Cables are rated for current draw at a given voltage.
Yes, but not how you imply! The voltage rating is a rating that relates to the insulation system in the cable. The current rating is a rating that has to do with self-heating of the conductor and the whole cable. The only interrelation between the two is that higher temperature insulation will usually let you pass more current for otherwise identical cable construction (thermal resistance to ambient).
Remember that self-heating power is only due to resistive losses caused by current flow: Power = Resistance * Current^2, it's irrespective of the potential on the individual conductor vs. any other conductor. A piece of wire never "sees" the 5V or 48V applied between it and some other conductor. It only sees the voltage drop across its length. From a perspective of a heating up piece of wire, the potential it's at is immaterial. So in your "V cross A", the "V" is R * Current, not 5V or whatever else is the potential difference between the positive and negative conductors in the power pair.
Since a high-power USB3 system would require a complete redesign of the hub or motherboard, it's not unthinkable that they'd engineer it to take care of the noise. I'm not saying it'd stay mechanically compatible with current cases -- it'd be a terrible idea, in fact.
I don't know where you got the #8 wire idea from, nor the idea that it'd run at 5V. Perhaps the article writers are clueless. I fail to find humor in most of everyday tech article stupidity. No one actually dealing with the spec would ever say something so silly.
Alas, you only need two power wires, so you'd need *two* #8 wires, not four. But even so, the current capacity is not everything. The wires must also have resistance low enough to maintain a voltage drop on the order of 0.05V per conductor (0.1V per cable), to have another 0.05V or so left for each of the connector pairs, for a total drop of ~0.2V.
The #8 wire has 2mOhm/m resistance. At 3m (a reasonable max for USB2), that's 12mOhm total per cable, and at 20A that gives you 0.24V drop -- about 2.5 times too much. So, you see, you need to be accurate for shit like this to be humorous -- humor in such technical discussions usually stems from people knowing WTF they're talking about. Had you said "Imagine a 3m long USB cable with two #4 wires", THEN I'd have thought "hey, he knows this shit". That would be hilarious. As it is, I fail to see humor in mistake upon mistake in the whole discussion. Sorry.
I think that due to power losses at interconnects, etc, the industry will develop a way of keeping it at 48V. Initially the motherboards may have step-up converters, those will consume about 10A @ 12V due to realistic estimate of 85% converter efficiency. In the long term, the industry will simply add an extra 48V contacts to a new motherboard power connector standard. It's also high time, methinks, that laptops switched to 48V supplies, ideally coming from a standardized brick.
If you think they'll be pushing 100W at 5V you're just plain silly. Stop it. Please. It'll be 48V, because that's the only choice that keeps it within limits for low voltage, and that's what you need without totally overhauling the USB connector. The voltage will be dynamically switched after standard enumeration at 5V -- that's the only sane design choice they have. Longer USB cables will gladly accept 2.1A without overheating. In fact, any USB3 cable at least 2m long must have #24 (24 AWG) power conductors to limit the puny voltage drop allowed for 5V, 900mA operation. Of course the host can't know if you have such a cable, so the cables will probably have some ID built-in, probably a resistor (not unlike PoE devices).
The cables are not rated for watts, they are rated for working voltage and current. 48V is within insulation ratings of all USB3 cables AFAIK. The current in existing USB3 spec is 900mA. If you'd push 48V at that current, you get 43W. You'd need a cable with a beefier conductor to use at 2.1A needed for 100W @ 48V. All you need is a longer, still standard, USB3 cable. Of course it's hard to enforce that, but cable-wise, 5m cables at 20AWG provide a 6x larger cross-sectional surface area than shortest cables at 28AWG. The longer cables use larger conductors to meet the minuscule voltage drops expected in USB3 operation at 5V.
Of course it'd be hard to enforce this without some special cable identification, but perhaps they'll figure something out. I'd think that the simplest solution may be to have a resistor between the power and ground wires added to one of the plugs. The USB host would measure this resistance differentially at voltages low enough to prevent startup of the target device (say at 0.1 and 0.2V). That's all it'd take. The existing plugs/connectors will survive 2A flowing through the power pins, and longer existing USB3 cables will too, so it's not a matter of reinventing the wheel, just an incremental improvement,
For all I know, a USB3 100W device will enumerate while powered via 5W, and only after it's confirmed, the output voltage will be raised, as well as output current limit.
The idea is, I believe, to save a bit of electronic waste and have a single line-interfacing power supply for desktop systems. This makes a lot of sense, and I applaud any efforts towards this end. The desktops will be designed in for a single 100W device powered via USB. I'd think a 150W power budget for USB devices available on the desktop will accommodate all of the wallwart powered stuff people typically have on their desktops, with exception of laser printers. A rather zippy and simple LJ P1006 consumes ~320W per my kill-a-watt while it prints.
While technically you could implement it as 100W per port, any affordable host will probably limit the overall USB power consumption to 150W or so, across all outputs. That lets you power a monitor, a hefty external hard drive, and a few small devices like mice, keyboards, flash drives. Power management is called just that for a reason, you know...
This article wins at making people jump to silly conclusions. I'm bookmarking it and will reference it every time someone asks why engineering is hard. It's "hard" because even slashdot posters often can't see the forest for the trees. All of the "problems" raised by various commenters here show obliviousness to everyday, consumer-deployed solutions that have been with us for a decade or more. Sigh.
Ever saw what connectors are used for PoE? RJ-45, no less. And that provides up to 90W in most recent incarnations. So obviously, if you choose silly numbers, you'll get silly results. I claim that they will spec cable and connectors for 2.2A, and the voltage will be negotiable up to 48V.
There's so much silliness in the comments on this article, it's beyond me.
Nope. It won't be 110V as that's a whole different ballgame when it comes to motherboard, connector and cabling design. It needs to be low voltage, and that means it will top out at 48V.
You could have gone with a small USB-toting microcontroller, probably not much more expensive than your ATTiny 13, and got the device to enumerate and switch to 500mA current limit.
PoE is nowhere to be found? WHAT? Even I'm designing instrumentation (transducers) that uses PoE, every IP phone in my workplace runs on PoE. A decent HP 2626-PWR switch with 24 PoE ports sells on eBay for $300 BIN. It's hardly expensive IMHO.
Nobody in their sane mind would do 100W at 5V for a consumer standard like USB3. The USB cables would probably sell at $50 in retail, and that'd be a sale price. Connectors that can pass 20A with a very small voltage drop even after years of use are quite expensive. My bet is that they'll ramp up the voltage as needed by the load, up to 48V. 100W at 48V is ~2.1A, and that's reasonable -- merely 4x the present limit of 0.5A.
Fire hazard? Huh? You seem like you know just about enough to be dangerous, but really just spew meaningless babble. You work in marketing?
My friend's gaming laptop has a 120W power brick. Nobody gives a shit. The supply provides a limited current, and the low-voltage cable has properly sized conductors and connector. That's all there's to it.
I can't see how providing a 100W, power-limited output from a USB port will cause a problem. The cabling obviously has to be designed to cope with the rated current of such a supply. Nobody would supply 100W over a 5V interface. Most likely USB3 will limit the current on the power pair to some sensible value, and then the power is upped by raising the voltage. To keep the cables affordable, I'm pretty sure that 100W will be provided at 48V, and that means about 2.1A of current. That's merely 4 times more current than the USB2 cables carry, and it means that the cross-sectional area of power conductors has to go up by the same factor. The diameter goes up by the square root of that. So the power pair will have conductors about twice as thick as the ones in the current USB cables, and perhaps the insulation will be a bit thicker, too. The overall diameter of a USB3 cable will be just a tad larger than current decent USB2 cables.
It had nothing to do with 5V, nor with the port being broken. It was an issue with electrical wiring (lack of proper PE - Protective Earth a.k.a. "ground"), most likely. Alternatively, there was no PE connection at all, and you were shunting power supply's leakage current to ground. Most PC power supplies have filtering capacitors between the case and the Live and Neutral conductors. Those capacitors form a voltage divider that puts the case at 50% of live voltage in absence of PE connection, that's the source of the leakage current.
Even if the wall is made of cement blocks, it should only take a good chisel and a 4lb hammer to get through. Perhaps if you're in shape a 6lb hammer will make the job quicker, but I don't recommend it if you don't use it regularly. Once you get two blocks out, the rest will be like eating cheesecake: smooth and easy goin'. Brick walls are easier once you start, but may be harder to break through the first brick or two. If there's two of you -- to start let one hold the chisel, while the other one uses an 8lb long demolition hammer. The bricks will pop right out. Yes, I've done some deconstruction...
Going through drywall can be pretty much noise-less. All you need is a good cast metal Stanley knife handle for W-shape blades, and a few spare blades. Score, cut through, remove. For cast prefab plaster walls (saw them in Europe in many places), the knife still works. Only when you face lath it's harder to keep it quiet.
A 6lb maul? You joking? I have an 8lb demolition hammer, and I wished I had something bigger when doing a rather "simple" remodel of a room and demolition of a deck. 8lb was barely enough to get a slightly curvy 6.5' 2x10 header in place...
I've seen plenty of doors where even a 24lb demolition hammer would perhaps dent them and scratch the paint, and not much else. Since I had to replace the front doors on my house, I did try the 8lb hammer on them. By my estimate, it'd take me half a day of pounding and sweating to get through. I would probably demolish the block wall those doors were mounted in before ripping the doors open. And those seem to be standard commercial steel entry doors. Not the cheap residential stuff, but nothing specifically designed for highly secure areas either.
I agree about Qt. Have been using it seriously since 2.x times.
The TRS-80 BASIC, and most other basics of that era, did not do any sort of JIT. It didn't even do what Python does -- no bytecode interpretation. The level 1 BASIC was a port of Tiny Basic, I remember that I first saw TB in Dr. Dobb's Journal of Computer Calisthenics and Orthodontia that my dad got his hands on -- behind the Iron Courtain, too. This was IIRC an untokenized, on-the-fly interpreter. The level 2 BASIC was MS BASIC, and it too doesn't use a bytecode, just tokens. There is a difference between bytecode and tokens: the bytecode represents assembly-type operations on a low-level virtual machine of some sort. The tokens are simply a more compact representation of the source code, and are fairly high level. The tokenized stream is fed to a parser that eventually calls upon the evaluator with the abstract syntax expression tree of some sort. Most of this is constructed on the fly, since the memory is tight.
Luxor's ABC-800 had a very fast BASIC that did use bytecodes, and was much closer to today's Python than other BASICs of the time. I have a bunch of ABC-802s and their BASIC is faster than MS-BASIC was on PC-XT.
Technically -- maybe, but in practice it works like this:
1. Install TwinCat 2. Reboot 3. Fire up the UI, set up the PLC runtime to run your code. 4. Done
You're still in Windows, everything else works the same. The user experience is completely Windows-centric, you're not aware of something going on behind the scenes unless you read about it in the documentation.
I'd say you eat both cheese and ham, with cheese under the ham. Ham took over.
Slashdot Mirror
Cables are rated for current draw at a given voltage.
Yes, but not how you imply! The voltage rating is a rating that relates to the insulation system in the cable. The current rating is a rating that has to do with self-heating of the conductor and the whole cable. The only interrelation between the two is that higher temperature insulation will usually let you pass more current for otherwise identical cable construction (thermal resistance to ambient).
Remember that self-heating power is only due to resistive losses caused by current flow: Power = Resistance * Current^2, it's irrespective of the potential on the individual conductor vs. any other conductor. A piece of wire never "sees" the 5V or 48V applied between it and some other conductor. It only sees the voltage drop across its length. From a perspective of a heating up piece of wire, the potential it's at is immaterial. So in your "V cross A", the "V" is R * Current, not 5V or whatever else is the potential difference between the positive and negative conductors in the power pair.
Since a high-power USB3 system would require a complete redesign of the hub or motherboard, it's not unthinkable that they'd engineer it to take care of the noise. I'm not saying it'd stay mechanically compatible with current cases -- it'd be a terrible idea, in fact.
I don't know where you got the #8 wire idea from, nor the idea that it'd run at 5V. Perhaps the article writers are clueless. I fail to find humor in most of everyday tech article stupidity. No one actually dealing with the spec would ever say something so silly.
Alas, you only need two power wires, so you'd need *two* #8 wires, not four. But even so, the current capacity is not everything. The wires must also have resistance low enough to maintain a voltage drop on the order of 0.05V per conductor (0.1V per cable), to have another 0.05V or so left for each of the connector pairs, for a total drop of ~0.2V.
The #8 wire has 2mOhm/m resistance. At 3m (a reasonable max for USB2), that's 12mOhm total per cable, and at 20A that gives you 0.24V drop -- about 2.5 times too much. So, you see, you need to be accurate for shit like this to be humorous -- humor in such technical discussions usually stems from people knowing WTF they're talking about. Had you said "Imagine a 3m long USB cable with two #4 wires", THEN I'd have thought "hey, he knows this shit". That would be hilarious. As it is, I fail to see humor in mistake upon mistake in the whole discussion. Sorry.
SQ, you made my day. Thank you!
a USB3 100W device will enumerate while powered via 5W
I meant to say 5V, not 5W, duh.
I think that due to power losses at interconnects, etc, the industry will develop a way of keeping it at 48V. Initially the motherboards may have step-up converters, those will consume about 10A @ 12V due to realistic estimate of 85% converter efficiency. In the long term, the industry will simply add an extra 48V contacts to a new motherboard power connector standard. It's also high time, methinks, that laptops switched to 48V supplies, ideally coming from a standardized brick.
If you think they'll be pushing 100W at 5V you're just plain silly. Stop it. Please. It'll be 48V, because that's the only choice that keeps it within limits for low voltage, and that's what you need without totally overhauling the USB connector. The voltage will be dynamically switched after standard enumeration at 5V -- that's the only sane design choice they have. Longer USB cables will gladly accept 2.1A without overheating. In fact, any USB3 cable at least 2m long must have #24 (24 AWG) power conductors to limit the puny voltage drop allowed for 5V, 900mA operation. Of course the host can't know if you have such a cable, so the cables will probably have some ID built-in, probably a resistor (not unlike PoE devices).
The cables are not rated for watts, they are rated for working voltage and current. 48V is within insulation ratings of all USB3 cables AFAIK. The current in existing USB3 spec is 900mA. If you'd push 48V at that current, you get 43W. You'd need a cable with a beefier conductor to use at 2.1A needed for 100W @ 48V. All you need is a longer, still standard, USB3 cable. Of course it's hard to enforce that, but cable-wise, 5m cables at 20AWG provide a 6x larger cross-sectional surface area than shortest cables at 28AWG. The longer cables use larger conductors to meet the minuscule voltage drops expected in USB3 operation at 5V.
Of course it'd be hard to enforce this without some special cable identification, but perhaps they'll figure something out. I'd think that the simplest solution may be to have a resistor between the power and ground wires added to one of the plugs. The USB host would measure this resistance differentially at voltages low enough to prevent startup of the target device (say at 0.1 and 0.2V). That's all it'd take. The existing plugs/connectors will survive 2A flowing through the power pins, and longer existing USB3 cables will too, so it's not a matter of reinventing the wheel, just an incremental improvement,
For all I know, a USB3 100W device will enumerate while powered via 5W, and only after it's confirmed, the output voltage will be raised, as well as output current limit.
The idea is, I believe, to save a bit of electronic waste and have a single line-interfacing power supply for desktop systems. This makes a lot of sense, and I applaud any efforts towards this end. The desktops will be designed in for a single 100W device powered via USB. I'd think a 150W power budget for USB devices available on the desktop will accommodate all of the wallwart powered stuff people typically have on their desktops, with exception of laser printers. A rather zippy and simple LJ P1006 consumes ~320W per my kill-a-watt while it prints.
I've seen plenty of high-power USB on POS systems in the U.S. Sure, there are serial links there, too.
While technically you could implement it as 100W per port, any affordable host will probably limit the overall USB power consumption to 150W or so, across all outputs. That lets you power a monitor, a hefty external hard drive, and a few small devices like mice, keyboards, flash drives. Power management is called just that for a reason, you know...
This article wins at making people jump to silly conclusions. I'm bookmarking it and will reference it every time someone asks why engineering is hard. It's "hard" because even slashdot posters often can't see the forest for the trees. All of the "problems" raised by various commenters here show obliviousness to everyday, consumer-deployed solutions that have been with us for a decade or more. Sigh.
Ever saw what connectors are used for PoE? RJ-45, no less. And that provides up to 90W in most recent incarnations. So obviously, if you choose silly numbers, you'll get silly results. I claim that they will spec cable and connectors for 2.2A, and the voltage will be negotiable up to 48V.
There's so much silliness in the comments on this article, it's beyond me.
Nope. It won't be 110V as that's a whole different ballgame when it comes to motherboard, connector and cabling design. It needs to be low voltage, and that means it will top out at 48V.
True, but misleading. USB1.0 does nothing less than USB2.0 in that respect.
You could have gone with a small USB-toting microcontroller, probably not much more expensive than your ATTiny 13, and got the device to enumerate and switch to 500mA current limit.
PoE is nowhere to be found? WHAT? Even I'm designing instrumentation (transducers) that uses PoE, every IP phone in my workplace runs on PoE. A decent HP 2626-PWR switch with 24 PoE ports sells on eBay for $300 BIN. It's hardly expensive IMHO.
Nobody in their sane mind would do 100W at 5V for a consumer standard like USB3. The USB cables would probably sell at $50 in retail, and that'd be a sale price. Connectors that can pass 20A with a very small voltage drop even after years of use are quite expensive. My bet is that they'll ramp up the voltage as needed by the load, up to 48V. 100W at 48V is ~2.1A, and that's reasonable -- merely 4x the present limit of 0.5A.
Current USB has 500mA current limit at a fixed voltage of 5V. That's 2.5W.
USB3 won't have a 4.5W limit, where did you get that.
Niche: you mean possibly millions of POS systems :)
Fire hazard? Huh? You seem like you know just about enough to be dangerous, but really just spew meaningless babble. You work in marketing?
My friend's gaming laptop has a 120W power brick. Nobody gives a shit. The supply provides a limited current, and the low-voltage cable has properly sized conductors and connector. That's all there's to it.
I can't see how providing a 100W, power-limited output from a USB port will cause a problem. The cabling obviously has to be designed to cope with the rated current of such a supply. Nobody would supply 100W over a 5V interface. Most likely USB3 will limit the current on the power pair to some sensible value, and then the power is upped by raising the voltage. To keep the cables affordable, I'm pretty sure that 100W will be provided at 48V, and that means about 2.1A of current. That's merely 4 times more current than the USB2 cables carry, and it means that the cross-sectional area of power conductors has to go up by the same factor. The diameter goes up by the square root of that. So the power pair will have conductors about twice as thick as the ones in the current USB cables, and perhaps the insulation will be a bit thicker, too. The overall diameter of a USB3 cable will be just a tad larger than current decent USB2 cables.
USB is not really CPU-intensive. The host chip does most of the hard work.
It had nothing to do with 5V, nor with the port being broken. It was an issue with electrical wiring (lack of proper PE - Protective Earth a.k.a. "ground"), most likely. Alternatively, there was no PE connection at all, and you were shunting power supply's leakage current to ground. Most PC power supplies have filtering capacitors between the case and the Live and Neutral conductors. Those capacitors form a voltage divider that puts the case at 50% of live voltage in absence of PE connection, that's the source of the leakage current.
Very nice. Thanks for the link.
Even if the wall is made of cement blocks, it should only take a good chisel and a 4lb hammer to get through. Perhaps if you're in shape a 6lb hammer will make the job quicker, but I don't recommend it if you don't use it regularly. Once you get two blocks out, the rest will be like eating cheesecake: smooth and easy goin'. Brick walls are easier once you start, but may be harder to break through the first brick or two. If there's two of you -- to start let one hold the chisel, while the other one uses an 8lb long demolition hammer. The bricks will pop right out. Yes, I've done some deconstruction...
Going through drywall can be pretty much noise-less. All you need is a good cast metal Stanley knife handle for W-shape blades, and a few spare blades. Score, cut through, remove. For cast prefab plaster walls (saw them in Europe in many places), the knife still works. Only when you face lath it's harder to keep it quiet.
A 6lb maul? You joking? I have an 8lb demolition hammer, and I wished I had something bigger when doing a rather "simple" remodel of a room and demolition of a deck. 8lb was barely enough to get a slightly curvy 6.5' 2x10 header in place...
I've seen plenty of doors where even a 24lb demolition hammer would perhaps dent them and scratch the paint, and not much else. Since I had to replace the front doors on my house, I did try the 8lb hammer on them. By my estimate, it'd take me half a day of pounding and sweating to get through. I would probably demolish the block wall those doors were mounted in before ripping the doors open. And those seem to be standard commercial steel entry doors. Not the cheap residential stuff, but nothing specifically designed for highly secure areas either.
I agree about Qt. Have been using it seriously since 2.x times.
The TRS-80 BASIC, and most other basics of that era, did not do any sort of JIT. It didn't even do what Python does -- no bytecode interpretation. The level 1 BASIC was a port of Tiny Basic, I remember that I first saw TB in Dr. Dobb's Journal of Computer Calisthenics and Orthodontia that my dad got his hands on -- behind the Iron Courtain, too. This was IIRC an untokenized, on-the-fly interpreter. The level 2 BASIC was MS BASIC, and it too doesn't use a bytecode, just tokens. There is a difference between bytecode and tokens: the bytecode represents assembly-type operations on a low-level virtual machine of some sort. The tokens are simply a more compact representation of the source code, and are fairly high level. The tokenized stream is fed to a parser that eventually calls upon the evaluator with the abstract syntax expression tree of some sort. Most of this is constructed on the fly, since the memory is tight.
Luxor's ABC-800 had a very fast BASIC that did use bytecodes, and was much closer to today's Python than other BASICs of the time. I have a bunch of ABC-802s and their BASIC is faster than MS-BASIC was on PC-XT.
Technically -- maybe, but in practice it works like this:
1. Install TwinCat
2. Reboot
3. Fire up the UI, set up the PLC runtime to run your code.
4. Done
You're still in Windows, everything else works the same. The user experience is completely Windows-centric, you're not aware of something going on behind the scenes unless you read about it in the documentation.
I'd say you eat both cheese and ham, with cheese under the ham. Ham took over.