<awygle> I have heard some variation on "wow, you look so young!" from almost every coworker today
<awygle> (I shaved and got a haircut)
<awygle> Given that I'm already the second youngest person here this may not have been a good plan
<whitequark> lol
lain has quit [Quit: brb]
Lord_Nightmare has quit [Ping timeout: 264 seconds]
<rqou> wait you don't want to look super young?
<awygle> rqou: not particularly?
<rqou> i guess it has problems of not being taken seriously
<awygle> i have a performance review this week (allegedly) where i'll be asking about becoming a senior engineer. looking like i'm 16 is not a plus for that interaction :p
<rqou> wtf you don't look 16
<awygle> no i don't
<awygle> but i look like, 21
<sorear> awygle: we speculated a while ago that it’s TRESET
<rqou> at least in your catphoto you look much older than that
<awygle> sorear: but there's no (external) JTAG
<awygle> why ball out TRESET but not the rest?
<rqou> although i guess i tend to overestimate how old white people are or something (the opposite of underestimating how old asian people look?)
Lord_Nightmare has joined ##openfpga
<whitequark> it's probably multiplexed with SPI pins
<awygle> rqou: lol
<awygle> whitequark: ah, good point. didn't think of that
<sorear> Lattice removed JTAG from their datasheets some years ago, they may or may not have changed the actual chips
<awygle> rqou: decap when :p
<awygle> it's dead week, make yourself useful
<rqou> either later this week or after finals
<awygle> nobody at berkeley studies
<awygle> (this is not true)
<rqou> yeah, i just walked into Cory earlier and everybody was studying
<awygle> rqou: racially-diverse clone check - do you have the experience i had where technical classes require little to no study for finals but humanities classes require an enormous amount?
lain has joined ##openfpga
<rqou> no, i kinda fit with the group that felt that berkeley humanities classes were easy AF
<whitequark> racially-diverse clone
<awygle> i recall in my sophmore year spending all of dead week alternating watching 90 minute history lectures and 30 minutes of playing FFX
<rqou> at least if you didn't care _that much_ about a really high grade
<awygle> well, i didn't go to/pay attention in the history lectures, you see :p
<rqou> lol i took history 127ac and barely remember studying anything :P
<rqou> i did consistently attend almost all lectures though
<rqou> this is the class where i also learned about the most awesome volume unit ever: acre-feet
<awygle> my grad student from my R1B class tried to convince me to switch to a comparative literature major
<awygle> i very politely did not laugh at him
<cr1901_modern> >30 minutes of playing FFX I hope you finger slipped and you meant a "T" where you put "X"
<cr1901_modern> yea yea Fast Fourier Transform (Final Fantasy Tactics)
<awygle> cr1901_modern: FFT:A almost caused me to fail the seventh grade
<awygle> well, that's not fair. but it would be a lie to say it did not contribute.
<cr1901_modern> Never played advance or the remake. Just the PS1 version :). Which is a game I'd take on a desert island.
<Bike> advance is completely different, except i guess you still sort of kill jesus
<awygle> yeah advance is very different
<cr1901_modern> Ever beat Velius w/o cheating? I've done it exactly once.
<awygle> i haven't played the first one
<whitequark> what the fuck
<awygle> i did play some War of the Lions
<Bike> i never got very far in tactics, unfortunately
<whitequark> I'm touching the cathodes of this LED and it lights up
<awygle> but couldn't get into it as much
<whitequark> quite appreciably
<awygle> O_o
<whitequark> just how high is the efficiency?!
<Bike> then i got confused hearing about a calculator-only run
<Bike> i think i've done that before, with the LED
<cr1901_modern> Bike: Calculator is broken
<cr1901_modern> As is Dancer
<Bike> cr1901_modern: yeah but calculator with no other class moves
<cr1901_modern> errr...
<Bike> so basically a stupid melee class
<cr1901_modern> yea that's dumb
<awygle> i statted out the perfect clan in FFTA on 30 index cards when i should have been learning science
<cr1901_modern> I always made main character a Priest until he masters the class
<cr1901_modern> Then switch back to Squire for the rest of the game
<rqou> oh wtf awygle you're totally me :P
<cr1901_modern> Holy + healing spells = <3
<rqou> i played FFTA too
<Bike> i was never very good at video games so i beat ffta without knowing any of the "broken" stuff like Concentration
<rqou> never beat it because young me didn't want to go back into jagd dorsa after i did very poorly in the first mission that required you to do that
<rqou> also instead of strategy everybody was a black mage :P
<cr1901_modern> jagd... I only know that term from 12 lol
<awygle> hahaha
<whitequark> nerds
<Bike> jags have permadeath
<Bike> v scary
<awygle> well, unless you played a lot of fire emblem
<awygle> which i did and still do
<Bike> you're powering lights with your literal hands
<cr1901_modern> Bike: In 12, jagd means "area where airships are incapable of flying"
<cr1901_modern> apparently it has a different meaning in FFT:A
<Bike> so like... mountains?
<cr1901_modern> No, I mean "an airship once it enters a jagd, just crashes. Loses power. Can't fly"
<Bike> huh
<cr1901_modern> One plot point is that you can't get to an area that's a lone island "on the edge of the world" b/c it's in a Jagd
<rqou> i was never a huge FF fan though
<Bike> in ffta the deal was that the world in general is under the force of magical law, which prevents death and also puts random restrictions on you, but jagds are lawless so people die when they are killed
<rqou> i did play a lot of pokemon, but who didn't
<Bike> all shadowy and shit
<cr1901_modern> and you need to fine a macguffin that allows airships to fly in jagds
<Bike> oh well that's bullshit
<cr1901_modern> Bahaha
<Bike> if you can't fly in jags you can't fly in jags
<cr1901_modern> http://finalfantasy.wikia.com/wiki/Jagd_Naldoa I haven't played this game in 10 years
<cr1901_modern> why do I remember this bullshit?
<cr1901_modern> 12 is basically "Medieval Star Wars" anyway
<awygle> i only played like six minutes of 12
<Bike> "A giant hole in the ocean. It is so deep, it draws in water from all around, creating furious tides" so they got a pump s ystem, or like, what
<awygle> 10 was one of my favorite experiences as a kid
<Bike> i did not know "cataract" could mean waterfall
<cr1901_modern> Bike: B/c plot, that's why
<cr1901_modern> awygle: I love 10's story. Gameplay not so much
<cr1901_modern> Break Damage Limit and Quick Hit breaks any semblance of strategy.
<awygle> eh gameplay is "JRPG"
<whitequark> >The tower is well over 100 floors tall, divided into three levels, and is infested with monsters, the perfect place to store the gift of the gods.
<whitequark> perfect indeed
<awygle> i haven't played a JRPG in years
<awygle> takes so much time
<Bike> a regifting
<awygle> did anybody play Tales of Symphonia
<awygle> ?
<rqou> nope
<cr1901_modern> No :(
<rqou> did anybody play runescape? :P
<whitequark> yes
<Bike> got you this two dollar gift basket. but it's in the death tower so in a few years you'll forget what it is and mythologize it
<whitequark> on dialup
<awygle> i loved that game because it made me actually hate the bad guys
<whitequark> it took forever to download
<rqou> lol
<rqou> i've done that too
<rqou> but mostly on broadband
<Bike> runescape had bad guys?
<awygle> not runescape lol
<cr1901_modern> Bike: Spoiler alert, part of the plot is to destroy this "gift of the gods"
<awygle> i didn't play runescape, the only mmo i played was wow
<whitequark> i also played lineageii c3/c4
<Bike> damn didn't see that twist coming
<awygle> and arguably dungeon fighter online
<rqou> i remember playing runescape on dialup in HK connecting to UK servers
<rqou> it worked ok
<Bike> what i remember about runescape is i downloaded a scripting thing that let you automate mouse movements and keyboard inputs
<Bike> because grinding in rpgs is stupid bullshit
<rqou> Bike: you cheater :P
* cr1901_modern never played runescape
<Bike> i might have gotten banned for it
<rqou> you're supposed to abuse MouseKeys instead :P
<Bike> which, pro, means i got to stop playing runescape
<rqou> hey, runescape is a11y-friendly :P
* cr1901_modern checks release date
<cr1901_modern> 2001? The only game I was playing back then was RCT
<rqou> um... don't check :P
<rqou> i was playing since 2005
<rqou> even got the fancy red cape :P
<awygle> i also hate grinding, which is why i always ended up beating pokemon games with hilariously low-level teams
<whitequark> around that time too
<whitequark> but i never really got anywhere in runescape
<rqou> runescape party? :P
<whitequark> maybe
<whitequark> so long ago
<cr1901_modern> I'd have to learn how to play
<Bike> is it seriously still going
<rqou> i've played for nostalgia a few times over the years
<rqou> Bike: yes
<rqou> they even have a native client now
<rqou> they finally killed java
<awygle> another thing that's still going - neopets
<rqou> lol
<rqou> but they prune accounts :(
<Bike> so you're saying my account has been deleted before it could become old enough to drink
<whitequark> oh wtf
<Bike> wow it has decent grafix now
<whitequark> my i2c core works in simulation but not on hardware :(
<Bike> aw
<whitequark> oh hang on
* whitequark slaps self
<whitequark> i didn't connect any clock to the FPGA
<whitequark> forgot that you need it, oops
<cr1901_modern> awygle: Thoughts on 13 (I've never played it) :)?
<awygle> "forgot to connect a clock" is way better than "mysterious synth/sim mismatch" though
<awygle> cr1901_modern: never played it either, i didn't own a PS3
<awygle> or 360
<cr1901_modern> I stopped following anything new after college started, tbh. Even by that point I didn't own a PS3 or 360
<awygle> yeah me too for the most part
<cr1901_modern> now, I only play SNES, Genesis, and VB stuff
<awygle> although i bought a switch this year
<cr1901_modern> (Yes I have a VB)
<awygle> i play SNES, N64, and occasional PS2 stuff
<awygle> oh and i play Luigi's Mansion and Metroid Prime usually about once a year on gamecube
<rqou> btw, anybody ever played maplestory?
<awygle> because those are both fantastic
<rqou> it was a big driver for me learning how to hax :P
<cr1901_modern> never heard of it ._.
<awygle> lol a friend of mine was working on hax for maplestory and then his australian business partner got arrested
<rqou> O_o
<awygle> err. _their_.
<Bike> i think watching someone walk through walls in pokemon was the first time it occured to me you could hack things
<awygle> gen 1 pokemon hax are so great
<whitequark> awygle: yes but it makes me look dumb, so, no one can say if it's good or bad
<Bike> because emulator hacking is nice and works by basic things like fucking with memory in the running machine
<awygle> did you know that if you go to the bike shop in gen 1, you can trick the game into making every text box appear instantly instead of scrolling?
<Bike> it doesn't make you look dumb! geez
<Bike> last time i was at a conference someone's demo had a problem because they were missing a file, and like who cares! the software is fine right
<rqou> i guess luckily for me i never actually made any hax for maplestory
<rqou> i just wanted to understand how hax worked
<awygle> rqou: i don't think the arrest was maplestory related
<awygle> i think the guy was into much more serious stuff
<Bike> i don't think nexon has arrest authority
<awygle> hey azonenberg which RS PSUs did you get again?
<Bike> cr1901_modern: i think i hit my limit of game hacks after that thing where they "played Portal" in super mario bros
<awygle> lmao tasbot is always great
<azonenberg> awygle: HMC8042
<azonenberg> i love them
<Bike> tasbot is great
<azonenberg> i'm actually going to be extending my PC control software for them in the near future to include I/V graphing
<awygle> azonenberg: $$$?
<Bike> and always a lil weird for me because one of the people they have is someone i used to talk to on irc
<azonenberg> awygle: about 1.2 kUSD each
<azonenberg> Plus the rack monut kit
<awygle> mm
<azonenberg> They're not cheap psus but they're worth every cent imo
<azonenberg> (As long as you actually need the features)
<cr1901_modern> Bike: dwangoac? endrift?
<azonenberg> ethernet control, 1 mV / 0.1 mA resolution on everything
<azonenberg> optional channel / whole unit shutdown on overcurrent, with adjustable trip delay
<Bike> hmm
<azonenberg> remote sense, although its on the back and a bit awkward to hook up
<Bike> maybe i'm misremembering, dwango is someone who knows someone who i used to talk to, which isn't quite as compelling
<cr1901_modern> Well they're both quite involved w/ tasbot afaik (I don't actually watch gdq)
<Bike> ah, no, ais523 was on tasbot plays pokemon plays twitch
<rqou> oh wtf
<awygle> lmao that was great
<rqou> so i'm doing the final lab report, and we have to measure semiconductor device parameters
* awygle remembers it's almost sgdq again
<rqou> the forward voltage drop of this random schottky diode at 5mA really is 0.1 mV
<Bike> boy, i hated that
<rqou> so the lattice vpp really doesn't work
<cr1901_modern> lattice vpp?
<Bike> awygle: whoa, it's my entire electronics course
<awygle> "Do you have any idea how hard it is to solder wires to germanium? I'll tell you: real goddamn hard."
<Bike> check this shit out (fig. 1)
<awygle> "I understand this is the same process by which the top quark was discovered."
<Bike> this was like every goddamn lab i took
<awygle> cr1901_modern: lattice, on their dev boards, power VPP (2V5) from VIO (3V3) with a diode (drop ~0.7V, in theory) except that they populated a Shottky
<Bike> building a radio, listening to crawfish neurons, doesn't fuckin matter, it's all a bunch of crap
<awygle> i just spent a bunch of time doing op amp problems and i've never used an op amp in anger once
<cr1901_modern> ahhh
<Bike> what does using an op amp in anger entail
<rqou> burning your finger? :P
<rqou> running out of time and slapping a heatsink on a to-92? :P
<Bike> that's just using an op amp in a way that induces anger
<rqou> lololk
<sorear> burning someone else's finger?
<awygle> using it for real
<awygle> to solve a problem
<awygle> (presumably a problem that makes you angry)
<Bike> i guess what i learned in electronics was mostly how to hook together passives with a few op amps, which isn't super realistic
<awygle> yeah. i've never had to build an integrator or a differentiator, for example.
<azonenberg> Look good? (planes shown outline only, same on both L2 and L3)
<awygle> and nobody ever explained loop inductance to me
<azonenberg> I'm going to run through the full design review checklist before sending it out later tonight
<awygle> lgtm, yeah
<Bike> i remember asking my professor about parasite inductance based on some stuff i learned in bio and he was like look, you wanna deal with that stuff, you be a grad student working at stupid high frequency and you don't ask me
<whitequark> lol
<whitequark> stupid high frequency of 10 MHz?
<Bike> iunno
<Bike> plus i dug up some paper from like the 1930s showing axons don't have appreciable inductance
<Bike> nobody has ever cared about the question since
<sorear> i'm not sure it's remotely useful to model an axon as a transmission line, given the massive distributed nonlinearity
<Bike> i think i asked about inductance from wires rather than components... i don't remember the details and am still inexperienced, so
<awygle> "parasite inductance", or the inductance of flatworms
<Bike> sorear: usually it's a transmission line with nonlinear components strewn throughout
<Bike> the myelinated parts are "sort of linear" "ish"
<gruetzkopf> "earthworm-on-flatworm coplanar waveguide"
<sorear> but earthworms (en_US anyway) are roundworms
<rqou> oh wtf did i do this right?
<rqou> i have a power mosfet with a datasheet-given max leakage of 25 uA
<rqou> the actual measured one is off-scale low
<rqou> something like 1nA
<whitequark> now heat it to 100°C
<awygle> roundworm on flatworm sounds like a monopole to me
<rqou> whitequark: not part of this lab :P
<whitequark> rqou: not with that attitude no
<whitequark> seriously, have you no curiosity
<balrog> rqou: school lab?
<rqou> yes
<balrog> don't they want you to, y'know, experiment? :P
<Bike> what, in a class lab? no
<balrog> like, plot leakage current vs. temperature :)
<rqou> don't melt the 4145 jig while you're at it :P
<awygle> rqou: whitequark's right, that datasheet value is at the extreme of its operating range
<awygle> gotta melt that shit
<rqou> there's a much higher number at Tj=150
<awygle> well then maybe you fucked up the measurement :p
<awygle> ime "results wildly inconsistent with expectations" has never prevented me from getting a good grade in a lab though so maybe not worth fixing it
<Ultrasauce> also verify that that specification is at Vgs=0 instead of something just sub-threshold
<rqou> it definitely is
<rqou> so maybe the FET is just better than i thought
<Ultrasauce> this one could also be down to process tolerances
<rqou> or colder than i thought :P
<Ultrasauce> ie they'll happily ship batches much shittier than the one that made its way onto your breadboard
<rqou> also, TO220 igbts are adorable compared to the modules you normally see :P
<awygle> woohoo, i made it on #ircnocontext again :p
<rqou> Vge just looks totally wrong to my eyes
<rqou> obviously not a power electronics guy
<awygle> haha
<rqou> what?
<rqou> have you done this?
<awygle> na, i get you
<awygle> igbts are cool
<awygle> but weird
<gruetzkopf> i'm kinda a GTO fan
<whitequark> wtf if you salt meat after roasting it and not before it's much better
<cr1901_modern> rqou: Is it really a "better" (n-type enhancement mode) FET if it's on (even a little bit) when Vgs=0?
* whitequark stares at the datasheet
<rqou> cr1901_modern: what do you mean?
<whitequark> this register is called IFCONFIG and it seems fake
<cr1901_modern> >so maybe the FET is just better than i thought
<cr1901_modern> If it's on when Vgs=0, that seems like that would be "bad" to me personally
<cr1901_modern> i.e. it's "not supposed to be on" when Vgs = 0
<rqou> it's not on
<rqou> or at least very close to being not on
<cr1901_modern> oh
<cr1901_modern> nevermind then lol
<awygle> whitequark: fake?
<whitequark> ifconfig(1)
<whitequark> what the fuck
<whitequark> it works now
<whitequark> that was too easy!!
<whitequark> i demand synthesis-simulation mismatch issues
<whitequark> also, AUGH MY EYES
<cr1901_modern> so no need for HX8 :D?
<whitequark> I wired the bit 0 of a register 0 to the retina-burning RGB LED
<gruetzkopf> (playing music on a 800kW-design-load threephase traction GTO inverter is fun, and a bit scary)
<Bike> nice work whitequark
<whitequark> thanks :3
<whitequark> awygle: any idea what USB PID we should take
<rqou> gruetzkopf: wtf?
<rqou> whitequark: <shill>0xf055</shill>
<rqou> oh wait, you said PID
<whitequark> no the VID is 20b7
<rqou> gruetzkopf: how did you get access to something like that?
<whitequark> Qi Hardware
<gruetzkopf> used, went out-of-spec
<rqou> am i just particularly bad at dumpster diving?
<gruetzkopf> naah, you don't have the right dumpsters around
<rqou> not even in/near silicon valley?
* jn__ is bad at gruetzkopfing too
<rqou> lol
<gruetzkopf> midi-to-locomototive-diagnostics is a far more fun project
<gruetzkopf> *-to
<rqou> why are german trains so fucky?
<gruetzkopf> they're not if you don't have autohotkey and the diagnostics software for the traction inverters
<gruetzkopf> but if you do, you can "easily" have a 24MW 16-voice synthesiser ;)
<whitequark> wtf
<gruetzkopf> https://www.youtube.com/watch?v=OUyjAD1H9A4 this model is really well suited for that
<gruetzkopf> but the whole siemens ES/ER/vectron family works, even the diesel-electric ones
<rqou> but but gruetzkopf, how do i use the Internet of Targets to hack it and crash trains into each other? :P
<rqou> you mean they actually want trains to be _safe_?
<rqou> "The S in IoT stands for security/safety" :P
<gruetzkopf> they're not permitted to use public networks anywhere NEAR the safety systems
<gruetzkopf> remote operation is not even done using a packet-switched network, all TDM
<jn__> rqou: 1. Intercept GSM-R, 2. ???, 3. ????, 4. ??????
<rqou> lol
<gruetzkopf> intercepting GSM-R gets you nowhere
<whitequark> cr1901_modern: still needed HX8
<whitequark> HX8K
<rqou> gruetzkopf: oh really?
<whitequark> this is just extremely basic usage
<whitequark> I don't even have any FIFOs
<rqou> what's gsm-r used for?
<gruetzkopf> only voice comms between operations and drivers
<rqou> huh
<gruetzkopf> aka never except for the daily test calls
<cr1901_modern> Ahhh :(
<rqou> can you intercept gsm-r and social-engineer your way into some confusion/collision?
<rqou> e.g. convincing a driver that they're supposed to pass a signal at danger?
<gruetzkopf> (oh, and updating timetables for the drivers, but those have offline backup)
<gruetzkopf> the network won't let you steal functional ids
<whitequark> rqou, the terrorist
<rqou> huh, so what about the osmocomm people that hacked a calypso phone onto gsm-r
<rqou> do they just get assigned a new id?
<whitequark> is GSM-R like ASMR
<rqou> maybe for you :P
<gruetzkopf> gsm-r has normal per-device phone numbers
<gruetzkopf> and then it adds abstraction over it
<rqou> ah ok
<rqou> what would operators see if a rogue phone joined the network?
<sorear> you might have better luck with the "hi-vis jacket and semaphore flags" method of disrupting comms
<rqou> lol
<rqou> probably
<gruetzkopf> the only flag still in use according to german regs is stop RIGHT NOW
<rqou> no tracks controlled by physical tokens? :P
<rqou> or does only the UK rail still have ancient stuff like that?
<gruetzkopf> and you can only get to the interesting functional IDs (signal box and similar) from landlines inside the rail networks phone net
<rqou> ah disappoint
<rqou> so it's not sufficient to have "that 5-band phone you've got lying around"
<gruetzkopf> it's quite easy to get a GSM-R sim
<gruetzkopf> but those will only be able to play train
<gruetzkopf> i do have such a 5-band phone, but no sim
<rqou> can you steal a sim from a signal box? or are signal boxes hardwired?
<gruetzkopf> hardwired
<whitequark> define hardwired
<gruetzkopf> (also not anywhere near the public network)
<whitequark> do they use an eSIM?
<rqou> iirc there was some city that managed to put sims in traffic cams or something that were somehow provisioned with voice+sms+data even though they only needed data
<gruetzkopf> ISDN landline
<rqou> criminals discovered this and realized they could steal the sims and make hard-to-trace calls :P
<whitequark> amazing
<gruetzkopf> muxed over DB-owned fibres to DB-owned phone switches
<rqou> O_o
<rqou> wow
<rqou> how does crossing borders work? is there a wire that connects DB to e.g. ÖBB?
<sorear> and as a side benefit this disables the traffic cams
<rqou> lol yes
<gruetzkopf> how calls between signal boxes on border crossings are done differs on a case-by-case basis
zkms has joined ##openfpga
<gruetzkopf> may be analog phone line, or even local-battery!
<awygle> Slowly bubbling butter looks very lovecraftian
<rqou> O_o
<gruetzkopf> there's supposed to be international calling, never had the chance to test that
<rqou> how extensive is it?
<gruetzkopf> drivers manually switch their radios to the new network at a defined point
<rqou> manually?
<gruetzkopf> yup
<rqou> i'm amazed cross-border services run as smoothly as they do
<gruetzkopf> i'm not up-to-date on GSM-R procedures, i only know the landline stuff well because i reimplemented their extensions to ISDN in linux-call-router
<rqou> O_o they have custom extensions?
<rqou> i guess of course they do
<gruetzkopf> its not a lot
<gruetzkopf> just defined priority levels for incoming calls
<gruetzkopf> (and *fun* group call extensions (VGCS, ASCI))
<gruetzkopf> those are carried over from GSM-R though
<rqou> i'm amazed all of these countries/companies managed to agree enough for this to all work
<rqou> at least in western europe
<gruetzkopf> a gsm-r landline phone will only ring with the emergency call sound ( file is called tng_redalert.wav and it's exactly that) if called with the right priority in SETUP
<gruetzkopf> afaik the landline side is not really standardises
<rqou> gruetzkopf: do you know how the heck they manage to handle timetables?
<gruetzkopf> that's fairly interesting
<rqou> e.g. if you wanted to try to do Lisbon->Beijing
<gruetzkopf> and also operations
<rqou> Lisbon->Beijing is possible, right? just not with one train?
<rqou> a decade ago some crazy people managed to do Vienna->Pyongyang
<gruetzkopf> bahn.de spits out a 9-day route for that
<rqou> which one?
<gruetzkopf> lisbon-hendaye-paris-mannheim-berlin-moscow-beijing
<whitequark> moscow-beijing
<rqou> how the heck does banh.de have the timetable for this?
<gruetzkopf> 6 days for moscow-beijing
<rqou> how did they manage to make that work?
<sorear> bering straight rail bridge when
<gruetzkopf> ask hafas
<rqou> sorear: too hard
<gruetzkopf> they're doing the black magic
<gruetzkopf> it'll also spit out routes for paris-moscow (using no long-distance-trains)
<rqou> i'm still amazed
<rqou> now try NYC->SF and watch our system fall flat :P
<gruetzkopf> https://www.youtube.com/watch?v=0Qf-3P4mALA another random traction inverter sound video
<gruetzkopf> this is class 425, and i hear that sound on my commute
<gruetzkopf> from that sound you can hear the driver is pushing it quite hard, i'd guess about 80% of 2350kW nominal output
<whitequark> train nerds
<rqou> all aspies here :P :P
<gruetzkopf> i'm not even into this for that long
<gruetzkopf> got nerd-sniped in physics 101
<sorear> i'm a lot more aware of the trains now than a month ago
<whitequark> train awareness day
<gruetzkopf> (if you're ever in a class 425 EMU, go to the middle, that bogey doesn't have a motor
<18VAD0BC3> [Glasgow] whitequark pushed 1 new commit to master: https://github.com/whitequark/Glasgow/commit/d54b222c56f169bcb6054351ef1c6d892d25d794
<18VAD0BC3> Glasgow/master d54b222 whitequark: Hook up some registers to I2C inside the FPGA.
<17SAA3KX8> [libfx2] whitequark pushed 1 new commit to master: https://github.com/whitequark/libfx2/commit/928751b7eedc1c250fde592c07fb0c9f47350d1e
<17SAA3KX8> libfx2/master 928751b whitequark: Remove useless statement.
<sorear> although I'm retrospectively discovering that san diego *did* have a connection to the national rail network
<awygle> I used to do amtrak from Camarillo to SD
<awygle> Probably doesn't count
<gruetzkopf> most liked by drivers and repair personell is class 420, which was developed for the commuter rail system of munich for the 1972 olympic games, built pretty much unchanged until 1995 or so
<whitequark> class 420 lol
<zkms> speaking of traction inverter sound; you can use h-field probe to get a clean recording of it, there's even a battery-powered h-field probe (that outputs signal on 3.5mm audio connector) designed specifically for train otaku
<rqou> wtf
<gruetzkopf> yeah electric trains are great EMI test generators
<zkms> there's even people who reverse-engineer the exact commutation patterns of various inverters from recordings from that type of h-field probe
<zkms> it's super endearing actually
<rqou> i assume nobody wants to record the noises on BART though? :P
<rqou> (which is mostly mechanical because we don't maintain the track for shit)
<sorear> thing about san diego's municipal trains I didn't appreciate properly: they're quite quiet
<sorear> now that i'm having to deal with trains running through 1897 tunnels
<gruetzkopf> that's fairly close to a class 440 or 1440 train
<rqou> gruetzkopf: you need to experience BART :P
<gruetzkopf> not sure i want to
<rqou> "we have the best trains, the very best, in the world" :P :P :P
<gruetzkopf> i can have shitty light rail over here too
<rqou> it's probably still better
<gruetzkopf> https://www.youtube.com/watch?v=HcyaEpBHF3I i'd guess that's the same SIBAS32 traction inverter controller as class 425, but the driver is *very* careful
<sorear> do you also have five mutually incompatible rail networks operating in the same city?
<gruetzkopf> i've seen 3
<gruetzkopf> maybe even 4
<gruetzkopf> and of course we have fun stuff like h-bahn dortmund and the wuppertal suspended railway system
<sorear> track and loading gauge issues just make me sad
<zkms> yeah there's plenty of trains in JP with Siemens inverters
<rqou> sorear: just imagine how many times you hit a break-of-gauge trying to do lisbon->beijing :P
<sorear> rqou: i've hit two breaks of gauge going back and forth in boston today
<rqou> oh wtf?
<rqou> how?
<gruetzkopf> meanwhile in karlsruhe: full interop between light rail and normal
<rqou> i guess the answer is always "becuase 'murica"
<sorear> technically everything is 1435mm but the platform height and loading gauge is completely different between the 4 subway lines
<rqou> > "we have the best trains, the very best, in the world"
<gruetzkopf> light rail units run off 750V DC or 15kV AC
<zkms> search for シーメンス 音 on youtube there's so much siemens traction inverter music ~
<whitequark> that's quite a difference
<sorear> unlike some cities I could name which also have incompatible rail gauges
<gruetzkopf> the internationally-operation DB class 406 runs on 1.5kV DC, 3kV DC, 15kV AC or 25kV AC
<gruetzkopf> krefeld has normal gauge coming from dusseldorf and 1m gauge locally
<gruetzkopf> plus normal db network
<gruetzkopf> germany has 3 different platform heights in common use by DB, but trains are made to cope with that
<gruetzkopf> extending steps and stuff like that
<rqou> meanwhile BART has indian gauge because "herp derp we're aerospace engineers, why should we listen to you" :P
<whitequark> who are aerospace engineers?
<zkms> how do they cope with the AC, is it just a "transform 15kV 50Hz down to a voltage suitable for rectification" thing or is there something more fancy
<rqou> whitequark: SF BART was originally designed by a team of mostly aerospace people
<zkms> BART is the product of some aerospace company deciding to build its first railway, iirc
<rqou> that's why it's so weird
<gruetzkopf> (i have first hand knowledge that a 406 would also run on 750V, but it'll loudly complain to the driver and there's no way it'd fit in the streets)
<whitequark> wtf
<zkms> no offence to aerospace people but buying trains from Japan would have been a *far* better idea than trying to rediscover basic railway principles the hard way
<gruetzkopf> no idea how they implement it
<gruetzkopf> california is trying to buy trains from germany
<gruetzkopf> i can vouch for siemens desiro ML being nice trains, but i'd rather have desiro HC
<sorear> *some* of the mbta commuter rail tracks have 25kV overhead wires, these are not used by anything in mbta and only exist because of trackage overlap with the NEC
<Bike> "commonly used in India, Pakistan, west of Bangladesh, Sri Lanka, Argentina, Chile, and on the BART" owns
<whitequark> hahaha
<Bike> trains here are... standard gauge. well that's boring
<zkms> BART was unclear about how to shape rails and train wheels until like, 3 years ago which i have been told is why their trains make a heinously loud screeching in curves
<sorear> took me a while to figure out why there were overhead wires at [station] when all of the trains it services are diesel
<gruetzkopf> standard gauge is extremely common here
<whitequark> oh they switched to 1435
<rqou> gruetzkopf: iberian gauge trololol
<gruetzkopf> hehe
<gruetzkopf> iirc talgo has bogeys that can switch between iberian and normal gauge
<Bike> brunel gauge baby
<Bike> wikipedia has a list of "almost-standard gauge" railways
<gruetzkopf> give me 1435, 15kV AC at 1/3 * 50Hz ;)
<Bike> hong kong mtr is three millimeters short
<rqou> wait it is?
<rqou> i always assumed it was standard gauge
<Bike> on most of the lines, claims wikipedia
<Ultrasauce> standard on a hot day
<awygle> A hot day in Hong Kong. Where will we find one of those?
<sorear> what they say around here about screeching in curves is that a few of the curves are way too tight for common sense
<Bike> http://www.2427junction.com/chinahongkongmtr.html i can't read this but i do see "1432" emphasis not added
<rqou> sorear: on BART or where you live?
<sorear> rqou: where I am now, not BART
<Bike> dc metro is 1429 like great
<awygle> The Seattle light rail is pretty good just not expansive enough
<awygle> I should be able to catch a train to the city dammit
<Bike> oh hang on, the actual trains here are standard gauge, but the subway is some other stupid bullshit, or rather ONE subway line is some other stupid bullshit but the other is standard
<Bike> so i guess they can't transfer or whatever, huh
<Bike> built by the same company, but the nonstandard line is twenty years older
<whitequark> so i'm wondering if i drive this fpga clock at 48 instead of 30 will it still work
<whitequark> yeah
* whitequark shrugs
<openfpga-github> [Glasgow] whitequark pushed 1 new commit to master: https://github.com/whitequark/Glasgow/commit/0e23d13a263f8b776beef41c0e3e9c3e8a89288f
<openfpga-github> Glasgow/master 0e23d13 whitequark: Add dedicated register read/write code....
GenTooMan has quit [Quit: Leaving]
digshadow has quit [Ping timeout: 240 seconds]
mumptai_ has joined ##openfpga
<whitequark> ok wtf
<rqou> O_o
<rqou> doing my lab report still
<rqou> and apparently using shitty power-wasting semiconductors can actually end up costing hundreds of dollars over the span of a few years
<rqou> the marginal cost of a better fet is much lower
mumptai has quit [Ping timeout: 264 seconds]
digshadow has joined ##openfpga
<sorear> why are indicator LEDs these days usually specced to be usable as emergency flashlights
<whitequark> hahahaha
<whitequark> i use the only led wired to the fpga on this board for testing if my i2c core works
<whitequark> when it's on my entire room gains a red tinge
Lord_Nightmare2 has joined ##openfpga
<openfpga-github> [Glasgow] whitequark pushed 2 new commits to master: https://github.com/whitequark/Glasgow/compare/0e23d13a263f...fc5d59eb056f
<openfpga-github> Glasgow/master fc5d59e whitequark: Use our official VID:PID pair, 20b7:9db1 (yay!)
<openfpga-github> Glasgow/master 683dd4c whitequark: Disable FIFO bus while configuring the FPGA.
<awygle> better really efficient and really bright than dim and inefficient
Lord_Nightmare has quit [Ping timeout: 248 seconds]
<awygle> I guess
<whitequark> so if you want unified diff and ignore whitespace
<whitequark> you use `diff -uwu` right
<awygle> Sure
Lord_Nightmare2 has quit [Ping timeout: 264 seconds]
Lord_Nightmare has joined ##openfpga
Lord_Nightmare2 has joined ##openfpga
futarisIRCcloud has joined ##openfpga
Lord_Nightmare has quit [Ping timeout: 248 seconds]
Lord_Nightmare2 is now known as Lord_Nightmare
<rqou> belegrade is far AF
<futarisIRCcloud> rqou: Yep, Belgrade isn't close. I'm Brisbane, so it's out of my reach. linux.conf.au 2018 was good though. https://www.crowdsupply.com/teardown/portland-2018 & http://communities.techstars.com/australia/brisbane/startup-weekend/12429 are in a couple of weeks...
<awygle> also far af, Gdańsk
<whitequark> yeah python package management is extremely bad
<rqou> ime as long as you use _only_ apt or _only_ pip it works fine
<rqou> anything else and all hell breaks loose
<whitequark> the alt-text to https://xkcd.com/1985/ is funny
<azonenberg> awygle: lol
<azonenberg> So i just did the schematic review on the magnetic characterization board
<azonenberg> It was the fastest design review i have ever done
<azonenberg> line by line "Doesn't apply... doesn't apply... doesn't apply"
<azonenberg> took less than a minute
<rqou> i'm amazed you bothered
<azonenberg> Still have to do layout review
<azonenberg> that should be quick too
<futarisIRCcloud> awygle: juliusb of #openrisc is organising ORConf at Gdańsk. Interestingly, pmezydlo of https://github.com/pmezydlo/BeagleWire is from Gdańsk University of Technology.
<rqou> nobody wants to be in the people's republic of california in the land of the orange buffoon? :P
<sorear> the orange buffoon who hates issuing visas, is the important part
<rqou> wtf who is this person? https://twitter.com/SaschaNaske
<rqou> another random hobbyist set up with a SEM and decap?
<rqou> how are there so many of these people?
Bike has quit [Quit: Lost terminal]
<whitequark> SEM?
<whitequark> oh
<rqou> also i see sem-like sample posts
<rqou> in some photos
<whitequark> erm
<rqou> how are there so many hobbyists getting into silicon?
<whitequark> what's the problem with it?
<rqou> no, not saying there's a problem
<whitequark> i mean
<rqou> it just seems really unusual
<whitequark> old equipment gets decommissioned
MoCoMakers has joined ##openfpga
MoCoMakers has quit [Client Quit]
<awygle> I am gonna try to go to ORconf
<awygle> Also, my second reflow in this oven went much better
<awygle> Also also, I should have made a real assembly drawing given that this board has no silkscreen
<whitequark> I think I'll present Glasgow at ORconf
<whitequark> got an invitation
<zkms> what's ORconf
<whitequark> https://orconf.org/
<awygle> I also got an invite
<awygle> Which was very flattering
<azonenberg> awygle: i just go thorugh the kicad design as i assemble
<azonenberg> i never export assembly drawings
<rqou> hey awygle fun request
<awygle> azonenberg: I like to assemble away from my laptop for various reasons
<awygle> rqou: okay I'm braces
<awygle> *braced
<rqou> awygle: can you check if CalNet SSO is working for you?
<rqou> seems to be borked for me
<awygle> No, because I lost my creds in the email migration
<rqou> oh ugh
<awygle> Bon chance
* awygle goes looking for a quarter
<whitequark> ok I think I should have the solder paste today
<whitequark> hypothetically
<whitequark> awygle: what should be the next priority
<azonenberg> awygle: i'm planning to throw a little SBC plus monitor down in the garage
<azonenberg> on every workbench
<whitequark> the O is for overkill
<azonenberg> for pulling up datasheets, viewing layout/sch during debug, etc
<whitequark> azonenberg: hm, the datasheet for ice40 fpgas says that reset time depends on the model
<whitequark> and it's proportional to memory size
<whitequark> do they have a small FSM inside that pushes zeroes through the configuration?
<azonenberg> that sounds likely
<azonenberg> a massive high fanout reset net would be hard to route
<azonenberg> so it probably just uses the config datapath
<whitequark> huh
<whitequark> ice40 delayer when
<azonenberg> When i get my new fume hood
<azonenberg> In my new lab
<azonenberg> And when i've cleared out the backlog of other chips i promised people i'd do once the lab was up and running
<awygle> Yes my usb connector is crooked. It did not have alignment pins, must be the wrong part number. Also, it's definitely going to snap off lol.
<rqou> wtf awygle why do you have a 1 yuan coin?
<awygle> It's a yen I'm pretty sure
<rqou> hmm it is
<rqou> i think
<awygle> The only time I carry cash is when I travel internationally
<rqou> so this was for weebmode? :P
<awygle> Lol well I went to Japan, yes
<awygle> whitequark: adc/dac?
<whitequark> awygle: right, good idea
<whitequark> hm
<whitequark> I kinda really hate the mousebites on oshpark boards
<whitequark> somehow dirtypcbs can give me a proper outline without any junk on it
<awygle> Me too tbh
<azonenberg> whitequark: they probably individually route them
<azonenberg> awygle: guess what i forgot on that board (and found during design review)?
<whitequark> azonenberg: wonder how they hold them
<whitequark> vacuum plate?
<azonenberg> never looked into it
<azonenberg> underpaid small children?
<azonenberg> awygle: i forgot to label the sma connectors
<azonenberg> :P
<azonenberg> Fixed now
<awygle> I would have guessed v score
mumptai_ has quit [Quit: Verlassend]
mumptai has joined ##openfpga
<whitequark> vscore?
<whitequark> rounded corners?
<whitequark> you pay extra for vscore too
<awygle> Ah rounded corners is fair
<awygle> I'm back to ignorance then
<awygle> Okay time for kicad stuff then bedtime
<awygle> It feels good to accomplish things I say I will accomplish lol
<rqou> wtf how does digshadow have so many micropositioners?
<rqou> why don't i have any :(
<whitequark> haha
<rqou> whitequark: ok, how many do you have? :P
<awygle> i had never heard of them, but now i desperately want one
<rqou> awygle: shoulda taken 143
<rqou> although idk if that would actually be a good thing or not since most of the 143 micropositioners were fucked up :P
<awygle> lol
<awygle> academia!
<rqou> the lab staff are amazing though
<rqou> but still, finite resources + students
<digshadow> rqou: I've been collecting them for a while. The best story is the S-930 which I got at the flea market for $5 from the guy selling honey...
<azonenberg> loool
eduardo_ has joined ##openfpga
<whitequark> wtf
eduardo__ has quit [Ping timeout: 255 seconds]
<azonenberg> lol oshstencils now lets you opt out of getting a paste spreader and sticker with every order
<azonenberg> i think i have enough of those right now...
<azonenberg> awygle: incidentally, this board will be useful if i ever want to test some cheap ethernet hardware, lol
<azonenberg> stick 30 dB in line and see if it still links up :p
<rqou> realtek will fail :P
<azonenberg> lol
<azonenberg> how about a ksz9031?
<azonenberg> or a ti industrial phy?
<awygle> i have a remote in this github repo called "calebns" and no memory of creating it
<awygle> apparently that's a person who was also working on kicad footprints
<daveshah> awygle and rqou: you found the secret hx8k pin by the looks of things ;)
<awygle> daveshah: indeed
<daveshah> The one NC pin is jtag enable IIRC
<awygle> ah
<daveshah> I found a really old schematic somewhere
<daveshah> But I think you also need to send a spi command to enable jtag too
<awygle> daveshah: be prepared for some questions in the next few days (maybe this weekend), i finished my LM board and intend to get icefuzz et al running now that i have something to test on
<daveshah> Sure, I'm very happy to help with that
<awygle> whitequark: so i did the MSOP and the DFN, i guess i'll just take the cypress footprint we used on the board and open a new PR?
<awygle> i thought there was a footprint pr languishing someplace but i can't find it
<whitequark> awygle: sec
<whitequark> I think our footprint isn't quite what they want
<whitequark> awygle: by the way, would you be up to also adding 3D packages that are currently missing upstream?
<whitequark> there's two of them that we can submit
<whitequark> actually I think I submitted one
<awygle> whitequark: i can, i just don't actually know any 3D modeling software. so it would take a bit.
<awygle> and you're right, our symbol isn't quite what they want. i like some of their ideas but their proposed paste layout is not good imo. i'm going to put a screenshot of each in that issue for the librarians to look at.
<awygle> do you know what "row/column spacing 8mm" means?
<whitequark> awygle: it's not about 3d modelling software
<whitequark> well, for the most part
<whitequark> the models are automatically generated in cadquery, and I even adjusted all the parameters
<whitequark> the xtal and molex connector models aren't so those will need to be done from scratch
Lord_Nightmare2 has joined ##openfpga
Lord_Nightmare has quit [Ping timeout: 248 seconds]
Lord_Nightmare2 is now known as Lord_Nightmare
<awygle> ah i see
<awygle> well how about i put that on my list behind the rev B stuff
<awygle> sound fair?
<whitequark> sure
<whitequark> you have something like 2-3 weeks at least to get rev B
<whitequark> made
<awygle> that's good to know
Lord_Nightmare2 has joined ##openfpga
Lord_Nightmare has quit [Ping timeout: 248 seconds]
Lord_Nightmare2 has quit [Ping timeout: 248 seconds]
Lord_Nightmare has joined ##openfpga
<awygle> okay, i dropped some things for comment into that thread
<awygle> whitequark: are there any other open issues besides the MSOP/MLF pull request, the ice40 symbol PR, and the cypress footprint?
<awygle> (i don't believe so and neither does the repo, just double checking)
<whitequark> awygle: nope
bitd has joined ##openfpga
Lord_Nightmare2 has joined ##openfpga
Lord_Nightmare has quit [Ping timeout: 248 seconds]
Lord_Nightmare2 is now known as Lord_Nightmare
Lord_Nightmare2 has joined ##openfpga
Lord_Nightmare has quit [Ping timeout: 268 seconds]
Lord_Nightmare2 is now known as Lord_Nightmare
indy has quit [Quit: ZNC - http://znc.sourceforge.net]
indy has joined ##openfpga
bitd has quit [Quit: Leaving]
digshadow has quit [Ping timeout: 240 seconds]
bitd has joined ##openfpga
bitd has quit [Quit: Leaving]
futarisIRCcloud has quit [Quit: Connection closed for inactivity]
digshadow has joined ##openfpga
bitd has joined ##openfpga
Bike has joined ##openfpga
bitd has quit [Quit: Leaving]
genii has joined ##openfpga
<whitequark> awygle: hm, I did a stupid thing
<whitequark> I should have ordered a laser-cut frame for positioning the board during stencilling
<Zorix> everyone else just tapes unused boards to the table for positioning
<Zorix> a makeshift frame
<gruetzkopf> i see you've been boiling galden
<gruetzkopf> this is already way less ghetto than my setup
eddyb has joined ##openfpga
<openfpga-github> [Glasgow] whitequark pushed 1 new commit to master: https://github.com/whitequark/Glasgow/commit/77dde5b1999147bcd014ef67b54afa5b31ffbab2
<openfpga-github> Glasgow/master 77dde5b whitequark: Add revA fab pdf.
bitd has joined ##openfpga
bitd has quit [Client Quit]
<q3k> whitequark: orconf looks cool, see you there ^^
<pie__> y'all going to orconf? neat
<pie__> or = operations reasearch?
<pie__> *research
<pie__> nevermind
<jn__> pie__: isn't it OpenRISC?
<pie__> i always end up thinking of the former first
<pie__> unrelated, i should try making a hexchat addon at some point that shows people you have common channels with
<jn__> social networking like it's 1995 :P
<pie__> seems like a missed opportunity for finding people with overlapping interests
<pie__> haha
<pie__> then again im probably the only person with like 50 open channels
<whitequark> pie__: i got invited
<pie__> cool
<gruetzkopf> pie__: i'm definitly over 100
<q3k> I'm honestly very pleasently surprised that orconf is in Poland.
bitd has joined ##openfpga
<awygle> Yeah definitely open risc
<awygle> Poland seems really cool. Hell of a flight for me though. Anybody based there?
<egg|zzz|egg> did someone say egg
<egg|zzz|egg> ah nevermind just double
<q3k> awygle: born and lived there for quite a while
<q3k> awygle: implr lives in Warsaw still
<awygle> q3k: awesome
<whitequark> awygle: you updated the resistor values
<whitequark> but you did not update Mouser_PN :( :(
<whitequark> at least as far as I can tell
<whitequark> because I have a lot of bags of resistors that aren't anywhere on the fab layer
<awygle> whitequark: I don't think I changed any resistors
<awygle> That sucks though :-(
<whitequark> awygle: R5, no?
<whitequark> maybe it was me
<whitequark> the rest are just LEDs so it's ok
<openfpga-github> [Glasgow] whitequark opened issue #17: R5 has wrong Mouser_PN https://github.com/whitequark/Glasgow/issues/17
<awygle> Wow I was going to suggest a 2k2 and a 1k but we didn't use a single 1k
<asy> is Orconf something "big"? Like, how many people will be there?
<awygle> Looks like 60 people last year. It's certainly not ccc or anything
<asy> Still looks nice.
<felix_> rqou, balrog: yep, since last month i have an employee working part-time on getting good ida support for the xtensa used in the ath10k chips
plaes has quit [Ping timeout: 276 seconds]
<openfpga-github> [Glasgow] whitequark opened issue #18: R16 has wrong Mouser_PN https://github.com/whitequark/Glasgow/issues/18
<whitequark> hm, I think just these two are wrong
<whitequark> and R16 is definitely my faut
<whitequark> fault
plaes has joined ##openfpga
plaes has joined ##openfpga
<openfpga-github> [Glasgow] whitequark opened issue #19: Move R12-R16 to top side https://github.com/whitequark/Glasgow/issues/19
<pie__> "This is huge if it holds up. Epstein Barr virus triggers up to 7 auto immune diseases."
<whitequark> awygle: so, everything except QFNs is a-ok
<whitequark> QFNs have bridges galore
<whitequark> que?
<openfpga-github> [Glasgow] whitequark opened issue #20: Y1 footprint is way too big https://github.com/whitequark/Glasgow/issues/20
<pie__> more random: https://github.com/IDArlingTeam/IDArling/ yet another IDA collaborative reversing plugin, maybe something will come of this one
<Zorix> id like to see somethin open source that can compete with ida..
<openfpga-github> [Glasgow] whitequark opened issue #21: J1 does not clear the board outline https://github.com/whitequark/Glasgow/issues/21
<pie__> Zorix, ask me in 5 years
* pie__ is *totally* passively working on that. totally.
<Zorix> hehe
<pie__> ; __ ;
<Zorix> theres one that was somewhat decent last i checked but i dont even remember the name of it
<pie__> ive seen a lot of stuff that does some very base things but i dont think ive seen anything thats gotten particularly far yet
<pie__> actually there was that one decompiler (iirc?) they open sourced relatively recently, not sure what one can glean out of that
<jn__> i'd like to see ScratchABit getting some users
<pie__> jn__, 's favorite ^
<pie__> i havent played with it yet
<jn__> it's a bit of a pain to even install (python spread over multiple submodules), but it gets some stuff right
<Zorix> looks interesting
<pie__> the guy that does panopticon.re is sidetracked with his ?masters thesis? for the moment, rust-falcon is getting worked on slowly
<jn__> and the DB is plaintext and can thus be tracked in git
<q3k> pie__: who isn't working on an ida clone ^^
<pie__> q3k, good point i guess
<pie__> jn__, i think it came up the other day in here that apparently git can offload to other diffing tools???
<pie__> (not sure)
<whitequark> yes it can
<jn__> pie__: i read about that in the context of that blogpost about PCB diffs
<whitequark> it's called filters
<pie__> anyway those are what ive got off the top of my head because i idle in the irc channels, but theres probably a bunch of other stuff
<pie__> oh im dumb lol ,well theres radare but ... :P
<pie__> whitequark, hm
<Zorix> thats it, radare is what i was thinking of
<awygle> whitequark: sounds like the stencil aperture is too big on the qfn pads. honestly it could stand to be reduced globally it looks like
<awygle> those don't look too bad, and they don't look like the kind of thing that indicates shorts to the thermal pad. some flux and either an iron or hot air will hopefully resolve them.
<awygle> mm... actually i was mostly looking at the cypress, the others are a little more concerning
<awygle> oh hey, you're getting twitter advice from an old friend of mine
<awygle> small world
digshadow has quit [Ping timeout: 264 seconds]
bitd has quit [Quit: Leaving]
mumptai has quit [Quit: Verlassend]
mumptai has joined ##openfpga
<whitequark> awygle: which?
<awygle> whitequark: Henry Hallam. Friend may be overstating it but I do know him and have for a long time.
digshadow has joined ##openfpga
<whitequark> ah, huh.
<whitequark> gonna touch it up now and solder the remaining caps
<whitequark> of which there are way too many
<rqou> oh wtf
<rqou> ebay is _forcing_ people onto sms 2fa
<rqou> no more totp
bitd has joined ##openfpga
florolf has quit [Remote host closed the connection]
<awygle> whitequark: i know you deal with python build stuff all the time, do you deal with it on windows?
<whitequark> awygle: unfortunately, yes
<whitequark> so
<whitequark> I assembled it
<whitequark> does it give up magic smoke? does it work? cast ur vote
<awygle> are you going to check power/ground impedances before you power it up?
<whitequark> welll I was going to
<whitequark> but I forgot my meter in the lab
<awygle> o
<whitequark> and I don't wanna wait
<awygle> yeah fair
<awygle> i predict "it does not melt but also does not completely work"
<whitequark> same tbh
<whitequark> AUGH MY EYES
<azonenberg> whitequark: 500 candlepower blue leds? :p
<whitequark> azonenberg: green
<azonenberg> but stupid-bright?
<azonenberg> and this was after i talked you out of running them at however many hundreds of mcd you were going to originally? lol
<whitequark> yes
<whitequark> this is 1 mA
<whitequark> 30 mcd
<azonenberg> lol
<azonenberg> i told you it was too much :)
<openfpga-github> [Glasgow] whitequark opened issue #22: Reduce LED intensity by at LEAST three times https://github.com/whitequark/Glasgow/issues/22
<azonenberg> whitequark: can we see a photo?
<pie__> azonenberg, no because it will be too bright
florolf has joined ##openfpga
<pie__> emphasis on allowed to see, but cant
<azonenberg> lol
<whitequark> azonenberg: sec
<whitequark> it's on my twitter :p
<whitequark> and no, it's not just the camera, it leaves holes in my retina too
<openfpga-github> [Glasgow] whitequark opened issue #23: Crystal pinout is inverted https://github.com/whitequark/Glasgow/issues/23
<awygle> i'm suddenly extremely paranoid about crystal pinouts on smolfpga and that lnb board
<pie__> whitequark, i never did ask where you got your catgirl from
<pie__> (on bird site)
<Bike> catullus 16
<azonenberg> awygle: i almost never use discrete crystals
<azonenberg> i use oscillators, which generally have a pretty standard pinout in my experience
<azonenberg> But it never hurts to double check
<awygle> i use discrete crystals when e.g. MCUs have crystal drivers, usually
<azonenberg> whitequark: lol
<azonenberg> ok that is bright
<azonenberg> i'm curious now, how bright is my standard LED
<azonenberg> 35 mcd @ 20 mA, Vf of 2V
<gruetzkopf> the indicator LEDs in my current project are too bright at 10k at 5V, 5% duty cycle..
<azonenberg> So if i run it at 3.3V with a 470 ohm resistor that's 1.3V over 470 ohms or about 2.8 mA
<openfpga-github> [libfx2] whitequark pushed 1 new commit to master: https://github.com/whitequark/libfx2/commit/b8510b16953fe9a6a85102b1d71d587a22b257cf
<openfpga-github> libfx2/master b8510b1 whitequark: Fix wrong exception class used in FX2Device.
<gruetzkopf> (also the cheapest i could get)
<azonenberg> Which according to the curve is... about 0.125x the nominal intensity
<azonenberg> or about 4.3 mcd
<azonenberg> That is still comfortably bright
<azonenberg> These are not "high efficiency" LEDs though
<azonenberg> if they were, i'd be running them probably in the tens of uA
<azonenberg> to get the same brightness :p
<whitequark> hm
<openfpga-github> [Glasgow] whitequark commented on issue #22: Correction: reduce PWR led intensity, and increase intensity of the *other* LEDs. I am not taking the internal resistance of the Cypress driver into account correctly. I am not smart. https://github.com/whitequark/Glasgow/issues/22#issuecomment-385778031
<openfpga-github> [Glasgow] whitequark commented on issue #22: Disregard the previous comment. I did not enable the output driver. "I am not smart" still applies. https://github.com/whitequark/Glasgow/issues/22#issuecomment-385778908
<whitequark> azonenberg: yeah, I've carefully matched the LED brightness and they're all very precisely equally eye-searing.
<whitequark> actually, WTF
<whitequark> leakage current from ... somewhere ... is enough to light up the LEDs
<azonenberg> through protection diodes?
<azonenberg> i've seen that happen
<azonenberg> weak pullups before boot?
<whitequark> no, the CY7C part doesn't have pullups
<whitequark> I think
<whitequark> this is after boot anyway
<whitequark> this is ridiculous
<whitequark> oh
<whitequark> oh I figured it out
<openfpga-github> [Glasgow] whitequark pushed 1 new commit to master: https://github.com/whitequark/Glasgow/commit/7dfcb30024cd394cf663122d1fd46e4e82655b0f
<openfpga-github> Glasgow/master 7dfcb30 whitequark: Actually set the LED_CY pin as output.
<whitequark> it's the *FPGA* pullups I think
<azonenberg> that would do it
<whitequark> I don't have them connected to the FPGA but it's on the same port...
<whitequark> no better ideas
<openfpga-github> [Glasgow] whitequark pushed 1 new commit to master: https://github.com/whitequark/Glasgow/commit/28bb2df1d5fc521df4a01592b1955ce0dd394d16
<openfpga-github> Glasgow/master 28bb2df whitequark: Make all endpoints byte-wide to avoid interfering with port D.
<whitequark> ok figured it out
<whitequark> that was leakage via tristated drivers
<openfpga-github> [Glasgow] whitequark pushed 2 new commits to master: https://github.com/whitequark/Glasgow/compare/28bb2df1d5fc...a9274fca428e
<openfpga-github> Glasgow/master a9274fc whitequark: Fix layer visibility.
<openfpga-github> Glasgow/master 5732ea2 whitequark: Add missing SYNCDELAY().
<azonenberg> awygle: btw
<azonenberg> what do you think of throwing a QDR-II+ footprint on the brain board next to the FPGA?
<azonenberg> rough budgeting says i'll have enough pins to throw one down, unsure if i can route it in 6L but it's worth a try
<azonenberg> it would have to use HR banks which limits fmax in -2 speed to 450 MHz / 900 MT/s
<azonenberg> Times 18 bits wide is about 16 Gbps of bandwidth
<azonenberg> and it's simple dual port so that's 16 Gbps in and 16 out
<azonenberg> and sram so no refresh overhead
<azonenberg> It's not a cheap part, 36 Mb of QDR-II+ is about $90, so we might choose not to populate it
<awygle> rather have it than not need it than need it and not have it
<awygle> if you can route it
<azonenberg> but i'm thinking if we start adding layer-3 switching capabilities, port mirroring, etc
<azonenberg> it might be nice to have some larger buffering
<azonenberg> my immediate use case is advanced port mirroring
<azonenberg> I wanted to do "pcap over TCP"
<azonenberg> so you can do nc switch 9001 > foobar.pcap
<azonenberg> If we want to support this at full 10Gbps line rate that means we need a tcp transmit buffer big enough to do TCP at 10 Gbps
<azonenberg> And i dont think we have enough block ram
<azonenberg> You can get 36-bit wide QDR-II+ as well but i dont think we have the pins for that
<awygle> what do you think of MatthiasM's thing about hacking the checksum btw?
<azonenberg> To decrement ttl?
<azonenberg> i thought that was standard practice
<awygle> ye[p
<awygle> it may be
<azonenberg> if you were doing cut-through i mean
<awygle> i hadn't heard of it
<azonenberg> Right now, re pin budgeting, the line cards will collectively use 96 HP and 39 HR banks
<azonenberg> then 7 HRs for the boot flash and UART to the ARM
<azonenberg> We have 185 HR and 100 HP so 4 HP, 139 HR available
<azonenberg> An 18-bit QDR-II+ needs 67 pins
<azonenberg> So we'd have 4 HP and 72 HR available after plopping it down
<azonenberg> While we could theoretically fit a 36-bit QDR-II+ i'm not sure it would prove to be routable on six layers
<azonenberg> And i want pins free for all of the other stuff that i didn't include in that budget
<azonenberg> like i2c to the SFP+ modules
bitd has quit [Quit: Leaving]
<azonenberg> module-present indicators, SFP tx/rx LEDs
Bike has quit [Ping timeout: 260 seconds]
<awygle> oh how i love python
<awygle> this module is not installed, but shows up in pip list, and can't be uninstalled with pip uninstall.
<q3k> python packaging is such a shitshow
<q3k> but otoh there's not many languages that actually get it right
<genii> Go seems interesting, it can be iterpreted, or compiled
<whitequark> I don't think Go is ever interpreted
<sorear> "interpreted versus compiled" is a fairly bs distinction these days
<whitequark> yes, but even within that context I don't think Go is ever interpreted
<sorear> pure interpreters like *old* Tcl are a dead art
<genii> Hm
<whitequark> not entirely
<whitequark> if you need extremely compact code and don't care about performance much, they're still best
<whitequark> oh, wait
<whitequark> *old* Tcl
<whitequark> nevermind
<pie__> i dont even know what that means
<pie__> interleaving parsing and interpretation?
<pie__> *execution
* Prf_Jakob looks at Lua
<jn__> lua is a parser + bytecode interpreter
<openfpga-github> [Glasgow] whitequark opened issue #24: Migrate to iCE40HX8K-BG121 https://github.com/whitequark/Glasgow/issues/24
<jn__> a setup that doesn't feel rare to me
Bike has joined ##openfpga
<q3k> genii: nobody really does 'go run'
<q3k> genii: since that limits you to single-file programs
<Ultrasauce> the best kind of program
<q3k> it's more of a convenience for quick tests/demos
<q3k> not to mention that go usually isn't the best language for one-off hacks
<q3k> unless you're doing something where easy concurrency is critical
<q3k> (like a web scraper)
<q3k> even then, python+gevent/asyncio might be less painful than go
<awygle> whitequark: did that package appear like... since yesterday?
mumptai has quit [Remote host closed the connection]
<whitequark> awygle: what
<whitequark> yesterday?
<awygle> the 121-ball 0.8mm BGA
<whitequark> i think it's at least a week old.
<awygle> wow
<whitequark> actually, no
<whitequark> it's at least oct 2017 old
<whitequark> since that's the revision of DS1040 that has it
<whitequark> but yes, it's very new
<sorear> `go run` can be used for multi-file programs if you do `go run *.go`
<whitequark> awygle: so basically what I want you to do is
<whitequark> I'll fix revA
<whitequark> I want you to fix the issues in revA and make revB
<whitequark> still with UP5K
<q3k> sorear: this actually works? dear god
<whitequark> this in case e.g. people want to do make it themselves and they don't want to bother with BGAs
<whitequark> and then a revC with 121-ball BGA
<q3k> sorear: they still all have to be in 'package main', right?
<openfpga-github> [Glasgow] whitequark opened issue #25: Add an USR LED connected to FPGA https://github.com/whitequark/Glasgow/issues/25
<cr1901_modern> Maybe I'll buy revB and try reflowing it myself as a prereq for C
<openfpga-github> [Glasgow] whitequark commented on issue #19: This would involve moving R1/R2/R11 to the right into the great empty plains of EEPROM-land. https://github.com/whitequark/Glasgow/issues/19#issuecomment-385812816
<whitequark> awygle: ^ see?
<sorear> q3k: yes, you pass the package main files
<mithro> whitequark: I'm actually pretty excited about your Glasgow!
<whitequark> mithro: nice!
<whitequark> I have quite a backlog of people
<whitequark> like uhhh I should make at least fifty
<awygle> whitequark: sure
<whitequark> a hundred would not be out of question but I don't think I have enough money on hand and I really don't want any crowdfunding
<awygle> seems reasonable
<mithro> whitequark: Do you need any hook ups with manufacturers or crowdfunding campaigns are you all good in that land?
<whitequark> mithro: manufacturers, possibly
<awygle> whitequark: by "i'll fix rev A" do you mean you'll continue what you're doing now and find and work around all the issues?
<cr1901_modern> I still have a GP breakout board _somehwere_ I should try reflowing
<whitequark> mithro: right now I'm thinking about using the cheap chinese options like dirtypcbs and seeed
<whitequark> this board is not very demanding, e.g. dirtypcbs design rules are sufficient
<Ultrasauce> I'd be happy to front it whitequark, probably not the only one
<awygle> whitequark: i can front the money if need be
<whitequark> I don't know if seeed can be trusted to do the job right
<whitequark> wow
<whitequark> okay
<whitequark> thanks :D
<mithro> whitequark: If you can get the BOM under $30 USD for 100 units, I'd be happy to front the money to get 100 initial too...
<q3k> whitequark: same if you're willing to get one more made for me
<whitequark> mithro: not sure about $30
<whitequark> or rather
<q3k> (front one, not 100 :D)
<awygle> lol suddenly we're building *counts* 400 boards
<whitequark> it can definitely go to $30 if we cut the analog frontend
<whitequark> but I don't like that.
<mithro> I haven't really looked at the BOM to see if there is anything I could recommend
<Ultrasauce> smaller led current limiters. gotta light up the room
<awygle> mithro: do you have a particular reason for that target number?
<whitequark> mithro: less decoupling lol
<mithro> awygle: More kind of $30 USD BOM ~= $50 USD sale price for crowdfunding, shipping, packaging, etc
<whitequark> right now it goes to C47
<whitequark> and those 0603 MLCCs aren't cheap
<whitequark> actually it's fucking impossible to get them too
<whitequark> I bet all the caps in my BOM are already evaporated from stock
<whitequark> mithro: we could cut the FPGA EEPROM
<mithro> whitequark: On the Opsis, I use the FPGA to emulate the EEPROM for the FX2
<whitequark> well
<whitequark> FX2 needs 8 bytes for C0, that's less than 50 cents
<awygle> price breaks might get us there all on their own, they can be fairly significant at 100 units
<whitequark> but the FPGA bitstream EEPROM is 128k
<awygle> or at least pretty close
<whitequark> and since in this system FX2 configures the FPGA, it has to be an I2C EEPROM
<azonenberg> Yeah, dont forget price breaks
<azonenberg> Things do drop off a lot in volume
<mithro> whitequark: Well, on the Opsis the FPGA is using SPI Flash, so I guess it's a different class...
<whitequark> it could certainly be redesigned to go the other way around with an SPI flash, but I'm not sure I like that very much
<whitequark> right now the FX2 is "the source of truth"
<whitequark> we could *possibly* cut the reset supervisor although this is not strictly legal per the datasheet
<whitequark> Lattice themselves don't bother on their devboards AFAICT
<whitequark> with power sequencing that is
<whitequark> the FX2 *needs* a power sequencer to be USB compliant though
<whitequark> reset sequencer
<awygle> let's revisit this after the rev B exists when we're planning for rev C
<whitequark> sure
<gruetzkopf> "usb compliant"
<gruetzkopf> i've yet to see the device that's completely usb compliant
<whitequark> gruetzkopf: well I'd like to at least try
<gruetzkopf> i can understand that
<whitequark> in general, I try hard to adhere to proper standards and procedures and not cut corners
<whitequark> this is really why people use my software
<whitequark> e.g. https://rubygems.org/gems/parser with 39 million installs...
<whitequark> this is just 1/3 as many as Rails
<whitequark> and it was an incredible pain in the ass to get right
<whitequark> awygle: this cut acrylic case is unexpectedly sturdy.
<whitequark> doesn't flex at all
<whitequark> even the top part weakened by cutouts
<q3k> acrylic cases are nice until you overtighten screws
<whitequark> indeed
<whitequark> that's why I want to ship these threadlocked
GenTooMan has joined ##openfpga
<gruetzkopf> aah, threadlock
<gruetzkopf> the saviour of many poorly designed automobiles
<gruetzkopf> (looking at you, fiat)
<whitequark> wtf
<q3k> that reminds me, I need to buy some proper blue locktite
<whitequark> I did not want to know that
<q3k> instead of using cyanoacrylate like an animal
<zkms> safety wire
<gruetzkopf> i'd love to
<q3k> zkms: I did once use kynar wire to help me thread a screw into a printed PLA part
<openfpga-github> [Glasgow] whitequark opened issue #26: Test/loopback PCB https://github.com/whitequark/Glasgow/issues/26
<q3k> although it wasn't locking it by the head, but threading it through the hole of the part so that the screw has something to bite into
<q3k> well, bolt, not screw
<whitequark> awygle: so hm
<q3k> that reminds me, never trust me with mechanical engineering of anything
<whitequark> double-sided reflow
<whitequark> can I simply reflow the back side take it out then reflow it both again?
<whitequark> it's not like a bunch of capacitors will go bad from being heated to 230°C twice
<awygle> whitequark: yup
<q3k> oh, is glasgow populated on both sides?
<whitequark> yes
<whitequark> all the decoupling is on the back side
<q3k> BGA?
<whitequark> QFN
<awygle> and for the record almost all ICs will not go bad from being heated twice either
<q3k> hum
<whitequark> q3k: it's really tight
<q3k> I can imagine
<whitequark> in revA I have every single pin used
<q3k> just that this will make your PCBA more expensive
<whitequark> on both FX2 and iCE40
<whitequark> actually
<whitequark> seeedstudio doesn't charge more for double-sided PCBA
<q3k> at 100 parts you might want to spend that extra NRE time for DFM
<q3k> oh?
<q3k> that's surprising
<whitequark> yeah
<q3k> one product a client of mine is shipping has a crazy layout because of how much extra the ODM would charge for double sided pcba
<q3k> but that's in a proper mass production run
<whitequark> there are more parts that would really benefit from having decoupling on the back
<whitequark> this would really mess up the level shifters
<whitequark> right now there are no vias in signal path
<whitequark> hmmm
<whitequark> q3k: I guess the real question is
<whitequark> is it worth moving *some* decoupling to the top
<whitequark> if I'm pretty sure I can't move *all* of it to the top
<q3k> well if you can't get rid of all of it then no
<whitequark> without some serious compromises
<q3k> i mean, the only difference is that you might have to have less feeders for pnp on the second side
<whitequark> awygle: yeah but I don't wanna reflow the top side 2nd time because all the shit is gonna fall off it
<gruetzkopf> i have some 90's grade SIL-4 IO boards here
<q3k> but they're probably reusing the same ones as on the top side so that doesn't change much
<q3k> feeders/reels
<awygle> whitequark: yeah obviously
<whitequark> q3k: also it's literally two components on the bottom side
<whitequark> 0.1u and 4.7u
<whitequark> ok also a 100R
<whitequark> because PLL
<gruetzkopf> 16I+16O, double sider everything, 6 layer eurocard, ~400 components front, 200 back..
<q3k> if you're doing ~100 units and they don't charge you extra for the second side then I wouldn't bother
<whitequark> gruetzkopf: wtf
<q3k> mmm eurocards
<whitequark> what's 16I+16O?
<q3k> input + output I guess
<gruetzkopf> 16 digital in, 16 digital out
<whitequark> ahh
<q3k> gruetzkopf: pix? ^^
<gruetzkopf> at 24V logic level, negative side switched
<gruetzkopf> not my design
* whitequark giggles
<whitequark> i should ship one of the boards with a 15 mA limited power LED
<whitequark> as a prank
<azonenberg> whitequark: lol
<whitequark> then watch google maps for sensor saturation
<whitequark> to see who got it
<azonenberg> :P
<gruetzkopf> can't find any published image where you can see more than a DIN41612 F48M connector
<azonenberg> whitequark: anyway, 2 side reflow is totally fine
<azonenberg> the only thing to be careful about is having heavy stuff on the bottom
<whitequark> azonenberg: well I was wondering about like
<whitequark> this paste doesn't seem to hold shit
<azonenberg> When in paste form, no
<azonenberg> You do two oven cycles
<whitequark> I wouldn't be comfortable populating a side and then flipping the board
<whitequark> ahh
futarisIRCcloud has joined ##openfpga
<azonenberg> The underside will melt but as long as you design the board right surface tension will hold them on
<cr1901_modern> why don't the parts fall off on the second cycle? Can it be done on a skillet?
<whitequark> I thought there's some way to do it in one cycle though?
<azonenberg> There's an equation to, depending on solder alloy
<whitequark> azonenberg: hmmmm
<azonenberg> and perimeter
<azonenberg> calculate the max mass of a part you can have on the bottom without glue
<whitequark> right now I use all the HandSoldering footprints
<azonenberg> cr1901_modern: I always do oven reflow
<whitequark> cr1901_modern: I always do vapor phase reflow
<azonenberg> whitequark: any small passive should be fine, period
<whitequark> :P
<cr1901_modern> whitequark: I don't have an ionpak to bastar- err, convert :P
<azonenberg> the things to not put on the bottom are like usb connectors, electrolytic caps, etc
<azonenberg> maybe a big qfp although 48-QFN and smaller are fine in my experience
<whitequark> cr1901_modern: that was just a 19" blade
<azonenberg> but i wouldn't do like a 144-QFP
<whitequark> doesn't matter that it's ionpak
<cr1901_modern> Ahhh. Hmmm...
<whitequark> after I make a production run of glasgow I *might* consider selling cheap vapor phase reflow machines
<q3k> whitequark: did you figure out the briding?
<q3k> *bridging
<gruetzkopf> it's a really nice process
<whitequark> q3k: probably just my shitty stencilling
<q3k> the large parts seem to have made it fine, but those qfns looked bad
<q3k> whitequark: your paste application looked okay to me
<q3k> whitequark: but lemme take another look
<whitequark> well... I have paste in a syringe so I didn't use a lot, and made many passes with the applicator
<cr1901_modern> whitequark: For shits and giggles I was looking for any stainless steel/metal I had on hand this morning to heat under a power resistor to see if I could get a hot enough surface for reflow
<cr1901_modern> (I don't)
<whitequark> that seems to cause it
<azonenberg> whitequark: single pass with stainless gives much better results
<whitequark> azonenberg: yeah but I don't have that much solder paste
<azonenberg> i do one big line across the board then one pass with the spreader
<whitequark> since all I can't use is wasted
<azonenberg> i have mine in a syringe too
<q3k> whitequark: yeah okay, those QFN pads look bad, I guess that's it
<q3k> this is from mylar?
<azonenberg> i do like a 3mm or so wide x full board length bead of paste
<whitequark> no, stainless
<whitequark> 120 micron
<azonenberg> i've been using this one syringe of paste for like a year now
<q3k> i think you didn't have it pressed well enough against the board
<azonenberg> it lasts longer than you'd think
<whitequark> q3k: quite possibly
<azonenberg> a solid contact fit is important, yes
<whitequark> this is the first time I've ever done it
<azonenberg> whitequark: incidentally this is where the jig i was designing comes into play
<whitequark> oh!
<azonenberg> For the second reflow cycle you have to hold the pcb down on something
<azonenberg> while stenciling the top
<azonenberg> and there's already parts on the bottom
<whitequark> azonenberg: yep
<azonenberg> so i want to make a jig kinda like the stickvise but bigger, and with support on all sides
<azonenberg> that has a flush top
<whitequark> you can do that with stickvise
<whitequark> you need new jaws
<azonenberg> Not big enough
<whitequark> I will design those for myself
<q3k> a steady supply of failed prototype pcbs from previous projects and some kapton tape are the best jigs
<azonenberg> i might scale up the design
<q3k> as long as you have a nice collection of different thicknesses
<azonenberg> q3k: this is what i use for smaller boards without thick stuff on the back
<azonenberg> but the latentred line card will have 6mm high magnetics on the bottom
<azonenberg> So i need an alternate solution
<azonenberg> (yes, they will be glued before the second oven cycle)
<q3k> when I have components on the back I arrange the old PCB scraps so that I have a hole between then where the packages go
<azonenberg> yes i do that
<q3k> it's shoddy but yeah
<azonenberg> It works fine for 1-2mm of thickness
<q3k> a proper jig would be nice
<azonenberg> But at 6mm you'd need a lot of them
<whitequark> I am going to make a jig from laser cut acrylic
<azonenberg> I have some quick cad sketches but havent had time to finish the design
<whitequark> dirtypcbs has real cheap laser cut acrylic
<whitequark> do you see a pattern here lol
<q3k> whitequark: no hackerspace around with a lasercutter?
<Ultrasauce> pretty convenient to get everything done in the same place
<whitequark> hk hackerspaces are useless
<q3k> oh yeah, dimsumlabs
<whitequark> mostly because you have shenzhen next door
<whitequark> lol dimsum
<q3k> they're, uh, special
<whitequark> oh?
<q3k> been there a couple of times
<q3k> first time was nice (4 years ago?)
<whitequark> i've been there once
<q3k> last time was... weird
<whitequark> wrote it off as a typical useless hackerspace
<Ultrasauce> has dirtypcbs improved since a couple years ago? last board I got from them had a pretty significant offset on the solder mask
<whitequark> Ultrasauce: definitely
<whitequark> this board has perfect mask and silk
<whitequark> even where silk overlaps vias etc
<Ultrasauce> cool
<q3k> whitequark: next time your in shanghai drop by xinchejian, it's pretty good
<whitequark> never been in mainland yet
<q3k> not even szh?
<whitequark> nope
<q3k> huh
<gruetzkopf> i'm pleasantly surprised by the quality of boards i get from "cheapest supplier on pcbshopper" and how cheap they are
<q3k> not interesting, or can't?
<q3k> *interested
<whitequark> q3k: extremely stressful for me
<q3k> oh
<whitequark> for that matter, glasgow is a way to cope with stress to a large degree
<q3k> sorry to hear that
<whitequark> since even when it breaks, it breaks in a predictable and interesting ways
<whitequark> instead of just random obscure shit that's someone else's fault
<whitequark> i'll take reading cypress' obscure 8051 manuals over JESD204B not initializing for "some" reason any day
<q3k> (via implr from another channel)
<implr> also, linked there, old but delivers every time: http://tomforsyth1000.github.io/blog.wiki.html#%5B%5BWhy%20didn%27t%20Larrabee%20fail%3F%5D%5D
<implr> >you download the game from Steam and they Just Work - they totally think it's a graphics card! But it's still actually running FreeBSD on that card, and under FreeBSD it's just running an x86 program called DirectXGfx (248 threads of it).
<implr> lol intel
<implr> >Remember - KNC is literally the same chip as LRB2. It has texture samplers and a video out port sitting on the die. They don't test them or turn them on or expose them to software, but they're still there - it's still a graphics-capable part.
<implr> lol lol intel
<gruetzkopf> i went from frantically debugging xtensa stuff to WTF why isn't this relay doing stuff when it's clearly getting voltage
<whitequark> relays
<whitequark> what century is it again
<gruetzkopf> 60s design
<whitequark> yeah I remember
<whitequark> just funny
<gruetzkopf> my xtensa code tells me that voltage is present there, my multimeter agrees but there's nothing happening
<gruetzkopf> and so my signal isn't turning green
<whitequark> wait
<whitequark> you're doing -what-
<gruetzkopf> building debuggers for 60s tech
<whitequark> you're connecting an xtensa core to 60s railway relays?
<whitequark> this is wild
<gruetzkopf> when this is finished it's a ~ 6000ch logic analyser..
<gruetzkopf> slowpoke (1ksps)
<q3k> what the actual fuck
<whitequark> hahaha
<sorear> so per nyquist limits you can detect 500 trains passing per track per second. seems a bit low
<gruetzkopf> the system isn't quite that fast
<gruetzkopf> these safety relays are reasonably slow
<gruetzkopf> button-push to signal green takes ~3s in the typical case, if all points are set right before you push the buttons
<q3k> bridge this to openttd
<gruetzkopf> else it takes n_points*0.4+5s longer
<q3k> so you can actually control trains with them again
<azonenberg> awygle: btw, i did some calculations
<azonenberg> at 10 gigabit line rate, to a first order estimate
<azonenberg> with a 36 Mb TCP transmit buffer, assuming only one socket active, you can handle up to 3.6 ms round trip time
<azonenberg> before you have to throttle transmits due to un-ACKed data
<azonenberg> That seems like more than enough for a 10G LAN (the intended use case of the sniffing mode)
<azonenberg> in practice, i probably would dedicate ~half of the ram to buffer (for 1.8 ms max RTT) and use the rest for other stuff
<whitequark> openttd is awesome
Xark has quit [Ping timeout: 256 seconds]
Xark has joined ##openfpga
Ekho has joined ##openfpga
<awygle> holy crap i've managed to run a python package
<awygle> there must be _someone_ out there whose brain is shaped the way the python package/module/whateverthefuck system is designed, but it sure as fuck isn't me
<q3k> it's a clusterfuck
<q3k> well, okay
<q3k> not as much as the js ecosystem
<q3k> but still pretty bad
<awygle> i'm not even talking about like, pip and easy_install and conda and all that
<awygle> i'm talking about basic fundamental things that just don't fit in my head at all
<awygle> like, why are packages and modules even two distinct concepts, and what is the difference between them
<q3k> like packages/modules/pythonpath/__init__.py?
<q3k> yeah
<q3k> packages are a unit of distribution, modules are a unit of decoupling/abstraction