egg|nomz|egg changed the topic of #kspacademia to: https://gist.github.com/pdn4kd/164b9b85435d87afbec0c3a7e69d3e6d | Dogs are cats. Spiders are cat interferometers. | Космизм сегодня! | Document well, for tomorrow you may get mauled by a ネコバス. | <UmbralRaptor> egg|nomz|egg: generally if your eyes are dewing over, that's not the weather. | <ferram4> I shall beat my problems to death with an engineer.
<bofh>
Ellied: yeah, it's only after UmbralRaptor's comment that I realized your classification: A) also somewhat makes sense on quarks, B) makes more sense if you only consider fundamental particles.
<bofh>
I'm a condensed matter theorist. "Fermions" encompasses a fairly huge menagerie.
<Ellied>
ohhh, gotcha, I thought it only referred to fundamentals
<Ellied>
we had a trivia night at the SPS zone meeting where the given answer to the question of how many fundamental particles there are was, somehow, 61
<bofh>
Yep, most of them with absurdly short decay lifetimes.
<tawny>
just eight more to go...
<bofh>
Like muonium is nice b/c the muon will exist for a full 2.2µs before decaying, which is basically forever (it's long enough to do chemistry w/it, in any case).
<UmbralRaptor>
tawny: the Nice model is more of astronomy than particle physics.
* UmbralRaptor
is only coming up with 26 particles. 6 quarks, 6 leptons, 8 gluons, 3 weak force, 1 photon, 1 graviton, 1 higgs. Should I be counting antimatter also?
<Ellied>
oh, so I fixed my toughbook's main USB port today (it had been working for data but not for power, turns out there was a bent pin in the internal ribbon connector)
<Ellied>
and it looks suspiciously like there are protection diodes on the data lines of each port
<bofh>
Good, there bloody should be.
tawny has quit [Ping timeout: 207 seconds]
<Ellied>
there's a little SOT-23-3 with one leg on each data line and one on the ground fillzone
tawny has joined #kspacademia
<Ellied>
Scanlime was telling me a few months back that most USB ports don't because the diodes' junction capacitance would fuck up the high-speed signalling?
<Ellied>
these are 2.0 ports that have demonstrated being capable of the full 480 Mbit/sec transfer rate
<bofh>
I guess this isn't USB 3.0
<Ellied>
yeah, not 3.0
<Ellied>
but I thnk she was saying this about 2.0
<Ellied>
it was in the context of the USB killer, which IIRC targets the 2.0 datalines (or did originally at the very least)
<soundnfury>
certainly work _can_ be extracted, since "x·ẋ/(|x||ẋ|) ~= 1" is one heck of a regularity
<soundnfury>
btw I finally realised the other day that "negentropy" is just another name for "mutual information"
<soundnfury>
!wpn -add:adj Shannon
<Qboid>
soundnfury: Adjective added!
<soundnfury>
!wpn UmbralRaptor
* Qboid
gives UmbralRaptor a nonsteroidal lance
<UmbralRaptor>
WTB 1 entrometer
<UmbralRaptor>
!wpn soundnfury
* Qboid
gives soundnfury a cathode ray tube
tawny has joined #kspacademia
<UmbralRaptor>
(… chaometer?)
<soundnfury>
UmbralRaptor: there can be no such thing, entropy is not a property of the actual state of a thing, it's a measure of the observer's degree of partial information
<soundnfury>
(if you haven't read E.T.Jaynes' paper "Gibbs' Paradox", you should do so)
<UmbralRaptor>
I'm not sure how an observer's knowledge scales with temperature, but okay.
<soundnfury>
the Second Law isn't really a law of nature (that would be Liouville's Theorem, under which phase space volume is _conserved_), it's just an artifact of "you are rounding off the boundaries of volumes of phase space when they get too wiggly to track"
<soundnfury>
UmbralRaptor: see also the Szilard Engine
<soundnfury>
if you have enough knowledge about the particles in a box, you actually _can_ run Maxwell's demon and extract just as much work as the mutual-information says you should be able to
<soundnfury>
(and in doing so, you destroy that mutual information, thus maintaining total entropy)
<soundnfury>
it's also why an irreversible operation on a CPU _must_ emit at least kT ln 2 of heat per bit erased
tawny has quit [Ping timeout: 383 seconds]
tawny has joined #kspacademia
tawny has quit [Ping timeout: 207 seconds]
tawny has joined #kspacademia
tawny- has joined #kspacademia
icefire has joined #kspacademia
tawny has quit [Ping timeout: 383 seconds]
StCypher has quit [Read error: Connection reset by peer]
StCypher has joined #kspacademia
ferram4_ has quit [Read error: Connection reset by peer]
ferram4_ has joined #kspacademia
icefire has quit [Read error: Connection reset by peer]
StCypher has quit [Ping timeout: 207 seconds]
egg|phone|egg has joined #kspacademia
egg|cell|egg has quit [Ping timeout: 186 seconds]
egg|cell|egg has joined #kspacademia
egg|phone|egg has quit [Read error: Connection reset by peer]
ferram4_ has quit [Read error: Connection reset by peer]
ferram4 has joined #kspacademia
egg|phone|egg has joined #kspacademia
egg|mobile|egg has joined #kspacademia
egg|phone|egg has quit [Read error: Connection reset by peer]
egg|cell|egg has quit [Ping timeout: 186 seconds]
tawny- has quit [Ping timeout: 383 seconds]
awang has quit [Ping timeout: 182 seconds]
awang has joined #kspacademia
egg|work|egg_ has joined #kspacademia
egg|work|egg has quit [Ping timeout: 180 seconds]
egg|work|egg_ is now known as egg|work|egg
egg|phone|egg has joined #kspacademia
egg|mobile|egg has quit [Read error: Connection reset by peer]
egg|cell|egg has joined #kspacademia
egg|phone|egg has quit [Read error: Connection reset by peer]
egg|phone|egg has joined #kspacademia
egg|cell|egg has quit [Read error: Connection reset by peer]
tawny- has joined #kspacademia
<Ellied>
bofh: on USB 3.0 the 'new' superspeed data lines are on completely new pin positions, and the original data line pin positions are still occupied by a 2.0 bus
egg|cell|egg has joined #kspacademia
egg|phone|egg has quit [Read error: Connection reset by peer]
egg|cell|egg has quit [Read error: Connection reset by peer]
<Ellied>
yeah, I have plenty of gripes with the USB connector design, but I do think it's pretty cool how they managed to add pins to it without breaking backwards-compatibility.
<Ellied>
even if the type-A 3.0 connectors are pretty much just a holdover while we move towards type-C on everything
APlayer has joined #kspacademia
<Ellied>
my guess is the electronics associated with talking at 5 gigabits per second don't work all the way down at 480 Mbit/sec, so they couldn't just make the super-speed mode an additional capability of the same data lines
<Ellied>
and while they were at it, they made it full-duplex, which is rad
<bofh>
Ehh, I'm not sure full-dupl3x is that useful with USB. It complicates the design a shitton, too.
<kmath>
<eggleroy> @chordowl Argh. It's no longer a Cayley table at this point, but if we say two partitions P and Q, such that for al… https://t.co/BfvcTnm3AI
<APlayer>
Heh
<APlayer>
Anyway, I need to go... See you tomorrow!
APlayer has quit [Read error: Connection reset by peer]
tawny- has joined #kspacademia
<Ellied>
bofh: it seems to me like full-duplex simplifies the timing a great deal, because the limitations of who can talk when are a lot less strict
<Ellied>
line drivers no longer necessarily need the ability to go high-Z while the other end talks (unless it's a 3.1 interface with half-duplex double-speed support)
<Ellied>
it also allows communications to be initiated from the peripheral without being constantly polled by the host IIRC
<Ellied>
and theoretically host-to-host links using crossover cables are a thing but that still appears to have shitty support everywhere
<Ellied>
I'm still perplexed by why host-host links weren't possible with 2.0; I understand that the protocol was designed to move as much control to the host as possible and keep the peripheral cheap, but it still seems like reprogramming a host controller to act as a peripheral would have been trivial
<Ellied>
it seems to me like adding the ability for a computer to set one of its USB ports to act as a virtual terminal or network interface shouldn't have been much additional cost, and it would've been quite nice because only one computer would need to support it for two computers to be able to talk to each other
<Ellied>
some devices can even do just that! but they're limited to mobile devices for some reason, and as far as I can tell no hubs support it.
<Ellied>
it seems like USB is designed with exactly one functionality in mind (connection to simple dumb peripherals) which is a bit odd for something called a "universal" serial bus IMO
<Ellied>
I guess most peripherals are simple and dumb though, that's true
<Ellied>
by my understanding of serial hardware, I almost feel like they should've just slapped some differential line drivers on SPI and used that. Would've been a few more conductors than USB does, but the peripheral-side hardware could have been extremely simple, just a single PISO shift register.
<Ellied>
HDMI shows that very fast synchronous serial interfaces work just fine
<bofh>
I mean the issue I find is that most SPI controllers don't support high clockspeeds.
<bofh>
(That and nobody outside of TI makes DMA-capable SPI controllers, which pisses me off somewhat).
<Ellied>
well I mean that seems like a comparatively small problem to developing a whole new serial interface from the ground up, no?
<bofh>
Does USB 2.0 predate or postdate SPI?
<Ellied>
oh. I assumed it postdated it but I guess I don't know that
<bofh>
(since I barely consider USB 1.1 a usable standard, USB for me starts at EHCI)
<Ellied>
mid 1980s for SPI. when was USB?
<Ellied>
1996, apparently
<Ellied>
2.0 was 2000
<Ellied>
SPI is so simple I figured it had to be pretty old
<Ellied>
like, if you gain a basic understanding of digital electronics and then sit there and think "how do you transfer data serially" you basically end up with SPI
<bofh>
Like, agreed.
<bofh>
it's honestly basically what happens when you slap a pair of FIFOs onto a 3-wire serial interface.
<Ellied>
data and clock is pretty obvious, and then just assigning a random GPIO to delimit transmissions is a pretty obvious easy solution
tawny- is now known as tawny
<Ellied>
AFAIK the SPI spec has no maximum clock speed; it's usually just a limit imposed by the controller
<Ellied>
there's a paper from TI about doing differential SPI even, but I'm not sure how often that's done or how many controllers that have differential inputs and outputs there are.
<Ellied>
ohh huh USB 3.0 is bit-balanced, like HDMI. that's interesting
<Ellied>
...so are SATA, FireWire, PCIe, many types of gigabit ethernet, DisplayPort, and Fibre Channel. guess that's not so uncommon lol
<Ellied>
god, why is it so hard to buy general-purpose small-signal diodes in surface-mount packages?? there are various versions of the 1N4148, but those are all listed as obsolete and usually on back order at Mouser and DigiKey
<Ellied>
I literally just need a thing that fits on an 0805 pad, conducts at least 10 mA one way, blocks at least 5 volts the other way, and recovers in under 25 ns.
<Ellied>
all I can find are big honking power Schottky diodes that take an amp or more and take 500 ns to recover
<Ellied>
at the rate I'm going, I might as well just use the body diode of a MOSFET with the gate tied to source
* UmbralRaptor
would not have expected size trends to go that way.
* UmbralRaptor
🔪 catalogs.
<UmbralRaptor>
So, I'm trying to see if one with derived parameters for M-dwarfs is relevant to my needs, and >10% of the stars have negative radii.
<UmbralRaptor>
Also, more than half are outside the parameter space that the authors define as good.
<bofh>
Ellied: wait, why do you need SMD diodes?
<Ellied>
LED blinker uses a charge pump to make sure that even very short pulses turn into visibly-long LED blinks
<Ellied>
but as we speak I'm refactoring the design to use only MOSFETs
<whitequark>
Ellied: "USB, SATA, FireWire, PCIe, GigE, DP, FC" it's an ongoing joke i tell that all interfaces slowly converge to "NRZI over differential pair"
<Ellied>
bofh: I think an open-drain-output logic gate would be better suited to my purposes here, but that would just be yet another part to find
<Ellied>
the most straightforward way of converting a push-pull output into an open-drain-like output was to just bolt on a diode
<bofh>
Ellied: so like, if it's just an LED blinker, why not directly use the push-pull output?
<Ellied>
I guess my silly physicist idea that "fewer junctions means simpler device" is as naïve as it sounds lol
<Ellied>
bofh: because if I use the output to drive the LED directly, then a 25-ns pulse won't produce a visible flash
<bofh>
Okay, but in that case why not increase the pulse width?
<Ellied>
that's what this is for
<Ellied>
the device is supposed to accept any pulse width that the counter can actually see, and the LED's brightness should not vary depending on the characteristics of the input signal
<Ellied>
I just want on=working, off=not working
StCypher has joined #kspacademia
<Ellied>
or, if the pulse rate is very slow, blink=got one
<bofh>
Wouldn't it be simpler to do that by altering the clocking? Oh, I see, you want it independent of the input signal characteristics. Gotcha.
<Ellied>
yes, this is not a nice well-behaved clock, this is probably the amplifier output from a PMT/APD/SiPM
<whitequark>
how does an open-drain output help you?
<whitequark>
also, your failure to find a diode is disturbing
<kmath>
<sarahemm> Our cats are learning to send faxes now. They seem to like the sound the machine makes when they press buttons, and… https://t.co/qsutD21fFP
<kmath>
<dribnet> bias of the day: stare downwards and ImageNet trained models may think a potter's wheel is nearby. (shown: class tr… https://t.co/mJV9XH35ER
<whitequark>
Diodes - Rectifiers - Single, then SOD-123 case
<Ellied>
hrm, perhaps my problem is that I don't know enough surface-mount packages
<Ellied>
I was looking for ones in the 0805 or MicroMELF packages, because those are the ones that my Adafruit ruler has pads for
<Ellied>
whitequark: also thanks
<whitequark>
diodes don't come in 0805 packages
<whitequark>
they come in SOD/SOT packages or SM[ABCD] ones
<Ellied>
well I mean, I did get *some* results for them in that package and kicad has a pad in its library for 0805 diode
<Ellied>
but yeah that makes sense now
<Ellied>
also I had no trouble finding LEDs in 1206
<whitequark>
because LEDs *do* come in passive packages
<Ellied>
wait why
<whitequark>
LEDs aren't really like small signal or power diodes package-wise
<whitequark>
why? because the teams that ship signal diodes and light-emitting diodes are completely unrelated
<Ellied>
gotcha
<Ellied>
I think I'm gonna keep the new design that dispenses with signal diodes completely in any case. The cost savings for using a diode over a MOSFET are actually negligible as far as I can tell; the DMG2302/2301 is only a couple cents more than those diodes you linked in quantity
<whitequark>
body diode characteristics are quite subpar typically
<whitequark>
unless your MOSFET has been specifically designed such that the body diode is usable (e.g. as a freewheeling diode) it's not recommended to use it