251–260 of 544 posts
> Making USB reversible to begin with would have necessitated twice as many wires and twice as many circuits, and would have doubled the cost. This is... not correct. You can make USB reversible with 1 extra pin and 1 extra wire. Grounds on pins 1 and 5, data on pins 2 and 4, and VCC on pin 3. Then have those pins on both sides of the plug and a socket with a single set of contacts on one side. That's BASICALLY what…
> Making USB reversible to begin with would have necessitated twice as many wires and twice as many circuits, and would have doubled the cost. This is... not correct. You can make USB reversible with 1 extra pin and 1 extra wire. Grounds on pins 1 and 5, data on pins 2 and 4, and VCC on pin 3. Then have those pins on both sides of the plug and a socket with a single set of contacts on one side. That's BASICALLY what…
Earlier quoted context omitted.
> All they had to do was make the connector have a non-symmetrical shape so that it's immediately obvious which way round it goes when you pick it up - you could do it without even looking. Think of how much time we'd have collectively saved with this minor design change. I disagree. When connecting an HDMI cable I sometimes have issues, especially if in a weird angle. However I do concede is faster to connect an HDM…
Yeah, I want the connector that can be attached by only moving the TV a few inches, just enough to get my hand to fit and feel around, not something I have to rearrange the furniture for and break out some headlight to see which direction the cable is oriented. With BNC cables, I could do it with my eyes closed.
My take on USB-A: aside from audio connectors that have axial symmetry (and have no meaningful orientation), I never used any reversible computer cables prior to USB-C, and that’s not the main issue. Nor is it the axis of symmetry of the casing allowing failed upside down plugin attempts. The shitty part of USB-A is that the tactile feedback of being slightly misaligned is identical to it being upside down. My experi…
The floating tongue of the USB connection is in the socket, not the on the cable. As in the most fragile part of the design is on the device side, not the cable side. This means that you can more easily break your $1000 device not the $2 cable. They repeated this mistake with USB-C as well.
It's not hard to make a port where the cable is the free floating tongue and the device is a more robust socket that wraps around that fragile piece. I know everyone's happy about the iPhone moving to USB-C but the tongue on the cable side that it had was much better. Anyone who's tripped on a cable and broken a USB socket can attest to this.
More interesting question for me is what exactly made USB possible? I understand that parallel communication wasn't possible because even for small differences in wire length, data on each wire will arrive at different times and that limits speed. So, people went with serial communication. But what exactly enabled high speed bit transfer? Why wasn't it possible before? Similarly, why USB2 speeds weren't possible befo…
I would argue that what USB (1.1-2.0) does differently from previous serial peripheral ports is mostly software and standardization, and really relate to making it cheap and simple for "normal people" to use:
1. USB 1.1 supports only two bit rates (1.5 and 12 Mbps), which is autoconfigured before device enumeration. 12 = 8 * 1.5, so the clock divider is cheap and easy.
2. USB limits cables to fairly short lengths (3. USB (pre-OTG extensions) rigorously enforced the idea of a "host" (upstream) and "device" (downstream), at both the protocol and physical connector level (which greatly simplifies things--for example, you can't create a loop, and don't need STP to detect it). A child can easily see that a B (device) socket doesn't fit an A (host) plug.
4. USB device enumeration and configuration are extensively software based, and USB defines a number of standard device classes so that many common types of devices (e.g., keyboards, mice) don't need specific drivers.
5. One thing that USB does that's less common among serial peripheral interfaces is to use a single differential pair for data going in both directions (it's time-shared: so the host polls the device then the device responds). The pin-count is less than a "regular" RS-422 port, but you still get the advantages of differential signaling.
6. USB carries power with only one more pin than needed to create a minimal bidirectional serial data connection, so "lite" devices don't need a separate power connector (ignoring very slow protocols like "1-wire").
7. USB does some funny things with packet framing (like NRZI encoding and bit-stuffing) and some things to help reduce device cost (like the JKJKJK packet preamble to sync the device baud-rate generator and using SE1/SE0 states for device disconnect and bus reset signaling), but none of that is really fundamental to making a 10Mbps-class serial interface.
Despite the obvious design drawback, I can't blame intel entirely, a big contributor to the problem was that the ports were still being positioned in hard to access areas by computer makers. The first iMac design was a bit more forward thinking in this regard, whereby USB ports were prominently positioned on either side of the keyboard, this made it very easy to plug in the USB mouse and use the other for something like a USB stick. The problem with the type A design isn't just the orientation issue, the rectangular port design provides very little tolerance for off-angle insertion, having the port in full view helped a lot with insertion.
I know everybody is still on the high from USB-C but I'm still going to go on the record it will be one of the biggest disasters in cable history. 1) There's going to be video-only cables; low-voltage only cables; etc and this time everything is usb-c so you can literally only tell which ables work by testing all of the dozen cables instead of the 1-2 USB-As you have. 2) The connector is symmetrical but the pins aren…
3) The most fragile part (the wafer in the center of the receptacle) is on the expensive-to-repair device, while the least fragile part is on the cheap-to-replace cable. HDMI makes this same mistake. With USB-A, at least the plastic part on the device is relatively robust, and with Lightning, both plug and socket were rather robust (albeit prone to debris incursion). With older pin-type connectors like VGA and DVI, t…
Earlier quoted context omitted.
> Then you implement auto crossover detection, dwhich had been around for years Mdi-x/auto crossover for ethernet was introduced in 1999, that is after USB (1996) was designed, not sure of the technique was known at that time
There were auto-crossover RS-232 boxes going back deep into the 1980s.
> Making USB reversible to begin with would have necessitated twice as many wires and twice as many circuits, and would have doubled the cost. This is... not correct. You can make USB reversible with 1 extra pin and 1 extra wire. Grounds on pins 1 and 5, data on pins 2 and 4, and VCC on pin 3. Then have those pins on both sides of the plug and a socket with a single set of contacts on one side. That's BASICALLY what…
>edit: Gah! you don't even have to do that just flip the flipping wires) Isn't that essentially what all ethernet cables do? it's not like this something never done before.
> Making USB reversible to begin with would have necessitated twice as many wires and twice as many circuits, and would have doubled the cost. This is... not correct. You can make USB reversible with 1 extra pin and 1 extra wire. Grounds on pins 1 and 5, data on pins 2 and 4, and VCC on pin 3. Then have those pins on both sides of the plug and a socket with a single set of contacts on one side. That's BASICALLY what…
There are also reversible USB plugs that are a single "tongue", inheriting that design from USB drives that don't have any plug to speak of but look more like a card-edge connector: https://www.pcgamer.com/youre-telling-me-we-could-have-had-r... https://2b.com.eg/en/2b-cv177-cable-usb-type-a-reversible-pl...