1
\$\begingroup\$

I am trying to power two USB 3 cameras off 24V, so I can power a LED circuit at the same time, instead of 5V and basically put a 5V regulator at each camera side. I am wondering if there were a good way to do this or I would just have to power it from the USB port.

  • Left the 5V from the USB host device floating and connected the grounds together
  • Both the device and computer are powered from the same 24V power supply
  • The system works at USB 3 without dropping frames when one camera is plugged in, but drops a bunch of frames when two are plugged in
  • Plugging both cameras in directly with a USB 3 cable works fine

Here is a basic schematic enter image description here

  • The SBC is powered off the same 24V as the USB3 camera
  • The two 4-layer PCBs are designed to have a 90 ohm differential impedance, but not impedance controlled
  • The connectors are made for USB 3 speeds
  • Never tested the loss of the entire path
  • System works at full bandwidth with a single powered camera setup when the SBC is replaced with a floating laptop

Thank you

\$\endgroup\$
10
  • \$\begingroup\$ The frame dropping issue likely is not related to the power question. We're you also asking about that or just remarking? \$\endgroup\$ – user1850479 Apr 7 '20 at 18:54
  • \$\begingroup\$ You certainly can "inject" the power externally, but you better make this injection to be gated by presence of VBUS from the host. In this case your injection will be USB-compliant, and would not cause any harm due to uncontrolled leakage through active pull-ups on DP or DM wires. \$\endgroup\$ – Ale..chenski Apr 8 '20 at 6:59
  • \$\begingroup\$ @user1850479 I was wondering if people had ideas of what could be causing the frames to drop in our setup. I think it may have to do with impedance mismatch or maybe even ground offset on the USB 3 lines between the two cameras since issues only occur when both externally powered cameras are plugged in. \$\endgroup\$ – echo Apr 8 '20 at 12:05
  • \$\begingroup\$ @Ale..chenski can I gate the injection by just turning on the device after the host is powered on? I can also test plugging in the cameras after the host is powered up. \$\endgroup\$ – echo Apr 8 '20 at 13:23
  • \$\begingroup\$ Why not power it from the USB port? \$\endgroup\$ – pjc50 Apr 8 '20 at 16:32
1
\$\begingroup\$

The easiest option would be to buy a powered USB hub. 24v is a common supply voltage for a larger hub, so you could simply plug the hub into your 24v source with no additional components at all.

Alternatively, you can disconnect the PC USB power lines and use your own 5v supply generated from a regulator. This is essentially what the powered hub will do.

\$\endgroup\$
1
  • \$\begingroup\$ Sorry, I didn't make it clear, but I basically put 24V through a USB cable instead of 5V such that I can power some higher voltage LEDs. I then step the voltage back down to 5V at the USB 3 camera. The camera seems to power on fine and connect when I test it with a floating laptop, but has problems when tested in the system where the SBC is also powered off the same 24V source and another camera is plugged in. \$\endgroup\$ – echo Apr 8 '20 at 12:37
-1
\$\begingroup\$

A USB endpoint can be self-powered without an explicit connection to VBUS. It’s useful to detect VBUS to get USB reset, but USB endpoints can and do sense reset from DP/DM instead, so even that isn’t needed.

So you could do a local power breakout for your cameras without issues, so long as you don’t back-drive host VBUS.

Not sure why you’re dropping frames. That may be a hub issue.

\$\endgroup\$
8
  • \$\begingroup\$ This is all misleading. USB devices must see VBUS to start connect process, regardless if they draw any power from VBUS or not.This is true for USB2 and USB3. See electronics.stackexchange.com/a/455854/117785 \$\endgroup\$ – Ale..chenski Apr 7 '20 at 20:47
  • \$\begingroup\$ Nope. The endpoint can detect a connect event from DP/DM. It can, but doesn’t have to, see VBUS. A self-powered device can force its local VBUS high. \$\endgroup\$ – hacktastical Apr 7 '20 at 21:16
  • \$\begingroup\$ hacktastical, did you try to read specifications I referenced in my link above? DP/DM cannot be pulled unless VBUS is present. Your understanding is unconditionally wrong. \$\endgroup\$ – Ale..chenski Apr 7 '20 at 22:35
  • \$\begingroup\$ Difference between cannot and should not. The only issue is leakage from a powered endpoint to a powered-off host. In other words, yes it’s not compliant, but is it a problem worth blocking the application in this case? Nope, it’ll work fine and no harm will come from it. \$\endgroup\$ – hacktastical Apr 8 '20 at 3:15
  • \$\begingroup\$ "it’ll work fine and no harm will come from it" - wrong again. The USB specifications explicitly require "no sourcing of any power if no connection is made" for a reason. One major reason is to support system functionality under various power saving states/transitions, S3/G3 etc., when unsolicited leakage from DP/DM pullups (made by rogue designers) would cause partial charge of PHY circuits and entire system, and the system might fail to restore the USB connection. Specifications are written for a reason, and you seem to be unamenable to it. -1 finally. \$\endgroup\$ – Ale..chenski Apr 8 '20 at 6:55

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.