0
\$\begingroup\$

In trying to create a device that provides a USB Hub (Microchip USB5537B) I noticed the evaluation schematic and documentation both reference the use of USB port power controllers (AP2511A in the schematic). All the downstream circuits would be high-speed USB2 for my application (upstream connection for the hub is USB3).

I've implemented USB2 before without a hub and I got away with just connecting the power directly to the 5V rail and haven't run into problems. What's the advantage of using port power controllers? Is it required for some devices?

\$\endgroup\$
2
  • \$\begingroup\$ Using USB3 hub silicon for USB2 devices is a meaningless waste, having USB3 upstream won't give you any advantage over a cheaper USB2 hub IC. \$\endgroup\$ Nov 8, 2021 at 16:56
  • \$\begingroup\$ Not sure where did you get the information about the devices inside the Intel RealSense box to be USB2. The realsense uses custom ASIC with upstream USB3.0 Gen1 interface, and 5 MIPI channels to handle video cameras, plus else. USB3 has no defined mechanisms of "transaction translation" unlike USB2 is doing for FS/LS devices. So each USB3 cable has a separate SINGLE USB2 interface, so all USB2 traffic has to split the 480Mbps bandwidth. \$\endgroup\$ Nov 9, 2021 at 1:28

3 Answers 3

1
\$\begingroup\$

What's the advantage of using port power controllers? Is it required for some devices?

One advantage is the protection against short circuits or overloads that Tom mentioned in his answer.

Another advantage is to prevent the fight between two different 5V supplies when, for example, one of the downstream ports of the hub is connected to another upstream port such as a PC. In this case, the 5 VDC coming from your circuit and the 5 VDC coming from the PC will be shorted and this may lead to serious problems.

It's also a good practice to implement a controlled switch for the downstream ports when there's a risk of a sudden overvoltage. Keeping the power to the downstream ports off until the supply gets stabilized prevents possible damages to permanently attached downstream devices. This is possible during the power up of the hub if it is supplied from a buck converter which is turned on and off a few times a day.

\$\endgroup\$
1
\$\begingroup\$

Think of a (faulty) downstream USB device suddenly drawing 30A (or simply a short circuit). Best case is your 5V will simply shut down. You want to protect your upstream power supply from such behavior.

\$\endgroup\$
0
\$\begingroup\$

What's the advantage of using port power controllers?

In normally-designed hubs (with no power control ICs) each downstream facing port is usually protected from overcurrent by cheap resettable polyfuses. And the short condition is usually fed into hub controller logic, which informs the upstream host of overcurrent situation, just as a separate IC will do. So using port power controller IC does not give you much advantage here, except a better/faster speed of recovery.

Regarding overvoltage and VBUS conflicts, usual port power switches do not protect from back flow, and the only "protection" they can provide is their body diode junction, giving some room of about 0.7 V for voltage mismatches. So the separate IC does not have much advantage here as well.

The main advantage of using individual port power controller in hubs is that when the hub is reset (by HW or SW), all power on all ports gets disabled, so any downstream devices will see true disconnect (no VBUS). When the hub ports will start enumerating by host, one-by-one, each downstream device will get clean connect signaling, and very likely will respond properly to default enumeration process. In addition, the power-on will be sequential, one port at a time, so there will be less surges in case of many high-power bus-powered devices attached. So the main advantage of having port power ICs on downstream ports is vastly improved system robustness to power-on-off cycling and brown-out conditions. And a better (instant) recovery from overload (while full recovery of polyfuse can take days). In addition, in theory you can better control power consumption of the system by selectively disabling ports and thus cutting power to bus-powered devices.

All the downstream circuits would be high-speed USB2 for my application

Using USB3 hub silicon for USB2 devices is a meaningless waste. Having USB3 upstream won't give you any advantage over a cheaper USB2 hub IC. This is true because USB3 hub has actually a true superspeed path, plus a slap-on USB2 hub. If downstream devices are solely USB2, the USB3 logic will be disabled/shut down, and does not play any role in data flow.

\$\endgroup\$
4
  • \$\begingroup\$ Where can I read more about "full recovery of polyfuse can take days" ? \$\endgroup\$
    – Ben Voigt
    Nov 8, 2021 at 18:13
  • \$\begingroup\$ @BenVoigt, web.archive.org/web/20150122051210/http://www.te.com/content/… \$\endgroup\$ Nov 9, 2021 at 1:08
  • \$\begingroup\$ @Ale..chenski The combined bandwidth of the 7 downstream USB devices exceeds the bandwidth of USB2. The devices support USB2. Am I wrong that a USB3 hub would be able to handle the combined bandwidth with USB3 upstream and just USB2 downstream? \$\endgroup\$ Nov 9, 2021 at 1:18
  • 1
    \$\begingroup\$ @JamesSuffolk, USB3 hubs do not translate USB3 traffic into USB2 traffic. Unlike USB2 does for FS/LS devices, USB3 does not define the "transaction translation" mechanism. \$\endgroup\$ Nov 9, 2021 at 1:33

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.