<cr1901_modern>
well if you click the link you get a PDF
* awygle
quietly adds to archive
<rqou>
so, i'm going to be brave and trust a kicad bga footprint
<rqou>
wish me luck?
<rqou>
oh wtf?
<rqou>
apparently that error i got earlier means "i ate your schematic symbol, fuck you"
<rqou>
huh, this is actually really bad
<rqou>
at any time a kicad lib refactor can just go and ruin your schematics
<rqou>
azonenberg why are your oscillator footprints all so bespoke?
<rqou>
awygle: how do i calculate the dimensions of the footprint for an exposed pad given the dimensions of the pad?
<azonenberg>
rqou: what do you mean about my oscilators?
<azonenberg>
awygle / rqou: just thickness
<rqou>
azonenberg: your OSCILLATOR_XxY footprints
<azonenberg>
yeah what about them
<azonenberg>
theres multiple sizes of oscillator
<rqou>
they're very bespoke and don't have any silk/fab/etc. layers
<azonenberg>
i never use fab layers
<azonenberg>
and i think they have markings for pin 1
<azonenberg>
but in general most of my designs are high density
<azonenberg>
heck, my default 0402 print has no silk either
<rqou>
azonenberg: how do i figure out the correct size for an exposed pad without trying to pirate the ipc spec?
<azonenberg>
because i expect you to be using it under a bga
<rqou>
*the footprint for an exposed pad
<azonenberg>
rqou: i go off the chip vendor dimensions
<rqou>
no recommended footprint that i can find
<azonenberg>
i meant the package dimensional drawing
<azonenberg>
if thats all i can get
<rqou>
azonenberg: also, what have you been doing the whole day? i didn't see any commits from you :P
<rqou>
azonenberg: so pcb pad size exactly the same as package pad size?
<rqou>
infineon appnote says do it that way
<rqou>
TI has an appnote recommending a SMD center pad rather than NSMB
<rqou>
*NSMD
<rqou>
hrm, it seem some people definitely expand the pad a little bit
<azonenberg>
rqou: i havent pushed everything but i was also doing cleanout in the garage
ironsteel has joined ##openfpga
<rqou>
azonenberg: so does "expanded by 0.05mm from max" sound like a proper size for the exposed pad footprint?
<rqou>
azonenberg: what's a good amount of paste coverage for an exposed pad?
<awygle>
rqou: the Kicad people prefer slightly larger thermal pad than the one on the chip
<awygle>
50-70% paste
<rqou>
yeah, i've done 0.05mm larger each dimension, copied from some random appnote
<rqou>
wow, that's very little paste
<awygle>
That sounds totally reasonable
<awygle>
70% is the standard
<awygle>
Ideally relatively evenly distributed
<awygle>
And not directly above thermal vias
<rqou>
huh, it turns out that my "eyeballed" paste pattern is actually 52% coverage
<rqou>
so good enough :P
<rqou>
no thermal vias in this footprint lol
<rqou>
if your 240 LE CPLD is dissipating more than 1W you already have a major problem
<rqou>
azonenberg: what oscillators do you normally use?
<rqou>
the full model number
<rqou>
observation: BGAs are smol :P
<rqou>
e.g. they're only slightly bigger than the area taken up by right-angle pin headers
<rqou>
QFPs are lorge :P
<awygle>
Neither of those is necessarily true lol
<rqou>
well, i have a qfp144 and a bga256
<rqou>
oh wtf the kicad pads are _rounded_
<rqou>
i guess that works
scrts has quit [Ping timeout: 256 seconds]
<rqou>
awygle: does kicad finally have the ability to save layer colors yet?
<awygle>
*shrug* no idea
<awygle>
I didn't change the defaults
<rqou>
the defaults are almost unusable imo
<awygle>
You're not necessarily wrong lol
<awygle>
Good luck with PCBs, fireworks are over, nobody's building furniture upstairs, I'm going to bed
<ZombieChicken>
Has Lattice made any comments about the open source toolchain for (some of) their FPGAs?
<rqou>
yes, but the symbiflow team won't share them
<ZombieChicken>
but apparently nothing like like a C&D?
<rqou>
no
<ZombieChicken>
That's good to know
<rqou>
ugh, all of these color palette pickers seem to suck for some reason
<ZombieChicken>
What do you need one for?
<rqou>
pcb layer colors :P
<ZombieChicken>
Just do it all in Red, Blue, and Yellow
<ZombieChicken>
iirc, that's enough for a map, so it should be enough for a PCB
<ZombieChicken>
Or did you need 4 for a map?
<azonenberg>
You need four
<ZombieChicken>
R, G, B, & Y then
<rqou>
i'm basically doing that
<rqou>
not quite exactly like that though
<rqou>
since there's a _lot_ of layers in a pcb
<rqou>
i'm currently doing red/purple/orange/blue for cu layers
<rqou>
yes, heretic i know :P
<ZombieChicken>
Purple is a nice color
<ZombieChicken>
think it clashes with orange, though
<rqou>
red/blue are the heretical colors
<rqou>
yes, i deliberately want color clashes
<rqou>
so i can tell what is supposed to be on what layer
<ZombieChicken>
sounds like me and my hot pink duct tape
<rqou>
anyways, red/blue for top/bottom is EAGLE default colors
<rqou>
red/green are kicad default colors
<rqou>
but i'm used to the eagle way
<ZombieChicken>
How are you liking kicad?
<rqou>
it's okay
scrts has joined ##openfpga
openfpga-github has joined ##openfpga
openfpga-github has left ##openfpga [##openfpga]
<openfpga-github>
[openfpga] cr1901 opened pull request #124: gpcosim VPI module and iverilog tests now compile on Windows (MinGW64). (master...vpi-win) https://git.io/fbpXc
<cr1901_modern>
Took me long enough
<rqou>
somebody sure wanted to shave some yaks today :P
<cr1901_modern>
Maybe, but seeing so many entries on my Todoist irritates me, so I'm trying to get old ones done before doing new ones
<cr1901_modern>
s/doing/making/
<cr1901_modern>
Oh and xptools currently needs 3 lines added
<rqou>
yeah i hate how everything is in separate repos
bitd has joined ##openfpga
<rqou>
anyways, azonenberg now just needs to actually make gpcosim useful :P
<rqou>
so, anybody have good ideas how to make this smaller?
<rqou>
azonenberg?
<rqou>
a huge problem is that PMODs must be on the edge of the board
soylentyellow_ has joined ##openfpga
soylentyellow__ has quit [Ping timeout: 256 seconds]
<rqou>
final size 5.3x4.1 in
<rqou>
hrm
<rqou>
i guess i can afford that
<rqou>
PMOD is a huge pain in the ass
<rqou>
since it has pretty precise mechanical specs
<cyrozap>
rqou: 1. Replace the giant TQFP parts with smaller BGA versions. 2. Move the PMOD connections to a cheaper, 2-layer daughterboard connected with an FFC or something.
<rqou>
but FFCs are a pain
<rqou>
also, one of those TQFPs cannot be switched
<cyrozap>
Also I see you're using an entirely-SMD micro-USB connector. Bold move...
<rqou>
i've had good luck with it?
<rqou>
i've used it many times
<cyrozap>
I like to be able to dangle devices by the cable without any fear of ripping off the connector. Not that I'd actually _do_ that, but I've had so many bad experiences with products that use those SMD connectors that I've decided to never use them in anything I design. :P
<cyrozap>
*cough* DIGILENT *cough*
<Prf_Jakob>
rqou, cyrozap: Use a SODIMM connector?
<rqou>
wrong thickness of pcb for oshpark
<Prf_Jakob>
Ugh right
<cyrozap>
rqou: You could also use B2B connectors and skip the cable. Give your PCB some wings ;)
<rqou>
lol
<rqou>
which company's marketing-speak was "wings" again?
<cr1901_modern>
papilio
<cyrozap>
Shields, capes, wings, hats...
<rqou>
it's so dumb
<cyrozap>
When is anybody going to make a board that uses skirts/dresses/bonnets/bows?
<cyrozap>
Call it "board-chan"
<rqou>
somebody already used bonnet
<rqou>
personally i'd prefer if we stopped using these silly names that are only cute the first time
<rqou>
also, azonenberg: have we decided yet what we want to do about the VPI headers in a post-oracle-vs-google world?
<rqou>
cr1901_modern: so my undef is not sufficient?
<azonenberg>
i'll let you two figure this out, but either way that PR is not getting merged
<azonenberg>
as it seems to be in the wrong place
<rqou>
azonenberg: and VPI in a post-oracle-vs-google world?
<cr1901_modern>
rqou: I would say no it's not sufficient
* cr1901_modern
makes a test
<cr1901_modern>
rqou: __WINPTHREADS_VERSION is defined on my machine
<cr1901_modern>
So move the #undef error out of that #if-block and into a new "#if defined(__MINGW32__)"-block or move the #if !defined(__WINPTHREADS_VERSION) test _into_ the existing #if-block
<openfpga-bot>
[jtaghal-apps] azonenberg pushed 1 new commit to master: https://git.io/fbjej
<openfpga-bot>
jtaghal-apps/master c169454 Andrew Zonenberg: jtagclient: Changed default verbosity for device-count message
<azonenberg>
LOl
<azonenberg>
LOL*
<azonenberg>
the coresight device ID for a cortex-a9