Hacker News new | past | comments | ask | show | jobs | submit login
Connectors That Changed the World (outfunnel.com)
56 points by standrews on April 19, 2021 | hide | past | favorite | 49 comments



Tim Hunkin, who was behind the British show The Secret Life of Machines, is releasing a YouTube series called The Secret Life of Components, and episode six is about Connectors:

* https://www.youtube.com/watch?v=q43tZ6DjuIE

His channel (which also has the original Machines series available):

* https://www.youtube.com/c/timhunkin1/videos

* https://en.wikipedia.org/wiki/Tim_Hunkin


I was hoping someone would mention him. I found out about that series recently and have really been enjoying it.


Thanks for the share!


Correction on #3: It was generally the brakemen, not the engineers, who were doing the coupling. (Engineers mostly stayed in the engine.) It’s horrifying how cavalierly this problem was treated; there’s an old joke about how, at conventions for brakemen, everyone had to pair up to applaud.

Brakemen also had the unenviable job of running along the tops of the cars while the train was in motion, jumping the gaps to get to the brake wheels on each car. In fact, I’d argue that more important than the automatic coupler was the automatic continuous brake—particularly the Westhinghouse air brake and its gladhand connector, still used today on railways and semi-trailers—which was considerably safer (both for the brakemen and also for everyone else, since trains would now fail safe and stop if they broke) and made longer and faster trains possible.


> It’s horrifying how cavalierly this problem was treated; there’s an old joke about how, at conventions for brakemen, everyone had to pair up to applaud.

I'll admit it took a bit for me to realise the horror of that joke, but boy howdy.

I'm rather glad we've moved beyond an era of business and government where traumatic amputations were considered an inevitable fact of life.


I'd be glad of that too, but we are not there yet.


USB had a quite successful precursor: RS-232. The serial standard is now 60 years old (!) and still much more widespread then USB in many industry machines.


RS232 is the electrical signal standard; the chunky connector usually used is a DB25 but basic RS232 can fit in DB9

RJ11 (eg landline) and RJ45 (eg ethernet) were also connectors that achieved very widespread use, as did RG6 (eg cable tv)

(I'm not going to say they changed the world and I'm also not so sure I like the grab-bag nature of the various things the article lists either

for example, 8, 14 and 16 (and more) pin DIP were pretty important, or was it the TTL signals they carried or was it the connecting solder that ultimately made USB valuable... dumb article to write as a listicle.


If you are interested in being a pedant, which you may or may not be:

What you call DB-9 is actually DE-9. (A DB-9 would be the size of a DB-25 but with only 9 pins, which would be weird.)

What you call RJ11 is a 6p2c. (Six position, two connectors.)

What you call an RJ45 is an 8p8c. (You can guess this one.)


To expand on the pedanticness:

The letter in the D-subminiature line (aside: they really didn’t anticipate how small connectors were going to become) refers to the shell size (A through E, though not in size order). “DB” is frequently used colloquially as a prefix for any size in the line. There’s also some interesting variants like the DB13W3, which is a B-size shell containing 10 ordinary pins plus 3 coax connectors.

RJ11/RJ45 refer to the connectors and the way they’re wired, together. For example, RJ14 is the same 6p connector as RJ11 but wired to support two phone lines instead of just one, and the same 8p8c connector we’d use for Ethernet but wired for 4 phone lines would be an RJ61.


I fully expect another pedant to respond to this, any minute now.



I was absolutely not being a pedant, because even though I was replying to a comment, everything I pointed out would have fit into the original article; whereas what you wrote fits only into my comment.

I went to the trouble to point out truly salient distinctions between physical/hardware layers, signal/electrical layers, and virtual/logic layers that were being ignored, and thoroughly pointing out technologies that were extremely important to changing the world thoroughly in the ways in which the article was discussing

And, here, you can buy yourself a DB9 https://www.l-com.com/frequently-asked-questions/what-is-a-d...


Firstly, I think you’ve misread: I didn’t accuse you of being a pedant, I suggested you might like to be. Behind a pedant on the internet is sometimes quite fun, but if you don’t want to, then don’t.

Secondly: a common mistake is still a mistake. That shop is selling things with DE-9 connectors, all of which they have mislabeled.


> misread: I didn’t accuse you of being a pedant

and nor did I say you did, who now is misreading?

why do you avoid replying to the the main thrust of my rebuttal?

again, I made a cogent argument about the main topic, and whether I labelled a old connector the way many others did is irrelevant to the topic at hand, and if anything hobbles your pedantry hobby horse. Where are your critiques of the more egregious errors in TFA?


You didn’t rebut a thing I said: the only substance of my initial comment was that you had mis-named some connectors... which you did. Your reply contained no proof otherwise, except the shop link which I explicitly addressed. If you’re really asserting that I’m wrong about connector names, here you go:

> Because personal computers first used DB-25 connectors for their serial and parallel ports, when the PC serial port began to use 9-pin connectors, they were often labeled as DB-9 instead of DE-9 connectors, due to an ignorance of the fact that B represented a shell size. It is now common to see DE-9 connectors sold as DB-9 connectors

from https://en.m.wikipedia.org/wiki/D-subminiature

There are plenty of other sources available, which I’m not going to bother looking up.

As for the rest of your initial comment, I don’t disagree: you made a pretty substantive contribution to the thread, but again I never said you didn’t.

The main thrust of your “rebuttal” is trying to fight an argument I’m not making.


thank you for directing my attention to the mistake.

I mistakenly took your phrase "if you want to be pedantic" to mean that I was being pedantic, but inadequately so, a common just desserts.

now, if you want to be pedantic, which you seem to, you might wish to say that I did not mistake your phrase but in fact I chose the most common usage of that phrase as its clear meaning, and you could have been more clear simply by saying "not to be pedantic, but...", as my comment was clearly not focused on model numbers but on real issues of substance, and your comment was focused on pedantry.


And yet I wonder how many of those machines have USB to RS-232 converters internally... not that I've ever been asked to design such a thing....

Properly implemented RS-232 is a thing of beauty, in a way that USB just can't be. It's a shame so many implementations (of both!) fall short.


The ones that barf out a debug console from the processor's first init steps are probably not USB internally. ;)

But yeah, USB never came up with a good way to do peer-to-peer. On-the-go is super driver heavy and just not useful in the way RS232 is. New TVs still ship with it, for home automation reasons.


And neither are they RS-232, unless you've got a transceiver (MAX232 etc) in there :)

Man... USB CDC is painful. It's so much "fun" that a serious solution on one product was two back-to-back FTDI transceivers. I think they finally designed that out, but I'm scared to look at what might be in its place now....




That is equally parts silly and useful. There's so many time's I wish I'd had one.


Is it two lines with ^S/^Q/BREAK, or hardware flow control, with or without ring, db25->db9 converters (never of the right gender), etc...

I was pretty happy to throw it out the window in favor of USB. Of course, firewire was better than usb and the whole cycle started again.


I've recently started working with firewire, and man what an interesting bus. You've got isochronous and asynchronous support simultaneously, awesome! But then you get this: "Voltage is specified as unregulated and should nominally be about 25 volts (range 24 to 30)" Looking at the nearest firewire device on my table, it will certainly blow up if I get near it with 25 volts. It's like USB-C without any negotiation!


I worked on the FireWire driver stack for OS X... and we encountered many incidents of managing to plug in the 6-pin connector upside down. Devices will actually catch fire. (FireWire carries up to 45W (!) at 30v)

Also, isochronous mode is great in theory, but in practice, at least in a desktop computer, the rest of the machine internals are not isochronous so you can still run into real-time delivery failures.


> Devices will actually catch fire.

A rare case of successful naming, then.


I know I shouldn't be surprised, but how can that keyed connector get forced in backwards? That is mind-boggling.


The metal shell of the 6-pin socket has a seam where it's closed, and that can open or loosen with use. That makes it easy (enough) to plug in a connector the wrong way.

Trivia: Pretty sure the 6-pin FireWire connector was inspired by the GameBoy link cable connector. https://en.wikipedia.org/wiki/Game_Link_Cable


One of the nice things about UARTs is that that's pretty much all there is to it. That is, it's about as much pain to implement as to use.

USB CDC is worlds of pain more complicated to implement, but quite easy to use. As long as you don't hit a bug. But USB stacks never have bugs, do they?


Actually the more appropriate predecessor of USB is 8-bit Atari SIO:

https://en.wikipedia.org/wiki/Atari_SIO


No love for Apple Desktop Bus?


Atari was way earlier.


This article is a list of connectors that doesn't go into much detail into how each changed the world.

The idea that APIs are a singular thing - or started as late as the article suggested - is also sufficiently wrong to put everything else written in the article into doubt.


I agree. The author laughably claims APIs are how applications take to each other and implies they are necessarily remote or on the Internet. Web APIs are one kind of API, but so are various C Library interfaces.


It is an ad. It worked. Don't expect a car advertisement to teach you how an engine actually works.


Don't expect a car advertisement to teach you how an engine actually works.

Why not? [1]

[1] https://youtu.be/7Z_TaWWG_NI


I knew this was coming, but those ads are not the feeling based, narcissism invoking ads of today.


So, am I crazy or does every header file define an API since the dawn of header files? Is API really meant to specify just the remote procedure interfaces like Rest, RPC, etc?


You're correct, but a lot of otherwise knowledgeable and professional people seem to understand the word different because of how they see the word used in context in IRL.


Is it bothering anyone else that several of the images are incorrectly labeled? The female Micro B is clearly a male Micro B and the female Lightning appears to be the female end of a Micro B to Type C converter.


Switchcraft 280 and XLR connectors deserve mention for longevity and ubiquity in music and recording.


In February 2000, Salesforce released the world’s first API into the wild

API's existed long before this.


It's written by a marketing guy, this is a puff piece.


I really wish that regular trailers would develop some sort of standard linkage system rather than the SAE ball. I want to back up to my trailer and have it hitch a load equalizing connection without having raise/lower and get grease all over my hands and pants.


There is. Tractors for semitrailers have a fifth wheel, of course. That's the 90mm size. There's a smaller 50mm size used for RV trailers. The hitch for that bolts into the truck bed.


Yes, but no. The thing is I don't want to give up the bed of my truck in order to hitch. I use the truck as a truck when it isn't attached to the trailer.

So something more in line with the rail car system is in order.


I was certain this was going to be a whole piece dedicated to molex connectors. Interesting read though.


As a software guy I have to feel that APIs are too general. How about REST or even JSON?




Consider applying for YC's Spring batch! Applications are open till Feb 11.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: