lordpil's blog


and i guess if i want to spray i can use circular and the half complete thing
and then i have to figure out how to turn that shit off
that sounds complicated fuck that
this aint even audio
tho neat i bet dma i2s is easy
wonder what kostix doing for his synth
k so tomorrow i do ACK shit, and if that gets sorted everything else should be easy
spray data to pi and write to files i can feed to gnuplot

Posted by renesis at 17:50 | permalink | 0 comments


timecop: k the hal driver for uart dma only wants to be called once, then i dunno something with the sync line from the pi was fucking shit up so i delet0red all that
and shit just works
no idea what happens if packets desync
maybe use the sync signal to deinit and reinit the uart
i wonder how tx works
yeah
surely its not going to continuously spray trash?
so just dont use circular mode?
cool

Posted by renesis at 17:45 | permalink | 0 comments


what is it

Posted by renesis at 16:23 | permalink | 0 comments


timecop: now i got it only working in debug mode, but one packet delayed
according to debugger dma is working
its like the problem is with the completion callback

Posted by renesis at 16:10 | permalink | 0 comments


am i only supposed to call that shit once?

Posted by renesis at 11:43 | permalink | 0 comments


timecop: how come my dma uart shit works in debug but not irl
like it gets a packet once and gives up irl
yeah i just changed that and still does same shit
works over and over in debug
wait now doesnt work in debug either
well its working more than without dma

Posted by renesis at 11:21 | permalink | 0 comments


wait i think it worked

Posted by renesis at 10:50 | permalink | 0 comments


w00t got dma rx working
timecop: keil does some weird shit
mostly w/ the live syntax checking, like if im outside of main itll spaz out on nothing
stuff that looks and compiles fine, shrug
timecop: when i hit debug it says eval mode 32k max
maybe i never put the activation?
shrug
wait lets see if dma works with >1 byte

Posted by renesis at 10:45 | permalink | 0 comments


i should buy an umbrella

Posted by renesis at 06:16 | permalink | 0 comments


anyway its dip, so easy to short out the amps later
also you should line up the edge of the power connector with the board edge because looks cool
and cap across the top right to pins is hardcore
there isnt one at all
on the upside youre going to do a great job of keeping your system noise out of the power supply

Posted by renesis at 06:08 | permalink | 0 comments


but you have not fixed the three noise generators
why do you think were saying put the cap and ground the amps
someones cell phone gonna ring and your shit gonna blow up
if you get the angle right they will do that even with caps
the 12V trace doesnt need to be parallel to the sensor trace for the horizontal run
you can bring the tracve straight up from the connector and keep the Q signal trace more isolated

Posted by renesis at 06:02 | permalink | 0 comments


555 is mostly just opamps
i think it has comparators?
so, shitty opamps
is this even getting fabbed?
what program is this?
learn pours, use them
well it made sense a few years ago
less now, schools usually lag reality by a few years
theres much worse than eagle, as least you are not using kicad or geda
why?
altium doesnt have standard library of parts?
anyway
right but grounding out amps amd adding a cap isnt end of the world
that said, you can do that better with pours for power and shielding around the sensor line
thats being that ass

Posted by renesis at 05:57 | permalink | 0 comments


jero32: is this for a prof or another student or some rando or what?
if its for a prof maybe dont fuck with the circuit, anyone else, fix it
if they dont want to buy the extra two cap they dont have to
fixit
you cant use jumpers?
whatever single layer is some mass production hacker shit
well you dont need any
?

Posted by renesis at 05:52 | permalink | 0 comments


also if you take 12v up and over you can keep it fat before splitting to quad amp and 555, and dont have to go between pins
then you can ground all the quad amps without a jumper
man fuck school i miss electronics
unless you think designer is a real mfkr, and hes going to rework the caps and shorts directly across the pins and have opamps available for rework sex

Posted by renesis at 05:47 | permalink | 0 comments


shawn|i7-: i dont know really, but reserved prob means unused in most application, so big company like OH YA RESERVED FOR US
jero32: i think you are talking about JP1 and he is talking about a vga cable
jero32: is the size of the board super critical?
i think you should bring ground from the right and use it to shield the sensor trace
same with ground from top left
but because right now you have motor flyback diode and motor power right on it
youre obviously pushing like .006 or .004 space trace, so might as well use it!

Posted by renesis at 05:42 | permalink | 0 comments


if Q1 burns traces you unground J1, which is maybe not ideal
also is more copper, and more star'd
really there should be a bigass cap right there and you should star to that but whatevers
also 12V trace is crazy close to tab
oh thats bottom thats cool
yeah move quad amp up and right, move r5 and r2 under it, put big cap there, and get a cap with wide pitch to jump traces from quad amp power pin to gnd on bottom
tell the circuit designer #electronics said so

Posted by renesis at 05:37 | permalink | 0 comments


and their flip out noise might get into your shit
:O
attach + to ground and put out to -
srs
catching it early means less rework tho
this needs to be single sided?
anyway, bump the quad amp right and 0R jump ground to the other side of the 12V trace and ground out the amps
dont be china
also i would run JP1 GND and Q1 GND to the power connector seperate

Posted by renesis at 05:32 | permalink | 0 comments


because people had a problem and someone had an idea
i like new bigger signal traces
ground them out
also moar copper pours
they can flip out and fuck up power and grounding in the chip

Posted by renesis at 05:27 | permalink | 0 comments


i dont think that will fit
timecop: hey so with f0 i can compile past 32kb i just cant debug?
yeah mdk activation page doesnt mention 32kb limit, so maybe its just debug

Posted by renesis at 04:50 | permalink | 0 comments


i guess i could make a Ks that based on angle instead of error and make it work and then make a block diagram and ask prof how to model it
i guess that doesnt break the one input rule, same input the error variable, and linearizing gravity seems like some useful irl shit
so i guess i have to make a sine to 2^16 table

Posted by renesis at 04:45 | permalink | 0 comments


guys where does one find tennis balls for free
tennis courts

Posted by renesis at 04:39 | permalink | 0 comments


Kp need sin function
*needs sin
because mg for the pendulum has a sine components
thats why enough p to keep it wobbling around tdc isnt enough to save it if you push it past maybe 20 deg off
i can just model spinny platform linear because prob everything else scales the same anyway
somehow in the end it all comes back to bacon sim technology

Posted by renesis at 04:34 | 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