<wpwrak>
yup. but the rest of silabs is of course different. you could reuse SWD and probably the AHB-AP but not the flash programming code. that would be specific to silabs.
dos1 has quit [Read error: Connection reset by peer]
<nicksydney>
cool...different silicon have different address right ?
<nicksydney>
so theoretically it is possible if i made a 'programmer' using MCU chip to 'program' Cortex-M series chip with SWD right ?
<wpwrak>
of course, yes
<wpwrak>
like in y-box the kl2x will program the cc2543 (via ti's oh so much simpler protocol)
<nicksydney>
cool....awesome!
<nicksydney>
thought the cc2543 firmware is proprietary ?
<wpwrak>
the 2543's interfaces are open. ti have other chips in the same family, with closed rf interfaces
<wpwrak>
of course, this still means that someone has to write the firmware ... ;-)
<nicksydney>
let me get this right...
<nicksydney>
the 2543 code is available to download ?
<nicksydney>
or it comes prebuilt from factory ?
<wpwrak>
dunno if ti have any code for the 2543
<wpwrak>
in any case it wouldn't be BTLE
<wpwrak>
because they have another chip, with closed BTLE code
<nicksydney>
right so what you saying the 2543 will be 'programmed' later on with bluetooth stack ?
<wpwrak>
but the 2543 should be able to speak BTLE regardless, although with a bit more difficulties (e.g., the whitening algorithm seems to be different. thus one has to transform the data stream at the bit level)
<wpwrak>
yup, at least the low-level part of it
<wpwrak>
higher-level stuff is probably better kept in the KL2x. the 8051 in the CC2543 is not the nicest environment (well, the core is okay even if it's very very antiquated. but the only free compiler we have for that has its issues)
<nicksydney>
aren't there any bluetooth stack available that we can reuse for cc2543 ?
<wpwrak>
ah, might be worth looking at at some point in time. what we need is BTLE and only BTLE. the rest of BT is a) the super-sized XXXL version of mortally obese bloatware, squared, and b) the CC2543 doesn't quite speak BT (though it may be able to communicate at the low rate, albeit with an accent)
<wpwrak>
ah no, not worth looking at
<wpwrak>
wrong license
<wpwrak>
"BTstack is available under a dual license. The code provided in this repository allows for non-commercial use."
<nicksydney>
in order to write your own stack you need to understand the BLTE spec first ?
<wpwrak>
royalty-free doens't mean open ...
<wpwrak>
yeah, understanding what one is doing ought to be useful. of course, one could also just hit random keys and see what happens :)
<nicksydney>
the spec is 2302 pages :(
<wpwrak>
something to look forward to for long winter days ;-)
<nicksydney>
out of my league :)
bartbes has quit [Remote host closed the connection]
<wpwrak>
what a blatant lack of heroism :)
CYB3R has quit [Ping timeout: 245 seconds]
xiangfu has joined #qi-hardware
luke-jr_ has joined #qi-hardware
Luke-Jr has quit [Ping timeout: 248 seconds]
luke-jr_ is now known as Luke-Jr
kristianpaul has quit [Ping timeout: 248 seconds]
kristianpaul has joined #qi-hardware
<nicksydney>
wpwrak: i aint superhero :)
<nicksydney>
wpwrak: but seriously how can you go through 2302 pages and implement something like that...you go mad !
<wpwrak>
isn't it that this is BT 4.x ? BTLE is just a part of it. and there's always a lot you can skip. some things you'll never need. some you don't need until later.
<wpwrak>
so you can build the monster step by step
<nicksydney>
the question i have in my mind now is..how to start digging into it..where to start ? what to look for ?
<nicksydney>
i'm still crawling when it comes to this kind of thing so be nice :)
<nicksydney>
hehehe
<wpwrak>
oh, if you haven't done such a thing before, then you probably want to start at page 1 :)
<nicksydney>
hahaah...like reading a fiction eh :)
* whitequark
wants to try making multilayer pcbs sometime
<whitequark>
everyone on the net keeps saying how hard it is, but
<whitequark>
you need a working electroplating process, a heated press and some epoxy
<whitequark>
and a cnc
<whitequark>
okay, I see how that would be too hard for your average hobbyist with a clothes iron
<DocScrutinizer05>
and a damn sharp and exact drill
<whitequark>
yep
<DocScrutinizer05>
or rather, microscopic headmill
<whitequark>
endmill?
<DocScrutinizer05>
yeah
<whitequark>
good thing I have two dozens of .2mm flying to me :p
<DocScrutinizer05>
drilling a via hole to contact a inner layer with your hole plating is not exactly easy
<whitequark>
well, not easy. but doable
<DocScrutinizer05>
maybe
<whitequark>
do I understand it right... to make such a blind via, you need to have a pad (filled area of copper) on the inner layer
<DocScrutinizer05>
depends on precision of your drilling process and quality of your PCB material
<whitequark>
then you drill from the outer layer right up to this filled area
<whitequark>
and then you do electroplating on that hole, connecting traces on outer layer with that area on inner layer?
<DocScrutinizer05>
well, yes, for embedded vias that's what you do
<whitequark>
there's something else?
<DocScrutinizer05>
usual (simpler) way is to drill all through the whole lamiate stack and thus expose a copper surface of the drilled-through inner trace to the hole's surface
<whitequark>
(copper surface) you expose the _side_ of the copper layer, right? not its _surface_ ?
<whitequark>
cross-section of it
<DocScrutinizer05>
and that copper surface of the inner layer trace in the drilled hole has to be quite clean and tidy to make for proper hole copper plating making contact to the inner trace
<DocScrutinizer05>
yes
<DocScrutinizer05>
that's how vias usually work
<whitequark>
right. just was making sure
<DocScrutinizer05>
you can crate an inner stack of just maybe 2 or 4 copper layers, do the vias, then laminate further stacks of FR and copper on top on both sides and create further vias when outer copper layers are involved
<DocScrutinizer05>
create
<DocScrutinizer05>
you shall fill the vias prior to laminating further layers, or your PCB will go popcorn
<whitequark>
fill the vias?
<DocScrutinizer05>
yes
<DocScrutinizer05>
you usually fill vias
<DocScrutinizer05>
with either copper or varnish
<DocScrutinizer05>
particularly when you need in-pad vias
<whitequark>
filling them with copper sounds painful
<DocScrutinizer05>
of course there are finer smaller ones, which are better suited
<DocScrutinizer05>
the trick isn't in the sawbalde but in the very exact and smooth moving of goods to cut vs the sawblade
<DocScrutinizer05>
to create an absolutely plane cute
<whitequark>
yeah, the result is pretty cute
<DocScrutinizer05>
and of course the sawblade mustn't have any tolerances e.g. in spindle axis Z direction
<DocScrutinizer05>
also for some materials the substance to get cut is simply too soft, so you maybe need to cool it down to very low temperatures to make it more rigid and thus create a smooth cut
<whitequark>
:nod:
<whitequark>
like with liquid N2
<DocScrutinizer05>
yep
<whitequark>
I've read insane stories about CNC mills using liqid N2 to cool workpiece
<whitequark>
won't the endmill shatter very quickly in such conditions?
<DocScrutinizer05>
you probably need special toolbits for such work
<whitequark>
sure. but it is hard to imagine metal regaining its properties at 70K
<DocScrutinizer05>
though the age old urban legend about cooling down steel of e.g. padlocks with N2 and then crack them open with a small hammer is just BS
<whitequark>
is it?
<DocScrutinizer05>
yes, sure
<DocScrutinizer05>
otherwise, which maerial would you use to *produce* and *contain* liquid N2 in it
<DocScrutinizer05>
freezing makes materials less elastic, but not weaker
<whitequark>
maybe you don't smash whatever you store N2 in with a hammer
<whitequark>
(not elastic) but I see your point
<DocScrutinizer05>
a steel rope able to lift 1t isn't breaking at 100kg when you cool it down
<whitequark>
but if you put it under stress, it handles it much worse
<whitequark>
so it will *hold* 1t but probably not be able to *lift* 1t under same conditions
eintopf has quit [Ping timeout: 253 seconds]
<DocScrutinizer05>
and I honestly doubt a steel rope would break on bending it, even when cooled down to next 0K
<whitequark>
you tempt me to get some N2 and show you
<roh>
DocScrutinizer05: there are milling concepts using ice
eintopf has joined #qi-hardware
<roh>
because properly cooled it seems ice could be harder than steel
<DocScrutinizer05>
that's the point
<DocScrutinizer05>
steel won't suffer much elasticity loss from cooling it down, I think
<DocScrutinizer05>
just like it's not getting considerably more elastic by heating it up a 250 or 400 °C
<DocScrutinizer05>
aka K
<DocScrutinizer05>
it softens considerably at the temperature where burning skyscrapers break down
<DocScrutinizer05>
which iirc is ~600..800°C
<DocScrutinizer05>
for usual "normal2 steel
<DocScrutinizer05>
a SF I once read however claimed that water has a number of additional unusual aggregate states under conditions like they are found on Saturn, some of those states making it the better material for building constructions than steel
pcercuei_ has joined #qi-hardware
<whitequark>
DocScrutinizer05: that's true
<whitequark>
well, if your "buildings" will always stand in like 100k atm
<DocScrutinizer05>
yep, that's been the idea of that SF, talking about buildings on saturn
<whitequark>
Ice XIAn orthorhombic, low-temperature equilibrium form of hexagonal ice. It is ferroelectric. Ice XI is considered the most stable configuration of ice Ih. The natural transformation process is very slow and ice XI has been found in Antarctic ice 100 to 10,000 years old. That study indicated that the temperature below which ice XI forms is −36 °C (240 K).[54]
<whitequark>
ferroelectric!
pcercuei has quit [Ping timeout: 272 seconds]
<DocScrutinizer05>
>>At even higher pressures, ice is predicted to become a metal; this has been variously estimated to occur at 1.55 TPa[44] or 5.62 TPa.[45]<<
<ysionneau>
what? I didn't know Ubuntu was willing to follow Debian in its choice
<ysionneau>
I thought Ubuntu would keep Upstart anyway
rz2k has joined #qi-hardware
<viric>
debian too?
<ysionneau>
well that's the burning topic for weeks
<ysionneau>
debian next init system
<ysionneau>
there has been several votes, including a vote for resignation of the chairman of the technical comittee
<ysionneau>
things went bad for debian
<ysionneau>
but then it calmed down a bit some days ago
<ysionneau>
then the last vote was not successful in answering the question : upstart or systemd?
<ysionneau>
then the chairman decided with his casting vote : systemd
<ysionneau>
(only for jessie)
<ysionneau>
but I wasn't aware that Ubuntu was willing to follow debian in this choice, is it breaking news? or was it known befor?
<ysionneau>
+e
<larsc>
surprised me
root_empire has quit [Ping timeout: 245 seconds]
arielenter has joined #qi-hardware
uwe_ has quit [Ping timeout: 252 seconds]
uwe_ has joined #qi-hardware
Luke-Jr has quit [Ping timeout: 272 seconds]
Luke-Jr has joined #qi-hardware
pcercuei_ has quit [Ping timeout: 246 seconds]
arielenter has quit [Read error: Connection reset by peer]
arielenter has joined #qi-hardware
lekernel has quit [Ping timeout: 260 seconds]
lekernel has joined #qi-hardware
dos1 has joined #qi-hardware
wpwrak has quit [Ping timeout: 240 seconds]
apelete_ has joined #qi-hardware
jekhor has joined #qi-hardware
apelete has quit [Ping timeout: 272 seconds]
<whitequark>
hm... why are there so many couples around
<whitequark>
oh right, valentine's day!
* whitequark
briefly considers sending a postcard to his CNC mill
<larsc>
make sure to use protection
<larsc>
for your eyes
<whitequark>
:)
CYB3R has quit [Quit: Page closed]
<whitequark>
... I just received a small package from Iceland
<whitequark>
addressed to "Whitequark" (sic!)
<whitequark>
I have no idea who sent this, how they got my address or what's inside
<whitequark>
and this is the first time I ever see this tracking# by royal airmail
<DocScrutinizer05>
xkcd LOL
<larsc>
whitequark: open it!
CYB3R has joined #qi-hardware
<CYB3R>
And my board finally works!
<larsc>
nice
<whitequark>
larsc: can't, it's just a... notification
<larsc>
oh
<whitequark>
bascially a piece of paper instructing me to come to post office
<whitequark>
and it's 1AM
<whitequark>
I'm most puzzled about address. I'm generally not careful about my home address because why bother--if you want it, come and get me--but I don't think I've managed to left this one anywhere public
<whitequark>
e.g. definitely not in WHOIS info
<whitequark>
there is three to five people overseas who know it, and of course vendors who sent anything to me
<whitequark>
though amazon won't give them my nickname, only full name, so it must be either ebay (but they would likely write "whitequark_ru", my ebay username, unless they know me more closely) or ... something else
<whitequark>
and I have never, to my best knowledge, worked with anyone from Iceland
<whitequark>
I have always dreamed of receiving a package from someone I don't know so there's that
<whitequark>
DocScrutinizer05: re our yesterday resist discussion: I've looked up DS for my resist and its resolution is ~40µm
<whitequark>
so I'm close to theoretical limit in that too
<whitequark>
I'm writing quite a bit of parametric gcode. Not to mention DRO on my screen and live toolpath preview are handy.
<whitequark>
I'll maybe write something on top of that sometime. Thanks for the link though!
<DocScrutinizer05>
that sucks
<DocScrutinizer05>
I just found it, hadn't any look into it, beyond what's been written in latest c't hw-hacks mag
<DocScrutinizer05>
ct.de/ch1401128
<whitequark>
sure--I understand. hence me being grateful for the pointing. I've looked into writing a proper gcode interpreter to be run on some device and it is not very easy.
<whitequark>
probably not something an atmega can do, even! (requires quite a bunch of RAM)
<whitequark>
my best bet today is porting EMC's interpreter to run headless/linuxless
<whitequark>
so... if I order that they'll send me the last issue too, right?
<DocScrutinizer05>
they did for me
<DocScrutinizer05>
and for me last issue been like 2 months ago
<whitequark>
nice!
<whitequark>
okay, ordered!
<whitequark>
I actually didn't use the translator once while ordering it. Words have close enough roots to be understandable :p
wej has quit [Ping timeout: 252 seconds]
<DocScrutinizer05>
ouch, 12EUR for shipping to Russia
<whitequark>
I'm used to ridiculous shipping rates :p
<whitequark>
last parcel I'm remailing from USA cost me $45 in just *aggregation*, not even shipping
<whitequark>
shipping would be somewhere around $100 I estimate, based on weight
wpwrak has joined #qi-hardware
<whitequark>
wpwrak: welcome back!
<wpwrak>
phew. i wonder what was wrong. it looked as if the modem had just gone offline but then maybe i just didn't see the LEDs (too much daylight)
<DocScrutinizer05>
it seems like you were offline for quite a while
<DocScrutinizer05>
[2014-02-14 21:59:03] <DocScrutinizer05> it's basically what wpwrak (away?? :-o) suggested to build
<wpwrak>
yeah, i though i may have lost one phase so i also turned off the PC, and that one always needs to be off for at least half an hours before it will power on again
<wpwrak>
and then i just had lunch and took a nap instead of sitting there, waiting :)
<wpwrak>
"Not supported: Tool radius compensation" that's just where the fun starts ;-) (well, you can take that from my cae-tools/cameo)
<whitequark>
wpwrak: I never figured out how to make tool radius compensation work in EMC2...
<whitequark>
so, always did that by altering coordinates. thankfully, EMC2 does support []expressions--that interpreter does not!
<wpwrak>
i think few programs use anything but straight lines without the mill doing something fancy
<wpwrak>
also your parameters could just be evaluated if you generated your toolpaths in some script (y occasionally use perl or python for that. though i now usually go via cameo for things like offsetting and such)
<DocScrutinizer05>
whitequark: can you already access hacks-archive articles?
<whitequark>
wpwrak: right--I mostly used straight lines by now!
<whitequark>
I really need to figure out how to work with that compensation.
jekhor has quit [Ping timeout: 248 seconds]
<whitequark>
from what I could tell, EMC2 only supports it for lathes--weird.
<whitequark>
DocScrutinizer05: it instructs me to "Download des Zeitschriftenartikels"
<whitequark>
but I'm not sure where is the link for that
<whitequark>
oooh, an overview of oscilloscopes. it pays off already! :D
<wpwrak>
in cameo, i just send the path and tell it the tool size. it also features tool selection by size so i can, say, pick a drill for small holes and use the mill for larger ones. cameo then tries to assemble and close all the paths, then orders them by "insideness". then calculates the offsets for all the paths. the one(s) most on the outside have the tool run outside the line, the next inside (opening) have it run on the inside, if ther
<wpwrak>
e's an "island", it's again cut on the outside, and so on.
<whitequark>
wpwrak: cameo... that's some custom software for your mill?
<wpwrak>
it's my gerber/excellon/gnuplot to gnuplot processor
<whitequark>
ah, you wrote it. I see
<wpwrak>
my mill comes only with a pile of windows manure, of course ...
<whitequark>
vendor of mine actually advertised it as linuxcnc-compatible
<whitequark>
and even has documentation on setting up on website
<whitequark>
anyway... I'll look into cameo. seems like it's not quite the workflow I'd like to have.
<whitequark>
or maybe it is
<whitequark>
DocScrutinizer05: now that I think of scopes. is it possible to freely buy one in DE ?
<whitequark>
or just anywhere in Europe actually
<DocScrutinizer05>
why not?
<whitequark>
well, some absurd regulations make it impossible to import a scope to Russia
<whitequark>
and I'm not exactly fond of homegrown ones
<whitequark>
and whatever imported ones they sell have 3x markup, which means an already expensive piece of equipment becomes absurdly hard to get
<whitequark>
sometimes even 5x
<DocScrutinizer05>
incredible
<wpwrak>
cameo is basically the rock bottom fallback :) if i'll find something nicer at some point i'll be glad to use it. but for now there really isn't much that would tempt me
<whitequark>
for 2d I'd really want a simple dxf/svg(?) cam
<whitequark>
for example that UV exposing thing DocScrutinizer05 imagined, I'd likely be better using a 2D CAD and my head/paper to draw it
<whitequark>
and then cutting it out of acrylic/plywood
<wpwrak>
i used fped for 2D. the free 2D cads are all non-parametric. unusuable.
<whitequark>
so it basically just needs to understand arc/straightline contours and translate that to gcode
<wpwrak>
fped does straight lines, arcs, circles :)
<whitequark>
you and your NIH software :p
<wpwrak>
what it doesn't do are more fancy geometric operations. like intersections, tangents, projections, and such. that's basically by design since i don't want to overload the footprint designer functionality. but sometimes i feel the pull of temptation ...
<wpwrak>
yeah, very NIH, i admit :)
<DocScrutinizer05>
hmm, openoffice can't do parametric?
<wpwrak>
its main use is to take gerbers and excellon files and turns them into machine toolpaths (in gnuplot, for syntax conversion by a further script/program)
<wpwrak>
it can also help with the generation of 2.5D toolpaths
<wpwrak>
e.g., it has an "area clear" algorithm
<wpwrak>
(used opencascade for the nastier bits. there are some battles i gladly walk away from ;-)
<DocScrutinizer05>
oo draw is maybe no real CAD, but for sure it can do "parametric"
<wpwrak>
DocScrutinizer05: parametric means x = 5mm then you work with x. if you find you actually need x to be 5.1mm, you change it and everything that depends on it gets recalculated
<DocScrutinizer05>
so?
<wpwrak>
in ME you need to make such adjustments quite a lot. so you don't want to tweak every little thing.
<DocScrutinizer05>
hmm?
<wpwrak>
#s/want to/want to have to/
<DocScrutinizer05>
I don't get it
<wpwrak>
let's say you have 256 BGA balls. so you entered each of them, with some radius value. then you find out you have to make them smaller/larger. how do you do that ?
<wpwrak>
dumb cad (qcad and friends): you go to each of the 256 balls and edit the radius. smart cad (fped): you edit the radius value you defined somewhere.
<wpwrak>
of course, in fped you also have loop constructs and the like
<whitequark>
I think freecad may work for 2d
<whitequark>
I mean, you could just stop at the "sketch" step
<whitequark>
maybe even extrude it to the thickness of your material if you like
<wpwrak>
yes, freecad may be a possibility. it may get hard to follow the logical structure but it has good features
<whitequark>
I just wish it crashed less often
<DocScrutinizer05>
wpwrak: I'd probably select all of those bga pads, then select "Size and Position" from context menu, and edit "radius"
<wpwrak>
(well, it didn't quite work back when i tried it but it looked encouraging)
<wpwrak>
ah, so it still does that ...
<whitequark>
it worked well for me, if you can tolerate occasional crashes
<whitequark>
perhaps it's easiest to just get gdb and fix that.
<wpwrak>
DocScrutinizer05: heh, okay. let's make it some spacing then ;-)
<whitequark>
it asserts on something dumb like 0-vector where a 3-vector is expected.
<DocScrutinizer05>
eh?
<DocScrutinizer05>
like in "relative postion" ?
<wpwrak>
DocScrutinizer05: let's say a cut-out for some connectors. let's assume you have lots of them. you work with, say, 20 mm. but then you notice it doesn't fit. so you increase to 20.5 mm. welcome to tweaking all the coordinates :)
<wpwrak>
yes
<DocScrutinizer05>
umm, while your example doesn't work, I think i know what you meant
<wpwrak>
and fped also has automated measurements. so it shows you immediately what dimensions you ended up with. rather convenient when comparing footprints with data sheet drawings.