This chart is a lie


Serial cables are neutral? No way. Chaotic evil. I had to make too many of them. DB9 or DB25, or some ghastly custom pinout some manufacturer saw fit to stick on their device? I’ve encountered lots where all you need is 3 pins — ground, transmit, and receive — but even then you have to worry about whether this is a straight pass-through cable or a null modem cable. Some devices require one or several of the handshaking lines to be enabled — but different machines require different handshakes. Do you need DTR or DCD? Sometimes yes, sometimes no. Then some of those handshake lines are completely redundant, and you can make it work just fine by shorting out the line to one of the other pins.

I remember the bad old days when you’d buy laboratory devices and they’d have some odd connector hanging off the back and there’d be a cryptic pinout diagram in the specs, and you’d have to solder up your own cable for it. It was not a happy time.

Comments

  1. says

    Standard USB should be lawful evil. There’s absolutely no legit justification for having a different connector on each end.

    I mean, I talked to someone who was on the USB design committee, and even *he* didn’t know why they did that.

    (The miniature versions at least have the justification of needing a really tiny jack for super-skinny devices.)

    Also, one of my primary functions used to be working out cables for connecting exotic devices to PC serial ports, so I hear your pain. I still have my copy of the IBM PC Technical Manual with all the official pinouts…

  2. says

    @#1, Woozle:

    I’d be willing to bet that USB cables have different connectors on each end because it somehow managed to shave a fraction of a penny off the manufacturing cost at the time. The whole original USB standard was designed to be as cheap as possible, and to heck with quality, and the upgrades to the spec have been paying the price ever since. (And, of course, the PC world shunned FireWire, which was superior in just about every way for the entire time that it was still under active development, and could still hold its own against USB 3 even now if anybody had built devices which made full use of the final version of the spec, which they didn’t.)

  3. brucej says

    They left off SCSI, which could probably populate the entire matrix. (nah it’s all chaotic evil. One time I had to put the terminator in the middle of the chain to get it to work! and don’t even get me started on jumprs….shudder

    We had a well-used DB-25 serial breakout box we connected with a logic probe to suss out connections.

  4. brucej says

    @3, firewire was doomed by the costs of implementing it; ironically, by Apple, mainly, who were the main driver to acceptance of USB (we were seeing USB ports on PC’s before the original iMac, but I never saw an actual USB peripheral in the wild until after the iMac debuted)

  5. Mark Dowd says

    I’d be willing to bet that USB cables have different connectors on each end because it somehow managed to shave a fraction of a penny off the manufacturing cost at the time.

    Except that makes no sense. You would need two different toolings and need to track two different inventory stocks for the different connectors. There’s no way that could be cheaper than using the same bin and the same tools for both ends of the cable.

  6. kaleberg says

    Firewire was a nightmare. I never did manage to get daisy chaining to work for copying one disk to another. The copy would always fail somewhere before completion. It was as bad as paper tape. You had to keep an eye on it.

    3.5mm headphone jacks weren’t too bad, at least until they moved to stereo and some of them started supporting microphones as well as sound out. You had to count the little metal rings to be sure.

    This chart seems confused by phono. Those RCA connectors were usually red or black and came in pairs, one for each stereo channel. The other end was often a pair of bare wires to clip to your speakers. Good luck guessing the polarity, so turn up the volume slowly, at least if you like your speakers. The yellow connector shown was for video, not audio or phono. That was video without sound. You still had to run at least one audio line and guess which channel or just run both.

  7. says

    A word in favor of DB25 connectors: These are still the standard connector for music studios that have to carry 16 tracks or more in either direction; i.e. just about any professional facility. Yes, tremendous strides have been made in making ethernet functional for that purpose, but that has proven to be an expensive and rather elusive goal, which only now is starting to emerge as a real possibility. If you are interested, you might look at this article from a recent issue of Sound on Sound magazine, exploring the problems in some depth:

    https://www.soundonsound.com/techniques/ethernet-audio

    But at this point, DB25 connectors are still the main choice for most studios.

  8. says

    @#5, brucej

    @3, firewire was doomed by the costs of implementing it; ironically, by Apple, mainly, who were the main driver to acceptance of USB (we were seeing USB ports on PC’s before the original iMac, but I never saw an actual USB peripheral in the wild until after the iMac debuted)

    The implementation costs of USB 2 were as high as FireWire, which was still viable at the time, and with worse return on investment. (The theoretical maximum throughput of USB 2 — which it never reaches in practice because of technical overhead — is still lower than FireWire 800, which can easily be reached.) What really killed FireWire was PC users who said “this belongs to Apple so we’re not going to use it even if it works better”. (Ever done video capture over USB 2 or lower? Dropped frames are normal there, as opposed to FireWire, where they are unthinkable.) Anti-FireWire crusaders usually used patent licensing fees on the connectors as an excuse to avoid it — which were $1 for most of the time during which it was a viable option and were lowered to $0.25 by the time of USB 2. It was a big smear campaign, and I’m amazed it wasn’t repeated with Thunderbolt, which was an Apple-Intel team-up.

    @#6, Mark Dowd

    Except that makes no sense. You would need two different toolings and need to track two different inventory stocks for the different connectors. There’s no way that could be cheaper than using the same bin and the same tools for both ends of the cable.

    Not if it turned out that, say, you needed at least one end of the cable to have [special quality A] which is only available with [connector type A], which is relatively expensive, but once one end had [special quality A] the other end could be anything at all, permitting you to use the much, much cheaper [connector type B]. Even if the expected cost savings came out to a fraction of a penny, if you’re planning to make a million zillion of the things it would be an appreciable savings, and the extra tooling costs would be dwarfed. (For that matter, it’s also possible that it’s the jacks, not the cables, which had to have some special quality.)

  9. consciousness razor says

    Lawful evil seems right for a lot of these…. I would put MIDI cables in that bin, then set up multiple charges which would all be carefully timed to explode at the right moment. Justice demands it.

    I don’t think the various forms of USB should take up two very different alignments. They should all go into true neutral, which in practice always seemed rather chaotic anyway.

  10. Azkyroth, B*Cos[F(u)]==Y says

    So what do we have for “‘chaotic good’ where ‘good’ is interpreted as ‘good-ish Lawful Neutral’ and ‘chaotic’ is interpreted as ‘silly and/or hyperactive?'”

    Fuck I (still) hate that one.

  11. andyo says

    Woozle,

    There’s absolutely no legit justification for having a different connector on each end.Standard USB should be lawful evil. There’s absolutely no legit justification for having a different connector on each end.

    From what I understand it was because the way USB transmits power (one way). USB Type A is always downstream-facing (host), and Type B is upstream-facing (device). Having different ports makes it impossible to connect 2 sources of power together, and that’s also why Type-A male to Type-A male cables are “illegal”. There is a bit of an exception with USB OTG with mobile devices, but that came much later. And if you think this validates your complaint that USB was badly designed, then that’s why USB-C came out, which tries to do “dual-role ports” and so much more, but it has its own share of complications (thus its position on that chart), which reading PZ’s posts, are strikingly similar to the serial port issues he mentions.

  12. cartomancer says

    It’s like a whole different language here. I can’t understand a word of it, or grasp even the most basic meaning from the joke. I know next to nothing about plugs and wires – even the concept that people can assign character to them is a revelation to me. Truly there is more in heaven and earth, Horatio…

    Carry on.

  13. mcfrank0 says

    I’m also still trying to understand why USB-C is “chaotic evil” — other than being relatively new and not yet present at the checkout counter at Walgreen’s. The plugs-in-any-orientation feature is greatly appreciated each time I recharge my cell phone!

  14. blf says

    andyo@14 has it correct, the different host-side and device-side USB connectors are due to the power distribution model. Or, put more simply, to prevent incorrectly-connected cables from blowing things up.

  15. Larry says

    In a drawer in my office, I have a plastic box that contains at least one of every type of RS-232 serial adapters known to mankind. DB-25 M to DB-9 F, F to M, M to M, F to F, null modems, DB-9 to DB-9 adapters of various sexes and kinds. Then there are the serial cables of a variety of lengths with their own connectors. They’ve all been used a one point or another in my career, though. And now they’re tanned, rested, and ready for the call if 1200 baud external modems ever become fashionable again.

    BTW, PZ, I though you solved the slow loading problem on this site. It was back with a vengence this morning. Buttons and links and the scroll bar took forever to load and become active. Maybe you need to borrow a serial adaptor??

  16. whheydt says

    Engineering nitpick… There is NO SUCH THING as a “DB-9” connector. What there is, is a DE-9.

    The series is… DA-15, DB-25, DC-37, DD-50, DE-9. There is also the HD-15 that is used for VGA.

    So far as I can tell, far too many people think that “DB” is some sort of unitary designation. Formally, it’s just the “D”. The second letter associates with the number of pins.

  17. says

    USB: fascinating, because it is a four dimensional connector.
    You try to insert, and fail.
    You rotate in space, try to insert and fail.
    You rotate in space again, thus adding movement in time to the process, try to insert and after movement in the fourth dimension…. Yay! Success! (Usually.)

  18. says

    I’m baffled by the idea of USB-C as chaotic evil.

    It’s the best implementation of USB to date. Vastly improved charging time for devices, it fits either way you turn it, what isn’t to like?

  19. latveriandiplomat says

    The original version of USB was driven by Intel. They wanted people to connect devices to their PCs, not connect devices to each other.

    So, they shunned the peer-to-peer firewire and pushed a cable with two different connectors and baked-in distinct host/device roles from the physical layer on up.

    The gymanastics needed later to adjust USB protocols to smarphones and tablets (which can be either a device or a host depending on what’s on the other end) are a tribute to this legacy.

  20. Rich Woods says

    Fortunately brucej has already mentioned SCSI, so my hands have almost stopped shaking by now.

  21. andyo says

    #24 sotonohito

    I’m baffled by the idea of USB-C as chaotic evil.

    It’s the best implementation of USB to date. Vastly improved charging time for devices, it fits either way you turn it, what isn’t to like?

    In theory, and if all manufacturers followed to the letter USB-C spec, then it probably would be great.

    As it is now, though, “chaotic” is a very accurate word I’d use. It’s been a couple of years now since it started becoming mainstream on Android phones, and it’s been quite a mess of a ride. Back then, a Google engineer who worked on the USB-C implementation of the Chromebook Pixel took it upon himself on his own time to wage war against dangerous cables and accessories, and he’s still at it, with a few other engineers joining along. Google Benson Leung, who was dubbed at the time the “USB Vigilante” because of his very useful posts on Google Plus and on Amazon reviews. One of his biggest and most beneficial victories was the banning from Amazon of Type A-to-C cables that used the wrong resistor to advertise and force “fast charging” at 3A, which would overwhelm most Type-A ports, especially on computers, which can be as low as 500mA (USB 2) or 900mA (USB 3).

    That was only one thing. If you google his name and USB type C, you’ll find lots of different real-world issues. Most of them boil down to manufacturers not following spec, but since USB-C tries to do so much with a single cable/connector, there are also other, completely spec-compliant, sources of confusion.

    For instance, USB Power Delivery is the charging standard used nowadays. It can charge at different voltages (up to 20V) and currents (5A max), and all C cables are required to support 3A. But, USB-PD goes up to 100W. So, with a regular charging cable, you only can charge up to 20V/3A. You need a special “emarked” cable with a chip to charge at 5A. You also need special cables for USB 3.1 Gen 1 (previously known as USB 3.0–another can of worms), and for 3.1 Gen 2. You want Thunderbolt support? Look for those cables. “Alternate modes” (HDMI, DisplayPort), who knows at this point. Analog vs digital audio is just now becoming an issue with Google’s new phones ditching the 3.5mm plug and only supporting digital on the C port, where previous phones supported analog Type C to 3.5mm adapters, now all adapters won’t work on all phones. It’s getting too long, but you see the issues in the real world.

  22. se habla espol says

    @whheydt #22, 17 October 2017 at 10:35 am

    Engineering nitpick… There is NO SUCH THING as a “DB-9” connector. What there is, is a DE-9.

    Yes, the usual 9-pin D-sub is the DE-9.
    There could easily be a DB-9: it would be a D-connector with a B-size shell and a 9-pin insert. Usually, such an insert would provide for larger “pins”, like small coax connectors or larger power connectors.

    The series is… DA-15, DB-25, DC-37, DD-50, DE-9.

    Those are indeed the most common combinations.

    There is also the HD-15 that is used for VGA.

    More properly known as a DE-15: E shell, 15-pin insert, necessarily using smaller pins with shorter inter-pin distances than the usual ones.

    So far as I can tell, far too many people think that “DB” is some sort of unitary designation. Formally, it’s just the “D”.

    IIRC, the D-sub (sub for ‘subminiature’, which was descriptive at the time) rack-and-panel connector and its nomenclature are from Cannon, now ITT Cannon, although it might have been Amphenol, Cinch or some other connector specialist. My ancient catalogs are long gone, so I can’t verify my recollection of the origin.

    The second letter associates with the number of pins.

    May latest Allied catalog is from 2009, but it shows a large number of specialty inserts in the standard DA/DB/DC/DD/DE shell sizes.

  23. zibble says

    I don’t get what any of this chart is supposed to mean, but coaxials are definitely the most evil cable. I can still remember how sore my child fingers used to get just trying to plug my SNES into the TV.