<czr>
damn, I wish I would have read the milkymist thesis couple of years back
<czr>
it has the best explanation of how DRAMs are interfaced that I've seen so far
<czr>
(like, not the most comprehensive, but best overview)
<wejp>
hello everyone, i already mentioned it here a while ago, but for those who might have missed it: i have release a new version of Gmu: http://wejp.k.vu/gmu/gmu-0-7-2_released
<wolfspraul>
not missed
<wolfspraul>
:-)
<wolfspraul>
wejp: did we uptick the version number in openwrt-packages already?
<wejp>
i'm not sure
<wolfspraul>
ok
<wolfspraul>
maybe not yet...
<wpwrak>
wolfspraul: quick update: chatted a bit with rsharpe yesterday. he'll need some boards before he can do much. (can't really recommend manually making them, for they have gotten quite difficult)
<wpwrak>
wolfspraul: (difficult) as in taking about 0.5-1 days to make one atusb board. atusd should be a bit quicker, though.
<wpwrak>
wolfspraul: by the way, if you run into any RF experts, please send them my way :) a review by someone with actual real-life experience with such stuff would be very useful.
<wpwrak>
wolfspraul: and also interpretation of measurement results. i feel a bit like an alchemist with my experiments ...
<wpwrak>
wolfspraul: on the good news side, i finally got two "new design" atusb boards to behave. dunno what happened with the first two i made - maybe they didn't like the hot plate. RF looks better on the new ones than in the first generation. there are still some kinks to work out, though.
<wolfspraul>
ok
<wolfspraul>
lots of work :-)
<wolfspraul>
RF experts, well, I definitely met some at 27C3 but missed the chance to try to get them hooked
<wolfspraul>
but that's not a lost cause, thanks for the reminder
<wolfspraul>
chat with rsharpe sounds good, seems that's moving
<wolfspraul>
I'm working on the kicad-scripted stuff again
<wolfspraul>
want to finish it
<wolfspraul>
currently I am a little surprised by how eeschema handles the auto/a4/a plot setting for postscript files
<wolfspraul>
whether I have a4 or a3 pages, they always come out right when I set the dialog to 'a4'
<wolfspraul>
but if I set auto for an a3 page, it's cut off
<wolfspraul>
auto for an a4 page is the same as setting a4 explicitly
<wolfspraul>
finally 'A' just seems to be auto + extra margin on the right side (white space)
<wolfspraul>
so I think I will just always default to the 'a4' setting inside kicad
<wolfspraul>
that's the only one that correctly prints an A3 page as well
<wolfspraul>
and I will not expose this auto/a4/a stuff via the cmdline, since the logic seems broken inside KiCad in 2448 at least
<wpwrak>
wolfspraul: maybe "auto" expect you to have physical pages of the corresponding size ?
<wpwrak>
wolfspraul: btw, how was 27c3 ? i already heard of the mess they made of sebastien's workshop :-( but how was the rest ? ah, and did you manage to sneak in ? ;-)
<wolfspraul>
it was very good
<wolfspraul>
of course I went there to meet people and get little tasks done, and that's exactly what happened and what I was able to do
<kristianpaul>
wpwrak: i forgot to mention yday that i got repeated data when reading data from a single addres in ths sram (let see how it fits with the read-pointer)
<wolfspraul>
so I met with 5-10 people, and all good
<wolfspraul>
maybe even 15
<wolfspraul>
if you want critical feedback, well the ticket handling was messy (but there are many pros and cons so that was just the best they could do)
<wpwrak>
kristianpaul: (repeated) and you don't expect repetition in this case ?
<wolfspraul>
I didn't manage to listen to a single talk in the big room, because of a very German handling of 'room full'
<wolfspraul>
'exit only'
<wolfspraul>
'door over there'
<wolfspraul>
bla bla bla
<wpwrak>
wolfspraul: (5-15 people) pretty good
<wolfspraul>
like a stupid program
<wpwrak>
wolfspraul: (room full) so not the egyptian way :)
<wolfspraul>
if (room_full) close_doors();
<wolfspraul>
if (people_slowly_leaving) ignore();
<wolfspraul>
keep_doors_closed();
<wolfspraul>
if (people at entrance door sent to exit door, where they see streams of people coming out) still_ignore_that full-people exiting == non-full
<wpwrak>
wolfspraul: or for that matter, often the argentine way. if you can stuff more people in and no authority figure not on your below-the-table payroll is watching, do so
<wolfspraul>
whatever
<kristianpaul>
wpwrak: nope, that why my initial data dump get bloated of zeros..
<wolfspraul>
:-)
<wolfspraul>
and so on
<wpwrak>
kristianpaul: could it be caching on the cpu side ?
<wolfspraul>
but the event was great, definitely
<wolfspraul>
I bought a ticket from someone who bought it early then got sick and couldn't come.
<wolfspraul>
half price, because I found that one only on day 3
<wolfspraul>
it was a big mess
<wpwrak>
wolfspraul: ah, perfect :)
<wolfspraul>
well
<wolfspraul>
many people get to suffer
<wolfspraul>
at least I am pretty good at surviving in chaos
<wolfspraul>
training in China...
<wolfspraul>
always hold onto your cash, as long as possible
<wolfspraul>
very important rule
<wolfspraul>
only spend the smallest amount that gets you one little step further
<kristianpaul>
wpwrak: catching could be, i was also reading and could be some setup in the EMC wich  SIE uses to interface sram-on-fpga, there are some cpu cicles are adding when reading/writing,
<wolfspraul>
do not try to understand the whole
<kristianpaul>
the best i got was 4 reapet data per reading :-)
<wpwrak>
kristianpaul: ah, so it does change eventually ?
<wolfspraul>
but the critique is just one part, overall the event was great
<wolfspraul>
all organized by volunteers, that's very admirable
<kristianpaul>
wpwrak: yes, well a counter implmentend on fpga was filling ram (from top to end)
<kristianpaul>
wpwrak: thats when i realized my *only* way to run out this problem was fill sram, then tell cpu that can read, wait for a ack-read and fill again
<wpwrak>
kristianpaul: hmm. may not be caching then. odd.
<wpwrak>
wolfspraul: sounds cool
<kristianpaul>
wpwrak: EMC allows to configure Setup(TAS), Wait(TAW) and Hold(TAH) cycles.
<kristianpaul>
wpwrak: i think is around that
<wpwrak>
kristianpaul: still odd. but yes, timing problems could cause all sorts of issues.
<kristianpaul>
wpwrak: but previous SIE examples with an ADC i saw a register wich points when sram is fill
<kristianpaul>
so following that desing rule it looks like you need:
<kristianpaul>
1. tell buffer to start collecting
<kristianpaul>
2. poll register in order to see when sram is full
<kristianpaul>
3. read sram
<kristianpaul>
4. go 1 again
<kristianpaul>
Any way i'll try thr ring thing
<wpwrak>
kristianpaul: good luck ! :)
<lekernel>
what's the best (i.e. with as little time as possible spent on sysadmin) wiki engine with captchas?
<lekernel>
mediawiki sucks
<lekernel>
pain to install/upgrade, no captcha, pain to modify, pain to transfer
<wolfspraul>
I have captchas on the qi wiki
<wolfspraul>
it is very easy to upgrade
<wolfspraul>
for transfer, indeed I need to do a bit more work, I have a weekly tarball thing running but I don't think it's really 100% thought through...
<lekernel>
yeah, it's easy, but time consuming and extremely boring
<lekernel>
it's like peeling potatoes
<lekernel>
I prefer things like wordpress that just work and upgrade and install plugins with a single click
<wolfspraul>
or just use github
<wolfspraul>
I think github has a pretty neat wiki now that is git based
<wolfspraul>
Werner would drool... :-)
<lekernel>
mediawiki needs a SQL database, so the tarball isn't enough..
<wolfspraul>
I understand. the 'transfer' thing is difficult since you need to think about the destination/purpose.
<wolfspraul>
offline viewing is one, cloning the server another
<wolfspraul>
there are also privacy issues since the account passwords are in there
<lekernel>
ah, yeah, github has an integrated wiki
<lekernel>
neat
<wolfspraul>
I still haven't gotten the openid login to work, that's next on my mediawiki todo list
<wolfspraul>
so in the long run I hope to have less and less passwords on the qi server
<lekernel>
I think i'll just put all the llhdl stuff on github only
<lekernel>
this is not a web/sysadmin project :)
<wolfspraul>
sure that's an option
<wolfspraul>
I don't agree with your broad 'mediawiki sucks' statement. Just to balance it out, I say 'mediawiki rocks'. :-)
<lekernel>
well, it has a few good sides too
<lekernel>
but it's a pain
<wolfspraul>
every tool has a 'minimum maintenance' load
<lekernel>
you can't even delete users
<wolfspraul>
if you are below that, it will drag you down and you will hate the tool
<lekernel>
so when some spam bot registers and vandalizes your wiki, you can't simply delete its account and remove all its edits
<wolfspraul>
but you can know that load before, and then focus on as few tools as possible
<wolfspraul>
I think mediawiki requires 1 full day / month maintenance, to keep it moving forward in a healthy way.
<wolfspraul>
I think roughly that's what I have at qi
<lekernel>
yeah, that's one day minus one minute too much for me
<wolfspraul>
I try to bring it down... but even that takes time...
<wolfspraul>
yes, and that's why you have mediawiki now
<wolfspraul>
hate
<wolfspraul>
s/have/hate/
<lekernel>
exactly :)
<wolfspraul>
I'm not aware of another wiki that is 'easier', not sure which features exactly you have in mind then.
<wolfspraul>
that's why I suggest just use github
<wolfspraul>
they seem to be doing a good job
<lekernel>
yeah, their git system is really great
<lekernel>
haven't really tried other features
<lekernel>
but it gets you going in no time and is really fast and convenient to use
<wolfspraul>
I'm very happy with the qi tools we have right now, let me see...
<lekernel>
you can do a lot with a single click... not like sourceforge which redirects you through 3 pages (select mirror etc.) when you want to download anything :)
<wolfspraul>
I will continue to work on the qi-hardware.com site to be like sourceforce/launchpad/etc., but for copyleft hardware, manufacturing process, etc.
<lekernel>
why not use github too?
<wolfspraul>
so the focus on kicad, qcad, etc.
<lekernel>
they're doing an outstanding job
<wolfspraul>
because things like schhist are the core of what I want and what I think will be valuable
<wolfspraul>
try to make that happen on github :-)
<wolfspraul>
there will be more along those lines. So I need to innovate on the server, to 'serve' copyleft hardware projects
<wolfspraul>
that has very different requirements from sourceforce/launchpad/github
<wolfspraul>
oh, and little things.
<wolfspraul>
like this channel being logged, searchable etc. really cool.
<wolfspraul>
I'm thinking about writing a mediawiki extension to have a scriptlet appstore
<wolfspraul>
but the kicad/qcad stuff is more urgent...
<zedstar>
wolfspraul: hi.....u see post about mit courseware/pdf search engine app? not sure if any interest in this kinda app or not but was hoping somebody might want to develop it further
<kristianpaul>
lekernel: how is usually sythesized a for loop in verilog ?(sorry for stealing your time :-))
<kristianpaul>
zedstar: the search is fast i saw
<zedstar>
kristianpaul: yeh it would scale to thousands of pdfs the search
<kristianpaul>
the for have just two blocking asigments inside
<wolfspraul>
zedstar: yes I saw it, and yes I am very interested. but it takes time to work down the pile of priorities...
<wolfspraul>
since the core of my work is hardware and manufaturing process related, I get to those higher level things all too little. hopefully more in the future...
<wolfspraul>
I'm a massive believe that we need to connect copyleft hardware to free data if we want to survive.
<wolfspraul>
free data requires free devices to run on
<wolfspraul>
so I don't want to get stuck on the low-level side, but that needs some fast progress now, so that's a priority
<wolfspraul>
zedstar: if nothing happens, just ping the list or here again in a month or two. We'll get to it :-)
<zedstar>
wolfspraul: sure.....i have to move on to other things also now but will see if anyone shows an interest to do all the messy build stuff to get it into an easy installable state
<wolfspraul>
what do you move to?
<wolfspraul>
how is the NanoNote shaping up for you, any wishes you have for next year?
<wolfspraul>
I think the base is coming together, so I'm fairly satisfied with the direction and progress.
<wolfspraul>
but... you are our biggest customer I think, so your feedback matters, not me padding myself on the shoulder :-)
<wolfspraul>
don't hesitate...
<zedstar>
wolfspraul: re: nanonote we gave a few out together with wikireaders to some indian students to evaluate for a pilot study....then aim is we going to nepal and i give them away there
<zedstar>
wolfspraul: me im personally very interested in seeing the RF additions!
<wolfspraul>
sure that's on! but the device itself needs a lot more and better software too
<wolfspraul>
don't you think?
<zedstar>
wolfspraul: yeh i think so.......for me i don't do GUI apps but that area could use some love......surprised nobody doing some SDL games from scratch or nintendo style apps
<wolfspraul>
work is still ongoin for nlove, that's sdl based I think
<kristianpaul>
reads xapp463 page 27
<wpwrak>
wolfspraul: (drool) by the bucket :)
<wpwrak>
wolfspraul: (sucks/rocks) it sucks rocks ? (-:C
<wolfspraul>
wpwrak: I think I can wrap up the next set of kicad scripting today
<wolfspraul>
for eeschema plotting, I will only do .ps and .svg for now
<wolfspraul>
.dxf is also nice but I'm not aware of any potential use right now, same for hpgl
<wolfspraul>
also, I will only do 'plot all pages', no ability to take out an individual one, or something similar to --list-layers in pcbnew
<wolfspraul>
I could do something like --list-schematics that prints a list of the daughterpages
<wolfspraul>
but we can wait until someone needs it
<wolfspraul>
as far as bringint this live on the server, unless there are objections I will create the files at the projects.qi-hardware.com/auto/project/latest url
<wpwrak>
wolfspraul: (eeschema) so you're basically rewriting/replacing my --plot ?
<wolfspraul>
I will keep --plot for backwards compatibility, no worries
<wolfspraul>
the only thing is that I am surprised that --plot works as well as it does
<wpwrak>
wolfspraul: (--plot) keeping the name would be good anyway. it's kinda intuitive :) i don't care much about my implementation
<wolfspraul>
as I wrote earlier I'm puzzled by the a4/auto stuff
<wolfspraul>
schhist is using it
<wolfspraul>
I don't want to break existing stuff
<wolfspraul>
so I want to keep --plot, the new one is --plotx :-)
<wpwrak>
(schhist) sure. but we can update it if needed. of course, as long as you don't pull the functionality :)
<wolfspraul>
definitely not
<wpwrak>
that sounds kinda ugly
<wolfspraul>
I'm just surprised about A4 stuff
<wolfspraul>
need to do some more testing
<wpwrak>
better to make a clean change. the only thing that depends on it is schhist anyway. and i think we know exactly where it's used ;-)
<wolfspraul>
ok then, --list-schematics added to the todo list :-)
<wpwrak>
or --list-sheets
<wolfspraul>
yes
<wpwrak>
maybe with hierarchical indentation :) maybe with filenames
<wpwrak>
things to watch out for: sheets can be nested more than one level deep. and they can be linked from more than one page !
<wpwrak>
i never tried all that weird stuff, so i'm not sure how my things break if you do :)
<wpwrak>
anyway .. gotta work a bit on the pyromania for tonight. electrical fire control. alas, i didn't have time to make an electronic launch system for the ben. maybe next year :)
<wolfspraul>
we have the power to define the 'good copyleft hardware' policies
<wolfspraul>
so we just outlaw more than one level :-)
<wolfspraul>
problem solved
<wpwrak>
(outlaw) heh ;-)
<wpwrak>
i think the multi-linked sheets could be used for repeated elements. e.g., if you have a robot with 5 H-bridges, you may have a H-bridge sheet and use it several times
<wpwrak>
but i'm not entirely sure if it works like this
<wolfspraul>
first I'll implement --list-sheets
<wolfspraul>
then we see what we learn from that, and what it outputs :-)
<wolfspraul>
I really hope the KiCad folks accept a few small patches
<wolfspraul>
just to cleanup logic
<wolfspraul>
in some places I have to hack into the kicad pretty ugly because the separation between gui and non-gui is not very good
<wolfspraul>
and there are many globals, static variables and functions, etc.
<wolfspraul>
but at least I know a lot of those spots now, so if they accept small logic patches we will slowly improve...
<wpwrak>
i think they like logic cleanup in general. may take a few iterations in case you're clashing with someone, though
<kristianpaul>
of course is ugly and not cleaned, not indented :S
<kristianpaul>
andres-calderon: Hi
<kristianpaul>
had you implemented a ring buffer in verilog not using directly xilinx lib of ram blocks?
<kristianpaul>
I just implemented something based on a xapp463 from xilinx but i will like something not so fpga-specific
<wolfspraul>
ok, --list-sheets also implemented
<wolfspraul>
4 different types of 'names', argh. you will see :-)
<emeb>
wolfspraul: quick question - do you know where to buy small quantities of the JZ4725 processor?
<wolfspraul>
yes
<emeb>
Cool. Any idea what the small qty price would be (~100pcs)?
<wolfspraul>
no, roughly guessing maybe 3 USD or so
<emeb>
wolfspraul: OK - thanks. Who sells them?
<wolfspraul>
Ingenic
<wolfspraul>
if you want to order, the problems will be in the practical details. first you need to contact them, talk about shipping, and how to pay.
<wolfspraul>
not sure how good their english is, and what shipping/payment options they can offer you
<wolfspraul>
I am not aware of a distributor, although they may exist. But I do know that several people, including Carlos I think, have contacted Ingenic and bought small quantities directly from them.
<emeb>
OK - sounds like it's still pretty informal.
<wolfspraul>
a normal Chinese company is all about sales. you call / email, say you want to buy something. great. they will serve you, as long as you pay :-)
<wolfspraul>
not informal, that's China
<emeb>
Heh.
<wolfspraul>
the problem will be how you get them paid, and how they ship to you
<wolfspraul>
inside china this is all super efficient
<emeb>
All very new to me.
<wolfspraul>
money can be wired within minutes, shipping within the country is fast and crazy cheap
<wolfspraul>
if you regularly order from them, you have their sales contact in your instant messaging client, for example msn or qq
<wolfspraul>
then it's a matter of a few minutes, and the package is on the way to you :-)
<wolfspraul>
all china = one big digikey
<emeb>
crazy
<kristianpaul>
we should live in china ;-)
<wolfspraul>
but they will only ship after the money is in their account, of course
<emeb>
but I wouldn't be able to read the street signs
<wolfspraul>
or you pay cash, as I do when I go to their office and need something :-)
<wolfspraul>
imagine go to Intel with some cash to buy chips, he he
<kristianpaul>
lol
<emeb>
good luck
<wolfspraul>
anyway this is how it works
<kristianpaul>
do you know who apart of sharism and carlos company sell/devel boards with Xbusrt chips?
<wolfspraul>
no
<kristianpaul>
:/
<kristianpaul>
is not for me but a friend is looking an alternative to arm or avr chips (because costs)
<kristianpaul>
k
<emeb>
wolfspraul: any idea what it would cost to get 100pcs of SIE?
<wolfspraul>
he
<wolfspraul>
where should I start...
<wolfspraul>
community projects are so much fun ;-)
<emeb>
true dat.
<wolfspraul>
I let Carlos and David speak for themselves wrt SIE, so I only give you strictly my part
<wolfspraul>
I will not manufacture any more SIE unless the things we learnt in the last run are improved.
<wolfspraul>
design problems, footprint problems, test software inefficiencies, etc.
<emeb>
eep!
<wolfspraul>
if those are fixed, I need a moq of maybe 200 or so
<wolfspraul>
100 is very hard
<wolfspraul>
at 200, the price / unit could be ca. 70 USD or so
<emeb>
Ah
<kristianpaul>
emeb: what do you need the FPGA in SIE (curios..) ?
<emeb>
kristianpaul: don't need the FPGA
<kristianpaul>
dont need SIE either then ;-)
<emeb>
just looking for a small/cheap JZ4725 board
<emeb>
for use in some audio products
<kristianpaul>
i wonder if ingenic have those kind of devel boards...
<kristianpaul>
ask then ! :-)
<wolfspraul>
correct :-)
<wolfspraul>
and how - easy - ask whether you can buy something - ask sales
<emeb>
Sounds simple
<kristianpaul>
mail sales@ingenic... < Dear Ingenic..
<emeb>
OK - thanks. Lots to learn about the Chinese way of doing business...
<kristianpaul>
please be carefull with enlighs
<kristianpaul>
english*
<emeb>
misinterpreting?
<kristianpaul>
i had some troubles when orderiong a RFM23B modules from hoperf :/
<kristianpaul>
because of that they missundertood my needs and declared the product including the shipping costs :S
<kristianpaul>
any way..
<kristianpaul>
confirm  twice !
<kristianpaul>
but ask first ;-)
<wolfspraul>
kristianpaul: he :-)
<wolfspraul>
you ordered yourself?
<wolfspraul>
great! congrats!
<emeb>
sounds a bit fraught.
<wolfspraul>
as I said, there are only 2 problems - payment & shipping
<wolfspraul>
because within china this is both crazy efficient
<wolfspraul>
so outside of china there may be issues - how do you get money to them, how do they ship/customs etc.
<emeb>
yep
<wolfspraul>
those are both nasty for 50 usd orders...
<wolfspraul>
so the overhead can easily be 3 times the value of the actual order
<wolfspraul>
which also they don't like, of course
<wolfspraul>
so the payment and shipping options they offer may not be sufficient
<emeb>
OK - I'm looking at 100-200qty, so the order would probably be a bit more than that.
<emeb>
(not so tiny that the overhead is disproportionate)
<emeb>
but... This is still in the 'blue sky' stage.
<wolfspraul>
wpwrak: alright, that's it, done and out :-)
<wolfspraul>
try it out and let me know (after you made it into the new year of course...)
<wolfspraul>
plus recovery
<wolfspraul>
I merged the cmdline stuff into one patch, scripted.patch, and moved that patch to the end of the quilt series file
<wolfspraul>
so if you uptick KiCad, and that will certainly break scripted.patch, then you just comment it out and I maintain it
<wolfspraul>
the patch adds 4 new files, and patches 17 others in KiCad
<wolfspraul>
total size of the patch is 1162 lines, of which 840 are in the 4 new files
<wolfspraul>
next step for me is to first continue on some other tasks, wait for your feedback regarding the new eeschema cmdline options, and then bring some of the cmdline stuff live on the server, automatically after commits...
<wolfspraul>
I'd say I should get to that within a week or two.
<wolfspraul>
if anything is missing or broken, let me know
<wolfspraul>
happy new year everybody, and hopefully another happy copyleft year as well...
<kyak>
indeed!
<wpwrak>
wolfspraul: (make into new year + recovery) you seem to know my routine by now ;-)
<qi-bot>
[commit] Wolfgang Spraul: broke scripted.patch into scripted-new.patch and scripted.patch for easier maintenance http://qi-hw.com/p/eda-tools/9b476d4
<qi-bot>
[commit] Cristian Paul Peñaranda Rojas: some basics utils to dump raw data to a file (work in progress) http://qi-hw.com/p/ben-gps-sdr/2f5e6fe