Lord_Nightmare has quit [Ping timeout: 255 seconds]
Lord_Nightmare has joined ##openfpga
promach has joined ##openfpga
futarisIRCcloud has quit [Quit: Connection closed for inactivity]
noobineer has quit [Remote host closed the connection]
<whitequark>
awygle: hi
<whitequark>
what *is* your timezone anyway?
<awygle>
whitequark: o hai
<awygle>
PST. Or PDT. Whichever one it is now...
<awygle>
West coast US, same as azonenberg
<whitequark>
kk
<whitequark>
so regarding next steps on glasgow-jtag, can you help me with the iCE40 footprint?
<whitequark>
I'll probably be fine with the smaller ones, but this one scares me
<whitequark>
lemme do something with issues so that we don't duplicate work
<awygle>
whitequark: sure, no problem
<whitequark>
"Non-Configurable Chip Select. This pin must be hard wired to logical 1 state (VCC). Device will not operate with this pin left floating or held to logical 0 (VSS).
<whitequark>
lol what
<whitequark>
oooh we can use a cheaper EEPROM
<awygle>
lmao Good Chip
<whitequark>
... hm no
<whitequark>
two EEPROMs are cheaper than one EEPROM with sane addressing
<whitequark>
awygle: do we try to add e.g. ON Semi CAT24C256 as an alias for MCP 24LC256?
<whitequark>
I wonder what the KiCAD policy is
<awygle>
that seems aliasable to me but i have no idea what kicad policy is
<rqou>
i still can't believe you trust kicad libs
<rqou>
let me know how it goes
<awygle>
it's amazing how good my fat cat is at just *poof* vanishing
<rqou>
lol cats
<rqou>
need more catpix :P
<whitequark>
rqou: you misunderstand
<whitequark>
in my case, "kicad libs will be trustworthy" isn't an observation
<whitequark>
it's a threat
<rqou>
heh, i personally wouldn't burn the time/money to make sure they're all correct
<rqou>
whereas if _i_ mess up my own footprints then it's definitely my fault
<whitequark>
that doesn't make any sense.
<whitequark>
even if I spend as much time validating them as I would validating my own footprints, that's still a win because a) I don't have to draw them b) I don't have to think of a standard to draw them to
<awygle>
argh kicad uses the "epad numbered as N+1" convention rather than the obviously superior "epad numbered as 0" convention
<awygle>
bummer
<whitequark>
yeah kicad has a few idiosyncrasies
<whitequark>
wait until you see how it connects several pads to the same logical pin
<whitequark>
now *that* is some serious bullshit
<rqou>
i use "epad numbered as N+1"
* awygle
tabs over to irc to complain about stacking pins - awww sniped
<rqou>
whitequark: oh, you mean the "one visible pin and several invisible ones at the same coordinate"
<awygle>
i grabbed a symbol from somebody online for smolfpga that did that and cursed them out
<whitequark>
(extremely opensource voice) what's the problem it works
<awygle>
sorry, "xess", whoever you are
<rqou>
i do it too
<rqou>
i thought that's the "correct" way?
<whitequark>
yes
<whitequark>
that's what i'm complaining about
<whitequark>
it's ridiculous
<awygle>
rqou: every professional board i've seen has a logical pin for every physical pin
<rqou>
also, making symbols for "mechanical" components _sucks_
<whitequark>
I'm still firmly in kicad over eagle for a good bunch of other reasons but it's absurd that these hacks are needed
<rqou>
"does your usb connector have an id pin? how about shield pads? how many pins should be on the schematic symbol for shield pads? 1 or 4?"
<rqou>
repeat for every other "mechanical" component ever
<awygle>
1 pin for "shield"
<awygle>
id pin is just a pin
digshadow has quit [Ping timeout: 260 seconds]
<awygle>
"pins not connected on the footprint may be omitted from the symbol" oh come on
<rqou>
now "does this sd card socket have a card detect pin? one pin or two?"
<rqou>
"does this pushbutton have 2 pins or 4 pins"
<awygle>
i don't understand your complaint
<awygle>
like... you _have_ a pushbutton in front of you
<awygle>
how many pins does it have?
<whitequark>
lol
<rqou>
it has 4, but the pairs are shorted together
<rqou>
inside the button
<awygle>
sure but there are 4 pins. so 4 pins. although in kicadland it would be two "real" pins and two hidden pins
<rqou>
goddammit
<rqou>
btw, the kicad footprint for this has two pins only
<whitequark>
awygle: hm
<rqou>
and the footprint just has two pads with the same pin number
<whitequark>
^ because that means kicad knows they're internally connected
<whitequark>
and lets you route to either
<awygle>
eh ok
<awygle>
that's not an unreasonable choice
<awygle>
whitequark: san check - the "overlapping pins" thing is not used for VCC, because it's a "power input", yes?
<rqou>
is kicad erc actually usable?
<rqou>
IME it never works right thanks to libraries not having their pins declared correctly
<rqou>
also, i just remembered discovering the other day that kicad's 74xx parts _still_ use the oldschool "implicit pwr/gnd" style
<rqou>
which is just nopenopenope
digshadow has joined ##openfpga
<awygle>
whelp, found a kicad bug
<rqou>
lol
<rqou>
that's not news :P
<rqou>
btw, are you two using nightly?
<awygle>
i am
pie_ has joined ##openfpga
rohitksingh_work has joined ##openfpga
<whitequark>
awygle: hmm I'm not sure actually!
<rqou>
be careful, i was told there are significant incompatible changes
<whitequark>
between nightly and stable? yes
<whitequark>
that's why we're using nightly
<rqou>
oh, i see that was a reply to something else
<awygle>
whitequark: there's already a 48-pin QFN footprint matching the ice40up
<awygle>
i did the symbol
<whitequark>
awygle: that was fast
<awygle>
what would you like me to do with it? a library in glasgow-jtag or a PR to kicad libs or both?
<whitequark>
hmm how good is kicad at replacing symbols with other symbols...
<awygle>
as far as i can tell, "not"
<whitequark>
awygle: I guess drop the library in hardware/ subdirectory if it's not too much work
<whitequark>
hopefully that one will never be actually used but if kicad drags their feet on merging PRs it will be useful
<awygle>
wilco
<whitequark>
feel free to take other footprint issues if you're bored / find drawing symbols exciting for some reason
<whitequark>
i'll occasionally grab some either
<awygle>
sounds good
<awygle>
completely off topic - do you have a suggestion for how to feed one of my cats drugged food without the other one eating it?
<whitequark>
what formulation
<awygle>
a powder mixed into the food
Bike has quit [Quit: Lost terminal]
<awygle>
called.....
<awygle>
panacour (fenbenzadole)
<whitequark>
*looks into crystal ball* sounds like antiparasitic
<awygle>
that is correct. lungworm.
<rqou>
:(
<awygle>
of course the _skinny_ cat is the one that got sick
<whitequark>
I don't really see any way other than like, closing the door
<awygle>
my poor cats/doors/rugs/ears
<whitequark>
what, is the fat cat so attracted by food?
<awygle>
well problem a) is the fat cat wants the food, problem b) is the cats don't like being separated, and problem c) is the skinny cat doesn't like the taste of the medicine and basically doesn't care about food at all
<awygle>
i closed doors last night and it took ~3 hours to get him to eat the damn food
<rqou>
sounds like "fun"
<rqou>
also, we need more catpix :P
<awygle>
lol
<awygle>
you're the only one who ever likes them on twitter :p
<rqou>
huh, hidden pins are in general not allowed, but "oldschool 74xx implicit power/gnd" is?
<rqou>
that's weird
<awygle>
ugh kicad has a CLA? weird.
<whitequark>
"KiCad library data is released under the CC-BY-SA 4.0 license (see more information here)."
<whitequark>
I'm ... not sure how this actually works
<whitequark>
do they just want to make it extra sure everyone understands they're submitting data under CC-BY-SA?
<rqou>
oh yeah, i found that to be really weird too
<rqou>
but they have some kind of exception
<rqou>
so that you can actually use them in commercial designs
<awygle>
my understanding is that by signing a CLA you license the stuff you submit to kicad in perpetuity, so that if they decide to change licenses later it's easier?
<awygle>
i didn't actually read theirs, probably should have
<whitequark>
this is literally all their CLA says
<awygle>
oh really? jeez lol
<awygle>
that does not seem necessary then
<whitequark>
yeah it's weird
<rqou>
hmm, the KLC guidelines don't handle bikesheds like "IPC minimal or hand-solderable"
<whitequark>
rqou: for passives you can freely choose the footprint to go with any symbol
<whitequark>
it's not either-or
<whitequark>
for ICs they go with whatever the datasheet says
<whitequark>
of course, nothing prevents you from having a local override
<rqou>
when i looked a few days ago there was only one type of "generic QFP"
<rqou>
*one type for each pin count
<rqou>
and it had smaller pads than at least ST's recommendation
<whitequark>
that's a bug
<rqou>
but maybe it agrees with some other vendor?
<awygle>
argh. how do i tell kicad to update the footprint library?
<rqou>
not even all the vendors even have consistent recommended footprints
<rqou>
and they're apparently based on micrel/microchip's recommendation
<rqou>
which for some reason was smaller than st's
<whitequark>
awygle: del /s /q %APPDATA%\kicad
<rqou>
although tbh the footprints that i personally use all have dubious origins anyways
<rqou>
even though some of them are board-proven
<whitequark>
rqou: kicad has vendor-prefixed footprints for this
<rqou>
i mean, ST's and kicad's are almost the same size
<rqou>
they're different by something like 0.05mm in the width
<rqou>
(of each pad)
<rqou>
so it'll probably fit
<whitequark>
oh
<whitequark>
that probably doesn't matter
<whitequark>
50 micron
<rqou>
also one of them had longer pads? but i don't remember which one
<rqou>
but yeah, even "generic QFP" can have variants and bikeshedding
<whitequark>
sure
<rqou>
so i wonder how kicad plans to balance between hobbyists asking for very-hand-soldering-optimized footprints (as an extreme, @bml_hubbard's) vs professionals wanting the smolest board possible?
<whitequark>
I dunno?
<rqou>
and this is why everybody NIHs them :P
<whitequark>
how about
<whitequark>
asking
<rqou>
we should have parametric footprints :P
<whitequark>
yes, I was going to build that software
<whitequark>
but then mental health happened
<rqou>
oh, that sucks :(
<whitequark>
awygle: ha, I just realized that we're building basically three Dangerous Prototypes device in one
<awygle>
The Most Dangerous Prototype
<rqou>
you know, that was exactly what i was originally trying to do
<awygle>
those are the violations i wanted/expected to have
<rqou>
O_o they have travis for libs?
<awygle>
i _think_ they're just because i did a multi-part symbol
<whitequark>
yes
<rqou>
is this actually built into kicad or a set of external tools?
<awygle>
so i'm gonna leave a comment on the PR and say "hey i think this is fine but please tell me what to do"
<rqou>
and this is another reason why i just do my own libs :P
<whitequark>
rqou: external tools
<rqou>
lol
<awygle>
sure would be nice if i could run it locally lol
<rqou>
no wonder the scripting apis suck, they don't even dogfood it :P :P
<whitequark>
you can?
<awygle>
i'm sure you can, i just wish there was a button in the schematic library editor
<whitequark>
yeah, that's true
<whitequark>
so I think what it doesn't like is
<whitequark>
VCCPLL (power output) being above VCC (power input)
<whitequark>
GND being a bidirectional pin
<rqou>
someone should force them to make the CI use only the kicad scripting apis so that the scripting apis improve :P
<awygle>
i fixed both of those issues, they're just not in the images
<whitequark>
oh
<whitequark>
okay
<awygle>
updating those now
<awygle>
updated
<whitequark>
nice, thanks!
<awygle>
that's me for tonight i think. i'll see if there's any comments in the morning, if not i'll ping some people on #kicad maybe.
<whitequark>
kk!
rohitksingh_work has quit [Ping timeout: 260 seconds]
bitd has joined ##openfpga
digshadow has quit [Ping timeout: 276 seconds]
scrts has quit [Ping timeout: 256 seconds]
scrts has joined ##openfpga
rqou has quit [Ping timeout: 276 seconds]
rqou has joined ##openfpga
bitd has quit [Read error: Connection reset by peer]
eduardo_ has joined ##openfpga
eduardo__ has quit [Ping timeout: 245 seconds]
scrts has quit [Ping timeout: 260 seconds]
Bike has joined ##openfpga
bitd has joined ##openfpga
scrts has joined ##openfpga
rohitksingh_work has joined ##openfpga
futarisIRCcloud has joined ##openfpga
bitd has quit [Remote host closed the connection]
rohitksingh_work has quit [Read error: Connection reset by peer]
Bike is now known as Bicyclidine
ZipCPU has quit [Ping timeout: 264 seconds]
ZipCPU has joined ##openfpga
soylentyellow has quit [Ping timeout: 276 seconds]
soylentyellow has joined ##openfpga
rohitksingh has joined ##openfpga
FabM has quit [Quit: ChatZilla 0.9.93 [Firefox 52.7.2/20180316222652]]
rohitksingh has quit [Quit: Leaving.]
rohitksingh has joined ##openfpga
bitd has joined ##openfpga
rohitksingh has quit [Quit: Leaving.]
rohitksingh has joined ##openfpga
rohitksingh has quit [Client Quit]
rohitksingh has joined ##openfpga
rohitksingh has quit [Quit: Leaving.]
rohitksingh has joined ##openfpga
rohitksingh has quit [Quit: Leaving.]
digshadow has joined ##openfpga
gruetzko- is now known as gruetzkopf
X-Scale has quit [Ping timeout: 240 seconds]
mumptai has joined ##openfpga
X-Scale has joined ##openfpga
sgstair_ has joined ##openfpga
sgstair has quit [Disconnected by services]
sgstair_ is now known as sgstair
xdeller has quit [Remote host closed the connection]
xdeller has joined ##openfpga
bitd has quit [Quit: Leaving]
eduardo_ has quit [Ping timeout: 256 seconds]
<whitequark>
awygle: hi
<awygle>
whitequark: hello
eduardo_ has joined ##openfpga
<whitequark>
awygle: I'm stuck in LLVM compilation hell so I'll probably try to assemble a prototype from my devboards and write bitstream download code
<whitequark>
any free time?
<awygle>
whitequark: this evening perhaps, and the weekend
<awygle>
I'm in the office for the next 4-5 hours
<whitequark>
curses! foiled again (by timezones)
<awygle>
:-). I can brainstorm and/or bikeshed if that's useful
<azonenberg>
boxed E and V = transceiver core and termination rails
<azonenberg>
black and white checkerboard = Vcore
<azonenberg>
black and white vertical box = block RAM supply
<azonenberg>
oh and gray/white diagonal = VCCIO for GPIO banks
<azonenberg>
whitequark: so yeah a thousand power/ground doesnt sound unreasonable if you look at the layout of the GTPs
<awygle>
whitequark: i think Person On Twitter is right, the a and b is differential pair related
<azonenberg>
whitequark: the kintex7 i am using in my switch has 285 GPIO and four GTXs in 484 balls, and the GPIO banks have a lower ground/power ratio per I/O than the transceivers do
<awygle>
(note that this puts the two parts of a differential pair on pins 12 and 21)
<whitequark>
azonenberg: that is fucking art
<whitequark>
awygle: yeah I agree
eduardo_ has joined ##openfpga
<awygle>
i look forward to hybrid BGA/QFN packages to get good low-inductance grounds to increasingly high-speed transceivers
<azonenberg>
awygle: unlikely to happen any time soon
<azonenberg>
modern packages are already integrating decoupling caps
<azonenberg>
at which point solder ball inductance becomes less of an issue
gnufan has quit [Ping timeout: 240 seconds]
<awygle>
that's good because the thing i said is bad and dumb and should never ever happen
<whitequark>
lol
<azonenberg>
awygle: that being said, it would not surprise me if we see a move to LGAs at some point?
<sorear>
Do those have better inductance?
<awygle>
they can
<awygle>
by making bigger pads than the commensurate balls
* awygle
bulls ahead with the use of the word "commensurate" despite not being sure it applies in this context
gnufan has joined ##openfpga
xdeller has quit [Quit: Leaving]
<qu1j0t3>
corresponding?
<awygle>
apparently commensurate means "corresponding in size or degree; in proportion." so commensurate land to the balls would definitionally have the same inductance
xdeller has joined ##openfpga
<qu1j0t3>
yeah, i know what commensurate means. i didn't dig fully into the semantics of your sentence tho
* qu1j0t3
should do some work :/
<awygle>
i apparently didn't know what commensurate means, at least not well enough :p
<rqou>
anyways, i believe my comment still stands that "'browser infosec people' are fucking awful at evangelism/marketing"
<rqou>
azonenberg, you're "infosec" right? thoughts on ^ ?
<awygle>
"developers bad at communication - news at 11"
<rqou>
mozilla at least has a "tech evangelism" team although idk if they can keep up with all the dumb ideas that come out of the rest of the company
<rqou>
afaik they couldn't even get people to use css prefixing correctly to the point where firefox now honors certain -webkit prefixes
gnufan has joined ##openfpga
gnufan has quit [Ping timeout: 240 seconds]
Bicyclidine is now known as Bike
gnufan has joined ##openfpga
<awygle>
mozilla is so weirdly split-brain in this regard. like the whole pocket thing, or the "opt in to experiements" or whatever
<rqou>
yeah, mozilla seems to really not understand how to do/sell any kind of strategic direction
<rqou>
they also have a _weird_ org structure with a for-profit private entity owned by a 501(c)(3)
<rqou>
also: "AOL renamed the Netscape Communications Corporation to New Aurora Corporation and transferred the Netscape brand to themselves. AOL sold the former Netscape company now known as New Aurora Corporation to Microsoft who in turn sold the former Netscape company again to Facebook; the Netscape brand remained with AOL. [clarification needed] The former Netscape company is currently a non-operating subsidiary of
<rqou>
welcome to the tech industry :P
<awygle>
TIL you can be charged per broken bone even if you don't get surgery
<rqou>
what
<rqou>
"maybe you should have made a choice to break fewer bones"
<rqou>
:P
gnufan has quit [Ping timeout: 260 seconds]
plaes_ has quit [Quit: Reconnecting]
plaes has joined ##openfpga
plaes has joined ##openfpga
gnufan has joined ##openfpga
gnufan has quit [Ping timeout: 264 seconds]
gnufan has joined ##openfpga
gnufan has quit [Ping timeout: 260 seconds]
<awygle>
i really want a reason to use silicon carbide semiconductors just so i can get them from a company called "wolfspeed"
<rqou>
shoulda taken ee113/213 :P
<awygle>
it turns out SiC is not very good under radiation
<awygle>
GaN is great tho
<rqou>
wtf are you building?!
<jn__>
you should make a processor and run wolfenstein on it
<awygle>
nothing currently, that's just what originally lead me to look at alternative FET materials
<rqou>
ever use even more exotic devices like HEMTs?
<awygle>
i ended up designing a charge pump gate driver made from discrete BJTs to drive an eGaNFET as basically just a switch
<rqou>
for what application?!
<awygle>
(BJTs deal with radiation quite a bit more gracefully than FETs, their beta just drops instead of the oxide rupturing)
<awygle>
cubesats
<awygle>
duh
<rqou>
why do you need a GaN fet?
<awygle>
and i've "used HEMTs" in the sense that i've used RFICs containing HEMTs, i'm pretty sure
<rqou>
also how are you so good at RF?
<awygle>
because GaNFETs have low Rdson and high radiation resistance (like 5 MRad iirc? smoething huge)
<awygle>
and the last thing you want is for your "power enable for the whole spacecraft" switch to fail open
<rqou>
btw do you have a signed agreement with the government that you will always abide by itar? dig apparently has and so can't touch certain RE like microbolometer firmware
<awygle>
i don't, no
<awygle>
but i intend to do so anyway because i enjoy not being fined or put in prison
gnufan has joined ##openfpga
<rqou>
just do the RE outside the US :P
<rqou>
(idk if that's actually sufficient)
<awygle>
do the RE outside the US... and never come back
<rqou>
what does itar actually restrict?
<lain>
the clue is in the name
<awygle>
Everything on the USML
<rqou>
but does "go to another country to perform RE" fall under itar?
<awygle>
Depends on if you're found guilty
<rqou>
lol
<rqou>
true enough
<awygle>
If I was prosecuting that case I'd argue that you're exporting technical secrets (covered) from an American asset (you, a citizen) to a foreign national
<lain>
the thing is a country can always just not let you back in, and realistically I'm not sure it's possible to fight that :P
<awygle>
balrog: it's not super clear but i think those are PSat=1W
<balrog>
ah
<balrog>
well good luck then
<awygle>
i guess if i'm just screwing around i can always design something smaller
gnufan has quit [Ping timeout: 256 seconds]
<lain>
I forget if I mentioned in here, we've been chasing a power anomaly at our place.. 11khz and 22khz noise (it looks like a half-wave rectified sine wave, several volts peak to peak, riding on the mains)
<lain>
we are now pretty confident we know what it is
<awygle>
oh? do tell
<lain>
we share a polepig with our neighbors, and they have a rather large house. oil burning furnaces are common in this area (though we don't have one). we think most likely it is a defective ignition system in an oil burning furnace, assuming they have one
<lain>
a friend thought of that after looking at the waveforms, and some googling confirms some oil burners will run the ingition system full time while burning, and often have defective grounding or transformers, such that they leak a bunch of 10-30 khz noise onto the lines
<awygle>
ah, that's cool
<lain>
considering it correlates extremely well with outside temperature, I'm content to believe that
<awygle>
you should send someone over to make them turn it off and verify :p
<lain>
haha
<rqou>
or just figure out a way to inject damaging EMI back into their house :P :P
<lain>
yeah I haven't quite figured out how to deal with the human side of it yet
<lain>
this should only happen if their igniter is defective in some way, which they're almost surely unaware of
<awygle>
"hi, we've never met before, but you're raising my noise floor by 0.1 dB, so please turn off your furnace and freeze to death?"
<lain>
^
<lain>
haha
<lain>
the noise from various appliances is pretty obnoxious