<kristianpaul>
woakas: do you think is proper to "bomb" mico32 processor with  512K interrupts per second?
<kristianpaul>
wpwrak: ^
<kristianpaul>
arggh, i'll disable this autothing..
<kristianpaul>
i implemented the method you suguested loong time ago,
<kristianpaul>
a circular buffer and a pointer that show me the current writing address
<kristianpaul>
overflow is not managed..
<kristianpaul>
buffer is just 2048 x 32bits
<kristianpaul>
word
<kristianpaul>
sige receiver sampling rate is 2.048~ MSPS, but i'm writing to buffer every 4 samples so i can be more efficiente with data bus wide
<kristianpaul>
but i noticed a posible problem, i was reading/polling from this buffer without feedback and noticed i was reading same data more than 20 times..
<kristianpaul>
hmm i need do some drawing here, minute
<wpwrak>
kristianpaul: (512 kInterrupts/s) you can probably bombard it - it just won't do very much ;-)
<wpwrak>
kristianpaul: i think you either need some data reduction in the path, DMA, or just make a loop that polls. even polling for "data ready" may be too slow, but you could probably add a "data not ready" signal that just stalls the core.
<kristianpaul>
yeah i was thinking add that "data not ready" signal
<kristianpaul>
and resume sampling signal
<wpwrak>
kristianpaul: you could get an estimate for how fast you can go by writing a loop that just reads some peripheral register a number of times.
<wpwrak>
kristianpaul: e.g., uint8_t buf[1000000]; for (i = 0; i != 100; i++) for (j = 0; j != sizeof(buf); j++) buf[j] = REGISTER;
<wpwrak>
then time how long it takes
<kristianpaul>
i thik time command in rtems is kinda no what is in linux..
<kristianpaul>
how i'm going to measire time..
<kristianpaul>
i dont have rtc..
<wpwrak>
do you have a watch ?
<kristianpaul>
what kind of watch? ;)
<wpwrak>
one that shows seconds :)
<kristianpaul>
yes i do wear one
<wpwrak>
that loop should take 100 s if your max sample rate is 1 MSa/s, 10 s if it's 10 MSa/s, etc. and you can always crank up the end value for i to add more precision to your measurement :)
<kristianpaul>
okay
<kristianpaul>
ha, wait i do have a clock, the one from sige chip is 4Mhz after a clock div i need for writing buffer
<kristianpaul>
okay i'll try wpwrak method
<wpwrak>
don't worry about the clock ;-) this is just to get an idea of how fast you can read a register. estimate that you'll need 2-3x the time for a loop of the type
<wpwrak>
for (i = 0; i != sizeof(buf); i++) { while (REG_BUSY); buf[i++] = REG_DATA; }
<wpwrak>
ah, and use -O9 ;-)
<wpwrak>
and make sure your register is declared volatile :)
<wpwrak>
gcc may still optimize away the store, but that's okay for now. we're just after an estimate.
<kristianpaul>
[register] volatile unsigned int *count_addr = (volatile unsigned int *)0xe0011004;
<kristianpaul>
yes
<kristianpaul>
the first loop with the buf[1000000]; took less than a second..
<wpwrak>
how many outer loops ? (i)
<kristianpaul>
100
<wpwrak>
wow. that's probably too good to be true.
<wpwrak>
that would mean that you can read a register and store away the value at > 100 MHz
<wpwrak>
well, it's not impossible. how long does it take for 1000 outer loops ?
<kristianpaul>
well, mm1 soc clock is 80Mhz.. mico32 is 6 stage pipeline.. and all soc uses same clock including memory remenber
<DocScrutinizer>
take your capacitive screens and ....
<kristianpaul>
wow, looks awesome
<kristianpaul>
(the part with the "pencil")
<DocScrutinizer>
now the *only* argument for c-ts is "glass"
<DocScrutinizer>
I don't need a blade-proof cocaine dispenser screen ;-)
<kristianpaul>
hum if you can correlate resistive measurements with some localizated vibrations on scree it could replace a kbd :-)
<DocScrutinizer>
there are projects to investigate this - along steam bubble buildup, electrofriction, piezo, ... duno what else
<kristianpaul>
Can we sample one for Ya?
<DocScrutinizer>
(sorry, reconnect)
<DocScrutinizer>
kristianpaul: dunno - never tried
<kristianpaul>
well, also wondering if worth the effort for a 320x240 lcm as currently ben chips, may be a N900-like size fits better for multitouch :-)
<wpwrak>
nice. now wait 20 years before it becomes affordable ...
<wpwrak>
or vote to abolish patents in your jurisdiction TODAY ! ;-)
<wpwrak>
i wonder if a country could sign a treaty with china to mutually ignore each other's patents. considering that everything gets made in china anyway, ... ;-)
<DocScrutinizer>
I wonder how to get a stantum-technology touchpanel from manufacturer $RANDOM
<kristianpaul>
wich jurisdiction?, the norh american and his pool of laws been transfered/force in every *free?* trade agrements... :S
<DocScrutinizer>
stantum headquarters: France Bordeaux
<DocScrutinizer>
not too far from here
<DocScrutinizer>
seems those are always built to order
<DocScrutinizer>
so no really free market :-/
<DocScrutinizer>
Stantum Asia Pacific: Taipei
<wpwrak>
kristianpaul: well, one of the first ladies expressed the only suitable policy for FTAs quite properly: "just say no" :)
<kristianpaul>
I agree !
<wpwrak>
DocScrutinizer: (not really a free market) let me rephrase this for you: cool new technology doesn't exist - unless you make it yourself
<wpwrak>
DocScrutinizer: that's why all the "cats" here don't help much ...
<DocScrutinizer>
cats?
<wpwrak>
(cats) from "look what the cat's brought in" (dead bird, mouse, etc.)
<DocScrutinizer>
mioww
<DocScrutinizer>
is missing the picture. Don't worry, it's me, as usual
<kristianpaul>
out of memory.. ah yes i forgot just 128Mb.. ;-)
<kristianpaul>
i need that ramdisk..
<kristianpaul>
xiangfu:Â Â !
<kristianpaul>
xiangfu: are you aware of the procedure for adding ramdisk to rtems?
<wpwrak>
(cats) people who find some "cool new product". then enthusiastically report it here, suggesting in one way or another "we" (i.e., someone else but them) should use it
<kristianpaul>
seems the current /ramdisk in flickernoise is just the name and not a separate filesystem
<whitequark>
what's wrong with free trade agreements?
<wpwrak>
whitequark: for one thing the bundle of IPR law requirements that usually come with them
<kristianpaul>
mianlly intellectual pooperty -_-
<xiangfu>
it is yaffs2. not separate
<DocScrutinizer>
lol
<kristianpaul>
xiangfu: yup,
<whitequark>
wpwrak: I don't quite get how IPR are related to the FTA's here...
<wpwrak>
whitequark: FTAs are contracts containing many items. some of these items are often requirements to adapt your legal system with regard to IPR to the one of the peer
<DocScrutinizer>
wpwrak: I still would like to investigate further on my age old plan to analyze HF pulse response of a dead simple 2 plane r-ts
<wpwrak>
whitequark: these things tend to be one-sided, e.g., colombia being forced to implement US IPR laws
<wpwrak>
DocScrutinizer: "just do it" ;-)
<whitequark>
wpwrak: thanks, I understand now
<DocScrutinizer>
what's been the appropriate phrase I used last time? "Where's my patent lawyer?!"
<DocScrutinizer>
1) measure wave speed of a r-ts
<wpwrak>
DocScrutinizer: no. "where do i post it ?" so that it solidly becomes prior art, and cannot be stolen from the public domain ;-)
<DocScrutinizer>
here, as you can see :-)
<wpwrak>
:)
<wpwrak>
so you want to find reflections, as the wave bounces off obstacles ?
<DocScrutinizer>
2) record response on plane 2 to a pulse in the sub-microsecond range injected to plane one
<DocScrutinizer>
3) compare pattern of that response for no touchpoint, 1 touchpoint, multitouch
<DocScrutinizer>
2a) record rsponse on all 4 wires of a standard touchscreen
<wpwrak>
hmm. sounds expensive. you need a very very fast ADC then.
<DocScrutinizer>
indeed
<DocScrutinizer>
modulo the expensive. Seems ultrafast ADCs are standard nowadays
<DocScrutinizer>
esp since you can do oversampling
<DocScrutinizer>
i.e. periodic signal & moving window
<DocScrutinizer>
so you only need a precise S&H, the ADC can be relatively slow
<wpwrak>
if you expect your wave to propagate with c = 200'000 km/s and you want a resolution of ~100 dpi (= ~4 dots/mm), you'd need a resolution of about one picosecond. oversample that ;-)
<DocScrutinizer>
C more like a tenth of that
<DocScrutinizer>
maybe less
<wpwrak>
10 ps resolution then
<DocScrutinizer>
I'm not convinced
<DocScrutinizer>
you can have cat5 cabletesters that tell a bend to the cm
<DocScrutinizer>
and probably that's just where the display stops as more resolution is pointless
<wpwrak>
they may work with interference patterns
<DocScrutinizer>
you got laser speedmeter guns
<DocScrutinizer>
those are definitely working with c=c
<wpwrak>
send wave, mix with return wave, rectify and low-pass filter, see if they cancel each other out
<DocScrutinizer>
we can do that?
<wpwrak>
laser measurement is usually digital, not analog. they're still not cheap, but a multi-GHz counter is easier to build than a multi-GHz ADC
<wpwrak>
(interference) yup, you could do that
<DocScrutinizer>
first find parameters of your system, the think about how to build the electronics to meet those requirements
<DocScrutinizer>
that's a large plane close to another (capacitor) and it has a really high series R
<DocScrutinizer>
the wave speed might get extremely low
<wpwrak>
try it: put a scope probe on each end, then send a pulse. look at the difference. if there's none, get a faster scope ;-)
<DocScrutinizer>
I'm more concerned about any wave passing thru at all ;-)
<DocScrutinizer>
the ersatzschaltbild looks like a really good low pass filter
<DocScrutinizer>
infinite number of series Rs, with an infinite number of Cs to ground
<DocScrutinizer>
(of course also the value of both R and C going to infinite low on this approximation)
<wpwrak>
well, simulate it then. qucs is your friend. 2-3 should be a good approximation for "infinite" :-)
<DocScrutinizer>
more like 10 to 20
<DocScrutinizer>
easy calculation: get the resistance of a plane, divide by 10. Get the distance beteen planes and the area, calculate capacitance, device by ten. Just good enough
<DocScrutinizer>
now just use an amp to feed back the response fromplane 2 back to plane 1 ;-)
<DocScrutinizer>
voila: working touchophon
<DocScrutinizer>
a pity I got no lab here anymore
<wpwrak>
DocScrutinizer: you could do the simulation, see if the results look promising, then ask on the list if there's anyone with suitable equipment within travel range
<DocScrutinizer>
sounds like a plan
<vladkorotnev>
hello everyone
<DocScrutinizer>
scrutinizes some of the disassembled frerunner touchpanels
<DocScrutinizer>
wonders whether to take useless nap or better get milk for a nice coffee
<wpwrak>
sleep refreshes the grey matter :)
<DocScrutinizer>
I'm serously doubting that
<wpwrak>
well, unless you're a zombie. then you have to eat that grey matter
<DocScrutinizer>
I guess a proper O2-tent would help better
<DocScrutinizer>
sleep is overrated
<lunavorax_frizzl>
DocScrutinizer, I wish it was
<DocScrutinizer>
plus I never feel better at waking up than I did when falling asleep
<DocScrutinizer>
briefly: I hate sleeping
<wpwrak>
"senile bettflucht" ? ;-)
<DocScrutinizer>
well, first get some supply, can sleep later anyway
<DocScrutinizer>
wpwrak: indeed
<kristianpaul>
pull-in match two prn satellites on view (acording gpredict) !! but no prn match... argh
<whitequark>
I'm sure that someone online knows enough SDRAM black magic to answer a simple question: is it possible to stack two chips one on another if I only have one DCS?
<whitequark>
the chips I have are 4 banks x 4 MBit x 16, and the CPU is 4725B/4755L
<xiangfu>
whitequark, the jzboot have included in last xburst-tools. there is a warning when create that package. no manual page for jzboot
<whitequark>
xiangfu: thanks for packaging, that's great for jzboot
<whitequark>
there's really no manual page, I should write one probably
<xiangfu>
I almost forget how I create this file :(, needs one command to debug I remember.
<whitequark>
oh... that looks quite perl'ish
<DocScrutinizer>
one CS per chip?
<DocScrutinizer>
should suffice, though I dunno how that will pan out nowadays with piggyback stacking storage, with all that narrow timing, and driver energy, and line ringing, and whatnot
<whitequark>
DocScrutinizer: no, that's the problem. one CS at all
<DocScrutinizer>
last time I did that was like 20 years ago
<DocScrutinizer>
well, I gather you got a CS pin on each chip
<whitequark>
the 4725B technically has DCS1, but it is not attached to a pin
<whitequark>
so I only have one CS
<DocScrutinizer>
you'll obviously need a 2->4 or 4->16 decoder to demux some higher order addr bits
<DocScrutinizer>
you can even get away with undemuxed addr ranges, like 01xx_xxxx for bank one, and 10xx_xxxx for bank 2
<whitequark>
hmm.
<DocScrutinizer>
but you really need to take care then about your addr ranges you use in your programs
<whitequark>
look, the DS on the memory chip says: A0-A12 are used for row address, and A0-A8 are used for column address
<DocScrutinizer>
mhm
<whitequark>
all the address lines wired to the chip are A0-A12, through I can theoretically use A13-A15 by unsoldering the flash
<DocScrutinizer>
and you got no chip select? only a address strobe?
<whitequark>
but I think that if I'd use high order address bits to select chips, the EMC will be confused and e.g. don't refresh them correctly, wouldn't it?
<DocScrutinizer>
whitequark: I can't answer these questions sufficiently correct, without having a look at the schematics of device and also the datasheets of chips
<whitequark>
and you probably have the one for 4725B already
<whitequark>
we here have a near-infinite supply of free 32MB SDRAM chips, and 64MB ones cost around $20
<whitequark>
given that a few people hacking the device ever used heatgun and will probably fry anything at the first attempts, it's not a very good idea to buy these
<whitequark>
there are $1.5 ones, of course. but they're BGA.
<DocScrutinizer>
use gates to distribute RAS/CAS to the two RAM chips
<whitequark>
so (keeping in mind that no one has a logic analyzer capable of debugging the chip or even scope) there's no simple way
<DocScrutinizer>
refresh afaik is using only one of those, and can go to both chips. you need to latch a higher order adrr bit of the other one to decide which of your storage chips gets the complete cycle of RAS+CAS of the next real r/w-operation
<DocScrutinizer>
not really
<DocScrutinizer>
modulo I haven't looked at any datasheet yet
<DocScrutinizer>
finally featuring this out to the gory detail takes like one manday
<whitequark>
okay, I won't ask you to explore this further, it's just not worth the result
<whitequark>
I was only hoping there is simple answer
<DocScrutinizer>
you need to check a lot of things, starting at fan-out of storage datalines and cpu data and addr lines, fan-in of all lines. Timing. Logic of RAS/CAS sequence, refresh scheme
<DocScrutinizer>
if you say you *could* connect a 64 storage chip, then probably best approach is to first compare the 32 and 64 and exactly check out the difference in operation mode. Then start from there 'emulating' the 64 chip with 2 32 plus additional gear
<DocScrutinizer>
have a short look at the fan-in/out situation, and then just give it a try
<whitequark>
I don't even know how to attach all those gates to the board
<DocScrutinizer>
hehe
<whitequark>
so that they'll work at 120MHz
<DocScrutinizer>
I'm famous for my "hedgehog" electronics
<whitequark>
wait a sec, I'll show you how I did the SD card connection...
<DocScrutinizer>
(120MHz) yeah, toldya I'm not sure those hacks will pan out nowadays at all
<DocScrutinizer>
it's for example a power supply for a z80 board (-12, +5, +12) built with discrete components like 2N3055 etc, completely without PCB, and about the form and size of a hedgehog, and also feeling like one
<whitequark>
oh. I can imagine that.
<whitequark>
I suppose that it was working. now that's impressive
<DocScrutinizer>
of course it worked, until somebody touched it and thus bend it to short some wires
<DocScrutinizer>
after each transport you had to "hairdress" it do remove all the shorts
<whitequark>
you should have embedded that in epoxy. like an insect in a piece of amber
<DocScrutinizer>
push a bit here, pull a bit there
<DocScrutinizer>
nice idea. That would've been ~2L though, and epoxy been rather expensive back when
<whitequark>
2L!!
<whitequark>
that's quite a big hedgehog
<DocScrutinizer>
toldy, size of a hedgehog
<DocScrutinizer>
well, I would have to use some semi-geometric form of any available box, or cup, or whatever
<DocScrutinizer>
that adds to the volume needed, as a ball isn't easy to form in epoxy
<whitequark>
hmm. can't you just use, eh, a ball?
<DocScrutinizer>
btw I'm not sure this would've worked at all, as one undeniable benefit of this "design" been a good ventilation
<whitequark>
so that was also a hot hedgehog
<DocScrutinizer>
indeed
<DocScrutinizer>
:-D
<DocScrutinizer>
and placing the thing into a ball would make "dressing" unfeasible, so I end up with a nicely encapsulated hedgehog that has shorts and doesn't work
<DocScrutinizer>
well, as I spent a little fortune for this first computer of my life, and also spent several weeks to solder all of it and trace the 3 missed solderpoints and finally start the whole z80 thing up, I eventually made up my mind and considered the hedgehog way too dangerous if eventually it would short e.g between emitter and collector of one of the regulator transistors. So I rebuilt the whole thing on a proper lab PCB
<whitequark>
it's good when you have access to some lab
<DocScrutinizer>
nah, just a PCB with those zillions of holes in it
<DocScrutinizer>
they came with either dots or stripes, or completely w/o copper
<DocScrutinizer>
and had a hole raster of 0.1
<DocScrutinizer>
aka 2.54mm
<DocScrutinizer>
they didn't change for like the last 40 years
<DocScrutinizer>
the lab back when was my dad's cellar (he's also been EE)
<whitequark>
was that a two-side pcb?
<whitequark>
I'm still failing at these
<DocScrutinizer>
no
<whitequark>
ah yes. then I have the skills for that pcb :)
<whitequark>
how many years of experience in electronics do you have?
<whitequark>
I should have started exploring that earlier :/
<DocScrutinizer>
when I was 8 I started reading my father's electronics magazines
<DocScrutinizer>
I started reading and drawing and building simple schematics some years earlier
<whitequark>
same here, but I was 12
<DocScrutinizer>
as my dad worked at SEL in semiconductor manuf for some time, we had lockers full of boxes with components
<whitequark>
what would you recommend to improve my skills? books maybe, etc.?
<DocScrutinizer>
when I was 12 I got some of the first TTL74xx chips Siemens produced in Germany (of course only those which failed QA, the others cost more than gold back when)
<DocScrutinizer>
I'm not sure I can give advice how to practice, as my approach surely wasn't the one you'd choose when you're an adult
<whitequark>
I'm still interested
<DocScrutinizer>
I think I learned most by reading electronics magazines, there's been "Elector" back when, which came with several schematics each month, and once a year they had a special issue with >>100 schematics - I've rad them all, and sometimes spotted bugs in them ;-)
<whitequark>
(first TTL chips) what year was that?
<DocScrutinizer>
finding the ones I could build with the components I got (maybe after modifying the schematics and values of components a bit to mach what was available) was also much fun and helped a lot to learn how to calculate those things
<DocScrutinizer>
(year) ~1973 iirc
<DocScrutinizer>
I think I always wondered if they've chosen the 74 part of the name because of the year
<DocScrutinizer>
maybe it's not been Siemens' first german chips, but some new fab nearby - can't remember exactly. Of course the 74 series existed quite a bit longer obviously
<whitequark>
it's good when you have a fab nearby
<DocScrutinizer>
I had an uncle "nearby" :-) - he worked at Siemens
<whitequark>
now, when there are almost no manufacturers alive here, it's very hard to buy materials like aluminium sheets (I'm trying to make some custom cases), or textolite (not the foiled one, I mean), or fluoroplastic in sane quantities
<whitequark>
when no one uses these, there are no scraps :)
<DocScrutinizer>
and I thought he's the friggin only true santa when he came and offered me this bag of chips
<whitequark>
that was truly a great gift
<whitequark>
I've just found a nearby company which offers a really good price for small quantities of laser cut materials
<whitequark>
like $1.5 for 1.5 meters of perimeter (bigger is cheaper)
<whitequark>
I hope this will finally make me able to do a simple, but good-looking case
<DocScrutinizer>
now that's cool
<whitequark>
yeah, I once thought that it only was possible in USA or Europe, everyone I found here wanted a minimal order of ~$300
<DocScrutinizer>
I guess complex shapes will quickly rise the pricetag
<whitequark>
why? I think they only compute the price by cutting machine time + preparation time
<whitequark>
isn't the actual shape indifferent for the machine?
<DocScrutinizer>
preparation time. how do you provide the shape data?
<whitequark>
dxf or solidworks
<DocScrutinizer>
aah k, then it shouldn't matter
<whitequark>
I didn't had experience with either of them
<whitequark>
can you recommend a good simple CAD to make a 2D shape?
<DocScrutinizer>
neither, only heard of it
<DocScrutinizer>
wpwrak: to the rescue! :-D
<whitequark>
also, there are some inherent problems with the design
<whitequark>
the thing I want to make is a WiFi-enabled clock. basically that's just my 4725 board with SDIO wifi card connected and in a good case
<whitequark>
the problem lies with the (hey, now I will mess up all the terms) fastening of the board on the case
<whitequark>
it should be in a shape of a triangular prism, lying on a rectangular side
<whitequark>
on the other rectangular side there will be a hole for the lcd
<whitequark>
and some for buttons
<whitequark>
I think that I may order the case itself in two parts: one would be two rectangular and two triangular sides welded together, and it will have two mounting tabs
<whitequark>
and the other will be a single rectangular part with mounting holes matching those tabs
<whitequark>
the problems are: 1) as you've seen, the board has no mounting holes (and anyway, I don't want to add any screws on the front panel at all)
<whitequark>
and 2) I'm not aware of a sane way to insert a protective plastic panel in front of LCD
<whitequark>
do you have any ideas?
<vladkorotnev>
hello everyone
<vladkorotnev>
is there a way to set OpenWRT encoding to KOI8-R?
<vladkorotnev>
whitequark: almost! blank chars instead of gibberish
<wpwrak>
DocScrutinizer: (2D cad) many people like qcad. i hate it. so i use fped for really simple things. if fped can't do it, i write a script. (e.g, to generate toolpaths, which is something fped doesn't know how to do)
<whitequark>
wpwrak: fped is for footprints, right? and I need to make a 2d contour for laser cutting
<whitequark>
do you know anything on this field?
<wpwrak>
whitequark: make three layers: 1) bottom, with 2-4 holes for screws. 2) top, with 2-3 holes for screws, transparent. 3) frame, with 2-3 holes for screws and a bit opening in the middle the size of the PCB. layer 3 has to have the thickness of the pcb (plus some tolerance)
<wpwrak>
whitequark: then put L1 + L3, align them, put the PCB inside and glue it in (glue, silicone, tape, whatever), then put L2 on top and put in the screws
<wpwrak>
whitequark: fped is for footprints but can also do some basic geometry. not as much as qcad, but it does it conveniently and parametric. has no toolpath generation, though.
<whitequark>
wpwrak: thanks for a great idea, I'll think about it. there may be some problems with thickness of 3rd layer. through
<whitequark>
wpwrak: (fped) can it save DWG's?
<wpwrak>
whitequark: so when i need a toolpath, i write a perl script that prints X/Y coordinates. then i verify with gnuplot. WYSIWYG is for girls ;-)
<wpwrak>
whitequark: (L3) if you can't find or can't machine a thick enough sheet, use multiple thin ones.
<whitequark>
wpwrak: yeah, just got the same idea
<wpwrak>
(fped, DWG) no, just postscript, kicad, and its own format. you could probably use kicad to generate DXF from it, though.
<DocScrutinizer>
who's sourcing for Qi?
<whitequark>
wpwrak: just installed it. version 0.0 and the rich interface are promising :)
<wpwrak>
DocScrutinizer: the things sharism makes ? that would be mainly adam
<DocScrutinizer>
I dunno who's sharism
<DocScrutinizer>
just asking who's the best man to find out e.g about OMAP4
<DocScrutinizer>
when it comes to production, sourcability, etc
<DocScrutinizer>
always thought that's wolfgang
<wpwrak>
(sharism = the company that gets manufactured bens and manufactures mm1s (and sells both to customers and distributors))
<DocScrutinizer>
aaah, so who or what is Qi-HW then?
<wpwrak>
(omap4) hmm. wolfgang may some things about sourcing those chips. probably no details, though
<wpwrak>
qi-hw is the community effort of which sharism is a part
<DocScrutinizer>
sure, as we never talked to TI so far afaik
<DocScrutinizer>
thanks, now things get clearer
<wpwrak>
(talk to ti) yeah. as wolfgang describes it, you probably need a direct line to TI anyway, if you want a realistic chance to get the chip even out of reset ;-)
<DocScrutinizer>
lol
<DocScrutinizer>
I'll copy that to my best quotes
<DocScrutinizer>
for now it seems more like you need a direct line to dinno whom, God?, to get an OMAP4
<DocScrutinizer>
wonder where panda get theirs from
<wpwrak>
good luck with the sourcing then ;-)
<DocScrutinizer>
do we know anybody at beagle/panda?
<wpwrak>
maybe ping goldendelicious if they have any contacts ? or at least war stories to share :)
<DocScrutinizer>
well, for now I'm just collecting options and trying to tell apart the existing stuff from vaporware, so we can do some educated decision about the preferable architecture for meegophone
<DocScrutinizer>
I already have that on my todo list - thanks nevertheless
<whitequark>
so, are you doing an open phone?
<DocScrutinizer>
as long as this laterego guy doesn't come up with some names and stuff, the project itself hasn't left the vaporware status in my book ;-)
<DocScrutinizer>
whitequark: we're planning to do so, according to this guy
<wpwrak>
whitequark: i guess he's been plotting the defect rate vs. time of his N900 and got worried that it may disintegrate before apple and android are both dead :)
<wpwrak>
whitequark: well, with apple and samsung at other's throat, things may move a little faster on that front. i wouln't expect samsung to do anything overly rash, but apple's another story.
<wpwrak>
whitequark: also, apple has already hat its near-death experience. maybe they liked it ;-)
<wpwrak>
s/hat/had/
<DocScrutinizer>
I need facts though. When I say "ok, send me some pandaboard for evaluation, they're cheap but I'm not sponsoring this enterprise" - well, we'll see what's happening
<whitequark>
wpwrak: I never heard about apple near-death experience. can you send a link?
<DocScrutinizer>
hah, wasn't that when Jobs resurrected?
<wpwrak>
whitequark: oh, i'm sure you remember. before they brought jobs back, they weren't doing so well.
<DocScrutinizer>
o/
<DocScrutinizer>
still pondering coffee or nap
<wpwrak>
DocScrutinizer: how many hours of uptime now ?
<DocScrutinizer>
40?
<whitequark>
wpwrak: ah yes. was that in 90's? just to clarify, I was born in 1993 :)
<whitequark>
so I doubt I can remember that :)
<wpwrak>
DocScrutinizer: hmm. i guess we'll know when nature has solved that problem when your head rolls on the Enter key ;-)
<DocScrutinizer>
tested it, nothing happens :-P
<DocScrutinizer>
IRC doesn't like empty lines
<wpwrak>
whitequark: hmm. childhood memories perhaps ;-) i'd google it up if i dared to move my mouse (it's recharging. so it'll be chat-only for me for a while)
<DocScrutinizer>
but, as you never know what can happen:
<DocScrutinizer>
time for 12648430Â Â Â Â
<wpwrak>
DocScrutinizer: while rolling towards the Enter key, your head will reach other keys .. ;-)
<DocScrutinizer>
bigvikp#+
<DocScrutinizer>
yeah that works
<DocScrutinizer>
not easy though
<wpwrak>
DocScrutinizer: and you'll roll a lot as your subconsciousness relives the fiery awakening that makes you afraid of sleeping now :)
<DocScrutinizer>
lol
<whitequark>
wpwrak: according to wiki, that has happened in 1997. I barely remember that time, and anyway I hadn't even touched my first PC :)
<DocScrutinizer>
actually I need to find a replacement switch for my mouse, this drives me mad
<DocScrutinizer>
this thing
<wpwrak>
DocScrutinizer: are you an elephant ? :)
<DocScrutinizer>
no, my mouse is a tick
<wpwrak>
whitequark: ah yes, a bit early then
<DocScrutinizer>
firggin Logi MX Revolution (some 80¬) and cheap chinese switches that break and drive you nuts after 3 years
<DocScrutinizer>
:-/
<wpwrak>
DocScrutinizer: you got what you asked for. or don't you find this behaviour revolting ? :)
<whitequark>
DocScrutinizer: iirc they're sending replacements for free or something like that
<DocScrutinizer>
replacement switches? or new mause? :-)
<whitequark>
i heard about new mouse
<wpwrak>
lifetime warranty ? that could be dangerous in places where electronics are expensive but killers are cheap
<whitequark>
heh
<Jay7>
vladkorotnev: check screen -u too for your application
<Jay7>
screen have some way to do with non-utf-8 apps
<vladkorotnev>
Jay7: fbterm solved my problems ;)
<Jay7>
ah, ok then
<Jay7>
DocScrutinizer: I've changed lot of switches in old MS mouse
<Jay7>
may be good idea to replace mechanical switches with something other..
<whitequark>
Jay7: capacitive sensors? optical sensors? put a webcam inside and capture the shadow of fingers?
<Jay7>
whitequark: btw, about latest
<Jay7>
I have Genius mouse with 'optowheel'
<Jay7>
it's optical sensor instead of wheel :)
<Jay7>
a bit crazy but after disabling horizontal scrolling in X it started to be useful again :)
<Jay7>
easiest way to replace mechanical switch with reed switch
<Jay7>
it even will do 'clicking' :)
<whitequark>
Jay7: another idea is to capture electromyogram (is it called so?..)
<whitequark>
you just think about clicking -- and voila -- it's already been done!
<Jay7>
we should implement this into Ya NN :)
<whitequark>
Ya?
<whitequark>
yet another?
<Jay7>
after-Ben :)
<Jay7>
it was reffered so some times
<whitequark>
sounds interesting
<whitequark>
can you tell me a few facts?
<Jay7>
not sure about history.. may be just next 'letter' in alphabet :)
<Jay7>
whitequark: it's just bunch of ideas, nothing real
<whitequark>
I hope it will be more interesting than current NN :)
<Jay7>
we all hope so :)
<wolfspraul>
viric: did you by any chance find the nanonixos screenshot we talked about yesterday?
<lunavorax_frizzl>
Fusin, I suggested once the making of a ZX Spectrum-like NanoNote as a better alternative to the Raspbery Pi for childrens to learn computing
<Fusin>
Well, a NanoNote with a good BASIC-Interpreter would be a good tool for Kids
<Fusin>
The BASIC does not need to be full fledged like Virtual-Basic. Any GW-Basic interpreter would do
<Fusin>
only needs basic Graphics and Sound support (for Games etc) :D
<lunavorax_frizzl>
Fusin, a Nanonote with a good documentation and exemple for kids will be fine, whatever the language. They just need a proper doc and exemples (and the tools pre-installed of course)
<Fusin>
right
<Fusin>
I'm with you :D
<lunavorax_frizzl>
Now we need money
<lunavorax_frizzl>
:(
<lunavorax_frizzl>
The sad part
<Fusin>
Can't we print our own??
<Fusin>
;)
<Fusin>
Nano$
<lunavorax_frizzl>
Haha
<Fusin>
hrhr
<wpwrak>
hmm. my clock measurement with a guaranteed maximum interval between extremal measurements seems to live up to its promise. min-to-max 0.7 ppm with a requested 1 ppm interval, so far. alas, that takes it about half an hour to measure. not too great for production.
<whitequark>
kyak: golded, oh. isn't fido dead?
<vladkorotnev>
whitequark: lol, it isnt
<whitequark>
wpwrak: I remember that you've posted a link to Qi site where there was some pictures of a 8-layer pcb took apart with sandpaper, can you link it again?
<wolfspraul>
my todo list now: ben-wpan production, azonenberg 4004 diy, reorder for better flow, maybe get a screenshot from viric :-)
<wolfspraul>
then there may still be other things I overlooked or that could be mentioned, but I'm getting tired
<wolfspraul>
2 months is too long
<wpwrak>
you have one more day anyway :)
<wolfspraul>
yes sure
<wolfspraul>
what do you think so far?
<whitequark>
just found the 8:10 vga thing. it is really weird. is that done purely in software via bitbanging?
<wolfspraul>
your stuff is excellent as always, vga & so much on ben-wpan...
<wolfspraul>
how does this work?
<wpwrak>
"Adam Wang reported on the latest Milkymist One RC3 production status". maybe add an executive summary ?
<wolfspraul>
yes, it works :-)
<wolfspraul>
2 minutes, hmm
<wolfspraul>
well, the link points to the mail, and it's hard to summarize
<wolfspraul>
so many things moving, so many details
<wpwrak>
wolfspraul: (hard to summarize) indeed ;-)
<wolfspraul>
executive summary is simple: DELAY
<wolfspraul>
we are freaking behind schedule, a lot
<wolfspraul>
so rc3 cannot sell, ben-wpan still slow as well
<wpwrak>
something like "several minor delays, ETA 8/11" ?
<wolfspraul>
do you want me to add that executive summary :-)
<wpwrak>
well, everybody has delays :)
<wolfspraul>
you know I want to free the valuable times of the executives for drinks and other important meetings
<wolfspraul>
so just "delay"
<wolfspraul>
come back later, nothing to see...
<wpwrak>
ah, tuxbrain contacted me just a few minutes ago. the fab is looking for some cad file (which they should have :) and they plan to smt this week
<wolfspraul>
oh, great!
<wpwrak>
also "Two threads on the mailing list discussed plans about adding a MMU to the Milkymist SoC." wants a summary
<wolfspraul>
agreed, but there is none
<wpwrak>
this one is an excellent example for a good entry: "Xiangfu Liu demonstrated Open Sound Control ("OSC") with his Milkymist One. OSC is another promising way to control Milkymist One, with many nice clients easily available for popular smartphones and tablets."
<wolfspraul>
quite a lot of planning in those threads, I'm not sure anybody is doing anything now - they left that open :-)
<wpwrak>
"what happened ? what does it mean ? what does it mean for me ?"
<wpwrak>
(mmu) yeah, that was the impression i got
<wolfspraul>
if you know OSC, now you know that m1 supports osc
<wolfspraul>
that's the point of that entry
<wpwrak>
but there seems to be some narrowing down on what's feasible and what not
<wolfspraul>
do you know osc?
<wolfspraul>
I don't
<wolfspraul>
:-)
<wpwrak>
nyet :)
<wolfspraul>
so I can only read what it says in wikipedia, seems to be a 'modern midi'
<wolfspraul>
I guess first of all it's a protocol to send simple messages back and forth
<wolfspraul>
just click on that video, then it's clear :-)
<wolfspraul>
although ironically sound is missing in that video, argh
<wpwrak>
(mm1 sold out) do you know if anyone else has stock ? so that you can divert prospective buyers there ?
<wolfspraul>
at least it makes you focus on the OSC :-)
<wolfspraul>
not sure, of course we can mention that
<wpwrak>
also, the news entry ought to say something about when more will be coming. it reads almost like saying "EOL"
<wolfspraul>
the little feedback I get seems to indicate that it may be very low or out of stock everywhere
<wolfspraul>
me too, reading the backlog tomorrow...
<wolfspraul>
n8
<wpwrak>
(4tH) maybe put "FORTH" or a link explaining what it is somewhere ?
<wolfspraul>
I think the interesting thing about OSC is (and I'm guessing a bit), that you can basically have one of many Android or iphone/ipad clients, and then control m1 from there
<wolfspraul>
it's easy to hook a patch into osc input (again guessing)
<wolfspraul>
so basically with osc, we can become an iphone/ipad accessory - isn't that great?
<wpwrak>
(sound track made by kaossilator) maybe use "with" ? it's an instrument, although admittedly one that doesn't require a lot of skills for something as simple as this
<wolfspraul>
you need to connect the m1 ethernet to a router though, and from your phone with wifi to the m1
<wpwrak>
(osc) whoa. heavy setup.
<wolfspraul>
changed [with]
<wolfspraul>
not really
<wolfspraul>
think of someone who already has an android phone
<wolfspraul>
how can he control m1 right now?
<wolfspraul>
keyboard?
<wolfspraul>
remote control?
<wolfspraul>
but why not with your phone, if you can download a free osc client with a few clicks
<wpwrak>
whitequark: it's not entirely sw. well, an earlier version was. now it uses the MMC controller to pump out the pixel data. line timing is still controlled by software. that's a lot more efficient than 100% sw.
<wolfspraul>
I'm not saying it's all OSC now, but there are OSC fanboys, and the entry is for those that we say ok, m1 supports osc. if you have an osc client, you can use it to control m1.
<wolfspraul>
there is a linux command-client oscsend utility, I think
<wolfspraul>
command-line (getting sleepy)
<wpwrak>
(osc) no, osc is good. it's just a bit of a messy setup. well, you could probably plug a WLAN stick into the MM1. (does rtems have a wifi stack and drivers ?)
<whitequark>
wpwrak: that's what I have expected. I've seen some interesting implementations on AVR using its SPI shift register
<wolfspraul>
no way, not supported now and very hard
<wolfspraul>
I stay realistic. if you want osc, you have to network your m1 over ethernet.
<wolfspraul>
otherwise your shiny phone cannot reach m1
<wolfspraul>
osc or not
<wolfspraul>
but the ethernet osc works today (see the video)
<wolfspraul>
that's the news
<wpwrak>
(no wifi) i was afraid it would be so. well, if someone was really into it and ported over the whole stuff from linux ... but yes, it's hard. barely works on linux itself.
<wpwrak>
(slashdot coverage) also made it to hackaday and dangerousprototypes. nicer feedback there.
<wpwrak>
(first impression of news) looks great !
<wpwrak>
do you want to mention jane's keyboard hack ? something frmo leftfield
<wolfspraul>
sure, I'm just still in my todo
<wolfspraul>
4004 diy, ben-wpan production
<wolfspraul>
two screenshots missing, nanonixos and debian wheezy
<wolfspraul>
mark (debian wheezy) did not reply, I don't want to stalk him
<wpwrak>
the smt fab is funny. they always start sending mails around 6 pm ;-)
<wolfspraul>
but I need to get the facts straight, need to read the #milkymist backlog carefully first
<wolfspraul>
somehow I think he buys a 4'' wafer for 35usd, and the rest is all diy toasters (exaggerating), and the end goal is to make 4004 processors
<wolfspraul>
like I said, I need to make this a little more solid :-)
<wolfspraul>
and maybe I shouldn't skip over everything saying "4004 diy", but instead talk about the next step, the one thereafter
<wpwrak>
the 4004 still seems to be quite at the horizon
<wolfspraul>
but that's even harder to get right then
<wolfspraul>
yes sure, I know
<wolfspraul>
but I can introduce the project like that, maybe
<wpwrak>
for now, he seems to struggle with increasing the field-of-view. the microscope only gives him a very narrow area to work in.
<wolfspraul>
like I said, need to read first, but I want to mention it in some way
<wolfspraul>
oh I can imagine
<rjeffries>
irx log is not reachable by me, for now
<wolfspraul>
and I want to document the process, not just the result. but for that I need to understand more.
<wolfspraul>
that's for tomorrow...
<wpwrak>
wish the mm1 video in video ha some colors in it ...
<wolfspraul>
we probably need a vga grabber
<wolfspraul>
showing videos of the end result will be important in marketing, and we have nothing
<wolfspraul>
the indirect recorded ones we have are of horrible quality
<wolfspraul>
and I doubt we will get onboard real-time encoding done anytime soon (theora or webm)
<wolfspraul>
so either vga grabber, or sebastien implements tv-out (over the vga connector), that's also relatively easy to record I think
<wpwrak>
if there's enough bw, just stream out raw dumps and encode off-line ?
<wolfspraul>
we'll get there :-)
<wolfspraul>
all work probably, this is not (yet) a general purpose computing platform
<whitequark>
wolfspraul: what's that semiconductor thingy on the pic? a delay line?
<wolfspraul>
don't ask me
<wpwrak>
heh, you just dumped all my ubb-vga pics ;-))
<wolfspraul>
I'm trying to say something about it that is not completely stupid.
<wolfspraul>
yes but I think they are nice
<wolfspraul>
first I though of picking some, then I liked the idea of putting them all in a gallery uncommented
<wolfspraul>
first they were chronologically (like they appeard in your mails), then I reordered a bit so that you can see the process, the work, back and forth, improvements, etc.
<wolfspraul>
all the way to the end result
<wpwrak>
should have one with the 16 color board. the ones you have are all with the first board, 8 colors and no MMC-capable
<wolfspraul>
maybe I overlooked some still
<wolfspraul>
possible - you sent so many mails
<wpwrak>
;-))
<wpwrak>
if you picked them from the mails, maybe link to the mail ? that way, they can work as real teasers ?
<wolfspraul>
already there
<wolfspraul>
the whole list of mails
<wolfspraul>
not sure about another link from each picture, maybe
<wolfspraul>
I can try tomorrow whether I find some overlooked pics or mails
<wolfspraul>
the mail archiver is also not perfect, sometimes mails disappear from the archives
<wolfspraul>
buggy, and unmaintained for 10+ years
<wolfspraul>
there may be more, I think I only looked at the thread starters
<wolfspraul>
anyway, more polish tomorrow...
<kyak>
whitequark: i think FIDO is more dead than alive, but golded was ported mostly for vladkorotnev, he seems to be excited about FIDO and actually using it :)
<wpwrak>
FIFO.userbase = 1 ;-)
<Saint>
Could anyone help me out with the specs of a gaming desktop?
<Saint>
Anyone?
<whitequark>
Saint: try asking at #hardware, this is not the channel you are looking for
<wpwrak>
hmm, is anyone here good with math/statistics ? specifically, linear regression with least-square fitting ?
<whitequark>
vladkorotnev: this (http://patchwork.ozlabs.org/patch/27367/) suggests that it may be a bug in UBI, and manual cleaning of internal structure is needed, due to UBI unable to restore the structure itself
<vladkorotnev>
wpwrak: wolframalpha.com maybe?
<whitequark>
vladkorotnev: I don't think that there is easy way to recover your partition. I may be wrong through.
<wpwrak>
vladkorotnev: ;-)) naw, i already found GSL :) what i need is an idea for how to properly weight my samples
<vladkorotnev>
whitequark: FFFFFFFUUUUUUUU
<wpwrak>
whitequark: there is ! it's called "restore the latest backup" (-:C
<vladkorotnev>
whitequark: however, not that much data, all from my computer :P
<vladkorotnev>
no backups :(
<whitequark>
wpwrak: every time there is one man who's sarcastic about someone's backups :))
<whitequark>
vladkorotnev: what's your kernel version? I've experienced some UBI bugs in the past
<vladkorotnev>
whitequark: idk what happened to me today. Debian boot loader crashed. Virtualbox doesn't run. UBI partition doesn't mount. fu-k!
<vladkorotnev>
whitequark: Linux BenNanoNote 2.6.32.27 #1 PREEMPT Wed May 25 05:49:30 CEST 2011 mips GNU/Linux this?
<whitequark>
ahhh 2.6.32
<whitequark>
that's the point.
<whitequark>
there is a lot of nasty UBI bugs in .32
<vladkorotnev>
whitequark: so&?
<whitequark>
upgrade to at least .35, or better to .38-.39
<vladkorotnev>
whitequark: how? and will it be ok?
<whitequark>
we've did some work of porting all the jz47xx changes to recent kernels with Lars
<whitequark>
my tree should work on Ben, and it's .39
<whitequark>
and yes, UBI will work fine, it was tested by me and tens of russian developers
<vladkorotnev>
whitequark: i am not that professional, i don't know how to upgrade kernel. and I meant will it mount after the upgrade, not will it work
<vladkorotnev>
plus, it's 2:03 AM here
<whitequark>
ah yes. no, it probably won't mount, but it won't also fail in the future
<whitequark>
you can ask someone here with Ben to compile the kernel probably
<whitequark>
maybe there is ever an 'officially' approoved version of recent kernels
<vladkorotnev>
whitequark: i'm on the 05/24/11 image if that matters
<whitequark>
vladkorotnev: I don't have Ben and I haven't worked with Qi Hardware images/packages. it's just my device having the processor of same family with Ben, and so I am here
<whitequark>
but I still have worked with ubi nevertheless
<vladkorotnev>
ok, bye everyone. have to wake up at 10:56 am tomorrow. then I think I will reformat the partition...
<dvdk>
wpwrak: just trying to write a freashmeat.net release announcement for the 05-28 firmware and wondering about how to call the avrdude chip-programming capabilities of nanonote.
<dvdk>
wpwrak: you think it's correct to call it "repository now includes a version of avrdude that turns the NanoNote into a high-speed chip-programmer"
<dvdk>
s/"re/"[..] re/
<wpwrak>
dvdk: it's actually relatively slow (at least it feels like that ;-)
<dvdk>
wpwrak: thought it should be at least faster than a parallel port programmer :)
<dvdk>
so you thiink i should strip the 'high-speed'
<dvdk>
?
<wpwrak>
dvdk: maybe "that turns NanoNote + UBB into a chip programmer" ? then you have about the same number of attributes
<wpwrak>
i don't really know how it compares speed-wise with others. i had to slow things down a bit to make it work, so i suspect it's not the fastest one around. but maybe the others don't work with default settings either ...
<whitequark>
hm, this sounds like a real use case for NN, interesting
<dvdk>
wpwrak: what about 'Optional software installable from the repository includes a version of avrdude that turns the NanoNote into a generic chip-programmer (the "UBB" adapter-board for accessing the I/O pins is available from Tuxbrain and other distributors).'
<dvdk>
it's a freshmeat announcement for 'nanonote firmware' so we ought to explain what a UBB is
<wpwrak>
dvdk: it's only for AVR, not "generic". of course, you could make programmers for other chips as well. e.g., the f32xbase project has a ben-based one for C8051F3xx chips
<dvdk>
'into a AVR chip-programmer'?
<wpwrak>
dvdk: i would also drop the "optional". that often means "at extra cost" :-)
<dvdk>
s/a/an'
<dvdk>
:)
<wpwrak>
yeah, sounds good
<dvdk>
s/optional/Â Â yeah and the more concise the better
<wpwrak>
or maybe "into a programmer for AVR chips"
<dvdk>
even better
<wpwrak>
hehe, ~0.1 ppm accuracy. soon i'll be ready to measure the gravitation of passing neutrinos ;-)
<dvdk>
wpwrak: everybody measures neutrinus.  try measure WIMPS :)
<dvdk>
s/neutrinu/neutrino  s/WIMPS/WIMPs
<dvdk>
time for bed
<dvdk>
cu
<wpwrak>
hmm yes. maybe i'll even stuble upon some even darker matter :)
<whitequark>
wpwrak: 0.1 ppm of what?
<dvdk>
"The latest firmware features WIMPsens that uses the Ben's microphone to sample the density of dark matter"
<dvdk>
gotta go
<wpwrak>
whitequark: the frequency of the atusb crystal oscillator
<wpwrak>
dvdk: ;-))
<dvdk>
cu
<wpwrak>
whitequark: in relation to an NTP-disciplined host clock
<whitequark>
wpwrak: is NTP that good, really?
<wpwrak>
whitequark: kinda hard to tell without a nuclear clock reference ;-)
<wpwrak>
whitequark: what i'm interested in is relative time. so all i need is the drift compensation. i don't care about the time offset.
<whitequark>
then, what shall it do with NTP? NTP is all about network synchronization, isn't it?
<wpwrak>
NTP also corrects drift. PCs usually have lousy crystals. ~+/-100 ppm tolerance.
<whitequark>
ah. yes.
<whitequark>
okay, I'm going to bed
<DocScrutinizer>
wpwrak: only for system clock, RTC usually is way better
<wpwrak>
maybe. but there the resolution is horrible. and synchronizing to an edge may be tricky.
<wpwrak>
(i.e., i'm too lazy to figure out that one as well ;-)
<DocScrutinizer>
notices he missed 90% of relevant backscroll
<DocScrutinizer>
maybe time for a nice walk
<DocScrutinizer>
it starts to get funy when the uptime shows ???
<DocScrutinizer>
still no head on <enter>
<wpwrak>
well, you'll crash for some 12+ hours :)
<DocScrutinizer>
but a huge milk supply by now :-D
<DocScrutinizer>
for *lots* of 0xc0ffee
<DocScrutinizer>
wpwrak: usualy not
<DocScrutinizer>
as mentioned before: sleep is overrated
<wpwrak>
it has its uses :)
<DocScrutinizer>
a crime to sleep on such a nice night
<DocScrutinizer>
heads out for a walk in the dark
<qi-bot>
[commit] Werner Almesberger: atusb/fw/: added free-running 48 bit 8 MHz counter for clock measurements http://qi-hw.com/p/ben-wpan/8c57277
<qi-bot>
[commit] Werner Almesberger: atrf-xtal.c (do_atben, main): moved board-specific code to separate function http://qi-hw.com/p/ben-wpan/5a6aae5