From what I understand, USB has 4 lines just like an audio port. So why can't we have a USB port which is circular? I feel it can be more convenient to plug in and also would have a smaller form factor. May be, we can have the same 3.5 mm jack port switch between analog audio output and a digital USB function. What am I missing?
Rotatability while the USB connection is already established is a feature you don't want:
While for audio frequencies, the electric ringing caused by imperfect contact are benign enough to not matter, I wouldn't trust that to be true for 1 Mbd (low-speed USB1), and certainly not at 12 Mbd (Full-Speed USB1.1), 480 Mb (High Speed USB2), or the gigahertzes of bandwidth that USB3.0 and later generations use.
In fact, for the latter, you need pretty well-matched connectors that are designed to conduct an RF signal instead of a current. Building such connectors even in rotatable form is, if at all possible (never saw a rotatable connector that has more than one waveguide, e.g. coax connection), extremely expensive. Think of > 20€ per connector, even in larger quantities, and, due to physics, probably not very compact.
I was surprised to find out there's actually USB3 slip rings! They look like this:
USB connector is specifically designed to allow for safe sequence of pins contacting firmly for hot plugging and disconnecting, passing the rated amount of current, and not being able to connect incorrect cables or devices.
None of this applies to TRRS connectors. It will short out various pins during insertion and removal. It will rattle and crackle after some amount of mating cycles. And since 3.5mm connectors have been used for headphones, mics, line level audio, IR transmitters, IR receivers, and even for power supplies, there is a good chance two incompatible devices are plugged together and it does not work or causes damage.
There isn't really any reason we couldn't have a round connector. But there are reasons why the existing connector designs are better:
It's hard to have a round connector with shrouded contacts. On all USB connectors it is physically very difficult to touch or short out the actual electrical pins. Even if the cable brushes against metal surfaces it is unlikely to happen. You would need to intentionally jam something metallic into the connector to achieve this. This was of particular importance when USB was first designed as ESD protection wasn't very good, and the power supply for the port was not switched (as it is now for USB-C). Having an unshrouded connector like the 3.5mm jack would have required extra protection for the pins and this would have added cost. There can also be more corrosion problems when contamination can be more easily transferred to the pins (e.g. Ligthning connector) and the pins are more susceptible to mechanical damage.
Hot-plugging is harder. This is not dissimilar to the first point, but in USB connectors the ground pins (and shield) are designed to connect before signal pins. If you have a 3.5mm type plug then the tip contact will run through the other contacts before it reaches the correct one. This can be dealt with, but it would require more protection circuitry and hence more cost.
The width of the connector isn't that important. Most devices you'd connect to are not round. They are normally rectangular and in most cases the critical dimension is thickness (e.g. phone, flash drive). If you had a round connector you'd have to make it longer to hold all the contacts at the expense of the width. For most types of products this is not a desirable tradeoff.
Why? Let us count the ways...
- Ease of manufacture (stamped pieces, single-plane wiring)
- robust (flat wiping contacts)
- safe contact sequencing (ground/power first, signals last)
- protected contacts
- overall shielding
- good signal integrity (defined impedance)
- low cost
- can be stacked (better backpanel density)
- can be a circuit board (dongles, converters)
and perhaps most important...
- USB adheres to a defined, open standard.
I have actually seen USB carried through a 3.5mm TRRS connector, which is also used for headphones, so it's possible. This is on a waterproof MP3 player, where adding another sealed socket would add cost and bulk. The other end of the cable is a standard USB-A, so both the device and the host computer must be able to handle hot-plugging. I have no idea of the data rate this device tries to obtain.
In general, running power over male T(R)(R)S connectors is a bad idea. I've got hardware in work that does, and is also designed to be used on metal tables (photodiode modules and optic tables). Geometry means shorts are rare, but they do happen, especially if the tables are earthed.