Hacker News Reader

This! Even just comparing USB-A 1.0 to what it replaced, it's hard to imagine a reversible plug like USB-C having been feasible/cost-effective to produce in the mid 90s. The PS/2 mouse and keyboard ports had pins of a similar size and spacing, but it was hard to orient the circular connectors. You could get the connector lined up and twist gently until the pins found their holes, but eventually you would bend a pin. I remember being very impressed at the way USB solved the issue of bent pins, but wondering why they didn't just key the connector like a Zip disk (which had beveled edges along the top of the disk, and wouldn't go into the drive upside-down).

Meanwhile we were connecting printers, scanners, and other peripherals using a 25-pin parallel (or sometimes SCSI) plug. Granted, the DB-25 was already pretty old at that point, but I have to imagine a 24-pin reversible USB plug would have been close to that size if it had been developed in the 90s.

This thread makes a convincing case for why the original USB connector couldn’t be reversible, but there’s a parallel question I’ve never seen addressed: Why couldn’t it have been asymmetrical?

The physical design of FireWire was one of its many advantages, in that you could instantly tell, visually and by touch, which side was up by the beveled corners on one side of the plug. I’m guessing the answer here was again cost, that manufacturing an interconnect with two 90-degree angles and four 45-degree angles was more expensive than one with four 90-degree angles, but I’ve never seen any confirmation of this.

> This thread makes a convincing case for why the original USB connector couldn’t be reversible

The sole reason as stated by Ajay Bhatt, the project lead at Intel, was cost. They wanted to encourage mass adoption of USB and every cent saved helped in that initiative.

The compromise for UX was to make the connector rectangular rather than circular, so that there was a 50% chance of getting it correctly orientated on first try.

FireWire 400 was a great connector! I resisted using FW for the longest time mostly because the cables were more expensive and less common than USB. Once most of my computers had it it started to make more sense for faster file transfers and external drives. They were so much easier to plug in without looking than USB, but the FW800 and mini connector went to a more rectangular shape and somewhat lost that advantage.

I never ended up getting a camera or anything else that supported FW, but USB always seemed to work well enough for my (relatively light) use case.

Good story about that decision here (they did think about making it reversible): https://www.theverge.com/circuitbreaker/2019/6/25/18744012/u...

A standard USB 3.1 type A connector has 9 pins. The reversible type C has 3 extra pins (SBU, CC, VBUS), not counting the doubling caused by making it reversible. So while the reversibility doubled the number of pins, the increased complexity of the standard itself tripled it (from 4 to 12).

Good catch. I should have either compared the 9 pins of 3.1 or been more specific in saying 4 pins for USB 2.

Fun fact: USB 2 Mini/Micro connectors actually have 5 pins (hence my original 4-ish statement). The 5th pin doesn't actually go over the wire, it's just used as a signalling pin to the controller that tells it if the plugged in device is compatible with USB OTG.

> A standard USB 3.1 type A connector has 9 pins. The reversible type C has 3 extra pins (SBU, CC, VBUS)

Not precisely. The type C connector has four distinct USBSS pairs; the type A connector only has two. This allows the type C connector to be used to carry DisplayPort or other high-speed data alongside USB.

Isn't that a direct outcome of the reversibility aspect? [0] Technically it has 2 of everything.

[0] https://upload.wikimedia.org/wikipedia/commons/thumb/0/07/US...

Not everything. The USBFS pair (D+/D-, at the center of the connector) is mirrored; the two D+ lines are connected together. This is not true of the four USBSS pairs -- TX1+/- and TX2+/- are separate, and can both be used at the same time for different purposes. The CC1/2 lines (which aren't mirrored either) are used as part of the process to negotiate how these pairs are used.

I meant it the other way around. The non-reversible type A does not have this extra pair regardless of standard. But the type C is reversible and short of having pins automatically reconfigure based on the detected position, the only straightforward way to achieve this reversibility is to double the number of pins. So one side of the type C connector perfectly mirrors the other (but flipped for reversibility) mechanically and electrically.

I'm fairly certain they are using double the number of USBSS because they're available due to the mirroring. It was the easiest way. Unlike the control channel or side band, you can actually benefit from using all the pairs for data transmission since they're already there.

Also, the circuit board can be made into a connector, which is not possible with USB-C.

Lots of cool applications using that technique.

Assuming the structural issues are solved, isn't it literally as easy as doubling each pin in a 180° fashion and shorting each pair? What additional controller complexity would this lead to (assuming termination issues are solved)?

That is exactly how usb 2.0 (LS/FS/HS) devices are implemented with USB-C.

I suppose there's no reason you couldn't do that. I imagine they didn't want to do that because cables (as a platonic ideal) should just carry signals from point A to point B without fudging anything. When you start to screw with that dynamic, you end up with lots of non-standard issues that people blame on the standard (I'm reminded of contractors jamming two 100mb/s lines through a single Cat5e cable instead of just switching on a single 1000mb/s line)

Turns out the 24 pins of USB-C are now all used for 20Gbps USB 3

> Compare the 4(ish) pins of USB-A to the 24 pins on USB-C

This is misleading.

If you want to only cover the USB-A spec with USB-C, you need only 8 pins.

I have a cheap USB C on my desk that got 8pins because it do only USB and charge.

USB-C can pass HDMI signal on the other pins, which your USB-A can't do.

Sure, it's misleading. That's why the next sentence reads...

> Not all of them are strictly necessary for parity with USB-A, but you could cut that number in half quite easily if the plug were not reversible

Is 4 not half of 8?

You imply that the USB-C could have only 12 cables. Only the USB2.0 pins are duplicated, there is 4GND and 4PWR cables for power need, and the rest of the connector are not duplicated. If it was not reversible, USB-C would be 22 connectors, not 12. Plus USB-A is the only one with 4 cables, what was used on most phones was USB-B micro which has 5 cables.

What are you trying to argue here? All pins on the connector are rotationally symmetric. They may not be useless, because they still increase overall bandwidth, but they are indeed doubled. A non-reversible connector would not use this design and there would be fewer pins as a result (as much as half).

Also: USB-B micro connectors might have 5 pins, but the cable itself has 4 wires. The OTG pin is grounded in the connector. This is actually why I said "4(ish)" in my original post.

The rotating pair can send independant signals.

Thats why USB-C have more or less the same speeds than HDMI, while the HDMI have 22 connectors.

Compare what is comparable, you can't use your USB-A to display 4K-60Hz video, but you can with your USB-C, or HDMI.