Even with my electrical engineering degree, four years of hard math, signals courses and understanding all of this at an abstract level, it all seems magical to me when it works in practice.
Thanks to all the tireless engineering of all the folks that manage to abstract this all away so we get clean zeroes and ones at the other end.
I'm definitely a digital, logical, bits-type-of-guy.
I certainly don't have your credentials but wasn't the point of the articles that it looked flashy while proving nothing about the capabilities of the chip?
One of the fun parts of this is when you start realizing digital methods of improving analog signaling exist.
He mentions one of these techniques briefly in this article: equalization. The condensed version is that you use an ADC for the receiver, sample the incoming waveform, then pass the resulting samples through a digital filter that amplifies the high frequencies and suppresses the lower frequencies.
This is actually quite good at counteracting the losses and phase shakeups that come with passing high speed signals through long cables.
>I'm definitely a digital, logical, bits-type-of-guy.
I think the fundamental takeaway from these stories is that even the digitalest of digital things still ultimately exist in an analog world governed by the analog laws of physics.
> In summary, the barbed wire had zero impact on signal quality. The signals went through perfectly undistorted. The only thing the barbed wire did was impress the heck out of Broadcom's customers.
Next time you look at a transmission line, I hope you'll focus on the big four properties: characteristic impedance, high-frequency loss, delay, and crosstalk. These properties determine how well a transmission structure functions, regardless of the physical appearance or configuration of its conductors.
FWIU from "The Information" by Gleick, Shannon entropy Shannon started out with digital two-state modulations on wire fences
If one wire in the pair is still connected, it's not really working because of capacitance (which is ~nil at 5mm), it's working because diff pairs are amazing. The remaining signal level on the single wire was probably enough to keep the receiver happy (at ~10kBaud in the Reddit post, versus the design ~100MBaud), and so, you win.
Any time you can spare the pins & wires to go differential, and have the slightest hunch you might need it, just do it. Diff pairs work!
Old analog tv over coax used to work with the cable poorly seated. You could even disconnect the cable entirely and just point the conductor at the plug and still get a(n admittedly quite noisy) picture as long as the cable end was near enough.
I had that happen as a kid and never figured out what caused it. My Atari was giving a very snowy image but I kept playing for a while. After dying I go to fiddle with the cable and it’s not even plugged in.
Honestly kind of messed with me at the time especially as I couldn’t replicate it.
Crosstalk or other interference coupled from the outside into the line. The telco grade twisted pairs do not really behave like twisted pairs (that cancel out the external coupling) at the frequencies used by xDSL.
Surely, although iron has 1/6 the conductivity of copper, even if severely rusted there’s far more metal than the wires in a decent ethernet cable, and far more separation between them
I’ll beat a bit of a dead horse with this one. This is why I’m not a big fan of water/pipe analogies when it comes to the study of electricity, electromagnetism and somewhere down the line electronics. While I understand early pioneers used “fluid” as a kind of hypothesis, I do not think they used them as analogies. I think they were trying to derive what was happening by comparing and seeing if their observations matched their line of thinking. All analogies break down relatively quickly the moment you attempt to work upwards from first principles (as we presently understand them which for all intents and purposes is “good enough” given we got many things to work just fine.)
The reality of the situation is far more impressive and engrossing if we attempt to truly get a handle on what is happening. Only then can we have a clearer idea of the nature of things like impedance and where/why/how the formulas that we use are derived from.
That analogy has gotten an uncountable multitude of people, including children, into electronics where "working upwards from first principles" is either not practical or impossible.
True. And many people don't care about the details. My wife, for example. She just wants to understand enough to wire up her derby cars. The water metaphor is perfect for those situations.
In light of me prefacing electronics by saying “somewhere down the line,” I don’t see the quality in this comment.
While there are professions where half-way through you kind of have to go back to the early things you thought you knew, and examine them in a more educated light, I’ve yet to see one as egregious as this. Nobody past a certain very early cut off limit benefits from using water analogies, and there is a push in education right now to move past water analogies because too many students enter first year post secondary with, simply put, incorrect ideas, and it has teachers baffled.
The water metaphor is fine for DC. The chain metaphor is fine for low-frequency AC. There's no really good analogy once you get to RF, or deal with the details of active components, then you just need to understand it directly. But the metaphors are fine for the simplified situations they're used with for teaching.
Simplified, Thinnet is just RG-58, and RG-58 is just coax that has a characteristic impedance of 50 Ohms, and 50 Ohm coax works just about like any other impedance of coax as long as the termination impedance is correct (or the length between terminations is 1/4 wavelength), and the attenuation is not too severe.
One can send the whole RF spectrum down a single length of coax.
The little pinkish-copper wires in your wifi gear are just coax. SATA cables: Also coax. Uncompressed HD video over SDI? Also coax.
(But you asked about successes, not theory. I've run 5G cellular services through thinnet wire. Does that count?)
The primary question is whether you want just the cable or the multi drop bus architecture.
And well, DVB-T2 at UHF frequencies works over RG-58 and the just parallel connected T-pieces just fine, and the fact that whole such system has completely wrong impedance does not seem to matter for ~4 devices.
It has the wrong impedance for MoCA, and any long split-off sections might cause nasty reflections, but other than that, MoCA (possibly through an impedance matching transformer) ought to work fine.
Why does the demo title within the image say "Gigabit Ethernet"? Is that a marketing lie of what they had hoped to eventually achieve, with the actual demo running 100mbit? Or was the same demo repeated for (what would become) 1000Base-T?
I’m not certain, but I believe the demo photo is from WideBand (Roger E Billings CEO).
I got certified as a “wideband network administrator “ at their (really) underground bunker headquarters at the International Academy of Science.
It was a little bit surreal, but the tech was really cool. We were pushing 1gbps over cat3 to 150 meters, so pretty respectable even today.
I think they still are building ultra-low latency switches that are favored by HFT and others that need nanosecond latency switches.
1000basetx eclipsed WideBand and my cert was basically useless, but the experience was really cool and it was great to hang out with a science cult for a while in their vast subterranean lair.
Ah, I had missed the author's caption on that photo dating it at 1998. Looking at it again, it's even more odd because the article was written in 2001 - meaning that the 100Base-T4 anecdote was twice as old as the gigabit demo photo.
The part that stands out to me is the author writing off the high frequency loss because each pair is only carrying 25 mbit, yet Gigabit is ten times the bandwidth which is around where I'd think things would start to get a bit wonky. But maybe my intuition is still just the result of single conductor flapping around in free space, rather than a "controlled" impedance balanced pair.
Gigabit brings in all four pairs, so it's only at 125 MHz, five times faster (which is in fact exactly why it does that). That's not all that much worse than 25 MHz. And as for intuition... Howard Johnson is something of the authority on signal propagation, so if he says that's what your intuition ought to be, I listen!
I'm just saying, scopes have a 20Mhz limiting switch for a reason and that's also around where I start to think something might be better off not on a breadboard. I recognize neither of those things is a controlled impedance pair. It just hits differently to write off 25MHz vs 125MHz as "low frequency".
Going over barbed wire was a common way to set up local telephone lines to the neighbours in rural U.S. So yeah, "does it work over barbed wire" may be a "yeah but is it still worse than the telephone?" question.
Great story. Seems like a prime example of "any sufficiently advanced technology is indistinguishable from magic." In this case the barbed wire was a sleight-of-hand trick.
One day I'll dive into networking technology. It's fascinating to me how going down the OSI model layers results in such different goals, requirements, and constraints.
If you think this is impressive look up Powerline ethernet, that shit is bananas. Though it's probably its own protocol with far more error correction.
Thanks to all the tireless engineering of all the folks that manage to abstract this all away so we get clean zeroes and ones at the other end.
I'm definitely a digital, logical, bits-type-of-guy.