>
wpwrak: well. just more reason to fix openscad i think
<wpwrak>
roh: give me a better parametric cad and i'll gladly switch ;-)
<roh>
wpwrak: you know that autocad stated that way and still can do parametrics?
<wpwrak>
roh: besides, fped is used by others (for footprints)
<roh>
fped is really poor compared to any 2d cad i know
LunaVorax has joined #qi-hardware
<roh>
no harm intended
<wpwrak>
well, it's not designed to be a full cad system :)
<roh>
its a footprint editor, and it may well be a good one, but for 2d cad i cannot even imagine breaking my legs that hard
<roh>
even the utterly broken qcad2.x does a quite well job for 2d cad.
<wpwrak>
oh, try it. it's not that bad :) you'll appreciate its capabilities when you need to adjust some parameters
<roh>
if you can make something which extrudes arbitrary dxf stuff and lets me assemble that, that would help
<roh>
i couldnt get openscad to extrude dxf for me yet
<wpwrak>
one weakness of fped is that it has no geometrical construction besides concatenation of vectors. for "real" cad, you often need a few more operations. and they're messy to do in fped.
<wpwrak>
(extrude dxf) that sounds like my 2D slices
<roh>
i tried extruding dxf with openscad, but it failed on my parts
<wpwrak>
;-)
<roh>
e.g. if i could make one dxf per sidepanel for the mm1 case, i could extrude and 'assemble' them in only a few lines
<DocScrutinizer51>
moo
<roh>
doing the dxf is easy. but openscad has some problems with the parts
<wpwrak>
roh: post the files and ask for help on the openscad list ?
<wpwrak>
or maybe try heekscad ?
<roh>
wpwrak: well.. just use the file in svn, cut everything away besides 'one piece' and save it as new file.
<wpwrak>
not sure if it has extrusion from dxf, though. but it might.
<roh>
then try extruding that in 3mm
<wpwrak>
yes, i understand what you want to do. but you need to explain that to the openscad developers. it's them who might fix it for you :)
<roh>
currently i am not sure if the sw is broken, the manual is wrong or i am too stupid
<roh>
that one doesnt work for me (like loads of stuff in the 'manual'
<wpwrak>
that's what the openscad list is for ;-) i don't think anyone here knows openscad well enough to be able to help you
<wpwrak>
and yes, i think the 2D extrusion approach holds promise
<roh>
well.. only for such simple stuff
<wpwrak>
btw, regarding fped, it's also nice to be able to do things with immediate graphical representation. i do almost all my drawing in the gui, not in the script
<roh>
maybe as a step on the way. but on the long run we need a proper fullfeatured 3d cad
<wpwrak>
yeah. and a parametric one at that
<roh>
currently we are really limiting our doing hard by using bad and broken tools
<wpwrak>
last time i checked, heekscad also has the issue of scripts having only access to a small subset of features
<roh>
heeks is b asically dead
<wpwrak>
is it ?
<roh>
yes. the maintainer said so and most of the people left long ago to other projects.
<wpwrak>
oh, i see
<roh>
its diced for code reuse, but i am happy that nobody wants the ui
<wpwrak>
pity. it looked quite active not too long ago
<wpwrak>
;-)
<roh>
the ui was unbearable.
<roh>
done by a moron. i mean.. come on.. using a self-built treeview in opengl because due to the use of wxwindows he couldnt use the gtktreeview?
<roh>
which means every time you klick ANYTHING in there it folds up again
<roh>
no keybinding work also.
<wpwrak>
i wouldn't call it "unbearable" :) that ws more my feeling with blender :)
<roh>
in short: a nice example why not to reinvent the wheel when it comes to ui toolkits. you will only fuck up.
<wpwrak>
i found the lack of scriptability more worrying
<roh>
blender is a artists tool. its not intended for mechanical work
<wpwrak>
e.g., how do i go from a 3D model to a toolpath ? click and type a lot ...
<roh>
that you can not avoid. toolpathes cannot be done fully automatic
wej has joined #qi-hardware
<wpwrak>
(blender) surreal-masochists ... :)
<roh>
'artists'
<roh>
;)
<wpwrak>
i'll gladly manually do every step of the toolpath setup the first time. but then, i don't want to do the damn thing all over again and again and again for each run
<roh>
true. that one should be possible
<wpwrak>
for 2D, i have at least my makefiles for it. i may be the only one who understands how these things works, but they do ;-)
<roh>
regenerating the toolpath after changes should be no problem. that worked even in heeks
<wpwrak>
as far as i recall it didn't save the tooling parameters
<wpwrak>
ah wait, what didn't work was combining that with model generation from a script
<wpwrak>
so maybe you could keep the toolpath parameters but you'd lose parametric modeling
<roh>
yeah. there were inconsistencies between the scripting if and the ui. but that was the result of uncoordinated hacking by different people
<roh>
i think you give 'too much' about the parametric part of it. sure its nice, but in the end it doesnt help a lot
<wpwrak>
anyway, so where did the heekscad people go ? in particular, where did they take the heekscnc engine ?
<wpwrak>
i find parametric extremely important
<roh>
tolerances alone will eat you alive if you think that parametric only will save you
<roh>
i think some went to the pycam and pycad stuff. some to freecad.. not so sure
<wpwrak>
parametric is what lets me compensate for machining tolerances ...
<roh>
no. tolerances are EXTRA in real cad.
<roh>
thats a propery of material
<wpwrak>
and the machine and so on
<roh>
exactly. and in the end, somebody will want some font differently or placed different.... so 'just moving it' is much easier
<wpwrak>
but also the model often needs to accommodate things like clearances. they're often hard to predict precisely. even if the machining is "perfect"
<roh>
there is nothing perfect, even in a cnc world. its always just approximation which will show its quality on first machining
<wpwrak>
you didn't "just move" the holes in the M1 case. i know that you didn't - i reverse-engineered the underlying parametric model ;-)
<roh>
on the m1 case nothing was parametric. its completely done by hand in qcad opensource edition
<wpwrak>
i think it's not harder to make or use a parametric cad than a non-parametric one
<roh>
just because something isnt parametric it doesnt mean i cannot select and group n things and then move em controlled by numbers
<wpwrak>
qcad is anti-parametric. but you had a parametric model in your head when you did the thing
<wpwrak>
yes, and gro(u)ping and moving loses the history. so you no longer know how they ended up where they are
<roh>
nah. not really. i constructed it from measurements and then fixed it in like 6 tries while using a real machine and cheap wood
<wpwrak>
and most 3D cad also completely loses the constructive history. intersect A with B and poof, you get a C. now go and change something in A ...
<roh>
eh.. well.. i keep my dxf files in svn.. so i got a history
<roh>
classic problem. answer is simple: it would b ecome to many objects to keep for real complex models
<wpwrak>
(from measurement) yes. i didn't say it was a complex model :) still, with a little bit of abstraction, i could do things like move connectors around for one-off board variations
wej has joined #qi-hardware
<roh>
i know of bmw card models in cad where they modeled the surfaces of knobs and upholstery and even the wire inside of lightbulbs.
<wpwrak>
yeah, sure. the world is soooo complex. let's close our eyes and stay in bed ;-)
<roh>
so you got gazillion of parts in the end. which consist of even more origin parts
<wpwrak>
(bmw) nice :)
<roh>
fix your workflow and you dont need 'all source objects' in your model
<roh>
csg is nice for simple stuff. but in the end its all triangles anyhow
<wpwrak>
i like to experiment. that means that i often have to go back a few steps. without a proper parametric/constructive cad, that means deleting everything and doing it again. possibly lots of times.
<roh>
thats what components are for.
<wpwrak>
and yes, i realize that a lot of people don't have a problem with doing things even a machine would find a little boring
<wpwrak>
you still have the "redo until it fits" problem inside the components
<wpwrak>
and if the components become part of constructive geometry, you still have the same issue if you have to change a component
DocScrutinizer has joined #qi-hardware
<roh>
wpwrak: if your tolerances are that small that things do not fit, even in cad, your concept is likely to be fucked
<roh>
reality is a bitch. make enough room for it to breathe
<wpwrak>
i was thinking of things fitting in real life
<wpwrak>
or feeling too loose, etc.
<wpwrak>
i don't have decades of experience in 3D design. so i make a best guess and then improve my model or my tolerances based on it
<roh>
its always trial and error. thats why rapid prototyping helps so much
<wpwrak>
it's subtle things. like whether a panel feels crowded. whether it's prone to operator error to inefficient placing of elemenst. whether connectors are places such that i can plug in things easily
<wpwrak>
and for trial and error i want parametric. qed :)
<wpwrak>
parametric basically lets you record the construction history. that way, it tells you - to some extent - why things are where they are. not just where they are.
<wpwrak>
and a good parametric cad would also now make you waste a lot of time with converting your construction steps into something the cad can process
<wpwrak>
that would also go against the concept - you don't want to hide the forest behind a lot of trees
<roh>
wpwrak: well.. show me a single working 3d cad which doesnt break after a few undo first
<wpwrak>
yeah, it starts even there :)
<wpwrak>
of course, with parametric design, you don't need undo all that often :)
<wpwrak>
e.g., fped has undelete (delete maintains referential integrity, so it can do quite a bit of damage if you're careless), but that's all
<qi-bot>
[commit] Paul Cercueil: MIPS: JZ4740: Drop support for LG LGDP4551 and SPFD5420A Smart LCD panels. (jz-3.2) http://qi-hw.com/p/qi-kernel/d42e289
<qi-bot>
[commit] Paul Cercueil: MIPS: JZ4740: use standard module_param() macro for the 'panel' parameter. (jz-3.2) http://qi-hw.com/p/qi-kernel/1a7fb60