[Pxtl]

To me, the biggest failing of micro-USB isn't the up/down thing, but its inability to back-power the same port that is being used as a host. That is, if I have a device with one port, I can't plug that device into a hub where it hosts devices and gets recharged at the same time. This is becoming an obvious flaw in the day of tablets-as-laptops - you can't make a "docking station" using a single USB port. Meanwhile, the iPod has been sporting docking stations that provide both audio/video out and charging since before the smartphone war.

[laumars]

Sure you can. USB has a dedicated pin for sending power down and it's been an integral part of the design since the start (USB 1.0). How do you think USB pen drives (et al) work when they have chips which need powering? Or USB "soundcards", wireless dongles, USB keyboards and mice? They all need powering and they all need powering while in operation.

You might not be able to send higher charges while using data - I've not followed the spec that closely - but you can supply at least enough charge to keep the phone running even if you cannot recharge it.

The reason you don't get docking stations for other phones is because every sodding phone puts their USB socket in a different place (and often it changes from model to model from the same manufacturer). This means whatever dock you build will automatically be fugly compared to iPhone/iPod docks because you'll need a USB cable. Which is why any such dock will either have a 3.5mm audio jack (TRS) and thus support any portable device, or use bluetooth for a wireless solution.

[mbell]

I think you're missing the issue here. The examples you've given are all with the 'host' powering the 'device'. e.g. Your computer powering your USB keyboard. 'Host' and 'Device' are specific modes of operation in the USB spec, it isn't multi-master.

The problem is when you want the 'device' to power the 'host', as is often the case when the 'host' is something with a battery, such as a tablet. There is no support built into the USB spec for this, it is always assumed that the 'host' provides power to the 'device'. An example of something that isn't supported would be a tablet 'dock' with a bunch of device connectors that could also charge the tablet, all via a single USB link.

[laumars]

> I think you're missing the issue here. The examples you've given are all with the 'host' powering the 'device'. e.g. Your computer powering your USB keyboard. 'Host' and 'Device' are specific modes of operation in the USB spec, it isn't multi-master.

That's the only way you'd ever want to power a USB device though. It makes no sense to power the host from the device. Plus this isn't even the case with iPod docks - they're all powered in the mains.

> The problem is when you want the 'device' to power the 'host', as is often the case when the 'host' is something with a battery, such as a tablet.

The host on an iPod dock isn't the iPod; it's the dock.

> There is no support built into the USB spec for this, it is always assumed that the 'host' provides power to the 'device'. An example of something that isn't supported would be a tablet 'dock' with a bunch of device connectors that could also charge the tablet, all via a single USB link.

In this example, the dock is the host and the tablet is the device. So the tablet would still be charged.

The thing with USB is, it doesn't matter which device is the server and which is the client at the software end as that's just an arbitrary software paradigm. You can have the USB host act as the client at the software end and the device as the sever (eg a music dock will be the host, but the phone is the server and the dock the client), all you need is the two devices to be able to speak to each other. And since USB is a two way protocol (else USB storage wouldn't work), it's pretty easy to have the host as the power source and the device as the item requiring USB power even when the device is the one sending signals back to the host (eg how USB mice and keyboards work).

[mbell]

> That's the only way you'd ever want to power a USB device though. It makes no sense to power the host from the device. Plus this isn't even the case with iPod docks - they're all powered in the mains.

You're still not getting it. I want the dock to power the tablet so it charges, the dock gets power from mains does not change that at all.

> The host on an iPod dock isn't the iPod; it's the dock.

You can't make this statement definitively. If one of the features of the dock is a keyboard, it's going to be acting as a HID interface device and will be acting in device mode....

> The thing with USB is, it doesn't matter which device is the server and which is the client at the software end as that's just an arbitrary software paradigm.

This is wrong. A 'device' can never initiate a transfer, all transfers are negotiated by the 'host'. The 'host' is the bus master and manages all transfers regardless of what they are or where they are going. e.i. USB is a 'polling' interface, not a 'pushing' interface. You can not have more than 1 host on a USB bus.

[laumars]

> You're still not getting it. I want the dock to power the tablet so it charges, the dock gets power from mains does not change that at all.

Actually it's you who's missing the point. I've already told you that that does happen, it's just some handsets require a higher charge than others - and docks will only ever output the lowest required volts/amps because they wouldn't want to damage other hardware that is designed to pull less power. IIRC iPod/iPhone users had similar problems when Apple changed the input power rating for newer devices a few years ago.

> You can't make this statement definitively. If one of the features of the dock is a keyboard, it's going to be acting as a HID interface device and will be acting in device mode....

lol How many music docks act as a keyboard? Plus there's a contingency for that in the spec anyway as you can run more than one USB interface concurrently over a single USB port.

> This is wrong. A 'device' can never initiate a transfer, all transfers are negotiated by the 'host'. The 'host' is the bus master and manages all transfers regardless of what they are or where they are going. e.i. USB is a 'polling' interface, not a 'pushing' interface. You can not have more than 1 host on a USB bus.

You're missing the point. It's just a software specification. Once the devices have linked, you can programmatically design the software to work whichever way you want. As I pointed out before, there's plenty of USB devices that behave this way already; devices that -from the user perspective- both initiate and receive data.

[simias]

Isn't that limited at 0.5 amperes @5V? I believe it's not enough for all devices (or to charge them quickly enough).

I believe there's a hack in the apple products that can be used to send more current through USB but it's not standard. I might be wrong though, it's been a while since I looked into that.

Also, it's the reason why certain devices (notably external hard drives alimented through USB) come with two USB connectors in a Y configuration to be able to leech enough current from the host. At least that's how I think it works.

[mbell]

As a baseline it's limited to 100mA @ ~5V (150mA for 3.0). Devices can request additional power from the host, 500mA max for 1.X/2.X or 900mA max for 3.0. The host isn't required to be able to supply this much current, it can say 'no'. There are also some quasi standards and I believe now a real standard now for 'USB charging' which allows higher current (and voltage I believe).