lordpil's blog

lookup table so you can tune it

Posted by renesis at 14:33 | permalink | 0 comments

like, make 4ms max velocity, and make 0x2 something like 44ms
so with a 40ms window, full 7 bits, its like 300us sample time
you can just ignore past debounce
right its like edge of doable im guessing
40ms window, 5 bits, which isnt bad really, is 1200us
which is what youre doing, right?
like, call anything shorter than 4ms 127, call anything more than 44ms 1, and divide up the timing in between, and play with it
like, try and squish really slow but hard

Posted by renesis at 14:24 | permalink | 0 comments

so 3ms
anyway for testing, just jack that shit up as fast as the code works, then try and do a very slow hit that trips both switches
the diff between those times is where your resolution comes from
the thing is this shit isnt really velocity dependent its pressure dependent
like, you can get diff slow times
so i dunno, you can just decide a slow time
just press slow, itll always register the second switch if you smash hard enough
do you still get clear initial transitions?

Posted by renesis at 14:18 | permalink | 0 comments

try and figure out the delta between hardest hit and slowest that flips both switches
then divide that up by resolution and that slice is your optimal sample time
with a 5ms delta, 6 bits, its 78us/12khz
2ms delta, full 7 bits, 64khz

Posted by renesis at 14:12 | permalink | 0 comments

guys i need four m3 screws and do not want to pay for a large bag or shipping, but i live in america
is that the velocity value?
what am i looking at, thats the loop time?
yeah its like 4ms event
did you hit it hard?
do the same thing but hit it very hard
pretend is it a minority
it is
k what if you hit it super soft?
can you flip just one switch?
so thats your 1 val

Posted by renesis at 14:07 | permalink | 0 comments

like squeeze some squish fiber to fuzz a beam on a detector
code loop?
only thing i could think of was something to do with the matrixing
i dont want to write the ethics class memo
some shit on sustainability
is it timed or just flying?
make him pay you
timecop: that shit scope still works
the encoders havent got worse =\
that makes sense
can you make it go way faster?
otherwise prob get like 2 bits of velocity

Posted by renesis at 14:02 | permalink | 0 comments

yeah think so
can do aftertouch with that
cheap piezo discs for velocity were really sensitive to glue, inconsistent part to part
its possible hard hits were breaking them a little
fast cap touch, like pcb - silicone honeycomb - flex pcb - silicone pads, is maybe reliable and full range and smooth
or some optical shit

Posted by renesis at 13:57 | permalink | 0 comments

unlikely but some stuff is up there
some audio trash its just like 2 or 3 levels
i think a lot of solutions get fucked in mass production
cap touch and controlled materials is prob cheap way to get smooth full res
thats prob as famous as velocity pads get
some akai mpc
i think resistive ink

Posted by renesis at 13:52 | permalink | 0 comments

that shit has to be crazy fast if you want actual resolution in the velocity
im guessing the whole event is like 5-10ms
timecop: what kind of velocity res do you want?
velocity sense is a pain
id prob try and do some optical shit
well because you need to have a ton of counts between switch closures
yeah that shit has to be fast if youre trying to fill 7 bits of velocity
a lot of stuff is really course, tho, like just a few levels

Posted by renesis at 13:47 | permalink | 0 comments

we used piezos w/ sample nd hold circuits at first job for this
incredible sensitive to glue

Posted by renesis at 12:01 | permalink | 0 comments

instead of switch times?
you can do it on both switches and average
also how well does that work across like 1000 things?
need like a calibrated poker
garlic bagels are too hardcore

Posted by renesis at 11:56 | permalink | 0 comments

yes thats why i think they undersized the holes
they can get away with it
timecop: maybe its sticking
clean with alc or use springier things
be less stupid

Posted by renesis at 11:51 | permalink | 0 comments

so do like 2.2 or 2.4
creak like a genesis
so ew
for the keys?
supposedly that is mad standard
oh nm, not the keys
timecop: on top of soldermask?

Posted by renesis at 11:46 | permalink | 0 comments

do your things match their things?

Posted by renesis at 11:34 | permalink | 0 comments

print it and cut out the holes with a scalpel, mount the paper
thats what jero says
so theyre plastics are good and theyre prob biting the PCB a bit
because they are plastics.
thats just what the guy made the dxf and you say
it says 2mm in solidworks

Posted by renesis at 11:27 | permalink | 0 comments

jero32: for the holes to be .002" larger than the screws, the plastics would have to be perfect or youre going to bite the pcb
and like, its a self-tap screw boss, thats not going to happen the same way every time
like a vw bug
its half an RCH
.002" is
naw would prob just chew the pcb a bit
they would make that shit work!
maybe this is the solution to the locating features in the plastic sucking

Posted by renesis at 11:22 | permalink | 0 comments

actually is 2mm, threads are tiny deep
bad for baby
is theyre trilobular wouldnt they measure diff and never full diameter?
well 2mm nominal would be undersized on reg machine screws anyway
but right if the mounting hole ID is basically same size as screw OD, the interference is intentional or they suck at manufacturing, which i doubt
or their plastics are like .001" perfect
which i doubt

Posted by renesis at 11:17 | permalink | 0 comments

thats crazy tight
m2 holes would be more
theyre maybe biting into the board to stabilize it
like, theres M-whatever you tell them to make
rab: well then his dims are off because thats like .002 gap
no its just the size in mm
clearance holes, im pretty sure
oh, nm
right but i dunno if thats throwing in some gap for alignment tolerances, or because the 75% or whatever thread diameter is finished size

Posted by renesis at 11:12 | permalink | 0 comments

its easy to injection mold screw bosses for self tapping screws
well theyre not machine screws
so isnt really a standard
right im saying just do what works and dont worry so much about the screw dims
as long as they cant fall through, youre good!
its either those or sheet metal screws
which are usually pointy or close to it
meh, 3d render is so 2006
i havent played with it much since ultiboard
that sounds neat

Posted by renesis at 11:07 | permalink | 0 comments

what is this for?
man fuck m2
m2.5 is bad enough
M2, metric 2mm screw thread
how do you do double sided?
seems like if youre doing 2, might as well do 3
in the aluminum?
jero32: arent they plastic self tap screws?
self tapp screws!
rab: someones going to do it, because make bank because no one does it

Posted by renesis at 11:02 | permalink | 0 comments

neat, pi and stm32disco connected, not dead
have to get uart up on both

Posted by renesis at 10:38 | permalink | 0 comments


Posted by renesis at 07:55 | permalink | 0 comments

thats not even 1/8"
aight lets get high and solder kittens to the ceiling
blank here

Posted by renesis at 07:50 | permalink | 0 comments

things fit
actually no, i gt two 6" usb cables and one of them need to be 12"

Posted by renesis at 05:59 | permalink | 0 comments

shrug, imported back into dicktrace fine
i dunno why it destroyed the gerber that first time
if i did something wrong it would be neat to know what
im tired
must drill the thing

Posted by renesis at 05:36 | permalink | 0 comments

have to

Posted by renesis at 04:07 | permalink | 0 comments

boxman with a beard
i gotta do fucking lab report

Posted by renesis at 03:25 | permalink | 0 comments

in the space movie like everyone on the planet is helping them
also they would definitely need to consider velocity at contact =\
personal pref
marking all pins with all alt fns would be reusable
making a block with only the pins you are using, maybe with alt fn labels and pin labels, is clearer and faster
i usually leave pin number enabled, and label with GPIO port/pin and used alt fn
alternate function

Posted by renesis at 03:20 | permalink | 0 comments

dis why you save and clearly label all the pdf to local!

Posted by renesis at 03:10 | permalink | 0 comments

irl has google and app notes

Posted by renesis at 02:14 | permalink | 0 comments

probably a lot of people did
so tube boss told me thing he read, that workers have 4 stages of competence
1) unconscious incompetence 2) conscious incompetence 3) conscious competence 4) unconscious competence
like, you realize youre a fuckup, so you get good, but eventually youre good enough you dont have to think to do things correct
thats when you fuckup
a lot of shit i fuckup because went too fast, misread, seems more an issue on tests than irl

Posted by renesis at 02:09 | permalink | 0 comments

and i pointed that it woundnt work unless the table was 2^n long
and i dont think she realized she didnt %=, and she spent i dunno how long explaining mod to me
like, ive been programming for 15 years
rand() % something is like, every programming course ever
and she doesnt like it when you interrupt
when she finally understood what i was saying, she said it wouldnt matter
because processors are very fast and you wouldnt hear it
clearly electromechanicals are not her thing, she was all confused by pot deadzones, too
anyway, processor is fast, but acoustic damped sprung mass system isn't, by design
making the tiniest kickdrums and shit
(impulse followed by decay at resonance before settling)

Posted by renesis at 02:03 | permalink | 0 comments

so i once read that a huge amount of spoken language is initially misunderstood
like, the fine details are lost quite often
at least in engineering, ive noticed this a ton
irl, mostly people are bullshitting because meningless social normsl so who cares if people walk away with two diff accounts if an interaction
engineering is like, making sure people who think they understand actually heard the right thing and know wtf your objective or point is
like, its hard to write hundreds of one or two sentence problems without ending up ambiguous sometimes
maybe its diff in your language, english has a reputation
so like
prof i had issues with did % instead of %= for some counter reset thing, for a wavetable counter

Posted by renesis at 01:58 | permalink | 0 comments

11:44:18 <+jero32> what generation of programming languages was assembler?
assmbler isnt a launguage, its a program that a compiler uses
kind of
its a category of languages, assembly isnt really a language either
its basically mnemonic machine code
its the 1st generation of code
if youre not including actual machine code
in which case its the second generation of code
because your fellow students are dumb
and maybe dont know anything, because prob an intro class
assembly isnt machine code
personally i think its debatable, is the point im making
was it multiple choice or what it fill in the box?
if it required full answers, usually you can bullshit enough to make it clear you understand and get the points

Posted by renesis at 01:53 | permalink | 0 comments

Top | Add to Technorati Favorites

© 2007 lordpil.   XHTML 1.0! CSS! Site design by GNAA  Blog Engine by pbx | MULTI2 | ian hanschen | lolwat