lordpil's blog

avr32 is a joke
id prob fuck with stm32 before doing arm7 again
it almost doesnt exist, its expensive, doesnt have much to do with avr hardware

Posted by renesis at 19:15 | permalink | 0 comments

writing interrupt code isnt so hard
actually i primary run in a timered interrupt
ill put super uncritical shit in a mainloop

Posted by renesis at 18:53 | permalink | 0 comments

hardly 10x capability, and same price in 100 qty
the avr is 10% cheaper at 25qty
i didnt look at opcode much but the architecture seems not as awesome for asm/realtime shit
also you dont have to crack your ide

Posted by renesis at 18:45 | permalink | 0 comments

how much?
yeah thats maybe the way to go
the logic + uC is prob within .50
prob as low as like .10
the decoders will be similar
so like ~.60 total
and you can prob get away with a 2 port uC
maybe $1.50?
i doubt it adds $2 in PCB space

Posted by renesis at 16:04 | permalink | 0 comments

one 3-8 decoder, 4 8bit shift regs, and you drive one of the column bits direct
thats 7 io
8 if you want addy latching on the decoder
tho i guess you could share the latch with the shift reg, so back down to 7
its likely cheaper than using a bigger micro
board space would be my secondary concern

Posted by renesis at 15:55 | permalink | 0 comments

shift regs need 3 io, decoder can be just 3 io
depends how much shit the micro is actually doing
and thats it?
yeah but it doesnt have to do comm or processing *and* display the message, does it?
uC can prob hack it at 100Hz refresh
like .50 in logic, maybe .75 in transistors, in short run quantities

Posted by renesis at 15:49 | permalink | 0 comments

so you would be driving 7*5 at once?
and splitting it up 5 ways?
oh so split 7 ways?
so you need 25 outputs?
or you want to like, multiplex the 5*5 too
how much io do you have?

Posted by renesis at 15:44 | permalink | 0 comments

or sorry i missed the 5x part
you have to buffer them
but that just takes 3904
yeah what moon said
welcome to electronics
you can always pay a PCBA house to do it for you
because it probably exceeds total current
but whatever thats easy
just use npn buffers
but yeah check how the matrices are wired up
kind of determines how you can split them up to multiplex

Posted by renesis at 15:39 | permalink | 0 comments

or shift regs
shift regs prob uses less io
multiplexing with the shift regs might be too slow but its prob the cheapest
its clunky as fuck too
not sure ive ever seen pro shit done like that
or any shit, just kind of thought of it now
ya single decoder chip will do that

Posted by renesis at 15:34 | permalink | 0 comments

21:27 < monkeyisl> ldkdjlkj this datasheet has typo :(
welcome to electronics
usually theyre way off
so its obvious
but yeah every once in awhile theyre close =(
over dimensioned drawings are my fav

Posted by renesis at 15:14 | permalink | 0 comments

samian: well it all depends how reliable the car is

Posted by renesis at 01:23 | 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