lordpil's blog

timecop: its a bit crashy
timecop: but im kind of doing like, all the things, and hasnt eaten much work
lots of 3d and multiboard

Posted by renesis at 20:46 | permalink | 0 comments

amazon wants me to subscribe to JST PH cables

Posted by renesis at 15:35 | permalink | 0 comments

rab: setting windows to shutdown at 5%, and its fixed
LED still turns on when i plug in charger, powers up without bootlooping a bunch of times
was set at 2% before
so yeah i dont think this shit has a bios battery

Posted by renesis at 12:46 | permalink | 0 comments

just get silver plated PTFE insulation
is it that big an issue?
Silver-plated stranded copper wire
ive never had issues with that shit in clamp terminated connections

Posted by renesis at 10:11 | permalink | 0 comments

naw, your kids are better off
like, building systems is cool, but component manufacturing is a lot of chemicals, weird processes
thats cool they kept it going with 30 exployees
3rd eye for 50% more QA throughput
microphones and planar speakers?
naw, they take very thin films and deposit gold conducting paths
oh that sounds like bs
but im pretty sure there is a deposition and etch method being used for some planar magnetic stuff
depends on how shit the terminations are
all that OFC stuff turns green
with the clear insulation
and the super fine strands

Posted by renesis at 10:06 | permalink | 0 comments

so i guess cornell dubilier was buying from the domestic panasonic factory, and when panasonic shut it down they bought it
this was 8 years ago tho
Panasonic said Cornell Dubilier decided to buy its foil division when Panasonic decided to end U.S. production of formed aluminum capacitor foil, which created a foil supply chain problem for Cornell.
wonder what caps were made in USA by panasonic
article says they were panasonics customer for 20 years

Posted by renesis at 10:01 | permalink | 0 comments

thats cool i guess
or they stole the panasonic pic because thats what they cloned for part specs
=\ =\ =\
rab: http://archive.knoxnews.com/business/panasonic-plants-end-foiled-cornell-dubilier-purchases-its-assets-ep-408128163-358646681.html
Panasonic spokesman Clark Brandon said Cornell Dubilier, which is headquartered in Liberty, S.C., restarted foil production Sunday using about 30 employees who formerly worked in the facility for Panasonic.

Posted by renesis at 09:56 | permalink | 0 comments

supercap kind of seems a lot simpler
i dont have to fuck with connectors and assholes stripping out sheet metal screw holes
rab, how come the dubilier cap has a matsushita symbol
like, in the future, all corporate logos are panasonic logo
just the pic?
do you?
i just see japan
i think a thing happened
oh fuck
yeah pasted in
doesnt look like the other
well, okay
so dubilier is panasonic caps

Posted by renesis at 09:51 | permalink | 0 comments

and i can get away with a coin cell since its a scheduled maintenance thing
oh shit
like, 5v or something?
i just put coin cell for now and maybe do that if i got time
PSU in this shit is going to be like, 25W mean well and some linear to-220 vregs
because retro cool
also i dont want to think much about thermal management
and i can make a prehating argument, for temp stability
its a weak as fuck argument
but its there
rab: yeah perfect! i can do 3.3v rail and a schottky or something, like that
because they would have to remove back panel for coin cell
or maybe i can just do panel mount 9V
i dunno
i will do a lot of drugs and ask the universe or something

Posted by renesis at 09:46 | permalink | 0 comments

rab: can you think of any reason why PLL boosted RC and ext TCXO wouldnt be okay?
seems to easy
high speed timers non critical, in terms of absolute timing
like, tcxo for long term alarm type functions with less user clock maintenance
RC for system loop, ADC and PWM clocks, out to max rec speeds
should be okay, right?
do you think supercap is fine?
industrial building
maybe just few days poweroff
i should read more about supercaps i dunno how long they last
and i can just write in change coin cell in annual maintenance or something
i dont think ive shopped supercaps in 10 years or something
blackmoon: right i just dont know what it costs really

Posted by renesis at 09:41 | permalink | 0 comments

yeah i got all sorts of altium PCBA and solidworks assemblies to do this weekend
rab: it might be a dead coin cell but im not about to heatgun the glass off to find out
but since it is a managed internal battery, maybe they said fuckit
like, shit gets so dead, the charge LED wont flip on for a few minutes
i trashed a cable over it
so like, i think this and the false negative tactile feedback on the keyboard are its super-china special functions
if someone called it trash over the keyboard thing, i would have no credible response

Posted by renesis at 09:36 | permalink | 0 comments

why i was like fuckit and got a non-w for $5
rab: my chuwi tablet/laptop has no button cell
i have the shutdown set low battery %
and like, i think its wiping the bios on power on
from like, a battery sag
resets the bios
like, itll boot loop until its charge enough to stay up
i think its just some battery plus watchdog shit
plus how low i have the shutdown charge % set
if at all, i wiped out the factory windows
it was some dual boot android trash, like bitch you dont have the flash for android
anyway it has an american megatrends bios

Posted by renesis at 09:31 | permalink | 0 comments

yeah if you just want to drive the GPIO connector, zero is fine
old one, no wireless, single usb
w was $5?
i though zero non-w was $5, zero w sells for under $10 now?
they were like $20 when they were barely in stock

Posted by renesis at 09:26 | permalink | 0 comments

jero32: yes
im figuring out multiboard so i can do one big schematic for the project
well, like, across multiple sheets with block diagrams for common stuff and shared net names
like, you just do regular projects, and do a System=Connector property for connectors between boards, and then you can associate a PCBA project with diagram block, and the connectors show up
i have to make a jst ph

Posted by renesis at 09:21 | permalink | 0 comments

omg i made it work
PCBA connector shows up in multiboard schematic

Posted by renesis at 09:08 | permalink | 0 comments

jero32: the extender or the stm32?
it has like, any interface you want, too
well you can use the controller to time it somehow

Posted by renesis at 08:44 | permalink | 0 comments

timecop: is module title the sheet association?
timecop im not sure why we keep you around if we have to look this shit up
using one as an io seems so straight forward and versatile its hard to justify app specific part
unless you need like, light speed
like, combinational logic latencies
im guessing you dont
its more firmware to track and burn so maybe doesnt make sense for mass production
or very fast prototyping

Posted by renesis at 08:10 | permalink | 0 comments

timecop: so is the PCBA rooms feature useful for multiboard design?
like am i going to be laying that shit out in one giant workspace?

Posted by renesis at 08:05 | permalink | 0 comments

thats the trade name
its not one product
like, you can get stuff that is almost like mylar sheets
almost rigid, not as much tho
or you can get shit that is like, fruit rollup
its basically like dried up paste in a sheet
right kapton is fine but everything has to be absolutely flat
kapton is strong but scissors slice right through it, is usually how i explain to people
if you clamp a chipamp to a coarse finish aluminum heatsink with a very thin thermal interface, you can bite through the insulation at the edges
ive heard of techs getting fired on the spot because LM3886 kept blowing up like that
its a fabrication and inpection issue, or its a interface selection issue if you cant afford that
i would go with the insulated package LM3886TF, because of that, and take the C/W hit
save you money in customer returns and bad PR

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

but there is about 100 versions of that stuff
some of it is solid, some of it is almost like cookie dough, falls apart
they make little open silicone boxes
you just throw over the to-220 and clamp
yeah you shouldnt really reuse thermal interfaces
its measurably worse like that
im just saying in general, when you have to redo it, youre supposed to clean it up pretty good
like, it will forms thick spots and the paste is actually really shit at transfering heat
its just better than air
yeah that sounds right
fingers are weird
static issues
think mylar will stick or repel from your fingers, randomly
if your fingers are really clean
because super insulator, and super light
*thin mylar

Posted by renesis at 06:55 | permalink | 0 comments

junction to case is like, temp with ideal heatsink
even with an aluminum planet as a heatsink, it will still heat up this much
to-220 is heroic in this datasheet, as in most
0.1 C/W
so with a heatsink, thats the TO-220 contribution to the temp rise
and you add them
lower is better
thermal resistance
its a series resistance
like, its very simplified, and only steady state, so kind of conservative
but if you go with those numbers and a bit of safety factor, youll do okay
you have to add the thermal interface, too
mylar, or transfer paste
that stuff is usually like 0.1 to 3 C/W
like, its usually a good idea to put safety factor in one place
not if the surfaces are very flat
but sometimes theyre not
so you can use mylar for insulation, and paste to make sure pits and scratches still get some transfer
it depends a lot on how its mounted, like how much screw or clamp torque, and how much paste
i forgot the trade name

Posted by renesis at 06:50 | permalink | 0 comments

thats basically what you are calculating
and directly, no
you know the junction temperature delta
assuming steady ambient temp, the junction temp is ambient temp plus the temp rise
like, if youre running in the cold, you may not be average power limited at all
in which case, you are probably peak current limited, which is probably when you burn up the bonding wire connections
hmm no
i think they called it virtual junction temp?
its a C/W number they came up with in test

Posted by renesis at 06:45 | permalink | 0 comments

so like, you put parts together, and you get thermal runaway conditions
not because the part is using more and more power
but because the parts are derating each other
like, their temp goes up which forces the ambient up which forces die temp up, etc
its like RF
the math is all very one dimensional until it gets in the air
then its very 3d, with a time component, and shit all adds up, literally
theyll add an air conditoner to the system

Posted by renesis at 06:35 | permalink | 0 comments

from a thermodynamics sense, its extremely one dimensional
its steady state so youre not even dealing with time
minimizing that is usually the job
thats why i like prototyping, you get something to work, then you start realing in safety factors
it basically sets your max ambient
like you will get a delta temp for your die
you subtract that from operating temp, and thats max operational ambient
but thats local ambient
so if you have a part running next to it heatink the air to 70C, thats ambient, not the 40C in the box

Posted by renesis at 06:30 | permalink | 0 comments

if the temps are too hot using junction to ambient thermal resistance, you calc using junction to case
and then you add the heatsink Rth*W temperature, and thats your system Rth
if you cant do that shit i wouldnt hire you
how do you select parts without knowing that?
cant do vregs, cant select heatsink, cant design with known temp headrooms
exception being SMD w/ specified pads for thermal considerations associated with specific operating conditions

Posted by renesis at 06:25 | permalink | 0 comments

so lets call it 9V dropped, so 1.8w?
23.5 C/W, junction to ambient, TO-220
so thats no heatsink
43C temp rise
150C operating temp
100C ambient is okay
no heatsink, 200mA, 12V to 3.3V
basically, at 200mA, if you can touch the case with spit on your finger and it doesnt sizzle, the junction temp is okay
so looks like youre friend is right
but you should teach him how to prove you wrong

Posted by renesis at 06:20 | permalink | 0 comments

link a mmbt3904 datashet
or any transistor
or any part
thats prob exactly where it derates to 0W with a heatsink the size of the universe
well, no usually not that bad
but sometimes
burny: its a matter of how much volume the watt is burnt in, and how much surface area is exposed to move the heat, and how it is coupled to the environment, and what that environments thermal characteristics are
jero32: naw screw you im just going to use TO-220 LM317
and maybe some LM78nn stuff
how much power?
like, whats load current, Vin and Vout?

Posted by renesis at 06:15 | permalink | 0 comments

oh and i wouldnt let you use altium either
thats totally unfair to your classmates
yeah i dont want to talk about project groups
like, theyre inexperienced and they never done this before and act as such
but like, thats the point of the project, so i cant be frustrated
i just have to like, do more work if i want it to be sexy
and try not to be mean while being honest
china is often like your classmates
WHY DID YOU DO THAT? We save you $3 on BOM, you say that mean $20. we test. BUT... NOOOOOOOOOOOOO
jero32: you have to show them C/W
show them how to do the math

Posted by renesis at 06:10 | permalink | 0 comments

you already know how to go File -> Open
the whole class should
now the whole class knows how to use eagles janked the fuck up UI
the old UI was waaaay less untuitive, but it was faster
fuck liking it, you make shit with it, do you like what you can make with it?
i liked what i could make with it
it was flexible enough i could workaround what it wasnt explicitely built to do
altium is 100x more awesome than eagle ever was, so far
but eagle will put whatever you need onto a pcb
for the money, eagle is fucking amazing
eagle: HOW WILL WE ROUTE THE PCB WE ONLY HAVE ONE PADS SEAT you can buy me eagle for $2k for this size board REALLY? JUST DO IT, WTF BUY THE THING
that was eagles price niche, basically

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

if you just use the peak currents, youll be throwing away a couple dollars maybe, but save you time and less likely to burn up

Posted by renesis at 05:50 | permalink | 0 comments

yeah they bought natsemi so theyre like, ultramega neat now
im not sure but i think the datasheet equations only pop out peak inductor current
they all like that now
page 11, TI chip has the same table, pretty sure
so like, 28 equations

Posted by renesis at 05:44 | permalink | 0 comments

but you want your thing to work
anyway using TI or Linear's supply designer stuff is pretty sure thing
there app notes usually have exact part numbers if the tools dont
like, buying a diff size inductor in the same part family is pretty safe
every switcher ive powered up that didnt splode, i felt lucky, and its because inductors
jero32: your book link can help you figure out exact rms power for the coil, like how to figure out the current offset and shape, thats what the datasheet equations dont have, besides the coil winding stuff

Posted by renesis at 05:39 | permalink | 0 comments

the datasheets are the opposite, its like one page of 10 equations
in most applications its either a DC resistance thing, so thermal
so a core saturation thing, im pretty sure that changes inductance vs freq
and yeah, any switcher that wants a transformer instead of a normal inductor is going to be more complicated to source
maybe not more complicated, but finding parts can be a problem to match what you want
if youre trying to make money

Posted by renesis at 05:34 | permalink | 0 comments

rab: what speed yours switch at?
i bet that sounds cool
i think i did mine like 14khz, little supplies tho
i just use internal switch
jero32: a lot of that stuff is condensed into a table of equations in a lot of the gen purpose pwm switcher datasheets
the example stuff at the end is pretty cool, tho
right thats why i said the stuff more towards the end is cool
but a lot of that is like, what is happening in the inductor in what switcher type, and how to pick inductor
jero32: well this actually explains it, and has nice graphs

Posted by renesis at 05:29 | permalink | 0 comments

i was going to use 12v and get a 10v LDO for the load cell bridge supply, but fuckit
just have a line of dinosaur parts
my VFD tube supply was like tiny rattlesnakes
and some SMD coils
that paper you posted was really good

Posted by renesis at 05:24 | permalink | 0 comments

$6 @ 100 not so bad
jero32: is this a prototype or a production design?
thats neat for prototyping
what about rework!?
does it need to work outside?
does the building have AC?
and heater?
air conditioner
is it like, comfy temp all the time forever
then if it works fuckit
if youre outside recommended operating conditions, usually problems show up at extremes of operating temp range
what is that
i think im going to use a 15V 24W supply and LM317 for everything, for my controller box thing

Posted by renesis at 05:19 | permalink | 0 comments

really that one shouldnt work with 3.3v
assuming 4.5 for vdd , 3.6 v is lowest high level
is this crazy high speed?
can just level convert with a bjt or something
i would prob just but another stm32
how much is that thing?

Posted by renesis at 05:14 | permalink | 0 comments

.5mm pitch
cmon .020" is huge!
drive a truck through those pins
or you can tin and heatgun until the chip blobs flat
but if you dont line up right and the pad is well isolated, you might have some problems
yeah what i says how you do a hand assembly for mass production board
i would use that one
the op conditions are a ratio of Vdd
for input levels
the other one, the levels were specified as voltage offsets
and the worset case offsets added up to more than jeros 3.3v
i might be confusing part numbers im just talking about the two parts posted

Posted by renesis at 05:09 | permalink | 0 comments

blackmoon: can i use this for a co2 nerf cannon?
i wouldnt
look the operating conditions
low level is 0-2v
high is Vdd-2 to Vdd
so like, the answer is maybe and if it stops working its your fault

Posted by renesis at 05:04 | permalink | 0 comments

if you didnt show pics of the box open and empty, i think it would be fair for ebay to consider if you misled them
if you showed a pic of the product it contained, alone, i think you should give the money back
like, your title and description should definitely include EMPTY BOX ONLY or something similar

Posted by renesis at 04:10 | 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