<pie_> its just like, my favorite math subject right now or something :P
<azonenberg> rqou: not a fan
<azonenberg> awygle: i saw an article of some folks who did vna measurements
<azonenberg> cap for cap, they did have better performance
<azonenberg> factoring in pcb area and increased cost (presumably due to lower production volume)
<azonenberg> you were better off paralleling two standard ones
<rqou> er, is this about x2y caps or the "axis flipped" ones?
<pie_> can you make a cap by putting two solder blobs next to eachother
<pie_> then again, technically yes, but i bet the performance is useless :P
<pie_> something something traces as caps
<pie_> </random>
<rqou> pie_: yes, that exists
<rqou> usually you call them "parasitics" :P
<pie_> the former or the latter :P
<pie_> hehe
<rqou> if you did it right, you might call them "distributed element filters"
seu has quit [Quit: http://quassel-irc.org - Chat comfortably. Anywhere.]
<awygle> rqou: +1. Or "rf magic"
<pie_> hehe
<pie_> interesting
<awygle> Lots of people have no idea how to attach big caps without ruining the esl
<pie_> actually that looks decidedly arcane
<azonenberg> awygle: i normally do a minimum of four vias on each pad of a large cap
<azonenberg> one on the middle of each side
<pie_> awygle, yeah?
<azonenberg> if i have ViP i do a nice big grid in the pad
<pie_> (also damn thoes some big solder blobs?)
<pie_> sidenote: where does one learn the right way to do things
<awygle> I do 3 or 6, and I put them in between the pads instead of going outward
<awygle> Shorter loop
<azonenberg> awygle: yeah sometimes i do more
<azonenberg> it depends on how critical it is
<awygle> Depends how big
<azonenberg> for 0402s normally i do one on each side above/below if i can fit
<awygle> pie_: work? Basically
<azonenberg> otherwise one side only
<azonenberg> Since you normally can't fit vias under the cap
<awygle> Yeah
<awygle> Or vip if you're doing vip obviously
<azonenberg> Well yeah vip is awesome
<pie_> " There is no precise frequency above which distributed element filters must be used but they are especially associated with the microwave band (wavelength less than one metre)." thats a much longer wavelength than i expected
<azonenberg> i just cant afford it for hobby boards most of the time
<awygle> 0402 is so small that you usually don't need to worry much unless you're at like 4+ GHz
<awygle> And you quickly swap to distributed around there anyway
<azonenberg> awygle: i think i've mentioned this before
<azonenberg> but i've often theorized about creating a ~300 THz power amplifier and oscillator
<azonenberg> then hooking it to a phased array of full-wave dipoles
<sorear> that's called a laser.
<azonenberg> Would you get a light source? :p
<awygle> I want to reclaim "taser" for this purpose
<awygle> Is 300THz light? I always swap over to nm before then
<azonenberg> awygle: yeah i think thats on the order of it
<azonenberg> its hundreds of THz, i forget exactly
<awygle> In that case you will be basically unable to make a dipole
<azonenberg> ok closer to 550 THz
<azonenberg> 300 is SWIR
<sorear> 300 is NIR it looks like
<awygle> You get completely swamped with parasitics
<azonenberg> awygle: why? can you not just make a 550nm long wire and drive it from one end?
<awygle> 300GHz+ is really, really hard
<sorear> I would argue that the output of the power amplifier, being an electromagnetic wave, is already light
<azonenberg> sorear: well thats the other fun bit
<pie_> <azonenberg> awygle: why? can you not just make a 550nm long wire and drive it from one end?
<pie_> :D
<sorear> and since light doesn't propagate in copper you can't use copper for this
<azonenberg> well that was the question
<azonenberg> is it possible to have a 500 THz sinewave in copper?
<azonenberg> it's not radiated, it's conducted
<sorear> what's the skin depth at that frequency?
<awygle> You *can*, kind of. But what do you drive it relative to, and what do you drive it into? You have to match the impedance etc
<azonenberg> awygle: yeah
<azonenberg> So, say 250nm of oxide and a solid copper ground plane
<azonenberg> These are within plausible IC dimensions
<awygle> I haven't ever tried but I did read this Heinlein novel and do some googling :-P
balrog has quit [Quit: Bye]
<sorear> tangentially, I feel like the definition of "RF" has shifted over the past 50 years
<azonenberg> The tricky part i think is the 500 THz oscillator and PA
<azonenberg> Not the antenna
<awygle> azonenberg: but silicon oxide is a shitty insulator at that frequency
<awygle> sorear: agree completely
<awygle> Look at what we call "hf" lol
<pie_> pfffff
<pie_> how high are you </badjoke>
<sorear> as I learned it RF went up to ~500MHz and past that was microwave
<pie_> rqou, thanks for mentioning this DIF stuff, its cool
<rqou> "the highest frequencies we can get! oh sh*t, now for 'very' high. oh wtf, now 'ultra' high. etc."
<pie_> *DEF
<awygle> For me RF is like, 30MHz to 300GHz
<azonenberg> rqou: lol band names are pretty funny
<awygle> Band names are funny because there are two different c bands
<qu1j0t3> like VGA resolutions
<pie_> should just use exponent
<rqou> also using wavelength instead of frequency for hysterical raisins
<pie_> ^
<awygle> Yeah I like how it goes wavelength frequency wavelength
<pie_> wavelength changes dependin on propagation media
<pie_> engineers pls
<awygle> 2m 5GHz 300nm
<sorear> is 2m that common? I think I see 150Mhz more often
<sorear> fortunately electronics rarely has to deal with the part where you switch to eV
<awygle> Not anymore, although I had a neighbor for a while with a 2m dipole
balrog has joined ##openfpga
<awygle> 70cm is pretty common
<pie_> wait...so 300THz would be even shorter in a material vs free space
<sorear> a…fairly large fraction of cars have 2m dipoles attached to them? (1m + ground plane reflector)
<awygle> pie_: yup. What's the Dk of SiO2?
<awygle> At 300THz
<pie_> Dk?
<pie_> dielectric konstant? :P
<sorear> and if you ask the owner of said car what their favorite radio station is in meters, they look at you blankly
* pie_ just made that up
<awygle> sorear: that's a quarter wave monopole
<awygle> For pedantry
<awygle> pie_: actually yes
<pie_> wait shit, is that from german?
<pie_> because it would make sense
<awygle> No idea but Dk is relative permittivity which is dielectric constant
<pie_> oh
<pie_> :P
<pie_> ask wolfram maybe i guess
<pie_> i dont actually know how to calc propagation speed from ...ah wait
<rqou> not a tensor just to make it more complicated? :P
<sorear> bulk Si (not SiO2) is transparent in IR and used surprisingly often for making optics
<pie_> c = 1/sqrt(epsilon_0 * mu_0) or what right?
<awygle> Then you have loss tangent, which might not apply in SiO2 since it's presumably isotropic?
<pie_> isotropic at what scale? :P
<awygle> fair
<pie_> sorear, yeah i was gonna bring up "visible light is an awfully small range so stuff might behave nonintuitively outside it and this is apparently 1000nm in free space"
<awygle> pie_: yes, although we usually shorten it to "divide wavelength by sqrt(Dk)" in nonmagnetic materials
<pie_> D * k or D_k?
<awygle> This is a fun trick to make your antenna smaller
<awygle> D_k
<pie_> oh wait thats not the same anyway
<pie_> lambda = 1/sqrt(D_k), never seen that
<pie_> oh
<awygle> "lambda' = lambda_0/sqrt(D_k)"
<rqou> btw i currently have to work with magnetic materials
<rqou> much fun
<rqou> f*cking magnets, how do they work?
<pie_> c = f * lambda => lambda = c/f => lambda = 1/sqrt(mu*epsilon)/f ...im guessin theres some engineers approximation involved here
<awygle> Yeah we assume mu=mu_0
<awygle> So relative mu is 1
<rqou> awygle: what happens if i hand you a metamaterial? :P
<awygle> And then Dk is relative epsilon
<awygle> rqou: I call my friend at kymeta
<rqou> wtf you have interesting friends
<awygle> rqou: kymeta is in like Bothell, it's really nearby
<awygle> rqou: also at my day job my "interesting friends" decap ICs with drain cleaner
<rqou> what
<pie_> guess not being behind gods back (probably a hungarian / this geographic area expression) helps
<rqou> not sure if you're referring to me or to some other friends
<awygle> I mean that in another context you are my interesting friend
<pie_> rqou, you work at awygle 's dat job? :P
<pie_> *day
<rqou> no, i'm just stuck here at ucb working on building an inductor
<awygle> Actually the cast of ##openfpga gets mentioned to our security guy pretty regularly when he's talking about threat models
<pie_> hehe
<awygle> (not by name)
<pie_> i mean come one
<pie_> *on
<rqou> btw pie_ in case you were curious this is what the datasheet for a magnetic material looks like: https://en.tdk.eu/inf/80/db/fer/e_32_16_9.pdf
<sorear> after seeing some of the problems you post I wonder if "just build the damn thing and then measure its electrical properties" would be the easiest way out
<awygle> rqou: start a magnetic pcb substrate startup
<pie_> if this isnt the closest community to cyberpunk novel garage hackers then what is?
<pie_> awygle, ^
<rqou> sorear: that's what I did
<awygle> sorear: "simulate and tweak" is a common RF design technique
<rqou> now i still have to write the actual lab report
<awygle> Math is hard
<awygle> FEM is easy
<awygle> (no its not)
<rqou> yeah, we had "the math says the inductor gap should be xxx mm, and we continue to shim it until we get the inductance we want"
<pie_> *er (?)
<sorear> FEM is someone else's problem, presumably
<awygle> Not at my old job
<pie_> rqou, hey look numbers!
<qu1j0t3> The shims will continue until inductance improves.
<awygle> lmao
<awygle> qu1j0t3 wins
<awygle> I don't know why but that got me good
<pie_> sounds like a reference to something but its going over my head
<rqou> this also has fun effects like the inductance changing as a function of current because it causes the shims to get more compressed
<awygle> pie_: the beatings will continue until morale improves
<sorear> _ancient_ joke genre
<pie_> rqou, real life is fun
<pie_> ahhhh
<sorear> most common form is (what awygle said)
<awygle> rqou: that's why you epoxy the shit out of it
<qu1j0t3> sorear: I am an ancient joke myself.
<awygle> I have successfully called the death of several transformers based on that they'd broken their epoxy and were audibly fatiguing themselves to death
<pie_> awygle, huh that actually works?
<awygle> pie_: eh, sometimes
<pie_> qu1j0t3, aww dont be so hard on yourself :(
<qu1j0t3> oh, what would explain the stripe of welding on this german made transformer i looked at recently?
<qu1j0t3> that would*
<awygle> azonenberg: do you have preferred Ethernet magnetics?
<pie_> Ethernet Magnetic almost sounds like a good album
<azonenberg> awygle: standalone or in a jack?
<awygle> "Hi, I'm D. K. and these are the Ethernet Magnetics! We're gonna start things off with our new single, Dielectric Properties! 1 2 3 4!"
<pie_> lol
<pie_> gg
<awygle> azonenberg: that is part of the question, really, do you prefer one or the other?
<azonenberg> bel fuse 0826-1g1t-23-f
<azonenberg> is the jack i've used in the past
<awygle> Mk, cool
<rqou> wtf, my calculations are totally f*cked
<rqou> welcome to "time to salvage the lab report"
<qu1j0t3> awygle: I was thinking more... ambient...
<awygle> qu1j0t3: I was basically picturing Scott Pilgrim-style "punk bands"
<awygle> (note the extreme scare quotes)
<pie_> awygle, another thing that seemed cool about phasors was that it was immediately apparent that you could use them with a fourier expansion
<pie_> though i suppose if one has any amount of practice with such that would also be apparent with the sin | cos form
<pie_> the way we brought it up was "quasistationary currents" (translation from hungarian) but that doesnt seem to yield anything nonobscure on google 0_o
<pie_> i dont remember what quasistationary is supposed to mean
<pie_> oh i guess its quasistatic
<pie_> it probably felt more deep than it was but that 10 minutes or so of class felt very insightful to me
<rqou> "The actual ESR was 195 mOhm yielding a copper loss of 0.78 W which significantly exceeds the design specification."
<rqou> welcome to "i suck at lab"
<pie_> i might just be too humble to i usually end up put stuff i think i may have done wrong if i get absurd results
<pie_> *humble but
<pie_> man i butchered that sentence somehow
<pie_> usually chalked up to hypothetical measurment procedure errors but eh :P
<pie_> (unrelated, ill just read this as IC=integrated circuit, TYVM https://twitter.com/catehstn/status/974977571728314368 )
<pie_> rqou, sidenote, you do uncertainty analysis right? :P
<pie_> measurements aint worth shit without that
<rqou> not really
<pie_> well unless "everything will be within tolerances" or something i guessss
<awygle> Nobody really cares if you suck at lab, just *how* you suck at it
<rqou> btw azonenberg how do you source inductors for smps designs?
<azonenberg> rqou: i use the datasheet-recommended inductor if there is one that i can source
<rqou> lol
<rqou> no custom magnetics for you? :P
<azonenberg> Failing that, i look for one with the inductance and current ratings i need, a sane package, a self-resonant frequency well above the switching frequence
<azonenberg> frequency*
<azonenberg> and then sort by price
<rqou> lol
<rqou> obviously you don't do "special" psus
<azonenberg> Yeah no
<azonenberg> power isnt my specialty or anything
<azonenberg> it's a necessary evil to accomplish my goal
<qu1j0t3> after all, power corrupts
<azonenberg> Imaginary power...
* azonenberg hides
* pie_ throws `i`s at azonenberg
<pie_> or `j`
<pie_> fucking EEs
<pie_> :P
<awygle> azonenberg: but do you sort by price or by package size first?
<azonenberg> awygle: generally i select a range of packages then sort by price
<azonenberg> i give bonus points to a series i've used in the past and have a board-proven footprint for
<azonenberg> like the yuden nr6028 series
<azonenberg> nrs6028*
<awygle> I try to use Murata for power and Coilcraft for RF
<azonenberg> i've had good results with those too
<awygle> But I almost never supply more than like an amp
<azonenberg> i have an fpga board in front of me pulling 4.3W and i dont know how mjuch of that is on A3V3 vs 1V8 or 1V0
<azonenberg> then the vcu118? i dont wanna know how much THAT thing pulls
<awygle> I would like to do some mains powered stuff but for several reasons I have no desire to do the actual psu design
<azonenberg> This is what COTS PSUs are for
<pie_> landauer limit pls
unixb0y has quit [Ping timeout: 264 seconds]
unixb0y has joined ##openfpga
digshadow has quit [Ping timeout: 260 seconds]
genii has quit [Quit: GO LEAFS GO !!]
Bike has quit [Quit: Lost terminal]
soylentyellow has joined ##openfpga
rohitksingh_work has joined ##openfpga
Marex has quit [Ping timeout: 240 seconds]
Marex has joined ##openfpga
digshadow has joined ##openfpga
* smkz PFC's azonenberg's power :|
* rqou sticks a big fat motor on smkz's grid
<rqou> hey, at least motors are easy mode :P
<azonenberg> awygle: so my 10G TCP/IP stack is progressing nicely
<azonenberg> I can reply to incoming IPv4 ARPs, and accept inbound IPv4 UDP traffic
<rqou> azonenberg what is your current priority?
<azonenberg> currently, the first byte of any udp packet sent to port 9999 is displayed on my LEDs
<azonenberg> rqou: re the stack? or what
<rqou> er, anything?
<azonenberg> rqou: lol, loong list
<azonenberg> research this week @ $dayjob
<rqou> i thought you originally (like half a year ago) wanted to do more automated silicon RE stuff
<azonenberg> construction on the house
<azonenberg> assembling boards for $sidegig1
<azonenberg> writing the ip stack so i can have a high-speed debug connection to the fpga firmware for $sidegig2
<azonenberg> automated silicon RE is a distant dream right now
<azonenberg> My focus is 100% on construction and things that pay for construction
<rqou> wtf two side gigs?
<azonenberg> #1 was supposed to finish before #2 started
<rqou> are these the same 2 from the last time we spoke about this?
<azonenberg> Yes
<azonenberg> #1 is not a major time commitment, they just did a board respin and i'm assembling the updated prototypes
<rqou> why isn't #1 done yet? i thought that was supposed to have finished months ago at this point?
<azonenberg> but i need more components
<azonenberg> They found bugs in rev 1, respun
<azonenberg> brought up rev 2, found another bug
<azonenberg> now i'm assembling several rev 2 boards with three different possible solutions for the bug that they want to characterize
<azonenberg> it's not a major time commitment, there's been a lot of lead time waiting for pcbs and engineering on their end
<rqou> how do they not have the in-house skills for this?
<rqou> it sounds like somebody sucked at managing this project
<azonenberg> me assembling the boards is just a time saver for them, vs having somebody else assemble the board then go to me for bringup
<azonenberg> it saves a shipping delay
<azonenberg> And i'm better at rework than their in house team apparently
<rqou> wtf
<azonenberg> you remember that crazy diagonal qfn thing i did?
<rqou> yeah
<azonenberg> I sent it to them and all their engineers were @_@ at it
<azonenberg> saying they couldn't have pulled it off
<azonenberg> Anyway, i'm now building the ip stack on my own time so that i can drop the permissively-licensed rtl into gig #2's fpga testbench without them owning the ip
<rqou> ah
<azonenberg> So it's not paid work but it's kinda-sorta for the gig
<azonenberg> as in, it jumped up several priorities b/c i had a paying customer who wanted to use it
<azonenberg> but the endgame use for it is still my own projects
<azonenberg> on a different note the VPS i host my dns on is down
<azonenberg> and has been dow nall day
<azonenberg> apparently the host node is having problems
<azonenberg> At my new place i am totally going to just host all of my services on an in-house DMZ
<azonenberg> dealing with these VPS providers is just silly
<awygle> azonenberg: cool yo
<awygle> I have kept up on your #fpga efforts
<awygle> *reports although the other works
<azonenberg> awygle: now that i have this stack coming together, i'll be much more inclined to do work on starshipraider and the switch once i have the new lab up and running
<azonenberg> one of the delays was me not wanting to / having time to buckle down and do it
<awygle> Lol sure
<awygle> Out of curiosity, are you doing any formal work on it?
<azonenberg> As of now, no - i'm focusing on pounding it out as quickly as possible with minimal verification
<azonenberg> but almost all of the state in each module is brought out to top level ports
<azonenberg> and i plan to do formal at some point
<azonenberg> i.e. when i have time, and before using it in production
<azonenberg> rather than a debug environment
<awygle> Sure
<awygle> Not sure how to ask this exactly but how "Xilinx-y" is it?
<awygle> IIUC the Xilinx switchbox doesn't too *too* much heavy lifting?
mumptai has joined ##openfpga
<azonenberg> you mean the serdes?
<awygle> Yeah idk how I came up with switchbox. Gearbox I think is what I meant
<azonenberg> it has an 8b/10b coder built in, but the 64/66 is just a gearbox with bitslip and you have to do sync/descrambling yourself
<awygle> I'm halfway to sleep lol
<azonenberg> (getting that working was a pain lol)
<azonenberg> So it should work with other stuff easily enough
<awygle> Cool
<azonenberg> My MAC works with XGMII (although i renumber the lanes so MSB is 31 and not 0)
<azonenberg> then the PCS currently talks to the xilinx GTY wizard but should be easily adaptable to other stuff
<azonenberg> i'm going to implement a XAUI PCS for artix-7 as well
<awygle> Yeah I remember
<azonenberg> then everything past the MAC is generic and just wants 32-bit data at $CLOCK_RATE
<azonenberg> it should work for 1gbps too easily enough
<awygle> Sounds great. I'll keep my ears peeled (is that a saying?). I'm going to sleep now though, goodnight all
<azonenberg> I may have to do some timing optimization to get it to work at 10 Gbps on artix-7
<azonenberg> with a 32-bit data bus
<azonenberg> but TCP/IP is kinda meant to run at a 32-bit bus width
<azonenberg> doing 64 gets awkward
<azonenberg> If i have to rewrite it that way i will, thoguh
<azonenberg> (right now it runs at 312.5 MHz)
thallia- has quit [Ping timeout: 264 seconds]
thallia has joined ##openfpga
pie_ has quit [Ping timeout: 240 seconds]
Xark has quit [Ping timeout: 264 seconds]
pie_ has joined ##openfpga
<rqou> > buys alginate
<rqou> > uses it to demonstrate encapsulating an enzyme rather than to make popping boba
<rqou> obviously nilered isn't the right type of ABC :P :P
<azonenberg> rqou: lol
* azonenberg begins work on transmit path of ICMP stack
<azonenberg> i should probably sleep at some point
Xark has joined ##openfpga
user10032 has joined ##openfpga
pie_ has quit [Ping timeout: 265 seconds]
dingbat has quit [Ping timeout: 256 seconds]
dingbat has joined ##openfpga
user10032 has quit [Remote host closed the connection]
wpwrak_ has joined ##openfpga
wpwrak has quit [Ping timeout: 248 seconds]
wpwrak_ has quit [Read error: Connection reset by peer]
wpwrak_ has joined ##openfpga
wpwrak_ has quit [Ping timeout: 264 seconds]
wpwrak_ has joined ##openfpga
wpwrak_ has quit [Ping timeout: 264 seconds]
wpwrak_ has joined ##openfpga
wpwrak__ has joined ##openfpga
<azonenberg> rqou: o_O
<azonenberg> guess what
<azonenberg> You can buy some broadcom switch chipsets (ex: BCM5387) on digikey in qty 1
<azonenberg> good luck getting a datasheet, but it's nice to see their sales getting a little more open
wpwrak_ has quit [Ping timeout: 240 seconds]
pie_ has joined ##openfpga
pie_ has quit [Ping timeout: 264 seconds]
<felix_> rqou: it seems that i'll be employing a student to write ida support for the xtensa used in the ath10k chips soon. i'll also try to get some funding for the work on the ath10k project together with 2 other people. have to finish some other projects before i have time to work on that though
pie_ has joined ##openfpga
<jn__> felix_: have a look at https://github.com/pfalcon/ida-xtensa2
<jn__> it might have some issues because it is mainly tested under ScratchABit, as far as i understand, but it seems like something to build on
<felix_> i'm aware of that, but we can't use that, since it's gpl
<jn__> oh :/
<felix_> probably writing ida processor support that's licensed under gpl can't be shipped without doing a license violation (at least if it was c code; not sure how that works for a python module)
eduardo__ has joined ##openfpga
gnufan has quit [Ping timeout: 240 seconds]
eduardo_ has quit [Ping timeout: 265 seconds]
gnufan has joined ##openfpga
SpaceCoaster has quit [Ping timeout: 264 seconds]
SpaceCoaster has joined ##openfpga
<whitequark> azonenberg: I found some BCM switch docs for you
<whitequark> openwrt has support for BCM5380
gnufan has quit [Ping timeout: 260 seconds]
gnufan has joined ##openfpga
rohitksingh_work has quit [Read error: Connection reset by peer]
pie_ has quit [Ping timeout: 252 seconds]
genii has joined ##openfpga
wpwrak__ is now known as wpwrak
rohitksingh has joined ##openfpga
rohitksingh has quit [Read error: Connection reset by peer]
rohitksingh has joined ##openfpga
<q3k> whitequark: nice vintage
<whitequark> q3k: oh?
<q3k> yeah, it's 2xGE + 4xFE
<q3k> sorry, 8xFE
m_w_ has joined ##openfpga
m_w_ has quit [Client Quit]
m_w_ has joined ##openfpga
rohitksingh has quit [Quit: Leaving.]
<q3k> that reminds me of when I actually had some access to broadcom datasheets at some point in my previous life
<q3k> and I noticed that one of the standards they were offering to program the flows was called 'Open$something'
<q3k> and in my young naivete I happily proclaimed something like 'oh, so it's open source and public!' next to a rep or architect or whatever
<q3k> they looked at me like a lunatic, with one of the guys visibly gathering all of his business professionalism to not laugh straight at my face
<q3k> have I mentioned I'm not a huge fan of broadcom?
<jn__> and what kind of "open" was it?
<q3k> (no, it wasn't openflow, something else they came up with)
<rqou> meh, roboswitch
<jn__> "based on a multi-vendor standard"? :)
<q3k> i think it was basically copying the openflow standard
<rqou> that's the dumb switch
<q3k> so they kinda copied the name
<rqou> let me know when you can get a strata switch :P
<q3k> iirc the explanation was that yes, it's something that you can buy middleware for from $partner_vendors
<jn__> ugh, that's slightly worse than i could make up
<whitequark> broadcom is garbage
<whitequark> i have no idea why azonenberg wants their switches
<whitequark> i have no idea why anyone wants their anything
stefanct has quit [Ping timeout: 256 seconds]
pie_ has joined ##openfpga
<rqou> hey, their chips for the most part actually work
<rqou> you don't really have a choice for fancy enterprise/datacenter switches
<rqou> azonenberg: at some point i want to do a massive pwning of all cryptocurrency hardware wallets
<rqou> because it somehow seems like none of them really know what they're doing
<q3k> whitequark: there's not much alternatives
<whitequark> q3k: really?
<whitequark> i'm pretty sure qca makes switches
<q3k> haven't seen them in any hardware tbh
<q3k> all of the 10GbE+ dc switches are broadcom or custom silicon (if you're cisco or juniper)
<rqou> qca switches also don't have open datasheets
<rqou> they just leak more
<q3k> that too
<whitequark> oh right, bcm5387 is 5GE
<rqou> q3k: and some of those custom silicon are made by broadcom :P
<q3k> the actual hw from broadcom is _okay_ from what I've heard, they just rip you off on the datasheets (because really, that's what they sell you)
<q3k> rqou: not surprising
<rqou> also, broadcom datasheets are f*cking garbage
<rqou> you want the architecture reference manuals instead
<rqou> also, if you want register names and addresses for the fancy switch chips, you just need to (*hint* *hint*) find a white-box switch that happens to give you access to the "SDK" cli and hope the vendor didn't disable the relevant command
<rqou> (this may be much harder for newer parts. the "LB4M" miiiiight be a good start)
<rqou> I'm not going to say any more than this
<whitequark> so basically you're saying that i can tie you to a radiator in my basement and then you'll spill it, right
<q3k> rqou: that reminds me I need to unbrick my LB4M
<q3k> rqou: any chance you could dump the flash off of yours? :D
<awygle> growing up in California plus too many movies as a youth means I think of a radiator as a prison first and a heater a distant second
<rqou> q3k: i never found a command that dumps the entire flash, sorry
<rqou> btw in case it's not clear, i worked at broadcom
<q3k> rqou: yeah, but there is a command that seems to overwrite the entire flash, and I accidentally used it :P
<q3k> I'm probably just end up getting another LB4M from the internet and just dump the flash off of that one by, uh, more invasive means
<whitequark> rqou: i think it's called a "SOIC-8 clip"
<q3k> it's not SOIC-8
<q3k> it's a TSSOP NAND or whateber
<q3k> *whatever
<q3k> but it's doable without too much trouble https://gallery.hackerspace.pl/pictures/IMG_0100.JPG
<whitequark> hm i don't think they make clips for these
<q3k> whitequark: these don't really work
<q3k> i, uh, might have had used one in a stock PS3 flasher
<whitequark> oh?
<q3k> and they just pop off
<whitequark> pop off?
<q3k> they're sanded off sockets that are supposed to take the whole chip
<q3k> and they don't really hold on to the chip when it's soldered down
* whitequark stares at q3k
<whitequark> just put something heavy on top of it
<q3k> well if I'm gonna dick around with it, I might just as well desolder it ¯\_(ツ)_/¯
<q3k> less frustration
<whitequark> depends
<whitequark> i wouldn't desolder stuff that's near one of those plastic flex cable connectors
<whitequark> or something similar
<whitequark> well i guess one can always use chipquik in that case
<q3k> eh, I'm quite handy with a hot air gun
<q3k> depends on your skill level and patience and steady hand
<whitequark> I don't have that many problems with a hot air gun but I just try not to touch that shitty plastic
<q3k> just perform a hecatomb of flux on the thing and you're good :P
<awygle> I would have magwired it in place probably
<whitequark> ... that's clever
<q3k> awygle: also works
<whitequark> i have too much hand tremor to solder magwire :/
<awygle> what's a hecatomb?
<whitequark> and even worse now after tricyclics
<q3k> awygle: silly reference https://en.wikipedia.org/wiki/Hecatomb
<awygle> whitequark: yeah I struggle with that too. I usually drag solder anything small, that seems to work ok
<q3k> whitequark: I'm sorry
<awygle> ah
<whitequark> i once soldered four consecutive pins on a 0.3mm pitch TQFP as a teen
<whitequark> but i don't think i can reproduce that anymore
<whitequark> that took a long time
<awygle> I used to pretty regularly have to solder to the "pads" of a QFN, which may contribute to my loathing for that package type
<whitequark> awygle: mad respect
<q3k> I just had to do that two weeks ago, too https://photos.app.goo.gl/Nljx9rDQ6pe8Yzb43
<q3k> (and yes, that is blu-tac)
<q3k> blu-tac is my new favourit engineering material for when you need to get a piece of electronics in a testbench-like setup
<q3k> it doesn't conduct electricity, sticks to things and is easy to remove
<awygle> What is blu-tac in this context? The stuff you use to hang posters in dorm rooms?
<q3k> yeah
<pie_> rqou, oh.
<pie_> also hi dudes ^.^
Dolu has joined ##openfpga
<awygle> Good morning pie_ (pretty sure we're not all dudes)
mumptai has quit [Read error: Connection reset by peer]
mumptai has joined ##openfpga
digshadow has quit [Ping timeout: 264 seconds]
<awygle> what if instead of a) installing everything in a global namespace or b) duplicating every dependency of every program we did de-duplication at the file system layer?
<qu1j0t3> heh
<qu1j0t3> or, ya know, hard links
<qu1j0t3> or symlinks for that matter
<qu1j0t3> there are systems that work like that
<awygle> that's a reasonable form of deduplication, but i'd want it in a convenient way
<qu1j0t3> also: Zones
<sorear> awygle: so after spending years pushing ISVs to bundle dependencies Apple has recently fast-tracked deployment of a CoW filesystem and I can't help but think these are related
<awygle> sorear: cool
<qu1j0t3> does APFS do any dedup though? of course ZFS does
<awygle> qu1j0t3: i'm only familiar with zones as "jails but for solaris" and i'm only familiar with jails as "VMs but cheaper", can you elaborate?
<qu1j0t3> awygle: Zones involve sharing parts of filesystems
<qu1j0t3> awygle: in one flavour
pie_ has quit [Ping timeout: 268 seconds]
<awygle> qu1j0t3: oh yeah, sure, i see the application now
<awygle> the above complaint brought to you by Windows: "I can't load that assembly, there's already one with the same name and version loaded!" ".... Do you think it might be the same one?"
digshadow has joined ##openfpga
soylentyellow has quit [Ping timeout: 264 seconds]
user10032 has joined ##openfpga
stefanct has joined ##openfpga
<whitequark> lol
wpwrak has quit [Remote host closed the connection]
wpwrak has joined ##openfpga
RaivisR has joined ##openfpga
RaivisR_ has quit [Read error: Connection reset by peer]
wpwrak_ has joined ##openfpga
wpwrak has quit [Ping timeout: 268 seconds]
wpwrak__ has joined ##openfpga
wpwrak_ has quit [Ping timeout: 240 seconds]
wpwrak_ has joined ##openfpga
wpwrak__ has quit [Ping timeout: 264 seconds]
wpwrak__ has joined ##openfpga
wpwrak has joined ##openfpga
wpwrak_ has quit [Ping timeout: 240 seconds]
Dolu has quit [Quit: Leaving]
wpwrak__ has quit [Ping timeout: 260 seconds]
wpwrak has quit [Ping timeout: 240 seconds]
wpwrak has joined ##openfpga
mumptai has quit [Quit: Verlassend]
<rqou> azonenberg: I don't remember, have you ever tried programming the unused memory addresses in the center of the flash array on xc2c32a?
<rqou> do you know if they have flash cells?
<rqou> this is difficult to test because there aren't jed addresses for these cells
<azonenberg> i dont think i tried
<azonenberg> i do see flash cells
<azonenberg> i see no reason they wouldnt be writable
<azonenberg> i conjecture they're usable for stego
<azonenberg> but i havent tested
<rqou> think you'll have any spare time to test?
<azonenberg> spare time? what's that
<azonenberg> the only reason things arent even MORE nuts is that i'm booked on research @ work this week
pie_ has joined ##openfpga
<pie_> awygle, obligatory nitpick eh :P
<rqou> anybody know if the nrf51 code protection bypass works on the nrf52?
Bike has joined ##openfpga
user10032 has quit [Quit: Leaving]
eric_ has quit [Read error: Connection reset by peer]
eric_ has joined ##openfpga