Show more

is awesome.

(Please be sure to CW spoilers, and be aware I won't read them for at least a longish while.)

robryk boosted

@robryk@qoto.org Yes. You can measure temperature by observing a crystal oscillator's frequency drift (with a known reference clock). Hewlett-Packard made this lab instrument in the 1960s, with a resolution of .0001”C. AFAIK it's still a really precise method by today's standard, but it went out of favor as the crystal needs to have a special cut with linear temperature coefficient. See HP Journal Vol. 16, No.7, 1965. http://hparchive.com/Journals/HPJ-1965-03.pdf Linear Technology App Note 61 shows how to build your own, see page 13: https://www.analog.com/media/en/technical-documentation/application-notes/an61fa.pdf

Fun fact: This phenomenon has been used as a deanonymization attack against Tor. By observing the 24-hour drift of the system clock, one can determine the longitude of the target server, latitude can further be determined by the change in day length.
https://www.freehaven.net/anonbib/cache/HotOrNot.pdf

robryk boosted

#TIL Möbius resistor. Take a double-sided copper tape (with its top insulated from the bottom), and join the tape onto itself as a Möbius strip. Now you get an ideal current-sensing resistor with almost no parasitic inductance. The resistor behaves like two short-circuited microstrip transmission lines with continuous ground planes, with an input port always at the center. An extreme form of the "folding back the wire" technique for making non-inductive wire-wound resistors... Cool idea, but practically it probably isn't too useful... https://en.wikipedia.org/wiki/M%C3%B6bius_resistor #electronics

robryk boosted

@freemo Our math rendering continues to do very weird things (the quoted post gets some expressions rendered into very tall boxes with superscripts at the top and with some white rectangles overlaid on top of everything).
QT: mathstodon.xyz/@glakeland/1093

Grant Lakeland  
Let \[ G = \left\langle a,b,c \mid a^2, b^2, c^2 \right\rangle \] and let H be the index 2 subgroup consisting of (reduced) words in a, b, c of eve...
robryk boosted

the thing that i'm still most pleased with:

LEXY'S LABYRINTH, a puzzle game that happens to 99% emulate CHIP'S CHALLENGE 1 and 2! a trademark of bridgestone media group

- more than 800 community levels
- can load stock CC1 or CC2 levels too (and CC1 is free on steam!)
- rewind and undo!
- works on a phone!
- built-in editor; share levels with just a link
- new art and soundtrack
- fixes for some glitchy behavior, but compat settings if you'd rather not
- dev tools with demo playback and more

robryk boosted

Lojban thread!!! 

Okay let's talk about Lojban! No, this first post won't be in Lojban itself (but you gain enormous bonus points if you translate it).

Lojban is a natural, spoken language (like English, French, Swahili, Toki Pona, or Esperanto). It's also a constructed language or "conlang" (like Toki Pona or Esperanto). It's very easy to learn!

- la karda: a great tutorial: mw.lojban.org/papri/la_karda
- a lojban dictionary: la-lojban.github.io/sutysisku/

Toki Pona is fairly popular on the fediverse because it's such a cute language, and is in many ways the polar opposite of Lojban. Where Toki Pona is a fuzzy conlang, Lojban is a very crisp one. While spoken, it's also:

- Syntactically unambiguous! The "roles" in a sentence are always clear and parse cleanly. There's no such thing as "Time flies like an arrow" where you aren't sure if it's about the movement of time resembling a fired arrow or these creatures called time flies, who really love arrows.
- A predicate language. You can translate it into something like a problem solver and easily do analysis on statements, Good Ol' Fashioned Artificial Intelligence style

What it *DOESN'T* give you, and this is frequently misunderstood by various community members, is meaning disambiguity. The language tries fairly hard to reduce confusion there, but there are certain problems in language which are *inescapable*... the world changes, vocabulary drifts, two agents have different conceptions of things based on their experiences.

I AM NOT GOOD AT LOJBAN! But I really like it. It takes me a long time to interact via Lojban.

@cadey is probably one of our more prolific Lojban speakers. Who else speaks it? Let's use Lojban as much as possible in this thread!

How popular is en.m.wikipedia.org/wiki/Cynefi?

I've learned about it a few days ago and I realized that the boundaries it describes are very often by people without explicit references to it, and the correlations it implies (e.g. that need for reasoning from first principles is strongly correlated with problems that are not precisely specified). Now I wonder if everyone learnt about it and uses it (and I just didn't get the memo) or if there's a different source for these ideas.

The reason I noticed that is that I find some of the consequences of that view of the world weird and/or suboptimal for good modeling of the world. In particular, I like well-specified problems where you need to reason from first principles. Also, I see a tendency in the environments that deal with things that Cynefin marks as "complicated" to _not_ make it easy to reason from first principles (by not caring about making the relevant first principles easy to specify).

(-adjacent, because cryptography is the area that seems to straddle complicated and complex) (well, if not that then what is the branch of philosophy that does the same to motivation that epistemology does to knowledge?)

robryk boosted

@ariadne Who is Natalie Nguyen? A quick Google search yields a French TV presenter, a history professor, and a few other people that I don't think treehouse.systems is referring to.

robryk boosted

So turns out setting up a self-hosted Mastodon instance on NixOS is fairly painless.

github.com/delroth/infra.delro is basically all that was involved, and the only reason why I even have to setup the nginx config myself is because I do the weird split-domain thing where mastodon-web runs on mastodon.delroth.net (but I want to be ).

Of course I'm sure the pain will be in the long run... hopefully not :)

@delroth While experimenting with self-hosting be aware that the state machine for following has no support for reconciliation (so if you mess up the state on your side, it might stay inconsistent forever without anyone noticing).

msri.org/people/staff/levy/fil

This is a:
a) tale of trying to teach young (preschool) kids actual mathematics (not counting),
b) description of an approach to that
c) book that contains problems that are simply stated and sometimes nontrivial (in particular games, where the winning strategy is very unobvious).

robryk boosted

Another standard electronics practice is slapping a cheap and huge electrolytic capacitor on the DC power line, even in low-power devices. Clueless audiophiles would tell you it's like a pool of water that buffers energy to smooth the voltage. Until you realize their high series resistance and inductance make them useless at > 1 MHz in modern electronics, and often makes no sense.

What's the true reason? Resonance damping and control-loop stability. The inductor and capacitor are the input of a PSU form an LC resonator. At its natural frequency, noise voltage is not filtered but is significantly increased, ringing like a bell. Also, a regulated power supply has negative resistance, resonance turns a power supply into an oscillator.

A cheap and huge capacitor has high series resistance, it works as friction to dissipate energy, making it unable to ring. The capacitance also needs to be huge, since resonance occurs at a low frequency, the AC reactance must be low. For bonus, you get DC energy storage as explained by "conventional wisdom". All from a cheap low-quality capacitor... "It's not a bug, it's a feature!"
#electronics

Show thread

I am once again slightly-amusedly-enamored by Freefall:

freefall.purrsia.com/ff3900/fc

Helix: How was my story telling?
Sam: Not bad. You'll get better with practice.
Sam: Storytelling is an art that comes naturally to humans and sqids. It starts almost as soon as we can form sentences.
Dialog: My dolly wanted a cookie.
Sam: Robots see the world as it is and naturally tell the truth. It's a bad habit and with my help, one you'll learn to overcome.

robryk boosted

@robryk@qoto.org

"it should be about as labor-intensive to wire wrap through-hole components as it is to solder them"

Mass production uses wave soldering. You simply insert all the parts into the board, soak the board into molten solder, and pull it out. bingo, the full board is soldered within seconds. You can't wrap all the wires within seconds.

"due to wire shape, unwanted capacitances are kept much smaller in wire wrap (compared to e.g. a 2-layer PCB),"

Due to physical construction, unwanted inductances are much higher compared to a PCB. Even for advanced wire-wrap boards with ground planes (people used them in the late 80s during wirewrap's last days), it's still difficult to match a PCB's performance. A typical 4-layer board's signal layer is 0.1 mm above the ground plane, a wire-wrap board it's typically 5 mm, causing a 40x increase of loop area.

"no need for explicitly making multi-layer PCBs: the problems of routing, possibly stacking layers (for >2 layer ones), coating vias, etc. just go away."

One huge motivation to use a multi-layer PCB in modern designs is not even extra layers for the signals, but to have solid power and ground planes that contain nothing but solid copper. A copper pour on a multi-layer PCB naturally forms microstrips, the simplest kind of microwave transmission line, which is required to support high-speed digital signal transmission. Likewise, the capacitance between the signal layer and a ground plane is a feature, not a bug. Because of its planar circuit nature of a PCB, characteristic impedance is tightly controlled and extremely reproducible. Overlapping wires over a ground plane do not have a consistent characteristic impedance.

Even in lumped circuits and analog designs when controlled impedance does not matter, a solid ground plane is often still important to minimize loop area and parasitic inductance. Modern power MOSFET's switching speed is reaching 100 volts per nanosecond.

Standard wire-wrap construction is ill-suited above 10 MHz. Even back in the late 80s when electronics is made of really slow through-hole parts by modern standard, wire-wrap already started to show serious signal integrity problems.

---

But... Believe it or not. All of these problems I listed have been solved already in the 80s by combining the best of both world. The final next-gen invention was called "Multi-Wire" circuit board. It looks like a normal PCB but its inner layer uses insulated wires for connections, allowing overlapping traces. The ground planes and outer layers are similar to PCB. Reportedly,

IT can convert a multi-layer PCB design with 12 signal layers to a six-layer board with just two signal (wiring) layers.

But it was an highly expensive technology, and did not see much uses outside niche high-end applications. Today Hitachi is one of the few companies that still offer this service.

https://www.swtest.org/swtw_library/2013proc/PDF/SWTW13-22.pdf

Show more
Qoto Mastodon

QOTO: Question Others to Teach Ourselves
An inclusive, Academic Freedom, instance
All cultures welcome.
Hate speech and harassment strictly forbidden.