| Electronics > Projects, Designs, and Technical Stuff |
| Evolution of usb connector? |
| << < (18/18) |
| IDEngineer:
--- Quote from: westfw on January 03, 2019, 09:49:50 am ---The TI and NXP chips don't seem to be compatible. NXP uses a pair for each signal, TI a single wire. --- End quote --- True. But I've since been made aware that Linear (now Analog Devices) also has an entry in this I2C extension contest. (It drives twisted pair for sure, might work single ended too.) When not one, not two, but THREE of the big chip houses are pushing out products in a given category, that tells me there's serious interest and money looking for solutions. It's early yet, kinda like the early days of IC opamps. Eventually, like opamps, if there's enough of a market, pinouts and specs will settle into compatibility. --- Quote ---No standard connector --- End quote --- Agreed. That needs to be resolved. There are a few popular choices but hopefully a clear winner will emerge. I kind of like the 6-pin version of the RJ-11/RJ-45... well known and understood, cheap and reliable, termination tools are commonplace (just a different jaw set), can carry power+ground+two twisted pairs, yet different from Ethernet's RJ-45 8-pin connector (though the I2C plug can be inserted into the Ethernet socket). --- Quote ---Assorted uC's that I keep seeing "having trouble getting I2C to work right" messages --- End quote --- Heck, I still see "having trouble getting UART to work right" messages! I suspect a lot of those are more about the person writing the complaint than the underlying hardware. That seems to be how most of those threads finish up IMHO. --- Quote ---it wasn't an inherent weakness of the UART protocols that made RS232 such a PITA in practice, it was the lack of ... formalization of all of the "other stuff" needed to communicate. --- End quote --- My career predates USB so I've watched the entire scenario play out, from "RS-232 is all we have" to "we'll use PS/2 and mini-DIN connectors for keyboards and mice" to "let's use USB for everything". During the rampup of the personal computer era I was Project Engineer at the largest keyboard manufacturer in the world so I was literally in the middle of all this stuff. What precipitated the push for USB, at least from our customers (literally every single PC "clone" manufacturer you *ever* heard of), was the desire to 1) reduce the number and variety of connectors coming off the motherboard, and 2) reduce the number of Customer Service phone calls. In the early days you had standard-sized DIN connectors for the keyboard (as defined by IBM's original PC), DB-9F's for serial, DB-9M's for video, DB-25's for parallel, etc. It was expensive to include all those connectors. Then IBM defined the second phase with the mini-DIN's for keyboard and mouse; that was great except that the connectors were only different by COLOR and lots of customers got them backwards, driving up Customer Service phone calls. USB was meant to solve those problems: A single connector type, with a bus topology, so anything could be plugged into any available connector. Now PC manufacturers could reduce the number of connectors to perhaps three (supporting keyboard, mouse, and one expansion port that could support a hub if necessary). And these connectors were small, and FLAT (round connectors are a PITA on things like laptop motherboards). Great in concept. Max speed of 11Mbps(?) was ridiculously high for its intended use, being two orders of magnitude faster than what it was replacing. There were (and are!) some problems with low quality drivers that, once installed, would only recognize peripherals if they were always plugged into the same physical USB connector, but overall USB delivered what the industry had been looking for. But since then, USB has been spaghetti-coded into a mess as discussed in this thread. Instead of a simple serial connection scheme, it's now trying to be everything to everyone while maintaining reverse compatibility with all the legacy devices out there. We want it to work with slow devices, but we also want it to work with the fastest devices. We want it to be a source of charging power (or even operational power, measuring in several amperes!). And we're trying to mash all of those new desires onto an existing spec and infrastructure and installed base. That very often doesn't work out well, as anyone who's failed to get a USB1/2 device to work in a USB3.0 port knows all too painfully. And as more laptops (in particular) stop including the older USB ports in favor of all-3.0+ style, reports of "my peripherals don't work with my new computer!" are becoming more common. All of this is because a well conceived, well designed, and successful serial standard has been shoehorned into becoming something it was never intended to be. How often does that really work out well? Fortunately, the industry appears to be realizing this. So far it's an informal effort, but with at least three IC manufacturers investing in it I suspect it has legs. It cannot arrive fast enough IMHO. |
| Navigation |
| Message Index |
| Previous page |