6
\$\begingroup\$

For typical AC to DC power adapters, I understand that the device draws the current it needs from the adapter. If it pulls more than the adapter can supply, this can cause overheating and lead to breaking the adapter.

Following this... I am always charging my phone or tablet on 1A USB chargers or in normal PC USB ports, yet I have never experienced a broken USB port resulting from this.

My question is:

  • Do USB ports or devices typically have a mechanism to regulate current to avoid overloading ports?

I'm mostly thinking in the context of devices which charge via USB.

Improve this question
\$\endgroup\$
4
  • 2
    \$\begingroup\$ What is the “2A device” you are referring to in the body of your question? The phone and tablet? Also, when you say, “If it pulls more than the adapter can supply, this can cause overheating and lead to breaking the adapter.” This is most likely only the case in very cheaply made adapters that don’t have the ability to negotiate power draw, thus working at full capacity, thus burning out. For example, an iPad comes with a 12v adapter but can still be charged by a 5v adapter. Just that 5v adapter will charge it more slowly than the 12v adapter. \$\endgroup\$ – Giacomo1968 Sep 18 '18 at 16:26
  • 5
    \$\begingroup\$ You can't draw more power than the port can supply. (It's possible that the port does supply more than it could safely supply). \$\endgroup\$ – Bergi Sep 18 '18 at 18:20
  • \$\begingroup\$ When I was saying "typical AC to DC power adapters", i meant non-usb adapters. This was unclear on my part, so sorry for that. The root of the question comes from my experiences breaking laptop chargers because I didn't first check the amp supply, and wondering why this principle doesn't apply to (most) USB devices I use. As mentioned by you @JakeGould and in the accepted answer, being aware of the negotiated power draw in USB protocol is the basic info I was missing, as confirming the existence of this mechanism and its specific details is what the question is about. \$\endgroup\$ – RTbecard Sep 18 '18 at 19:33
  • \$\begingroup\$ @Bergi But if a cheap USB power adapter just operates at full capacity, it will definitely die out sooner rather than later. \$\endgroup\$ – Giacomo1968 Sep 18 '18 at 22:40
21
\$\begingroup\$

To be compatible with the original standard, USB devices should not draw more than 100mA (which is plenty to power the logic interface), until they have negotiated with the host, to find out what it can supply. After successful negotiation, they can draw up to 500mA. This is to protect the operation of a 4 port hub, should it be plugged into a PC with all its downstream devices already attached.

Not all USB devices are compliant to the standard, but just draw full current anyway, USB toys commonly do this. Most PCs provide 500mA anyway, so it all generally works.

Dumb power supplies generally hold their data lines in particular states, to signal to the device being charged that they are a power supply, with a certain capability. Later standard revisions allow USB-C and PCs higher currents, and higher voltages (eek!) to be negotiated.

Improve this answer
\$\endgroup\$
3
  • \$\begingroup\$ USB 3.0 devices can negotiate to draw up to 900 mA. \$\endgroup\$ – nekomatic Sep 18 '18 at 15:40
  • \$\begingroup\$ There're also fail cases that can result in excess draw. I haven't tried dismantling it to find the root cause, but after a year of good function, I recently had a USB-A to C cable go bad and trigger excess current draw alarms any time I plugged it into my laptop or docking station. \$\endgroup\$ – Dan Is Fiddling By Firelight Sep 18 '18 at 16:04
  • \$\begingroup\$ *"until they have negotiated with the host, to find out what it can supply. After successful negotiation, they can draw up to 500mA. *" - you repeated here a common misconception. USB device doesn't "negotiate" with host and doesn't "find out what it can supply", it only informs the host of what it needs This info is in device descriptors. The host then either admits the device, or stops enumeration process and rejects the device (in rare cases of bus-powered hubs or low-powered hosts). Power Delivery protocol, being totally independent from USB, is a different thing. \$\endgroup\$ – Ale..chenski May 12 '19 at 18:25
16
\$\begingroup\$

Historically:

  • really old mainboards connect USB power pins to the 5V power rail, with no protection
  • power on by keypress was added, which added a jumper or a BIOS setting that decided whether USB ports would be powered from the standby power or from the regular 5V rail. Since standby power was introduced in ATX, this does not exist on AT mainboards.
  • USB port power control was added to root hubs, allowing the host to turn off power to ports programmatically (with the hub controller switching an external "power" FET). These supply a lot of power to the ports, and have no meaningful protection besides using the FET as a current limiter, which is generally a bad idea for prolonged time as these are not usually cooled.
  • the control FET was later integrated into the root hub as manufacturing processes improved to a point where you could run a few hundred mA through what is otherwise a logic IC. This setup has lower current carrying capability, and shorting USB ports will usually destroy the southbridge IC and/or cause a reboot.
  • current monitoring and emergency shutdown were added to the controllers as well as processes permitted.

There are also some older mainboards that implement current monitoring as discrete components (increasing board cost, but giving robustness), but on consumer boards, expect the cheapest possible approach.

Some modern boards also use the same kind of integrated voltage/current controller that usually provides CPU and chipset power to control other circuits, as these ICs are sufficiently cheap that duplicating the logic around them saves enough engineering effort to make them a good contender to "dumb" FETs. On such boards you'd probably be able to draw exactly 2.000A, but current monitoring and reporting may be limited as communication between the USB root hub and the power controller is just "enable" and "error" signals.

Improve this answer
\$\endgroup\$
5
  • 4
    \$\begingroup\$ +1 for reminding us that USB1.0 and AT motherboards actually got to meet each other in active service for a very brief window. I've never personally seen an AT motherboard with a USB port... what a unicorn those must be! \$\endgroup\$ – J... Sep 18 '18 at 18:25
  • 2
    \$\begingroup\$ @J... Saw a Compaq once with this. It also predated Windows 95 OSR2 which was the first to implement “support” for it. Forget plug-and-play your USB thumb drive! Install drivers and boot with device connected and just maybe it would work. Good old days! \$\endgroup\$ – winny Sep 18 '18 at 21:32
  • \$\begingroup\$ @winny Was this a Compaq Presario 4505 by chance? I had that machine. :-) \$\endgroup\$ – Brad Sep 19 '18 at 2:59
  • 1
    \$\begingroup\$ This is highly inaccurate account of USB port power control. Power on/off control was always in hub controller's ICs, but is optional in USB. Integration of power control FETs into "root hub" is nonsense: power electronics requires vastly different technology and node features to carry hundreds of mAmps, and deep submicron digital technology can't afford mixing. The power control was always implemented in industry as stand-alone high-side switches. This post is mostly misleading. -1. \$\endgroup\$ – Ale..chenski Sep 19 '18 at 3:47
  • \$\begingroup\$ @Brad No, way before that one. Desktop. 100 or 133 MHz. \$\endgroup\$ – winny Sep 19 '18 at 8:16
6
\$\begingroup\$

My question is:

  • Do USB ports or devices typically have a mechanism to regulate current to avoid overloading ports?

The answer is yes, they do.

USB includes a fairly elaborate protocol that allows devices and hosts to negotiate the amount of power that the device can use.

Improve this answer
\$\endgroup\$
1
  • \$\begingroup\$ This is formally incorrect. USB doesn't include any negotiating protocols, only informative. You apparently mean BatteryCharging (BC1.2) or PowerDelivery (PD over Type-C) specifications, they are completely outside the main USB protocol. \$\endgroup\$ – Ale..chenski May 12 '19 at 18:30
3
\$\begingroup\$

Why doesn't my computer USB port break when I plug in a 2A device?

and

Do USB ports or devices typically have a mechanism to regulate current to avoid overloading ports?

Yes, USB ports do have a mechanism, but not to "regulate current", but to protect USB ports from overcurrent. Overcurrent leads to port disconnect, not to "regulate it".

And (not "or") USB devices (with internal batteries) typically have a mechanism called "port signature detection", which makes a USB device not to "regulate current", but "limit current" the device will take from a port. If the port is a regular USB data port, connected devices will limit their consumption to 500 mA (and to 900 mA if the device can detect USB 3.0 port).

That's why your computer ports don't break, because the "2-A device" can detect this, and becomes a "500/900-mA device".

Improve this answer
\$\endgroup\$
2
  • \$\begingroup\$ How well does the overcurrent protection work (lets limit this to usb 2.0 or 3.0 ports on PCs)? Should I view this as "my ports will protect themselves from damage", or should I still exercise some vigilance before plugging in a device that potentially has a high load (high load == 2A for instance) and which I suspect may not be compliant to the USB standard. \$\endgroup\$ – RTbecard Sep 24 '18 at 18:22
  • 2
    \$\begingroup\$ @RTbecard, usually the mainstream PC mainboards protect their USB ports by means of cheap polyfuses (resettable fuses), and more expensive (business-class) PC have ports with overcurrent-protected high-side switches. Usually the cut-off is set for about 0.8-0.9A for USB 2.0 ports, or ~1.5A for USB 3.0 ports. Some boards have "ganged" polyfuse for two or more ports, so you might have 2-3A from a single port. No manufacturer wants to be liable for setting your PC into flames, so they do reasonable work to protect their business from garbage devices and other experimentalists. \$\endgroup\$ – Ale..chenski Sep 24 '18 at 18:43

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.