lordpil's blog

you could trigger the switch with a differentiator driving parallel RC hooked onto a comparator
if the shit dont change for awhile, the RC after the differentiator drains and the comparator turns off the fet
or maybe use the RC to drive the fet direct
wtf why to simulate my code without some low wage worker counting to 4 and hitting a button all day
anyway mine would loop back through the samples, and then subtract the numbher of times it loops back from the value it loads into the hold timer

Posted by renesis at 18:02 | permalink | 0 comments

eh it looks fine
it holds a peak for 4 seconds, then it looks for new peaks
slow falls simulated the diode in series with parallel RC
actually yeah mine is vanilla sample and hold with reset button
haha 4 second idle watchdog
ive already replaced the song
w/ loadstar - link to the past
mine takes a sample at whatever moment the signal is higher than whats held
its a diode in series with a cap with a switch in parallel with the cap

Posted by renesis at 17:57 | permalink | 0 comments

this works any signal
could do it on an 8b micro?
it just requires a bunch of parallel buffers
its a spectrum analyzer
i know i saw
i adjusted
but you see why it doesnt work for my plan
because mine you have to preserv samples for as long as the hold time
to look back for the next highest peak
so for like 4 second hold @ 25 fps and 10 bands you need to preserve like 1000 samples
yeah if you were paying attention
you would notice it doesnt hard drop

Posted by renesis at 17:52 | permalink | 0 comments

because latency
okay to do this awesome
when your timer for your current peak value runs out
you go back in the rolling average
which has to be at least as long as your peak hold time
and you find the highest value
then subtract the time between that value and now from your peak hold time
and use that as your timer
and itll always show the highest peak in the hold time
um you have to impliment a buffer for a rolling avg anyway
um what
okay well my shit is perfect and functional and all that bullshit
did you read what i said?
itll hold the 0.8 until the hold time expires
because when hold time expires it looks back in the buffer

Posted by renesis at 17:47 | permalink | 0 comments

peak just means highest value and hold is how long it holds the peak value
how is that wrong you can do it either way
shit you could do it with a diode into a parallel RC
no thats an accurate peak detection circuit
yours is just a cool effect
if you set the hold time to 5sec with mine, itll show the highest peak in the least 4 seconds
well, no not really
but i could make it do that with a rolling average
no you clear it
but not really
its a parallel RC with a diode
which isnt a LPF
you can call it assymetric LPF, its a sample and hold with poor retention =)
its a diode in series with a parallel RC
that follows a rectifier circuit

Posted by renesis at 17:42 | permalink | 0 comments

lpf what?
its a spectrum analyzer
k i need a window comparator
this is gonna be ugly

Posted by renesis at 17:25 | permalink | 0 comments

infinite hold is lame
and i fixed it
i need to decrease verticle res now
macegr's makefair thinger

Posted by renesis at 17:20 | permalink | 0 comments

haha i commented out your peak amplitude decay lines, and now the peak hold is stuck

Posted by renesis at 17:07 | permalink | 0 comments

oh i see how you did it
the color is based on height

Posted by renesis at 16:54 | permalink | 0 comments

that shit is functional, useful, yours is like, very slow fire bounce
also there is more ) in that but whatever
macegr: i want individual px control whats with the antialiasing
im trying to figure out how to make the spectrum analyzer function draw blocks

Posted by renesis at 16:49 | permalink | 0 comments

trying to figure it out
then i have to make the native stuff work at output resolution
omg wai
if ((new_sample > peak_value) or (hold_time = 0) peak_value = new sample;
if ((new_sample > peak_value) or (hold_timer = 0) {peak_value = new sample; hold_timer = hold_time;}

Posted by renesis at 16:44 | permalink | 0 comments

macegr: wait and you peak hold droops
like it doesnt hold
macegr: it should ever fall
it resets
then gets pumped up
dude yours looks lame and overprocessed
and yes it does
it looks like its supposed to

Posted by renesis at 16:39 | permalink | 0 comments

use something with pin change (pc) ints and scan or just scan all on a timered interrupt

Posted by renesis at 12:10 | permalink | 0 comments

hmm also the peaks are filling the next pixel space up
whats up with that

Posted by renesis at 12:05 | permalink | 0 comments

macegr: ok we got problems
your spectrum analyzer fades
thats weird
also your peak delay is not long enough
but i can prob fix that
no im against weighting
haha i destroyed it

Posted by renesis at 11:57 | permalink | 0 comments

ur a blog

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

pipe flexes
the cnc router i worked on had a box iron tube frame, like 6" tubes
then it had the X rails bolted to that, solid stainless rods, maybe 3" thick
you could tell it was on ball bearings from how it rumbled, but i dont think it was like, ball bearing assemblies
not the normal bearings, like for spindles

Posted by renesis at 06:12 | permalink | 0 comments

my machine doesnt have any ball bearings except for the spindle
and im pretty sure most big machines are box ways, no bearing ways
they usually have recirculating bearings
and a lot of bigger machines its just flat metal on flat metal with lubricant bearings

Posted by renesis at 06:00 | permalink | 0 comments

ac-130u: why cant you find a local mill?
and wtf man how are you going to just buy a machine you havent seen

Posted by renesis at 05:55 | permalink | 0 comments

now shes a coworker i thought she was your boss
does she pay you
but digikey is like 1000 people probably
you are 1

Posted by renesis at 01:25 | permalink | 0 comments

im not going to work

Posted by renesis at 00:42 | 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